Aliment Pharmacol Ther 2010; 32: 859–871
Background The evidence concerning the use of isosorbide-mononitrate (IsMn) for oesophageal varices is equivocal.
Aim To assess the effects of IsMn for patients with oesophageal varices and no previous bleeding (primary prevention) or previous variceal bleeding (secondary prevention).
Methods Systematic review with meta-analyses of randomized trials on IsMn alone or with beta-blockers or endoscopic therapy for oesophageal varices. Electronic and manual searches were combined. Randomized trials on primary and secondary prevention were included. The primary outcome measure was mortality. Intention-to-treat random effects meta-analyses were performed. The robustness of the results was assessed in trial sequential analyses.
Results Ten randomized trials on primary and 17 on secondary prevention were included. Evidence of bias was identified. No apparent effect of IsMn on mortality compared with placebo or beta-blockers or IsMn plus beta-blockers vs. beta-blockers was identified. Compared with endoscopic therapy, IsMn plus beta-blockers had no apparent effect on bleeding, but did seem to reduce mortality in secondary prevention (RR 0.73, 95% CI 0.59–0.89), but not in primary prevention. The effect of IsMn plus beta-blockers on mortality in secondary prevention was not confirmed in trial sequential analysis.
Conclusions Isosorbide-mononitrate used alone or in combination with beta blockers does not seem to offer any reduction in bleeding in the primary or secondary prevention of oesophageal varices. Compared with endoscopic therapy, there may be a survival advantage in using IsMn and beta-blockers, but additional large multicentre trials are needed to verify this finding.
The prevalence of oesophageal varices is about 50% among patients who are diagnosed as having cirrhosis of the liver and at least two-thirds of patients with cirrhosis will develop oesophageal or gastric varices.1 The risk of bleeding, once the varices are developed, depends on the severity of the liver disease assessed by the Child–Pugh class, the variceal size, severity of red colour signs and the portal pressure.2, 3 Without treatment, about one-third of patients with oesophageal varices will experience their first bleeding episode within 2 years after the diagnosis is made. The subsequent risk of rebleeding after 1 year is about 65% if treatment is not initiated. Treatment with nonselective beta-blockers or endoscopic therapy reduces the risk of bleeding considerably.4, 5 Still, the prognosis remains poor. About 50% of patients are nonresponders to beta-blockers or cannot continue treatment due to adverse effects.6 A meta-analysis of randomized trials on patients with oesophageal varices and no history of bleeding has found overall mortality rates of about 34% after banding ligation and 27% after beta-blockers.7 Although both interventions appear to reduce mortality compared with no treatment, the mortality rates suggest that additional therapeutic options should be assessed. Isosorbide-mononitrate (IsMn) has been suggested to be used alone or in combination with beta-blockers.5 However, one randomized trial found that IsMn increased the risk of bleeding compared with placebo.8 Other randomized trials on IsMn vs. beta-blockers or IsMn plus beta-blockers vs. beta-blockers or endoscopic therapy have reached equivocal results.9, 10 Recent meta-analyses including between four and 25 randomized trials have found that the combination of endoscopic and pharmacological therapy with beta-blockers alone or combined with IsMn may be more effective in reducing the risk of rebleeding than either of the therapeutic options used alone.11–13 No clear effect on mortality was identified. The meta-analyses also found considerable inter-trial heterogeneity and some evidence of bias. Accordingly, we performed a systematic review of randomized trials to evaluate the strength of the evidence on IsMn alone or in combination with beta blockers or endoscopic therapy for the primary or secondary prevention of bleeding in patients with oesophageal varices.
The objective of the present review was to assess the effects of IsMn for patients with oesophageal varices and no previous bleeding (primary prevention) or previous variceal bleeding (secondary prevention). Randomized trials were included irrespective of publication status or language. The treatment comparisons assessed included IsMn alone or with beta-blockers vs. placebo, no intervention, beta-blockers, endoscopic therapy (banding ligation or sclerotherapy), transjugular intrahepatic portosystemic shunts (TIPS), or the combination of IsMn plus beta-blockers and endoscopic therapy. The primary outcome measure was mortality. Secondary outcomes included upper gastrointestinal bleeding, variceal bleeding, bleeding-related mortality and adverse events. The review is based on a protocol published in the Cochrane Library.
