Alimentary Pharmacology and Therapeutics

Systematic review with meta-analysis: the haemodynamic

effects of carvedilol compared with propranolol for portal
hypertension in cirrhosis
E. Sinagra,,1, G. Perricone,,1, M. DAmico, F. Tine* & G. DAmico*

*Gastroenterology Unit, A.O. Ospedali SUMMARY

Riuniti Villa Soa Cervello P.O.
Vincenzo Cervello, Palermo, Italy.

Internal Medicine Unit, A.O. Ospedali Background

Riuniti Villa Soa Cervello P.O. Propranolol is recommended for prophylaxis of variceal bleeding in cirrho-
Vincenzo Cervello, Palermo, Italy. sis. Carvedilol is a nonselective beta-blocker with a mild anti-alfa-1-adren-

Hepatology and Gastroenterology ergic activity. Several studies have compared carvedilol and propranolol,
Unit, A.O. Ospedale Niguarda Ca yielding inconsistent results.
Granda, Piazza Ospedale Maggiore,
Milano, Italy.

Endoscopy and Gastroenterology Aim

Unit, Fondazione Istituto San Raffaele To perform a systematic review and meta-analysis of the randomised clini-
Giglio, Cefalu, Italy. cal trials comparing carvedilol with propranolol for hepatic vein pressure

Radiology Department, Universita di gradient reduction.

Palermo, Palermo, Italy.
Methods
Studies were searched on the MEDLINE, EMBASE and Cochrane library
Corresponding author:
Dr G. DAmico, Gastroenterology Unit,
databases up to November 2013. The weighted mean difference in percent
A.O. Ospedali Riuniti Villa Soa hepatic vein pressure gradient reduction and the relative risk of failure to
Cervello P.O. Vincenzo Cervello, via achieve a hemodynamic response (reduction 20% of baseline or to
Trabucco 180, 90146 Palermo, Italy. 12 mmHg) with each drug were used as measures of treatment efcacy.
E-mail: [email protected]
Results
1
These authors equally shared the rst
Five studies (175 patients) were included. Indication to treatment was pri-
authorship.
mary prophylaxis of variceal bleeding in 76% of patients. There were overall
three acute (6090 min after drug administration) and three long-term
Publication data (after 790 days of therapy) comparisons. The summary mean weighted
Submitted 23 November 2013 difference in % of reduction in hepatic vein pressure gradient was: acute
First decision 9 December 2013 7.70 (CI 12.40, 3.00), long-term 6.81 (CI 11.35, 2.26), overall
Resubmitted 3 January 2014 7.24 (CI 10.50, 3.97), favouring carvedilol. The summary relative risk
Accepted 6 January 2014
of failure to achieve a hemodynamic response with carvedilol was 0.66 (CI
EV Pub Online 26 January 2014
0.44, 1.00). Adverse events were nonsignicantly more frequent and serious
As part of AP&Ts peer-review process, a with carvedilol. However, quality of trials was mostly unsatisfactory.
technical check of this meta-analysis was
performed by Mr M. Siddiqui. Conclusions
Carvedilol reduces portal hypertension signicantly more than propranolol.
However, available data do not allow a satisfactory comparison of adverse
events. These results suggest a potential for a cautious clinical use.

Aliment Pharmacol Ther 2014; 39: 557568

2014 John Wiley & Sons Ltd 557

doi:10.1111/apt.12634
E. Sinagra et al.

INTRODUCTION Compared treatments

Portal hypertension is an almost unavoidable complica- Oral carvedilol was compared with either oral or i.v. pro-
tion of cirrhosis1 and variceal haemorrhage is its most pranolol for the acute hemodynamic effect, or oral carv-
dramatic consequence, bearing a mortality of 1015% edilol compared with oral propranolol for the long-term
per episode.2 The hepatic venous pressure gradient effect. The assessment of the hemodynamic effect was
(HVPG) is signicantly associated with the risk of vari- considered acute when performed during the same
ceal bleeding, liver decompensation and hepatocellular hepatic veins catheterisation. When a second catheterisa-
carcinoma (HCC).35 A reduction in HVPG by 20% of tion was required (after days or months of treatment), it
baseline results in a signicant reduction in the bleeding was classied as assessment of long-term treatment
risk and a reduction to less than 12 mmHg virtually effect.
abolishes the risk.3, 6 Propranolol, a nonselective beta-
blocker (NSBB), signicantly reduces HVPG and is cur- Treatment outcome
rently recommended for the prevention of primary and The primary measure of the treatment effect was the dif-
secondary variceal bleeding. However, it achieves the tar- ference in % HVPG reduction achieved with each of the
get HVPG reduction only in approximately one-third of two compared drugs. The proportion of patients failing
patients, possibly because the increase in porto-collateral to achieve a hemodynamic response to the treatment
resistance impairs the reduction in portal pressure.79 was considered a secondary outcome measure.
This effect may be reduced by adding a vasodilator such Hemodynamic responders were considered those
as isosorbide mononitrate, which in fact rescues most of patients achieving a HVPG reduction 20% of baseline
nonresponders to propranolol, although at the expense value or in whom HVPG was lowered to
of increasing side effects.10 12 mmHg.3, 13, 14 The number and severity of adverse
Carvedilol is a potent NSBB with a mild anti-alfa1 events were also assessed as well as the difference in
adrenergic activity, which confers to the drug a potential mean arterial blood pressure reduction (MAP) with each
for a higher portal pressure reduction compared with drug as an indicator of systemic treatment effect.
other NSBB. Several studies have compared the acute or
long-term hemodynamic response of carvedilol and pro- Search strategy
pranolol yielding inconsistent results.11 Medline, Embase, Cochrane Controlled Trial Registry
The aim of the present study was to assess the hemo- and The Cochrane Library were searched up to Novem-
dynamic effects of carvedilol compared with propranolol ber 2013 to retrieve pertinent studies. Databases were
for portal hypertension. With this aim, we performed a searched by the following search terms: carvedilol, pro-
systematic review of randomised controlled trials (RCTs) pranolol, portal hypertension, randomised trial. A man-
comparing the two drugs for their effect on hepatic vein ual search of the reference lists of pertinent studies,
pressure gradient in cirrhosis. articles, reviews, editorials from the English language
medical literature was also performed. Proceedings of the
METHODS most relevant international congresses were hand-
searched.
Study selection and participants
All the studies that met the following inclusion criteria Data collection and analysis
were included: (i) the study was an RCT comparing Decisions on which trials to include were taken indepen-
carvedilol with propranolol for acute or long-term effect dently by two of the authors (ES and GP), who were
on HVPG in cirrhotic patients; (ii) reduction in HVPG unblinded with regard to the names of the authors,
under treatment was a measure of the treatment effect; investigators, institution, source, and results of the iden-
(iii) participants in the included studies were cirrhotic tied trials. Disagreements were resolved by discussion
patients with clinically signicant portal hypertension as with all the authors. Excluded trials were identied with
indicated by HVPG >10 mmHg and/or presence of the reason for exclusion. Data were extracted by two
oesophageal varices, according to the Baveno III crite- independent authors (ES and GP) and discrepancies
ria.12 Randomised trials were included irrespective of were solved by discussion. The following data were
blinding, publication status or language. Trials using extracted from each trial: (i) trial characteristics: time
quasi-randomisation, based on case record number, date interval between trial drug administration and hemody-
of birth, day of the week or similar, were excluded. namic effect assessment; denition of the outcome of

