We read the article by Sersté and colleagues1 with great interest. The authors carried out a single-center, observational, prospective study and evaluated the effects of nonselective beta-blockers (NSBBs) on the survival of 151 patients with cirrhosis and refractory ascites. Seventy-seven patients (51%) received propranolol (113 ± 46 mg/day) for the prevention of gastrointestinal bleeding. Their crude conclusion was that the use of NSBBs in these patients was associated with a detrimental outcome because the median survival time was dramatically reduced in the propranolol-treated group (5 months; 95% confidence interval = 3.5-6.5 months) versus the non–propranolol-treated group (20 months; 95% confidence interval = 4.8-35.2 months; P < 0.0001). In a multivariate analysis, the administration of propranolol remained an independent predictor of death, and this strengthened the new concept of NSBB avoidance in patients with cirrhosis and refractory ascites.
This intriguing conclusion deserves comment because NSBBs are currently considered to be the cornerstone of treatment for portal hypertension. First, because of the lack of random treatment assignment, clinicians must be very careful in interpreting the results of observational studies, which are much more vulnerable to methodological issues such as selection bias or the presence of hidden confounders. Randomized controlled trials are considered the best way of proving causality and confirming what has been found in previous observational studies. Here, the apparent deleterious effect of NSBBs on the survival of patients with a high degree of portal hypertension may simply have been the effect of higher portal hypertension per se, and this may also have been responsible for larger varices (an indication for NSBBs) and may have had an impact on prognosis independently of the Model for End-Stage Liver Disease or Child-Pugh scores. The authors stated that similar hepatic venous pressure gradients (HVPGs) were observed between the two groups, but HVPGs were measured in only a subset of this cohort (37%); this precluded the extrapolation of the measured values to the true mean HVPG value for each group. Besides the two main well-recognized contributors to portal hypertension (i.e., the increased resistance to portal blood flow within the liver and the development of a hyperdynamic splanchnic circulatory state), the role of angiogenesis (the growth of new blood vessels from a preexisting vascular bed) has recently been pointed out.2 This extensive network of portosystemic collateral vessels, among which gastroesophageal varices represent only the tip of the iceberg, pours high concentrations of toxins or bacteria into the systemic circulation, which contribute to complications of cirrhosis (mainly sepsis). The assessment of the magnitude of portosystemic collaterals is still an unresolved issue, and whether or not the network of collateral vessels is well correlated to the portal pressure estimated by the HVPG is still under debate.
Second, the authors dismissed several issues that can have a major influence on outcome. Abstinence should have been mentioned because more than half of their patients were alcoholic. Whether their patients had been subjected to long-term antibiotic administration or had good compliance with NSBBs is also questionable in this study. The authors attributed the deleterious effect of NSBBs to more frequent or more prolonged hemodynamic impairment after paracentesis, but they did not indicate more frequent renal impairment in the group receiving NSBBs, which would have confirmed such a hypothesis. Moreover, one would expect more frequent paracentesis procedures if NSBBs had caused more pronounced postparacentesis-induced circulatory dysfunction.
Third and most importantly, the authors did not discuss recent favorable properties of NSBBs that may balance their pessimistic conclusion. Indeed, NSBBs have been shown to shorten the intestinal transit time by stimulating the β-adrenoreceptor–mediated pathway3 and thus lead to a decrease in bacterial overgrowth and, consequently, bacterial translocation, which plays a key role in the complications of portal hypertension.3,4 Over the last few years, polymerase chain reaction–based detection of bacterial DNA has been proposed as a surrogate marker of bacterial translocation because it has been detected in the blood and ascites of one-third of patients with cirrhosis and culture-negative ascites.5 More recently, Zapater and colleagues6 found that the presence of bacterial DNA in serum and ascites in such patients resulted in earlier and more frequent deaths in comparison with patients without bacterial DNA. They also assumed that bacterial DNA–induced nitric oxide could provoke hemodynamic instability, with a low mean arterial pressure leading to lethal acute-on-chronic liver failure. Moreover, the benefit of NSBBs in decreasing bacterial translocation may be observed without the achievement of a hemodynamic response associated with a reduction in variceal hemorrhaging7; indeed, these authors showed that an 11% reduction in the HVPG from the baseline is sufficient for preventing spontaneous bacterial peritonitis. This is markedly less than the 20% reduction threshold conferring protection against variceal hemorrhaging.8 A recent meta-analysis demonstrated that the use of NSBBs had a significant role in preventing spontaneous bacterial peritonitis independently of the hemodynamic response.9 Therefore, the sickest patients with cirrhosis under NSBB therapy walk a fine line between the detrimental effects (e.g., lowered arterial pressure) and beneficial effects (e.g. reduced bacterial translocation) of these drugs. The acknowledgment of this lifeline will surely allow NSBBs to be administered to those patients with cirrhosis and refractory ascites, but better discrimination of good candidates for NSBBs remains an important challenge. One tool for such discrimination could be the measurement of the activation of systemic vasopressor systems, as suggested by Sersté et al.,1 because this causes renal vasoconstriction, which contributes substantially to the mortality risk in such patients.
Finally, the primary strength of this observational study is that it opens our eyes for the first time to the potential detrimental effects of NSBBs in patients with cirrhosis and refractory ascites. However, these drugs, which are currently considered to be the aspirin of hepatologists, warrant further investigation with more valid randomized controlled trials before they are definitively avoided in such a critical situation. Another concern is the safety of band ligation as a replacement for NSBBs in such patients because fatal iatrogenic bleeding has been reported.10 Furthermore, endoscopic band ligation cannot prevent bleeding from portal hypertensive gastropathy, whereas NSBBs can. We need to learn more about the physiopathological mechanisms that could counteract the thoroughly comprehensive beneficial effects of NSBBs before we reject them definitively in patients with cirrhosis and refractory ascites.