Eligible trials were identified through searches in The Cochrane Hepato-Biliary Group Controlled Trials Register, The Cochrane Library, MEDLINE via PubMed and SilverPlatter, EMBASE and Science Citation Index Expanded. The last search update was January 2010. The searches were performed with help from the Cochrane Hepato-Biliary Group, Denmark and included the terms ‘isosorbide-mononitrate’ and ‘varices’, combined with a search strategy developed for the identification of randomized trials (http://www.mrw.interscience.wiley.com/cochrane). Manual searches including scanning of reference lists in relevant papers, conference proceedings, correspondence with experts and online trial registers through controlled-trials.com were also performed. Two authors (LG and AK) searched for eligible trials and extracted data. The primary authors of the included trials were contacted for additional information if data were not included in the published trial reports.
Based on previous evidence, the primary assessment of bias control focused on randomization methods (allocation sequence generation and allocation concealment).14, 15 The allocation sequence generation was classified as adequate if based on computer-generated random numbers, a table of random numbers, or similar. The allocation concealment was classified as adequate if patients were randomized through a central independent unit, serially numbered indistinguishable drug containers or bottles, serially numbered opaque sealed envelopes, or similar. Additional measures of bias control included blinding (whether the trial was double-blind or single-blind, whether patients, outcome assessors or other persons involved in the trial were blinded and whether the effect of blinding was assessed), handling of missing outcome data, reporting of outcome measures (reporting bias) and other apparent biases including sample size calculations and whether the preset sample size was reached. The risk of reporting bias was assessed through comparisons among trial protocols, abstracts and full paper articles.
All analyses were performed in revman version 5 (The Nordic Cochrane Centre, Copenhagen, Denmark) and stata version 10.0 (Stata Corp, College Station, Texas, USA). Analyses were performed using the intention-to-treat principle including all patients, irrespective of compliance or follow-up. For patients with missing outcome data, carry forward of the last observed response was used. Due to expected clinical heterogeneity, all meta-analyses were performed using random effects models. Results were expressed as risk ratios (RR) with 95% confidence intervals (CI) and the degree of intertrial heterogeneity as I-square values (I2). Meta-analyses of all outcome measures except adverse events were performed stratified for subgroups of trials on primary prevention and secondary prevention. The analyses of adverse events were performed without stratification for previous variceal bleeding. To estimate the potential influence of bias, the primary meta-analyses were repeated including only trials with adequate randomization. Regression analyses of funnel plot asymmetry were performed to assess the risk of bias (or small trial effects). As we identified discrepancies between the number of patients and events in abstracts and full paper articles, we also performed post hoc analyses including data from the abstracts. We originally planned to perform multiple intervention comparison meta-analyses, but were unable to perform these analyses due to the limited number of trials. To account for multiple comparisons associated with the design of our review, we performed trial sequential analysis for statistically significant outcome measures with alpha set to 5% and power to 80%.16
Forty-nine of 451 potentially relevant references identified in the electronic and manual searches were retrieved for further assessment (Figure 1). Forty-five references referred to randomized trials that fulfilled our inclusion criteria. One ongoing trial (ISRCTN16334693) had to be excluded (data not yet available). In total, 27 randomized trials (described in 44 references) were included in our analyses.8–10, 17–42 Six trials were published in abstract form.21, 26–28, 33, 35 Remaining trials were published as full paper articles. One trial was translated from the Chinese.32 Remaining trials were published in English. The trials were performed in Argentina, China, Egypt, France, Great Britain, India, Italy, Pakistan, Spain and Taiwan. The maximum duration of follow-up ranged from 8 to 91 months (median 24 months). Ten trials assessed primary prevention and remaining trials secondary prevention (Table 1). Patient characteristics were reported in all full paper articles. In two trials, 85–86% of included patients had cirrhosis.34, 40 The remaining trials only included patients with cirrhosis. The mean age of included patients ranged from 51 to 66 years and the proportion of men from 53% to 85%. In one trial, all included patients had ascites.29 In remaining trials, 16–67% of included patients had ascites. The proportion of patients with large varices ranged from 36% to 100% in trials on primary prevention and from 77% to 100% in trials on secondary prevention. The reported exclusion criteria included contraindications to IsMn or beta-blockers (heart disease with aortic stenosis or atrioventricular block, peripheral ischaemic disease, or chronic pulmonary disease), chronic renal failure and malignant disease.