558 Aliment Pharmacol Ther 2014; 39: 557-568

2014 John Wiley & Sons Ltd
 Systematic review with meta-analysis: carvedilol vs. propranolol for portal hypertension

interest; (ii) patients characteristics: number of patients, Analyses were performed by STATA 10 (College Sta-
mean age, sex, aetiology of cirrhosis, Child-Pugh score, tion, TX 77845 USA, 19842009) by a xed-effect
proportion of patients with previous variceal bleeding; model if no statistically signicant heterogeneity was
(iii) outcome: mean HVPG change under trial treatment; found and by a random-effects model in case of signi-
number of patients achieving a hemodynamic response; cant heterogeneity (P < 0.10). The treatment effect was
adverse events observed during the treatment period. calculated as the weighted mean difference (WMD)
between the % HVPG reduction from baseline achieved
Assessment of study quality with propranolol and the % reduction achieved with
Two of the authors (GP and ES) independently assessed carvedilol. Data are expressed as % HVPG reduction
bias risk of the trials, without masking the trial names. with propranolol minus % HVPG reduction with carv-
For this purpose, the risk of overestimation of interven- edilol. Therefore, negative values indicate a higher
tion effects in randomised trials due to inadequate meth- HVPG reduction with carvedilol. The pooled WMD is
odological quality was assessed using the following reported as the summary statistic of treatment effect
domains.15, 16 together with 95% condence interval (CI). The pooled
Generation of the allocation sequence was dened ade- relative risk (RR) of failure to achieve a hemodynamic
quate if sequences were generated by table of random response with either treatment was also assessed and
numbers or computer-generated random numbers or reported with 95% CI. Inter-trial heterogeneity was sta-
similar; unclear if it was not described; inadequate if tistically assessed by a chi-square test and was expressed
sequences could be related to prognosis. using I2 values.17 If signicant heterogeneity was found,
Allocation concealment was dened adequate when a potential reasons for heterogeneity were explored and
centralised randomisation or sequentially numbered, combinability of trials was reassessed accordingly. Analy-
sealed, opaque envelopes were used; unclear when studies ses were performed using the intention-to-treat principle
did not report any concealment approach; inadequate including all patients randomised as they were reported
when the following modalities were used: alternation, the in each single study. Trials were rst combined indepen-
use of case record numbers, dates of birth or day of the dently of whether they assessed acute or long-term he-
week, and any procedure that is entirely transparent before modynamic effect of the study treatment. However,
allocation, such as an open list of random numbers. separate analyses according to the timing of assessment
Blinding was considered adequate when the study was were also performed.
double-blind or single-blind (HVPG measurements were To explore the robustness of estimates, the following
blindly evaluated); unclear when no information was sensitivity analyses were performed: (i) random-effects
reported; inadequate when an open design was used. model; (ii) standardised mean difference (SMD) instead
Handling of dropouts and withdrawals was considered of WMD to overcome potential measurement differ-
adequate when numbers of patients withdrawn or lost ences across trials; (iii) excluding trials with acute
and reasons were described; inadequate otherwise. assessment; (iv) excluding trials with long-term assess-
Intention-to-treat analysis was considered adequate ment; (v) excluding trials with inadequate control of
when all randomised patients were included in the analy- bias.
ses and inadequate otherwise. The weighted mean difference in arterial blood pres-
Following the above denitions, an included trial was sure with the two drugs was also assessed as a measure
judged as achieving inadequate or unclear control of bias of a major treatment side effect.
when the methodological quality was considered as inad-
equate or, respectively, unclear in at least 1 domain. If Study protocol
the quality was judged adequate in all the domains, then The full study protocol is available by contacting the cor-
the control of bias was considered adequate. responding author of this article.

