Recently, Lebrec and colleagues from Clichy, France, reported an increased mortality in 77 patients with cirrhosis and varices and refractory ascites in whom propranolol was administered, compared to 74 patients with refractory ascites but no varices, who were not taking nonselective beta-blockers (NSBBs).1 During follow-up lasting a median of 8 months (range 1-47 months), the probability of death was 59% at 1 year and 72% at 2 years: 81% of patients taking propranolol died during the follow-up, and use of propranolol was the third cause of death, with odds ratio = 2.61 (95% confidence interval = 1.63-4.19).1
These findings are potentially very important, but are difficult to reconcile with some of the published literature. Although it is true that most randomized controlled trials (RCTs) comparing beta-blockers to placebo or other pharmacotherapy for prevention of bleeding from varices excluded patients with advanced cirrhosis and refractory ascites, this was not universal. Moreover, despite more rebleeding, an increased mortality with propranolol has not been reported in comparative trials versus banding ligation. Indeed, the recent trial by Lo et al., with extended follow-up, showed better survival with beta-blockers than with banding despite more rebleeding.2 Second, when we reviewed the literature to explore the potential beneficial effect of propranolol in preventing spontaneous bacterial peritonitis (SBP) in patients with cirrhosis, we included three RCTs and one prospective study comprising 644 patients, 468 with ascites, and 257 receiving propranolol.3 Among these, 125 patients had Child C fibrosis (101 were taking NSBBs). The average hepatic venous pressure gradient was comparable to that documented in the study from Clichy. Moreover, in the prospective study, 67 of 134 (50%) patients had tense ascites requiring paracentesis. However, the overall mortality in the four studies was 21%, which is significantly lower than in the group in Clichy, despite a much longer follow-up: 8 years in two RCTs and 5 years in one RCT3 (Table 1).
|Turnes et al.9||Gonzales Suarez et al.10||Abraldes et al.11||Cholongitas et al.12|
|Type of study||RCT||RCT||RCT||Retrospective|
|Type of publication||Full length||Full length||Full length||Full length|
|Number of patients||71||230||73||134|
|Mean age ± SD (years)||57.8 ± 1.7||59 ± 1.2||54.5 ± 8||66 ± 12|
|Etiology of liver disease (alcohol/nonalcohol)||15/56||110/120||37/36||20/106|
|Severity of liver disease, Child A/B/C (number)||58/13/0||50/131/49||32/36/9||17/116/67|
|% of patients with ascites at inclusion||35||64||31||24.6|
|% of patients treated with propranolol||100||50||100||25|
|Control group||nonresponders||sclerotherapy||nonresponders||not treated|
|Follow-up (months)||76||23 ± 1.4||70||36|
|Death NSBB/ controls (%)||32/34.7||34/41||3.5/31||6/13.8|
|Causes of death||NA||35 gastrointestinal bleeding, 17 infection, 6 HCC, 8 other causes||4 gastrointestinal bleeding, 4 hepatorenal syndrome, 2 EPS, 1 other causes||NA|
In addition, the causes of death were different in the reviewed studies compared to the present study. Gastrointestinal bleeding was the most important cause of death, followed by hepatocellular carcinoma (HCC). Hepatorenal syndrome, which was not described as cause of death in the Clichy study, was described in four patients, all nonhemodynamic responders to beta-blockers. It is important to note that propranolol appeared to confer a reduced risk of developing SBP, which is a frequent infection in patients with refractory ascites. In contrast in the patients reported in the Clichy study, infection was the most important cause of death.1
In our review,3 NSBB conferred a statistically significant relative risk reduction of 12% (95% confidence interval = 5.5%-18.8%) in the future occurrence of SBP, which was not closely related to the hemodynamic targets of hepatic venous pressure gradient reduction, whether to less than 12 mm Hg or to a 20% reduction from baseline. The protective effect of NSBBs could be due to a reduction of bacterial translocation, due to an increase in intestinal motility and/or by a decrease in intestinal permeability consequent to the reduction of portal pressure.4, 5 Indeed, a model of splanchnic sympathectomy in the cirrhotic rat has demonstrated prevention of bacterial translocation of E. coli.6 Moreover, a significant decrease in the incidence of postsurgical infections has been shown in a cohort study of patients with cirrhosis treated with propranolol and ciprofloxacin, compared to ciprofloxacin alone after laparoscopic surgery (14.7% versus 42.4%).7 This effect may be due to the possible mechanisms described above, but also due possibly to an improvement of host defenses by NSBB, through inhibition of the stress-related cyclic adenosine monophosphate protein kinase A pathway, which has an inhibitory effect on the immune system.8
A possible important difference between the studies we reviewed,3 and that from Clichy, is that the mean dose of propranolol used was below 100 mg in all the former studies, but in the latter French study, 47% of patients were taking propranolol at a dose of 160 mg/day, except for one patient who was taking propranolol at a dose between 100 and 160 mg/day.1
It would have been interesting, after removing the causes of death due to HCC in the Clichy study, to evaluate the incidence of infections during follow-up, and the deaths related to this, according to whether propranolol was taken or not.
Thus, it is important to understand whether the increased mortality attributed to propranolol is solely due to the dose used or to the specific clinical setting of refractory ascites in patients treated at Clichy (or both), or if the increased mortality occurred by chance, or due to an associated factor not captured by the authors. An urgent survey of patient databases, and of current patients treated with NSBBs is needed, to confirm or refute the potential danger of prescribing or continuing to prescribe NSBBs in patients with cirrhosis who have or who develop refractory ascites. It is important to establish if hepatologists need to be hydrophobic when prescribing NSBBs for patients with cirrhosis!