|Trials on pharmacological therapy in primary prevention|
|Garcia-Pagan et al.8||IsMn vs. placebo|
|Angelico et al.20||IsMn vs. propranolol|
|Borroni et al.29||IsMn vs. nadolol|
|D’Amico et al.25||IsMn plus nadolol vs. placebo plus nadolol|
|Garcia-Pagan et al.23||IsMn plus propranolol vs. placebo plus propranolol|
|Merkel et al.9||IsMn plus nadolol vs. placebo plus nadolol|
|Deplano et al.27||IsMn plus nadolol vs. nadolol*|
|Trials on pharmacological and endoscopic therapy in primary prevention|
|Wang et al.37||IsMn plus nadolol vs. banding ligation|
|Lui et al.30||IsMn vs. propranolol vs. banding ligation|
|Abdelfattah et al.21||IsMn vs. propranolol vs. banding ligation*|
|Abulfutuh et al.33||IsMn vs. propranolol vs. banding ligation*|
|Trials on pharmacological therapy in secondary prevention|
|Patti et al.26||IsMn plus nadolol vs. placebo plus nadolol*|
|Gournay et al.10||IsMn plus propranolol vs. propranolol|
|Zhang et al.32||IsMn plus propranolol vs. propranolol|
|Trials on pharmacological and endoscopic therapy in primary prevention|
|Agrawal et al.28||IsMn plus propranolol vs. banding ligation*|
|Lo et al.38||IsMn plus nadolol vs. banding ligation|
|Patch et al.31||IsMn plus propranolol vs. banding ligation|
|Sarin et al.34||IsMn plus propranolol vs. banding ligation|
|Villanueva et al.18||IsMn plus nadolol vs. banding ligation|
|Villanueva et al.17||IsMn plus nadolol vs. sclerotherapy|
|Romero et al.36||IsMn plus nadolol vs. banding ligation plus sclerotherapy|
|Escorsell et al.22||IsMn plus propranolol vs. transjugular intrahepatic portosystemic shunt (TIPS) for secondary prevention|
|Garcia-Pagan et al.24||IsMn plus nadolol vs. IsMn plus nadolol and banding ligation|
|Lo et al.38||IsMn plus nadolol vs. IsMn plus nadolol and banding ligation|
|Kumar et al.40||IsMn plus propranolol and banding ligation vs. banding ligation|
|Shiha et al.35||IsMn plus propranolol vs. propranolol vs. banding ligation*|
|Ahmad et al.39||IsMn plus propranolol vs. propranolol vs. IsMn plus propranolol and banding ligation vs. banding ligation for secondary prevention|
All trials used a parallel group design. The number of comparison groups was two in 21 trials, three in five trials and four in one trial (Table 1). Trials on primary prevention assessed IsMn vs. placebo, beta-blockers, or banding ligation or IsMn plus beta-blockers vs. beta-blockers or banding ligation. Trials on secondary prevention assessed IsMn plus beta-blockers alone or with banding ligation vs. beta-blockers, endoscopic therapy, or TIPS. The initial dose of IsMn ranged from 20 to 40 mg/day. The dose was increased to a maximum of 40–80 mg/day. The mean dose administered in the trials ranged from 30 to 73 mg/day. In trials on IsMn and beta-blockers, IsMn was initiated after the maintenance dose of beta-blocker was achieved. The beta-blockers assessed included nadolol and propranolol. The initial dose ranged from 40 to 80 mg/day. In most trials, the dose was adjusted to achieve a resting heart rate of about 55–60 beats/min, a 20–25% reduction in the resting heart rate, or a maximum of 160–240 mg/day if tolerated. The mean dose of beta-blockers administered in the trials ranged from 40 to 125 mg/day. Most trials were initiated after the introduction of multiple band ligation devices in 1995. Patients randomized to endoscopic therapy with banding ligation or sclerotherapy initially underwent endoscopy every 2–4 weeks until variceal eradication was achieved. Subsequently, control endoscopy was performed monthly to every third month. In trials with long-term follow-up, patients randomized to endoscopic therapy alone underwent control endoscopy with about 6 months’ intervals. The TIPS procedure was performed using radiological/ultrasonographic guidance using wall-stent endoprosthesis. The diameter was initially dilated to 8 mm in diameter and then to 10 mm if the portocaval pressure gradient remained more than 12 mmHg.
