To the Editor:

Thank you for allowing us to reply to a letter by Triantos et al. in this issue of HEPATOLOGY. Compared with rubber band ligation (RBL), the propranolol group in our study had significantly higher rates of (1) first esophageal varix hemorrhage, (2) failure (e.g., first varix bleed and/or severe complications requiring stoppage of therapy), and (3) mortality.1 The direct costs of medical therapy were not significantly different. Liver transplantation (OLT) rates were high in both treatment arms but did not account for the outcomes.2 In our study, portal hypertensive gastropathy or banding ulcers did not cause severe bleeding in any patient. For a discussion of statistical issues, including an overlap of the confidence intervals of our results with other published trials, the reader is referred to our article.1

Several specific issues in Triantos et al.'s letter merit clarification regarding primary prevention of first variceal hemorrhage. First, we agree that prophylactic therapies should be safe and effective. They also should be well tolerated. Propranolol is not very well tolerated, nor easy to use, nor very effective. Schepke et al.3 reported that 16% of patients had severe side effects, 5% had contraindications to its use, and only 52.8% achieved 25% reduction in resting heart rate as a surrogate for effectiveness. In other pharmacological studies, only about 60% of at-risk patients with cirrhosis on maximally tolerated doses of propranolol had true responses with reduction of their hepatic venous pressure gradients to 12 mm Hg or lower; therefore, 40% remained at high risk for variceal bleeding.4–6 These data do not support the premise that propranolol is very effective. Also, documenting hepatic venous pressure gradient response is very expensive and far exceeds the cost of RBL for varix eradication.7 Our position is that medical therapies that are more effective and safer than nonselective beta-blockers are needed as prophylactic therapies for high-risk patients. Meanwhile, propranolol should not be regarded as significantly superior to RBL with regard to safety, efficacy, tolerability, or cost.

Second, technical differences, study design, and luck may account for the lower complication rates with RBL in our trial1 compared with others,3, 8 and these are important factors to understand. Some differences between our study and those of Schepke et al. and Triantos et al. are: (1) our use of proton pump inhibitors until complete varix obliteration; (2) the increasing time between RBL sessions to reduce complications and increase ulcer healing—4 to 5 weeks in our study1 versus 1 week in the Schepke et al. study3 and 2 weeks in the Triantos et al. study8 (at 4 to 5 weeks banding ulcers were rare in our study); and (3) standardizing RBL techniques and varix obliteration among investigators before initiation of the study, which was part of our trial1, 9 but was not discussed in the other two studies.3, 8

Third, there may be subgroup differences in response to these therapies.1 Our study results are particularly relevant to nonalcoholic patients with cirrhosis but without prior variceal bleeding who are awaiting OLT and have high-risk varices. We recommend that physicians consider RBL as a treatment that is easier to apply, better tolerated, and potentially safer and more effective than propranolol in such patients.


  1. Top of page

Dennis M. Jensen M.D.*, * VA Greater Los Angeles Healthcare System, Los Angeles, CA.