Primary prevention of variceal haemorrhage: A pharmacological approach

Abstract

The Royal Free Sheila Sherlock Liver Centre and University Division of Surgery, UCL, and Royal Free Hospital, London NW3 2QG, UKGastro-intestinal bleeding from oesophago-gastric varices or por-tal hypertensive gastropathy due to cirrhosis is a life-threateningcomplication. During the last decades, mortality rates of varicealbleeding in patients with cirrhosis have been falling, but portalhypertensive bleeding continues to be amongst the leadingcauses of death [1].For primary prophylaxis of variceal bleeding, non-selectivebeta blockade (NSBB)and endoscopic band ligation (EBL) are bothused as they are effective therapies [2,3]. Which therapy shouldbe the one of first choice can be considered a controversy. How-ever there are several issues that go beyond the initial assess-ment of differences in survival, for which there are none or firstbleeding, and for which EBL has a statistical advantage overNSBB.The safety of the endoscopic procedure is the first consider-ation. EBL can cause fatal iatrogenic bleeding [4,5], also reportedin the largest trial with the longest follow-up [6].The increasedexpense and need for specialized staff of EBL compared to NSBBis the second consideration. As mortality is not different betweentherapies, the cost-effectiveness could be mainly related to thecost of the extra variceal bleeding episodes in the NSBB treatedpatients versus the equipment and staff costs for EBL. This hasbeen evaluated prospectively in trials comprising patients onliver transplant waiting lists with mixed results, as a US trialfound a NSBB strategy more expensive [7] and a European trialfound the EBL to cost more [8]. Fatal iatrogenic bleeding wasnot ‘‘costed” and it is unclear how it could be evaluated in suchan analysis. The third consideration is that EBL cannot preventbleeding from portal hypertensive gastropathy, which is the casefor NSBB [9]. The fourth consideration is that EBL has only beenevaluated in patients with medium/large varices. Patients withsmall varices and severe liver disease are candidates for primaryprophylaxis [3]. In a single trial based on patients with contrain-dications to NSBB, some of whom also had small varices, fatal iat-rogenic bleeding occurred with EBL and the trial was stoppedprematurely [5]. Conversely NSBB are effective in patients withsmall varices [10] and are effective independent of cause andseverity of cirrhosis, the presence of ascites and variceal size[11]. Despite this long-standing evidence, the common percep-tion amongst endoscopists is that NSBB are less effective inpatients with decompensated cirrhosis. Another common mis-conception is an increased risk of hepatic encephalopathy withadministration of beta-blockers. This has not been documentedin placebo-controlled trials and is not a contraindication toreceiving beta-blockers.A further consideration is that preventative therapy, particu-larly in asymptomatic patients, should be easy to administer,have few and no serious adverse effects, and be effective. NSBBon paper fulfil these criteria better than EBL, but unfortunatelythese aspects have not been evaluated in randomised compara-tive studies. Although a survey on patient preferences in the pro-phylactic setting has been published favouring EBL, the potentialiatrogenic bleeding with EBL was not part of the questionnaire[12].There are 16 randomised trials in patients with medium tohigh risk varices comparing EBL with propranolol [6–8,13–25].In the meta-analysis of the 16 studies, EBL significantly reducedthe risk of first variceal bleeding compared to propranolol (rela-tive risk difference 9.2%, 95% CI 5.2%-13.1%, and POR 0.5, 95% CI0.37–0.68). However, mortality in the same meta-analysis wasnot statistically different (POR 0.94, 95% CI 0.70–1.28). Recentlynadolol and isosorbide mononitrate combined were found tohave similar efficacy to EBL [26]. A summary of the meta-analyt-ical data is set out in Table 1 and Fig. 1A and B. The number ofpatients needing treatment with EBL to save one episode ofbleeding is 11 (95% CI 7–21 by random effects model). It requiresan average of three endoscopic sessions to eradicate varices, sothis means, on average 33 endoscopic procedures discounting fol-low-up endoscopies after eradication. Conversely NSBB remainscheap, as haemodynamic monitoring is not required in this set-ting [27].The important conclusion when considering the merits ofNSBB or EBL for primary prophylaxis against variceal bleedingis that mortality is no different. Survival following variceal bleed-ing, including that of more severely ill patients is improving [28].As several of the randomised trials were undertaken before theuniversal use of antibiotics, which improve survival in acute var-iceal bleeding [3,29], it is likely that mortality from bleeding isalready better and will further improve. Moreover, data is emerg-ing data that NSBB may also have other beneficial therapeuticeffects in patients with cirrhosis, for example by reducing the riskof spontaneous bacterial peritonitis [30]. Improved survival withNSBB compared to EBL has been shown in a long term follow upof a randomised trial of secondary prophylaxis [31,32] leading toJournal of Hepatology 2010 vol. xxx