Methodological quality of included trials
Randomization methods (allocation sequence generation and allocation concealment) were classified as adequate in 17 trials (Figure 2). Two trials had adequate allocation sequence generation, but unclear allocation concealment.34, 40 In the remaining trials, randomization methods were unclear. For two trials,27, 38 there were considerable differences between the characteristics of patients randomized to the treatment and control groups (number of patients and proportion with large varices respectively). Adequate double blinding of patients and investigators using IsMn placebo was used in six trials.8, 9, 20, 23, 25, 26 One trial was described as single-blind without specification of the method of blinding.20 None of the trials assessed the effect of blinding.
Sample size calculations were reported for 18 trials.8–10, 17, 18, 22–26, 30, 31, 34, 36–38, 40, 41 Two trials performed simultaneously at one clinical site were terminated prematurely after interim analysis of a similar study suggested that IsMn increases mortality in patients with oesophageal varices.25, 26 Another trial, which was planned to include 352 patients, was terminated due to low recruitment rates.37 One single centre trial included 137 patients, although the reported sample size calculation suggested that the planned size was 120 patients.34 The authors did not describe whether the trial was continued beyond the planned size due to unexpectedly low event rates or for some other reason.
Three trials were included in clinical trial registers.24, 40, 41 All trials were registered after their completion date. Two trials described the same primary outcome measures in the protocol and published report.24, 40 The protocol for one trial defined rebleeding as the primary outcome measure and complications as the secondary outcome measure.41 In the published paper of the trial, the primary outcome measure was variceal rebleeding and the secondary outcome was mortality. The protocol and published report for the second trial defined upper gastrointestinal bleeding due to causes not related to portal hypertension and adverse events requiring treatment discontinuation as the secondary outcomes assessed.40 An abstract describing the trial reported that the secondary outcomes were complications of cirrhosis including encephalopathy, hepatorenal syndrome, spontaneous bacterial peritonitis and jaundice.42 The abstract also reported that the number of patients randomized and the overall mortality rates in the treatment and control group was 9/79 vs. 5/92 respectively. In the full paper article, the numbers were 2/88 vs. 3/89 respectively. The reason for the discrepancy including the total number of patients randomized was not clear. There were no notable differences between the abstracts and published reports of remaining trials.
Ten trials on primary prevention reported mortality (Figure 3). No apparent effect of IsMn was identified for the intervention comparisons assessed. No differences in mortality were seen between IsMn and beta-blockers vs. beta-blockers (49/277 vs. 50/275; RR 0.95; 95% CI 0.68–1.32), or banding ligation (6/31 vs. 8/30; RR 0.73; 95% CI 0.29–1.84). There was little evidence of intertrial heterogeneity (I2 0–6%).
Two meta-analyses included trials with unclear randomization methods.20, 21, 33 Excluding these trials from the meta-analyses did not change the overall conclusions. Regression analyses of funnel plot asymmetry found little evidence of bias (P = 0.323–0.993).
Nine trials reported bleeding (Figure 4). IsMn increased the risk of bleeding compared with placebo (RR 2.34; 95% CI 1.10–4.97) or banding ligation (RR 4.33; 95% CI 1.57–11.92). There were no apparent differences between bleeding rates of patients randomized to IsMn alone or with beta-blockers vs. beta-blockers or banding ligation. Considering the small number of trials and patient, we did not perform trial sequential analysis for this outcome measure.