Statistical methods RESULTS

Trials and patient characteristics are reported as number
of observations, proportions or means  standard devia- Trial retrieval
tions. Where standard error of mean was reported in the Overall, 39 potentially eligible references were retrieved
original article, the standard deviation was derived from in the literature search. Thirty references were excluded
the standard error. by reading the title and abstract, because they did not

Aliment Pharmacol Ther 2014; 39: 557-568 559

2014 John Wiley & Sons Ltd
E. Sinagra et al.

38 references One additional

identified through reference identified
database serching by handsearch

39 references
screened
30 excluded
2 RCTs exluded because did not include
hemodynamic assessment or -blocker control
9 references group:
assessed - one compared carvedilol with band ligation of
for eligibility esophageal varices for primary prophylaxis
of variceal bleeding
- the other compared carvedilol with nadolol plus Figure 1 | Flow diagram of the
7 references to 5 isosorbide mononitrate for secondary prophylaxis
pertinent trials of variceal bleeding selection process of
randomised controlled trials
for inclusion in the meta-
5 RCTs included analysis. RCT, randomised
In quantitative synthesis controlled trials.
(meta-analysis)

refer to randomised trials. The remaining 9 references propranolol was associated with isosorbide mononitrate
were fully assessed for inclusion in the study (Figure 1). 20 mg in the control group.
Among these, two referred to full articles and were In the evaluation of the long-term hemodynamic
excluded for the following reasons: one was a rando- response, treated patients received a mean dose of 31 mg
mised controlled trial of carvedilol vs. variceal band liga- of oral carvedilol and controls received a mean of 73 mg
tion for the prevention of the rst variceal bleed;18 the of oral propranolol in one study;22 a xed oral dose of
other was a randomised controlled trial of carvedilol vs. 12.5 mg of carvedilol was compared with a xed oral
nadolol plus isosorbide mononitrate for the prevention dose of 80 mg of propranolol in another;23 and a mean
of variceal rebleeding, which did not include hemody- dose of 14 mg of carvedilol and 122 mg of propranolol
namic assessments.19 Seven references2026 to ve perti- in the third.26
nent studies were eventually found: two of these referred A second HVPG measurement was obtained in all the
to the preliminary report20 or to a subgroup analysis25 of included patients in three studies.21, 24, 26 In one study,22
a fully published study26 and were therefore excluded. discontinuation of treatment caused missing second
The remaining ve references2124, 26 referred to rando- HVPG measurement in two patients in the carvedilol
mised controlled trials, which fullled our inclusion cri- group (heart failure one and pericardial effusion the
teria and were included in the quantitative meta-analysis. other) and in three in the propranolol group (portal sys-
temic encephalopathy 1 and fatigue 2). In the other
Description of included studies study,23 discontinuation of treatment hampered a second
The characteristics of the ve included studies are sum- HVPG measurement in one patient in the carvedilol
marised in Tables 1 and 2. In two studies, the acute group because of severe hypotension and in two patients
hemodynamic effect was assessed 60 and 90 min after in the propranolol group because of bleeding.
trial drug administration21, 24 respectively. In two stud- Control of bias was adequate in one trial, unclear in
ies, the treatment effect was assessed after a mean of three and inadequate in one trial (Table 2). In the four
11 weeks22 and 92.7 days.26 In one study,23 both acute trials with inadequate or unclear control of bias, this was
and long-term effects were assessed at 90 min and mostly due to the generation of the allocation sequence
1 week after initial trial drug administration, respectively. or to allocation concealment.
Overall, therefore, there were three acute and three The characteristics of the 175 patients included in the
long-term evaluations available for meta-analysis. ve studies are summarised in Table 3. Males were 74%.
In the acute comparisons, all treated patients were The number of patients with ascites was reported only in
given a tablet of carvedilol 25 mg, while control patients two studies22, 23 and was 44 of 87 (51%); patients with a
were given propranolol 0.15 mg/kg i.v. followed by a previous bleeding episode were 42 of 153 (27%; informa-
continuous infusion of 0.2 mg/kg/h in one study,21 tion not reported in one study24). Mean age was 55 years
40 mg p.o.24 or 80 mg p.o.23 In one acute study,24 and aetiology was mostly viral or alcoholic in all the

560 Aliment Pharmacol Ther 2014; 39: 557-568

2014 John Wiley & Sons Ltd
 Systematic review with meta-analysis: carvedilol vs. propranolol for portal hypertension

Table 1 | Characteristics of included trials

N of patients with 1st/2nd

Time of HVPG measurement
Author, N of randomised Interventions outcome
year Participants patients C/P C; P* assessment Carvedilol Propranolol
Banares, 1999 Cirrhotic patients 14/14 C: Carvedilol 60 min 14/14 14/14
with oesophageal 7 more patients 25 mg p.o.
varices, with or were randomised P: Propranolol
without previous to placebo (0.15 mg/kg
variceal bleeding i.v. followed by
a continuous
infusion of
0.2 mg/kg/h)
Banares,2002 Cirrhotic patients 26/25 C: Carvedilol 11  4.1 weeks 26/24 25/22
with oesophageal 31  4 mg/day
varices, no previous p.o.(range, 12.550)*
bleeding, and P: Propranolol
HVPG >12 mmHg 73  10 mg/day
p.o. (range, 10160)*
De, 2002 Cirrhotic patients 18/18 Acute: Acute: 90 min Acute: 18/18 Acute: 18/18
with oesophageal C: Carvedilol Long-term: 1 week Long-term 18/17 Long-term 18/16
varices, with or 25 mg p.o.;
without previous P: Propranolol
variceal bleeding 80 mg p.o.
Maintenance:
C: carvedilol
6.25 mg p.o.
twice daily
P: propranolol
40 mg p.o.
twice daily
Lin, 2004 Cirrhotic patients 11/11 C: Carvedilol 90 min 11/11 11/11
with HVPG 12 with 25 mg p.o.
or without previous P: propranolol
variceal bleeding 40 mg p.o +
isosorbide
5-mononitrate
20 mg p.o.
Hobolth, 2012 Cirrhotic patients with 21/17 C: Carvedilol 92.7  13.6 days 21/21 17/17
HVPG12 mmHg, 14  7 mg*
independent of varices P: Propranolol
or previous bleeding 122  64 mg*

C, Carvedilol group; P, Propranolol group; p.o., orally; i.v., intravenously.

