We read with great interest the article by Tripathi et al.1 In this randomized controlled trial, carvedilol, a noncardioselective vasodilating beta-blocker, was compared with variceal banding ligation (VBL) for primary prophylaxis of variceal bleed. The authors found carvedilol to have lower bleeding rates than VBL, with no difference in survival by intention-to-treat (ITT) analyses. In addition, this was the first reporting of drug therapy having an advantage over VBL. Although their results provided important data for primary prophylaxis of variceal bleeding, several issues deserve further discussion.
First, the most important finding was the greater efficacy of carvedilol in the prevention of the first variceal bleed by ITT analyses. However, the significant statistic disappeared after per-protocol analyses. The dropout rates of the two treatment arms were around 30%, suggesting that only 70% of the initial subjects really completed the study protocol. In addition, the relative higher rate of first variceal bleed in the VBL arm could be due to several reasons, such as bleeding before the first endoscopy following randomization in three patients, noncompliance with the VBL protocol in five patients, and the use of propranolol as rescue therapy in patients with discontinued intervention, which was less effective in primary prevention of variceal hemorrhage than VBL, as confirmed by two meta-analyses.2, 3 In contrast, the rescue treatment for patients with discontinued intervention in the carvedilol arm was VBL. Taking these lines of evidence together, we should be very careful to interpret the results from the ITT analyses.
Second, bleeding due to banding ulceration was one important complication of VBL, thus decreasing the efficacy in primary prophylaxis for variceal bleed. Patients with decompensated cirrhosis with bleeding tendency obviously had increased risk of bleeding from banding ulceration. In this study, two-thirds of patients in the VBL arm had decompensated liver reserve and almost 40% of them were classified as having Child C cirrhosis. Therefore, the risk of bleeding from banding ulceration could be higher in this population, which could contribute to the higher rate of first variceal bleed in the VBL arm. Therefore, whether patients with Child C classification, severe coagulopathy, or bleeding tendency could benefit from VBL for primary prophylaxis needs further studies to confirm.
Third, variceal eradication is an important endpoint to prevent variceal bleeding. This study adopted short intervals of 2 weeks between banding sessions, according to the guidelines suggested by the American Association for the Study of Liver Diseases.4 A previous report showed that longer interbanding interval (more than 3 weeks) had lower risk of rebleeding in secondary prophylaxis.5 Therefore, whether short intervals between banding sessions could accelerate eradication of esophageal varices or contradictorily increased the risk of bleeding remained unclear.
In summary, carvedilol and VBL are alternative options for primary prophylaxis of esophageal varices. Defining the optimum banding protocol and benefit-to-risk ratio in patients classified as having Child C cirrhosis are needed to improve the efficacy of VBL.