Meta-analyses of variceal bleeding found a detrimental effect of IsMn compared with banding ligation (RR 3.31; 95% CI 1.01–10.84). There was no apparent difference in variceal bleeding for the remaining treatment comparisons (Table 2). No effects on bleeding-related mortality were seen for any of the treatment comparisons assessed.
|Treatment comparison||Exp*||Ctr*||RR (95% CI)|
|Variceal bleeding in trials on primary prevention|
|IsMn vs. placebo||15||67||7||66||2.11 (0.92–4.84)|
|IsMn vs. beta-blockers||44||181||33||193||1.41 (0.94–2.11)|
|IsMn vs. banding ligation||28||124||7||88||2.82 (1.29–6.16)|
|IsMn plus beta-blockers vs. beta-blockers||28||279||38||273||0.71 (0.45–1.13)|
|IsMn plus beta-blockers vs. banding ligation||7||31||3||30||2.26 (0.64–7.93)|
|Bleeding-related mortality in meta-analyses of trials on primary prevention|
|IsMn vs. placebo||4||67||1||66||3.94 (0.45–34.33)|
|IsMn vs. beta-blockers||12||146||12||152||1.04 (0.33–3.26)|
|IsMn vs. banding ligation||1||62||2||44||0.35 (0.03–3.79)|
|IsMn plus beta-blockers vs. beta-blockers||11||279||8||273||1.30 (0.52–3.25)|
|Variceal bleeding in meta-analyses of trials on secondary prevention|
|IsMn plus beta-blockers vs. beta-blockers||44||141||65||148||0.71 (0.52–0.96)|
|IsMn plus beta-blockers vs. endoscopic therapy||146||445||133||449||1.07 (0.87–1.32)|
|IsMn plus beta-blockers alone or with banding ligation||60||173||38||177||1.61 (1.14–2.27)|
|IsMn plus beta-blockers vs. TIPS||17||44||6||47||3.03 (1.31–6.98)|
|IsMn plus beta-blockers and banding ligation vs. banding ligation||11||88||11||89||1.01 (0.46–2.21)|
|Bleeding-related mortality in meta-analyses of trials on primary prevention|
|IsMn plus beta-blockers vs. beta-blockers||13||99||13||100||1.01 (0.49–2.09)|
|IsMn plus beta-blockers vs. endoscopic therapy||22||299||26||298||0.87 (0.50–1.52)|
|IsMn plus beta-blockers alone or with banding ligation||11||138||5||140||1.95 (0.71–5.33)|
|IsMn plus beta-blockers vs. TIPS||4||44||3||47||1.42 (0.34–6.01)|
|IsMn plus beta-blockers and banding ligation vs. banding ligation||4||79||1||92||4.66 (0.53–40.82)|
Sixteen trials on secondary prevention reported mortality in trials on IsMn plus beta-blockers alone or with endoscopic therapy (Figure 5). Random effects meta-analyses found that IsMn plus beta-blockers reduced mortality compared with endoscopic therapy (111/496 vs. 137/502; RR 0.73; 95% CI 0.59–0.89; I2 0%). The findings were confirmed in sensitivity analyses excluding trials on sclerotherapy (RR 0.69; 95% CI 0.55–0.88) and trials without adequate randomization methods (RR 0.76; 95% CI 0.62–0.95). No differences in mortality were seen between remaining intervention comparisons. Regression analyses revealed no clear evidence of bias (P = 0.141–0.291).
We performed a trial sequential analysis to assess the potential effect of the effect of IsMn plus beta-blockers vs. endoscopic therapy on mortality. The priory intervention effect and the control group event rate were based on the findings of the random effects meta-analysis. The analysis confirmed that the cumulated z-curve (the estimated intervention effect) crossed the traditional boundary of 5% significance. However, the z-curve did not cross the trial sequential monitoring boundary, implying that there is no firm evidence of a survival benefit of IsMn plus beta-blockers when the analysis was adjusted for multiple testing.
Fifteen trials reported overall bleeding (Figure 6). The risk of bleeding was higher among patients randomized to IsMn plus beta-blockers compared with TIPS (22/44 vs. 8/47, RR 2.94; 95% CI 1.46–5.90). No other differences in bleeding were seen for the remaining treatment comparisons. IsMn plus beta-blockers increased the risk of variceal bleeding compared with banding ligation or TIPS (RR 1.61; 95% CI 1.14–2.27 and RR 3.03; 95% CI 1.31–6.98, Table 2). No differences in variceal bleeding were seen for the remaining treatment comparisons. The risk of bleeding-related mortality did not differ for any of the treatment comparisons assessed.