* Mean dose  standard deviation.
Interval between baseline and second HVPG measurement, mean  s.d.

studies but one where only patients with cirrhosis from in the carvedilol-treated patients. The mean % HVPG
HBV and/or HCV were included.24 Child-Pugh class C reduction under treatment was 15.6 (median 12.6, range
patients were overall 31 of 153 (20%) in four studies and 10.123.2) with propranolol and 22.2 (median 19.8,
the mean Child-Pugh score was eight in the fth.24 No range 18.627.7) with carvedilol (crude mean difference
appreciable differences in major patient characteristics 6.61%).
were observed between the two study groups across the The pooled WMD between % HVPG reductions with
ve studies (Table 3). the two drugs was 7.24 ( 10.50, 3.97; xed-effect
model) indicating a signicantly higher HVPG reduction
Outcome evaluation with carvedilol than propranolol. No heterogeneity was
HVPG reduction. Mean baseline HVPG was 18.3 mmHg found (P = 0.99; I2 = 0.0) (Figure 2). A sensitivity analy-
(median 18, range 16.620.4) in propranolol-treated sis performed by using the standardised mean difference
patients and 18.8 mmHg (median 19, range 17.619.5) between the % HVPG reductions with the two drugs

Aliment Pharmacol Ther 2014; 39: 557-568 561

2014 John Wiley & Sons Ltd
E. Sinagra et al.

Table 2 | Methodological quality assessment of included trials*

Generation
of the allocation Allocation Dropouts Intention-to- Overall risk
Author, year sequence concealment Blinding and withdrawals treat analysis of bias
Banares, 1999 Unclear Unclear Adequate Adequate Adequate Unclear
Banares, 2002 Adequate Adequate Adequate Adequate Adequate Adequate
De, 2002 Unclear Inadequate Adequate Adequate Adequate Inadequate
Lin, 2004 Adequate Unclear Unclear Adequate Adequate Unclear
Hobolth, 2012 Unclear Adequate Adequate Adequate Adequate Unclear
* Denition used for rating of each quality item are reported in the method section.
Blinded reading of HVPG measurements.
Double-blind.

Table 3 | Characteristics of participants in the included studies

Alcoholic/ Child-Pugh Oesophageal Previous

Treatment Patients Age, Males viral class A/B/ varices large/ Ascites variceal
group N mean  s.d. N aetiology N C, N small, N N bleeding, N
Banares, 1999 C 14 54.6  8.8 NR 8/NR 8/4/2 NR 7* 9
P 14 51.4  8.5 NR 8/NR 5/6/3 NR 7* 9
Banares, 2002 C 26 57.9  7.6 19 6/18 13/10/3 10/16 10 0
P 25 58.4  11 15 9/16 15/6/4 14/11 6 0
De, 2002 C 18 42.3  11.9 15 5/9 5/9/4 NR 12 7
P 18 47.3  12.9 17 10/5 0/13/5 NR 16 7
Lin, 2004 C 11 59  13.3 8 0/11 8.7  3.0 NR NR NR
P 11 61  13.3 9 0/11 7.3  2.7 NR NR NR
Hobolth, 2012 C 21 58.2  6.8 12 18/2 8/7/6 4/10 NR 5
P 17 56.2  6.1 12 12/1 6/6/4 1/11 NR 5

C, Carvedilol; P, Propranolol.
* Previous ascites.
Child-Pugh score, mean  standard deviation (derived from the standard error reported in the article).

conrmed the principal analysis: pooled SMD 0.50 (CI The pooled relative risk of failure to achieve a hemody-
0.78, 0.23; xed-effect model) without signicant het- namic response with carvedilol was 0.66 (CI 0.441.00)
erogeneity (P = 0.5; I2 = 0.0). The other sensitivity (Figure 3).
analyses conrmed that either the acute or the long-term
% HVPG reduction is signicantly higher with carvedilol Adverse events. Mean arterial blood pressure (MAP)
than with propranolol (Figure 2) and superiority of carv- was reduced more by carvedilol (WMD 10.40; 95% CI:
edilol held true also by excluding the study with inade- 13.90, 6.90; I2 = 0.0%, P = 0.477) than by proprano-
quate control of bias22 (WMD 7.41, CI 10.79, 4.03; lol (WMD 6.35; 95% CI 9.86, 2.83; I2 = 0.0%,
data not shown). P = 0.936) (Figure 4). The lack of complete information
The mean % HVPG reduction was 20% in 2 of 6 on the difference in MAP reduction between the two
comparisons with propranolol and in 3 of 6 with carv- drugs hampered obtaining a summary statistics. How-
edilol. Overall, a reduction in HVPG of 20% or to ever, as the CI of MAP reduction achieved with each
12 mmHg was achieved with propranolol in 33 of 87 drug largely overlapped, it may be concluded that the
evaluable patients as compared to 57 of 94 with carvedi- difference in MAP reduction between the two drugs was
lol (including patients with both acute and chronic not signicantly different (Figure 4). On the other hand,
assessments; P = 0.003, Fishers exact test) (Table 4). even if not statistically signicant, this difference was

562 Aliment Pharmacol Ther 2014; 39: 557-568

2014 John Wiley & Sons Ltd
 Systematic review with meta-analysis: carvedilol vs. propranolol for portal hypertension

Acute assessment
Banares (1999) 7.70 (13.24, 2.16) 34.70

De (2002) 4.69 (23.94, 14.56) 2.88

Lin (2004) 8.50 (18.48, 1.48) 10.71

Subtotal 7.70 (12.40, 3.00) 48.29

(I 2 = 0.0%, P = 0.942)

