Carvedilol for preventing recurrent variceal bleeding: Waiting for convincing evidence


  • Jaime Bosch M.D., Ph.D., F.R.C.P.

    1. Hepatic Hemodynamic Laboratory Liver Unit Hospital Clínic-IDIBAPS University of Barcelona and Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas (CIBERehd) Barcelona, Spain
    Search for more papers by this author

  • Potential conflict of interest: Dr. Bosch consults for and received grants from Chiasma and Microtech. He also consults for Gilead and Norgine.

  • This work was supported, in part, by grants from the Instituto de Salud Carlos III (ISCiii), Ministerio de Economía y Competitividad (PS 09/01261). The CIBERehd is funded by the ISCiii.

Lo GH, Chen WC, Wang HM, Yu HC. A randomized, controlled trial of carvedilol vs. nadolol plus isosorbide mononitrate for the prevention of variceal rebleeding. J Gastroenterol Hepatol 2012;27:1681-1687. (Reprinted with permission.)


Background: Carvedilol has been shown to be more effective than propranolol in decreasing portal pressure. Sufficient data from controlled trials remains limited. This trial compared the relative safety and efficacy between carvedilol and nadolol plus isosorbide mononitrate in preventing variceal rebleeding. Methods: After successful control of acute esophageal variceal bleeding, eligible patients were randomized to Carvedilol group, 61 patients, using carvedilol 6.25-12.5 mg daily or N + I group, 60 patients, using nadolol 40-80 mg plus isorsorbide-5-mononitrate 20mg daily. The end points were rebleeding from varices, adverse events or death. Results: After a median follow up of 30 months, recurrent upper gastrointestinal bleeding developed in 37 patients (61%) in the Carvedilol group and 37 patients (62%) in the N + I group (p = 0.90). Recurrent bleeding from esophageal varices occurred in 31 patients (51%) in the Carvedilol group and in 26 patients (43%) in the N + I group (p = 0.46). Recurrent bleeding from gastric varices occurred in 2 patients (3%) in the Carvedilol group and in 8 patients (13%) in the N + I group (p = 0.05). Severe adverse events occurred in 1 patient in Carvedilol group and 17 patients in N + I group (p<0.0001). Fifteen patients of the Carvedilol group and 17 patients in the N + I group died (p = 0.83). Two patients in the Carvedilol group and 3 patients in N + I group died of variceal bleeding. Conclusions: Carvedilol was as effective as nadolol plus isorsorbide-5 –mononitrate mononitrate in the prevention of gastroesophageal variceal rebleeding with fewer severe adverse events and similar survival.


Carvedilol, as propranolol, nadolol, and timolol, is a nonselective beta-blocker (NSBB) that has been proposed for the treatment of portal hypertension (PH). Unlike other NSBBs, carvedilol has the unique property of exerting a mild vasodilatory effect because of its intrinsic antialpha1 adrenergic activity and its capacity to enhance the release of nitric oxide (NO). These unique properties mean that carvedilol can reduce portal pressure both by decreasing portal-collateral blood flow (as all other NSBBs) and by diminishing the functional component of hepatic vascular resistance (the hepatic vascular tone) that is increased in cirrhosis and contributes to portal pressure elevation.1 Because of these combined effects, carvedilol has the potential of causing a more-pronounced decrease in portal pressure than propranolol or nadolol, the current standard NSBBs for the treatment of PH. However, because its vasodilatory effect is not liver selective, carvedilol carries the risk of causing or intensifying systemic hypotension, a drawback for any vasodilator in patients with advanced cirrhosis because it may enhance sodium retention.1, 2

Previous studies in patients with cirrhosis indeed demonstrated that carvedilol causes greater falls in portal pressure than propranolol, and that this effect is achieved with relatively low doses (12.5-25.0 mg/day) that usually do not cause systemic hypotension and are well tolerated.1-3 Reduction of the hepatic venous pressure gradient (HVPG) in these studies ranged between 16% and 43% with carvedilol versus 12%-13% with propranolol. Because the clinical efficacy of NSBBs decreasing the risk of variceal bleeding or rebleeding is tightly correlated with their capacity of decreasing the HVPG by more than 20% of baseline values or below 12 mmHg,4 the available evidence strongly suggested that carvedilol had potential to achieve better clinical results than propranolol/nadolol and thus was well suited to be used in randomized, controlled trials (RCTs) for PH.1

The fact that only two relatively small RCTs with clinical endpoints have been performed so far probably reflects the difficulties in conducting investigator-initiated RCTs, because the pharma companies did not sponsor therapeutic studies using carvedilol in patients with cirrhosis. The funding difficulties encountered explain some of the limitations of these studies that go from the lack of HVPG measurements to being open comparisons of carvedilol versus other treatments, instead of double-blind face-to-face comparison versus other NSBBs, alone or on top of endoscopic therapy. Label indications for carvedilol are limited to arterial hypertension and heart failure, for which there is regulatory approval. This is very much the same for all other NSBBs that, despite being recommended in all international guidelines as therapy for PH complicating cirrhosis, have to be used off-label.4

The first published RCT evaluated the use of carvedilol versus endoscopic band ligation in the primary prophylaxis of variceal bleeding in Scotland.5 This is the only trial reporting a statistically significant advantage for drug therapy over endoscopic therapy decreasing the probability of bleeding, although this did not translate into a survival advantage. Tolerability of carvedilol appeared good, becasue only 13% of the patients stopped taking the drug.