On the basis of identified discrepancies between the published report and abstract for one of the included trials, we repeated our meta-analyses including data described in the abstract.40, 42 These post hoc analyses did not change our overall conclusions.
A considerable number of adverse events were registered (Table 3). There was little evidence of intertrial heterogeneity (I2 0% in all analyses). There was no clear difference between the total number of adverse events compared with placebo (14/67 vs. 7/66; RR 1.97; 95% CI 0.85–4.57) or beta-blockers (30/62 vs. 26/66; RR 1.23; 95% CI 0.83–1.82). No differences in the number of withdrawals were seen between IsMn vs. placebo (10/67 vs. 3/66; RR 3.28; 95% CI 0.95–11.40) or beta-blockers (19/146 vs. 32/156; RR 0.63; 95% CI 0.38–1.05). Compared with beta-blockers, IsMn plus beta-blockers increased the total number of adverse events (96/251 vs. 57/251; RR 1.65; 95% CI 1.25–2.17) and the number of withdrawals due to adverse events (49/322 vs. 18/325; RR 2.60; 95% CI 1.55–4.38). IsMn used alone or with beta-blockers mainly increased the risk of headache, but was not clearly associated with the risk of hypotension (Table 3). As expected, the risk of developing hepatic encephalopathy was lower for patients randomized to IsMn plus beta-blockers compared with TIPS (6/44 vs. 18/47; RR 0.36; 95% CI 0.16–0.81). The risk of ascites was not different for these two treatment groups (RR 1.07; 95% CI 0.16–7.26). Among patients randomized to banding ligation, 28 of 442 developed clinically relevant oesophageal ulcers (RR 0.10; 95% CI 0.03–0.31) and 10 of 216 dysphagia (RR 0.12; 95% CI 0.02–0.68). One case of perforation of the oesophagus and one case of glottis oedema was registered following banding ligation. One trial reported adverse events following sclerotherapy.17 Among 43 patients randomized to sclerotherapy, bleeding ulcers were seen for seven patients and oesophageal strictures for two patients.
|Treatment comparison||Exp*||Ctr*||RR (95% CI)|
|IsMn vs. beta-blockers|
|Ascites, 1 trial||12||57||10||61||1.28 (0.60–2.74)|
|Asthenia or dizziness, 3 trials||15||146||12||152||1.34 (0.66–2.74)|
|Diarrhoea, 1 trial||3||62||0||66||7.44 (0.39–141.26)|
|Dyspnoea, 2 trials||5||89||25||91||0.23 (0.10–0.54)|
|Heart failure, 1 trial||0||57||2||61||0.21 (0.01–4.36)|
|Hypotension2 trials||2||84||3||86||0.72 (0.14–3.65)|
|Impotence, 2 trials||0||84||2||86||0.33 (0.04–3.12)|
|Insomnia, 1 trial||4||62||3||66||1.42 (0.33–6.09)|
|Nausea, 1 trial||3||62||4||66||0.80 (0.19–3.42)|
|Poor memory, 1 trial||0||62||3||66||0.15 (0.01–2.88)|
|Headache, 1 trial||16||119||0||127||11.69 (1.21–113.15)|
|IsMn plus beta-blockers vs. beta-blockers|
|Ascites, 3 trials||29||155||23||152||1.23 (0.74–2.03)|
|Angina or stroke, 3 trials||5||251||2||250||1.94 (0.48–7.80)|
|Asthenia or dizziness, 3 trials||21||240||22||240||0.94 (0.53–1.67)|
|Bradycardia, 3 trials||4||256||7||262||0.67 (0.19–2.35)|
|Arrhythmia, 1 trial||1||30||0||27||2.71 (0.12–63.84)|
|Cutaneous reaction, 1 trial||0||175||4||174||0.11 (0.01–2.04)|
|Dyspnoea, 2 trials||4||76||1||76||2.46 (0.33–18.12)|
|Hypotension, 1 trial||3||30||0||27||6.32 (0.34–117.09)|
|Hepatic encephalopathy, 1 trial||2||27||4||30||0.56 (0.11–2.80)|
|Impotence, 1 trial||0||46||1||49||0.35 (0.01–8.49)|
|Nausea, 1 trial||1||30||0||27||2.71 (0.12–63.84)|
|Raynaud, 1 trial||0||46||1||49||0.35 (0.01, 8.49)|
|Headache, 5 trials||56||385||17||363||2.92 (1.77–4.82)|
|IsMn plus beta-blockers vs. endoscopic therapy|
|Ascites, 2 trials||59||104||57||103||1.06 (0.85–1.32)|
|Asthenia or dizziness, 2 trials||14||161||0||155||9.04 (1.69–48.50)|
|Bradycardia, 2 trials||4||100||0||95||4.55 (0.53–39.28)|
|Dyspnoea, 3 trials||8||160||0||165||6.43 (1.16–35.57)|
|Headache, 3 trials||16||174||0||174||9.86 (1.81–53.72)|
|Hypotension, 1 trial||5||51||0||51||11.00 (0.62, 193.90)|
|Hepatic encephalopathy, 1 trial||44||61||39||60||1.11 (0.87–1.41)|
|Impotence, 2 trials||2||100||0||95||2.87 (0.30–27.10)|