Long-term assessment
Banares (2002) 7.00 (12.26, 1.74) 38.48

De(2002) 4.72 (21.99, 12.55) 3.58

Hobolth (2012) 6.80 (17.31, 3.71) 9.66

Subtotal 6.81 (11.35, 2.26) 51.71

(I 2 = 0.0%, P = 0.970)

Heterogeneity between groups: P = 0.789

Overall 7.24 (10.50, 3.97) 100.00

(I 2 = 0.0%, P = 0.998)

20 10 0 10 20
Better carvedilol Better propranolol
Weighted Mean Difference: fixed effects

Figure 2 | Weighted mean difference (WMD) (xed-effect model) of the % hepatic vein pressure gradient (HVPG)
reduction between propranolol and carvedilol in the subsets of RCTs assessing the acute and, respectively, the long-
term treatment effect together with the overall estimation of the whole set of studies. Each study is identied by the
name of the rst author and year of publication (references in the text). Squares indicate the WMD per each trial and
the size of the squares is proportional to the weight of trials. The horizontal bars denote the 95% condence intervals
of WMD. The vertical solid line is the equivalence line, where WMD is equal to 0. Differences are given as values in
the propranolol group minus values in the carvedilol group. Therefore, negative values (WMDs on the left of the
equivalence line) denote superiority, whereas those on the right denote inferiority of carvedilol. The diamonds
represent 95% CI of the WMDs per each subset and in whole set of RCTs. The vertical dashed line represents the
pooled WMD of the whole set of trials. The study by De (2002) provided estimation of either acute or long-term
treatment effect.

clinically relevant, with at least one symptomatic episode DISCUSSION

of hypotension reported with carvedilol and a total of 14 The present meta-analysis shows that, in the available
patients with orthostatic hypotension out of 65 patients RCTs comparing propranolol and carvedilol for portal
treated with carvedilol, compared to 9/60 with proprano- hypertension, carvedilol reduces HVPG signicantly
lol (P = 0.37) in three of the ve studies.22, 23, 26 more than propranolol. Superiority of carvedilol is con-
Other adverse events, even if not fully reported in all sistent across all the trials, regardless of acute or long-
the studies, were the following: need for increasing term assessment. The summary weighed mean difference
diuretics in 14/47 patients treated with carvedilol and in between the percent HVPG reduction achieved with pro-
6/42 with propranol;22, 26 shortness of breath in 12/47 pranolol and carvedilol is 7.24 (CI 10.50 to 3.97) in
and 7/42;22, 26 portosystemic encephalopathy in 3/26 and favour of carvedilol. No heterogeneity was found and all
4/2522 respectively. Overall carvedilol was withdrawn in the sensitivity analyses conrmed the principal result.
6/65 and propranolol in 8/60 treated patients22, 23, 26 Patients in the two treatment groups were fairly com-
because of adverse events. None of these differences was parable across the included studies, regarding demogra-
statistically signicant. phy, aetiology, indicators of disease severity, baseline

Aliment Pharmacol Ther 2014; 39: 557-568 563

2014 John Wiley & Sons Ltd
E. Sinagra et al.

Table 4 | Main results of the included studies

Carvedilol Propranolol
Weighed
Baseline % HVPG Hemo-dynamic % HVPG Hemo-dynamic Mean Difference
HVPG, mmHg reduction, responders Baseline HVPG, reduction, responders in % HVPG
Author, year mean  s.d.* mean  s.d.* n/N (%) mean  s.d.* mean  s.d.* n/N (%) reduction (CI)

Acute assessment
Banares, 1999 19.5  4.9 20.4  7.5 9/14 (64) 20.4  4.1 12.7  7.48 2/14 (14) 7.70 ( 13.24, 2.16)
De, 2002 19.0  3.8 27.7  31.5 11/18 (61) 16.60  3.9 22.9  27.4 9/18 (50) 4.69 ( 23.94, 14.56)
Lin, 2004 18.9  6.0 18.6  11.9 NR 17.6  4.0 10.1  11.9 NR 8.50 ( 18.48, 1.48)
Subtotal 7.70 ( 12.4, 3.00)
Long-term assessment
Banares, 2002 19.1  5.4 19  9.8 13/24 (54) 20.3  4.2 12  9.4 5/22 (23) 7.00 ( 12.26, 1.74)
De, 2002 19.0  3.8 27.7  31.5 11/17 (65) 16.6  4.0 22.98  20.1 10/16 (62) 4.72 ( 21.99, 12.55)
Hobolth, 2012 17.6  4.2 19.3  16.1 13/21 (62) 18.4  3.6 12.5  16.7 7/17 (41) 6.80 ( 17.31, 3.71)
Subtotal 6.81
( 11.35, 2.26)
Total 7.24 ( 10.50, 3.97)

* s.d., standard deviation of mean. In the studies by Banares 1999 and 2002 and by Lin 2004, the standard deviation of the mean
was calculated from the standard error of the mean reported in the original articles.
Patients achieving a HVPG reduction 20% or HVPG 12 mmHg.
Hemodynamic response was dened as a reduction in HVPG of 20% of baseline in the Des study.
n, number of responders; N, total number of evaluable patients.
The difference is calculated as % HVPG reduction with propranolol % HVPG reduction with carvedilol: negative values favour
carvedilol.