The second trial has recently been published by Lo et al.6 The study was conducted in Taiwan and evaluated the use of carvedilol versus nadolol plus isosorbide mononitrate in preventing variceal rebleeding. This was an open study conducted at a single center over 4 years including 121 patients that, if stable, were randomized at day 5 after bleeding to receive treatment with either a low dose of carvedilol (6.25-12.5 mg/day) or of nadolol (40-80 mg/day) plus isosorbide mononitrate (20 mg/day). The trial was powered to detect an absolute difference in rebleeding of at least 25%. Contrary to published guidelines, patients did not receive endoscopic band ligation unless they experienced repeat variceal rebleeding while on drug therapy. There were no differences between the two treatment groups in the number of rebleeding episodes or in actuarial probability of remaining free of rebleeding, either from gastroesophageal varices or from any source. Of note, rebleeding rates in this study were higher than expected in both arms, even taking into consideration the lack of endoscopic therapy, suggesting suboptimal efficacy probably related to the low doses administered. Actually, the observed rebleeding rates of 61% under carvedilol and 62% under nadolol plus isosorbide mononitrate after a median follow-up of 30 months are not much better that the 62% 2-year rebleeding with no therapy reported 30 years ago by Graham and Smith.7 Such insufficient dosage is also suggested by the relatively high heart rates once on stable doses of carvedilol (78 ± 9 bpm) or nadolol (75 ± 10 bpm). Survival probability was almost identical in both groups, of approximately 75% at 2 years. Only 5 deaths were the result of rebleeding (of a total of 31 deaths and of 100 variceal rebleeding episodes), indicating that though therapy was not good in preventing rebleeding, patients received successful treatment for their rebleeding episodes. The researchers claim that their results indicate that “carvedilol was as effective as nadolol plus isosorbide mononitrate in preventing gastroesophageal variceal rebleeding.” However, because the trial was not powered to detect absolute differences in rebleeding of less than 25%, there is large room for a type 2 error, with a fair chance of not detecting a substantial difference in rebleeding rates (for instance, of 15%, should this be the real difference in efficacy between the two treatments). Because of these limitations, the trial should at best be considered inconclusive with regard to efficacy, rather than showing noninferiority of one treatment versus the other. With these data in mind, it is regrettable that the researchers did not incorporate measurements of the effects of treatment on HVPG, because this would help in the understanding of whether the apparent lack of differences in clinical efficacy were the result of achieving an insufficient decrease in portal pressure. Incorporating HVPG measurements in such RCTs is actually recommended by recent guidelines4; hence, the lack of these measurements in this RCT represents a missed opportunity.

When two treatments appear to be of similar value in terms of efficacy and mortality, then other considerations, such as the profile of side effects and cost, become relevant. In this case, the cost of therapy is low and almost equivalent with both treatment options. However, a major difference was found with regard to adverse events and tolerability. Both regarding mild and severe adverse events thought to be treatment related, carvedilol fares better than nadolol plus isosorbide mononitrate; adverse events were noted in 8% versus 38% of patients, and severe adverse events were noted in 1 versus 17 patients. Interestingly, ascites did not worsen in any patient and no patient in the carvedilol group had to discontinue drug therapy because of intolerable side effects. The rate of adverse events and tolerability of carvedilol in this study are the best reported so far, because in the RCT by Tripathi et al. (that also used quite low doses of carvedilol), 13% of the patients discontinued the drug because of perceived side effects.5 Even with such contrasting results, it appears that at low doses of carvedilol are quite safe and easy to administer. Unfortunately, the clinical efficacy of carvedilol at these doses remains to be proven. In light of these studies, carvedilol should not be used to prevent variceal rebleeding outside clinical trials. Yet, it is likely that carvedilol would be finally proven to be extremely useful in PH. At least our own unpublished observations and recent presentations at international meetings coincide in suggesting that it is effective in primary prophylaxis, where it may become the beta-blocker of choice (if these preliminary reports are confirmed in fully published investigations). Certainly, the issue of its potential use for secondary prophylaxis requires the matter to be addressed by additional clinical studies. I favor a design comparing endoscopic band ligation (EBL) plus propranolol/nadolol versus EBL plus carvedilol, preferably in double-blind conditions and assessing the effect of treatments on HVPG.


The author acknowledges Mrs. C. Esteva for her expert support.