|Headache, 3 trials||16||174||0||174||9.86 (1.81–53.72)|
The available evidence presented in this review suggests that IsMn used alone or in combination with beta-blockers has no effect on bleeding in primary or secondary prevention of oesophageal varices. In secondary prevention, IsMn plus beta-blockers appeared to reduce mortality compared with endoscopic therapy. The potential improved survival benefit did seem related to a reduced risk of bleeding. The combined evidence does not support the use of IsMn for prevention of oesophageal varices, but suggests that additional large randomized trials with long-term follow-up are warranted.
The need for additional trials is underlined by the identification of reporting bias. Previous studies have found discrepancies between protocols and full paper articles regarding the defined primary and secondary outcomes.43 We found similar evidence when comparing published protocols, abstracts and full paper articles. Abstracts often report preliminary data, but may also indicate dropouts or withdrawals. We found evidence of losses to follow-up based on the comparison between the abstract and full paper article for one of the included trials. Accordingly, performing intention-to-treat analysis in systematic reviews without the original data may be difficult. Access to individual patient data would also make subgroup analyses possible. Questions such as the benefit of IsMn for patients with gastric as well as oesophageal varices would have been interesting to analyse. Unfortunately, the available data did not allow such analyses to be performed in a reliable manner.
The present review found that IsMn plus beta-blockers reduced mortality compared with endoscopic therapy, but had no clear effect on overall bleeding. This suggests that the medical therapy had beneficial effects not directly related to a reduced risk of bleeding. For beta-blockers, nonhaemodynamic effects such as protection against infections may explain our findings.6, 44, 45
The haemodynamic effects of the medical therapy are related to a reduction in the hepatic venous pressure gradient (HVPG). Clinical studies found that the risk of bleeding is considerably reduced when the HVPG is reduced to <12 mmHg or by at least 20% compared with baseline values.46 The effects of beta-blockers on the portal venous inflow include a decrease in cardiac output and splanchnic vasoconstriction. IsMn decreases the intra-hepatic resistance through vasodilatation. The addition of IsMn to beta-blockers has a synergistic effect on the portal pressure, but no additional effect on the collateral blood flow.47 Therefore, a combination may not translate into a clinical effect on the risk of variceal bleeding.
Sample size calculations are an essential part of the planning of randomized trials. When the calculation is not reported, the reader does not know whether the planned sample size was reached or the trial terminated at an arbitrary time. Two of the trials in the present review were terminated based on concurrent evidence from a concurrent trial suggesting a detrimental effect of the interventions assessed.25, 26 Another trial was planned to include 352 patients, but was terminated after inclusion of 61 patients due to low recruitment rates. Less than one-third of included trials reported sample size calculations or achieved the planned size. We cannot exclude that the total number of patients in the individual subgroups was too low to determine intervention effects with sufficient power. The combined considerations suggest that additional trials are warranted before IsMn can be recommended for prevention in oesophageal varices.
L. Gluud and A. Krag contributed to study concept and design. L. Gluud, E. Langholz and A. Krag did data extraction, analysis and interpretation of data. L. Gluud drafted the review. All authors have read and approved the paper. Declaration of personal and funding interests: None.