HVPG and history of previous bleeding. A second mea- carvedilol and 9/18 and 10/16 with propranolol.
surement of HVPG was available in all the included Although no individual patient data are reported in the
patients, but three in the Des study23 and ve in the article,23 it may be assumed that the acute hemodynamic
long-term study by Banares.22 Only two of the missing response tends to be maintained.
second HVPG measurements were due to bleeding (both Superiority of carvedilol over propranolol in reducing
in proporanolol-treated patients), while the others were HVPG is accredited to be a consequence of the alfa-1
caused by adverse events requiring treatment withdrawal. blocking effect reducing intra-hepatic resistance. In this
Control of bias was considered inadequate in only one respect it remains unclear the better (although not statis-
trial and the benecial effect of carvedilol over proprano- tically signicant) effect of carvedilol in the trial by
lol was conrmed also when excluding this study. On Lin,24 comparing carvedilol with propranolol plus isosor-
the other hand, control of bias was adequate in only one bide mononitrate. The authors adduce viral aetiology or
study: this nding calls for caution when interpreting the the known hyper-responsiveness to alfa-1 blockers in cir-
present results. rhosis as potential explanations, although equivalence of
The number of patients achieving a reduction in drug doses with respect to the expected hemodynamic
HVPG to 20% or to 12 mmHg was reported in 4 of effect should also be accounted for. On the other hand,
the 5 studies and was also markedly higher with carvedi- the only RCT comparing the clinical efcacy of carvedi-
lol (57/94 vs. 33/87). However, the reduction in the lol with nadolol plus isosorbide mononitrate showed that
pooled relative risk of failure to achieve a hemodynamic the two treatments did not differ in the prevention of
response failed to reach statistical signicance with carv- variceal rebleeding after a median follow-up of
edilol (0.67, CI 0.441.01), conceivably because of a type 30 months.19
II error. It may also be worth noting that in the only Adverse events were satisfactorily reported in only one
study assessing both acute and long-term effects,23 the study.21 The most serious reported adverse event was
number of patients with acute hemodynamic response orthostatic hypotension (14/65 carvedilol- and 9/60 pro-
was almost identical to the number of patients with pranolol-treated patients). Although not statistically sig-
response after 7 days: respectively 11/18 and 11/17 with nicant, the higher incidence of this serious adverse

564 Aliment Pharmacol Ther 2014; 39: 557-568

2014 John Wiley & Sons Ltd
 Systematic review with meta-analysis: carvedilol vs. propranolol for portal hypertension

RR (95% CI) %Weight

Acute assessment

Banares (1999) 0.22 (0.06, 0.85) 8.00

De (2002) 0.82 (0.45, 1.48) 25.10

Subtotal 0.48 (0.13, 1.85) 33.10

(I 2 = 71.3%, P = 0.062)

Long-term assessment

Banares (2002) 0.42 (0.18, 0.99) 16.09

De (2002) 0.97 (0.58, 1.62) 28.55

Hobolth (2012) 0.67 (0.34, 1.29) 22.27

Subtotal 0.71 (0.44, 1.13) 66.90

(I 2 = 35.1%, P = 0.214)

Overall (I 2 = 39.9%, P = 0.155) 0.66 (0.44, 1.00) 100.00

.2 .5 1 5 10
Better carvedilol Better propranolol
Relative risk: random effects

Figure 3 | Summary relative risk (RR) of treatment nonresponse (reduction in HVPG of 20% of baseline or to
<12 mmHg not achieved) with carvedilol compared with propranolol (random-effects model) in the subsets of RCTs
assessing the acute and, respectively, the long-term treatment effect together with the overall estimation of the whole
set of studies. Each study is identied by the name of the rst author and year of publication (references in the text).
Squares indicate the RR per each trial and the size of the squares is proportional to the weight of trials. The
horizontal bars denote the 95% condence intervals of RR. The vertical solid line is the equivalence line, where RR is
equal to 1. RRs on the left of the equivalence line denote superiority, whereas those on the right denote inferiority of
Carvedilol. The diamonds represent 95% CI of the RRs per each subset and in whole set of RCTs. The vertical dashed
line represents the summary RR of the whole set of trials. The study by De (2002) provided estimation of either
acute or long-term treatment effect.

event calls for caution in clinical practice when using by oliguria requiring treatment withdrawal and active
carvedilol. Moreover, the overall mean weighed reduction intervention for recovering.23 Importantly, the hypoten-
in the mean arterial pressure was 6.66 mmHg (CI sive effect of carvedilol was markedly reduced by starting
10.17 to 3.15) with propranolol and 10.40 (CI with a very low dose and careful titration by stepwise
13.9 to 6.9) with carvedilol. Thus, the mean arterial increases every 47 days.22, 26
pressure was signicantly reduced by both treatments In the study assessing the 11-week effect of the
and the reduction was in the order of one-third more drugs,22 a signicant increase in plasma volume and
with carvedilol than with propranolol. The difference in body weigh was noted with carvedilol, but not with pro-
arterial pressure reduction between the two drugs was pranolol, and an increase in diuretics dosage was needed
not statistically signicant as shown by the overlapping in seven carvedilol compared with two propranolol
95% condence boundaries of the reduction with each of patients because of new onset or worsening ascites or
the two drugs, although it is clearly clinically signicant. ankle oedema. This effect, similar to that observed with
Of note, in at least one carvedilol patient (with ascites), propranolol combined with prazosin,27 has been
orthostatic hypotension was symptomatic and followed attributed to a possible increase in sodium retention,22

Aliment Pharmacol Ther 2014; 39: 557-568 565

2014 John Wiley & Sons Ltd
E. Sinagra et al.

Propranolol WMD (95% CI) Weight

Banares (1999) 4.00 (12.32, 4.32) 17.87

De, acute (2002) 3.60 (14.41, 7.21) 10.57

Banares (2002) 4.80 (15.51, 5.91) 10.77

De, chronic (2002) 5.70 (15.32, 3.92) 13.35

Lin (2004) 9.00 (17.32, 0.68) 17.87

Hobolth (2012) 8.00 (14.47, 1.53) 29.56

Overall 6.35 (9.86, 2.83) 100.00

(I 2 = 0.0%, P = 0.936)

20 10 0 10 20

Carvedilol WMD (95% CI) Weight

Banares (1999)
16.00 (25.80, 6.20) 12.75
De, acute (2002) 10.50 (18.26, 2.74) 20.36
Banares (2002) 10.20 (17.56, 2.84) 22.62
De, chronic (2002) 15.10 (24.12, 6.08) 15.05
Lin (2004) 8.00 (22.13, 6.13) 6.13
Hobolth (2012) 5.00 (12.28, 2.28) 23.09

Overall 10.40 (13.90, 6.90) 100.00

(I 2 = 0.0%, P = 0.477)

20 10 0 10 20
MABP reduction, mmHg

Figure 4 | Weighted mean difference (WMD) of arterial pressure from before to after treatment. Data are expressed
as baseline minus post-treatment value. The top section of the gure represents the arterial pressure changes with
propranolol and the bottom section changes with carvedilol. Each study is identied by the name of the rst author
and year of publication (references in the text). Squares indicate the WMD per each trial and the size of the squares
is proportional to the weight of trials. The horizontal bars denote the 95% condence intervals of WMD. The vertical
solid line is the equivalence line, where WMD is equal to 0. WMDs on the left of the equivalence line denote
reduction in arterial blood pressure after treatment, whereas those on the right denote increase. The vertical dashed
line represents the pooled WMD of the whole set of studies in the arms of propranolol (top) and, respectively,
carvedilol (bottom). The diamonds represent 95% CI of the WMDs.

and might require some weeks to present, as it was not prophylaxis of variceal bleeding,28 where carvedilol
noted after 1 week.21 It is, however, worth noting that in was used as a rescue treatment for propranolol hemody-
two randomised clinical trials of carvedilol, one for namic nonresponders, no increase in new ascites
primary11 and the other for secondary19 prevention of development was reported under carvedilol, although 27
variceal bleeding, no difference was noted in the new of 67 patients given carvedilol received a dose 25 mg/
onset or worsening of ascites between the study groups. day.
The carvedilol dose in these studies was 12.5 mg/day A nal consideration is with regard to publication
suggesting that worsening of sodium and water retention bias. We may not exclude that some negative trial has
may be not relevant for relatively low carvedilol doses, been performed and not published. On the other hand,
which, on the other hand, may be effective in clinical the low number of available studies with low and similar
practice. However, in a more recent study of primary numbers of patients included hampers achieving a

566 Aliment Pharmacol Ther 2014; 39: 557-568

2014 John Wiley & Sons Ltd
 Systematic review with meta-analysis: carvedilol vs. propranolol for portal hypertension

reliable estimation by Funnel plot asymmetry tests29, concomitant presence of portal and arterial hypertension,
which, in such a situation, have unsatisfactory power. while liver decompensation and serum creatinine above
In conclusion, this meta-analysis shows that carvedilol the upper normal limit should be considered at least rel-
reduces HVPG signicantly more than propranolol either ative contraindications. If a decision is made to use carv-
after one single administration or in a more long-term up edilol, the starting dose should be very low with stepwise
to 1 week to 3 months. This effect is achieved at the titration up to 12.5 (or 25 if tolerated) mg/day, prefera-
expense of marked reduction in arterial pressure, which bly in two daily doses. The dose may be lower in decom-
becomes symptomatic in an appreciable proportion of pensated patients in whom a better hemodynamic effect
patients. Yet, worsening of water retention is to be may be expected. Side effects should be strictly moni-
expected in several patients, although the lack of rando- tored, particularly hypotension, and worsening of sodium
mised trials comparing the clinical effects of carvedilol and water retention and renal function.
with propranolol hampers drawing conclusions for clinical
practice on this potentially harmful effect. Moreover, the AUTHORSHIP
two recent RCTs showing that carvedilol is as effective as Guarantor of the article: G.DAmico
banding ligation for primary prevention of variceal bleed- Author contributions: E. Sinagra: study design, data col-
ing (and even superior in the intention-to-treat analysis)18 lection, results interpretation, rst draft of the article. G.
and as effective as nadolol plus isosorbide mononitrate for Perricone: study design, data collection, results interpre-
secondary prevention19 support a potential for a clinical tation, correction of draft article and tables. M DAmico:
benet from carvedilol. This consideration is much data checking, results interpretation, draft article correc-
strengthened by a recent study showing that carvedilol tion. F Tine: checking statistical analysis, results interpre-
may achieve a hemodynamic response in 56% of patients tation. G DAmico: research idea, study design, statistical
failing to respond to propranolol,28 although further RCTs analysis, results interpretation, nal article writing. All
should be encouraged. authors approved the nal version of the manuscript.
Available data suggest that carvedilol may be consid-
ered for portal hypertension at least in patients failing to ACKNOWLEDGEMENT
reach a hemodynamic response to propranolol28 or for Declaration of personal and funding interests: None.

REFERENCES
1. Bosch J, Berzigotti A, Garcia-pagan JC, 5. Ripoll C, Groszmann R, Garcia-Tsao G, pressure reduction during the
Abraldes JG. The management of et al; Portal Hypertension Collaborative administration of beta-adrenergic
portal hypertension: rational basis, Group. Hepatic venous pressure blocking agents in a portal hypertensive
available treatments and future gradient predicts development of model. Hepatology 1985; 5: 97101.
options. J Hepatol 2008; 48(Suppl. 1): hepatocellular carcinoma independently 10. Garcia-Pagan JC, Navasa M, Bosch J,
S6892. of severity of cirrhosis. J Hepatol 2009; Bru C, Pizcueta P, Rodes J.
2. DAmico G, Luca A. Natural history. 50: 9238. Enhancement of portal pressure
Clinical-haemodynamic correlations. 6. Feu F, Garcia Pagan JC, Bosch J, et al. reduction by the association of
Prediction of the risk of bleeding. Relation between portal pressure isosorbide-5-mononitrate to
Baillieres Clin Gastroenterol 1997; 11: response to pharmacotherapy and risk propranolol administration in patients
24356. of recurrent variceal hemorrhage in with cirrhosis. Hepatology 1990 Feb; 11:
3. Groszmann RJ, Bosch J, Grace ND, patients with cirrhosis. Lancet 1995; 2308.
et al. Hemodynamic events in a 346: 10569. 11. Tripathi D, Hayes P. The role of
prospective randomized trial of 7. Vorobioff J, Picabea E, Villavicencio R, carvedilol in the management of portal
propanolol versus placebo in the et al. Acute and chronic hemodynamic hypertension. Eur J Gastroenterol
prevention of a rst variceal effects of propranolol in unselected Hepatol 2010; 22: 90511.
hemorrhage. Gastroenterology 1990; 99: cirrhotic patients. Hepatology 1987; 7: 12. DAmico G, Garcia-Tsao G, Cales P,
14017. 64853. et al. Diagnosis of Portal Hypertension:
4. Ripoll C, Groszmann R, Garcia-Tsao 8. Lebrec D, Hillon P, Munoz C, Goldfarb how and when. In: de Franchis R, ed.
G, et al; Portal Hypertension G, Nouel O, Benhamou JP. The effect Proceedings of the third Baveno
Collaborative Group. Hepatic of propranolol on portal hypertension International Consensus Workshop on
venous pressure gradient predict in patients with cirrhosis: a denitions, methodology and therapeutic
clinical decompensation in patients hemodynamic study. Hepatology 1982; strategies. Oxford, UK: Blackwell
with compensated cirrhosis. 2: 5237. Science, 2001; 3664.
Gastroenterology 2007; 133: 9. Kroeger RJ, Groszmann RJ. Increased 13. Garcia-Tsao G, DAmico G, Abraldes J,
4818. portal venous resistance hinders portal et al. Prognostic indicators in cirrhosis.

Aliment Pharmacol Ther 2014; 39: 557-568 567

2014 John Wiley & Sons Ltd
E. Sinagra et al.

In: de Franchis R, ed. Portal mononitrate for the prevention of with viral cirrhosis. Am J Gastroenterol
Hypertension IV. Oxford: Proceedings variceal rebleeding. J Gastroenterol 2004; 99: 19538.
of the fourth Baveno International Hepatol 2012; 27: 16817. 25. Hobolth L, Bendtsen F, Hansen EF,
Consensus Workshop on Methodology 20. Hobolth L, Moeller S, Groenbaek H, Moller S. Effects of carvedilol and
of Diagnosis and Treatment. Blackwell, Keiding S, Bendtsen F, Feldager E. propranolol on circulatory regulation
2006; 47100. Carvedilol or propranolol in portal and oxygenation in cirrhosis: a
14. Feu F, Garcia-Pagan JC, Bosch J, et al. hypertension? A randomized controlled randomized study. Dig Liv Dis 2013;
Relation between portal pressure trial J Hepatol 2010; 52: s59182. doi: 10.1016/j.dld.2013.10.013 [Epub
response to pharmacotherapy and risk 21. Banares R, Moitinho E, Piqueras B, ahead of print].
of recurrent variceal haemorrhage in et al. Carvedilol, a new nonselective 26. Hobolth L, Moller S, Gronbaek H,
patients with cirrhosis. Lancet 1995; beta-blocker with intrinsic anti- Roelsgaard K, Bendtsen F, Hansen EF.
346: 10569. Alpha1-adrenergic activity, has a Carvedilol or propranolol in portal
15. Kjaergard LL, Villumsen J, Gluud C. greater portal hypotensive effect than hypertension? A randomized
Reported methodological quality and propranolol in patients with cirrhosis. comparison Scand J Gastroenterol 2012;
discrepancies between large and small Hepatology 1999; 30: 7983. 47: 46774.
randomised trials in meta-analyses. Ann 22. Banares R, Moitinho E, Matilla A, et al. 27. Albillos A, Garcia-Pagan JC, Iborra J,
Intern Med 2001; 135: 9829. Randomized comparison of long-term et al. Propranolol plus prazosin
16. Gluud LL. Bias in clinical intervention carvedilol and propanolol compared with propranolol plus
research. Am J Epidemiol 2006; 163: administration in the treatment of isosorbide-5-mononitrate in the
493501. portal hypertension in cirrhosis. treatment of portal hypertension.
17. Higgins JPT, Thompson SG, Deeks JJ, Hepatology 2002; 36: 136773. Gastroenterology 1998; 115: 11623.
Altman DG. Measuring inconsistency in 23. De BK, Das D, Sen S, et al. Acute and 28. Reiberger T, Ulbrich G, Ferlitsch A,
metaanalyses. BMJ 2003; 327: 55760. 7-day portal pressure response to et al. Carvedilol for primary
18. Tripathi D, Ferguson JW, Kochar N, carvedilol and propranolol in cirrhotics. prophylaxis of variceal bleeding in
et al. Randomized controlled trial of J Gastroenterol Hepatol 2002; 17: 1839. cirrhotic patients with haemodynamic
carvedilol versus variceal band ligation 24. Lin HC, Yang YY, Hou MC, Huang non-response to propranolol. Gut 2013;
for the prevention of the rst variceal YT, Lee FY, Lee SD. Acute 62: 163441.
bleed. Hepatology 2009; 50: 82533. administration of carvedilol is more 29. Egger M, Smith DG, Schneider M,
19. Lo GH, Chen WC, Wang HM, Yu HC. effective than propranolol plus Minder C. Bias in meta-analysis
Randomized controlled trial of isosorbide-5-mononitrate in the detected by a simple, graphical test.
carvedilol versus nadolol plus isosorbide reduction of portal pressure in patients BMJ 1997; 315: 62934.

568 Aliment Pharmacol Ther 2014; 39: 557-568

2014 John Wiley & Sons Ltd