TIPS vs. Endoscopic Therapy to Prevent Variceal Rebleeding

Abstract & Commentary

Synopsis: In a prospective comparison of transjugular-intrahepatic-portosystemic shunt (TIPS) vs. endoscopic treatment plus propranolol for patients with esophageal varices, rates of rebleeding at one and two years were less than half in patients who underwent TIPS.

Source: Roessle M, et al. Lancet 1997;349:1043-1049.

The transjugular-intrahepatic-portosystemic shunt (TIPS) procedure was introduced in 1988 as a less invasive, less morbid alternative to surgical shunts for portal hypertension. The TIPS procedure consists of creating an anastomosis between a hepatic vein and an intrahepatic branch of the portal vein. Although it appears to have a lower operative mortality and similar efficacy to surgical shunts, TIPS has been put into wide clinical use without having been subjected to a randomized control trial vs. the more traditional procedure.

This prospective, randomized study from the University of Freiburg in Germany compared TIPS not with surgical shunt but with endoscopic therapy plus propranolol in a series of patients presenting with bleeding esophageal varices. Of 190 eligible patients, 126 were randomized and underwent TIPS (61 patients) or endoscopic therapy (65 patients; sclerotherapy only in 33, sclerotherapy plus banding ligation in 31, banding ligation alone in 1). The groups were well matched in terms of age, Child-Pugh class, number of previous bleeds, and amount of transfusion.

All 61 TIPS patients were successfully shunted, with a reduction in mean portosystemic pressure gradient from 22 to 8 mmHg. There was no statistically significant difference in length of hospitalization following the initial bleed in the two groups. Interventional revision of the shunt was required in 18 (30%) patients after a mean of 7.6 months; mean follow-up in the TIPS patients was 14 months. The 65 patients randomized to endoscopic therapy underwent 581 procedures (mean, 9.7 per patient) and 331 interventions (mean, 5.3 per patient) during a mean of 13 months’ follow-up.

Recurrent upper gastrointestinal hemorrhage occurred during the follow-up period in 15 TIPS patients and 33 endoscopically treated patients. A total of 56 variceal rebleeds occurred, in nine (15%) TIPS patients and in 29 (45%) endoscopy patients. Rebleeding rates were 15% vs. 41% at one year and 21% vs. 52% at two years in the TIPS and endoscopy patients, respectively (P = 0.001).

During follow-up, 25 (38%) endoscopically-treated patients required 272 units of packed erythrocytes, as compared to 12 (20%) TIPS patients who required 96 units transfusion (P = 0.031). Patients who underwent TIPS had a higher incidence of development of hepatic encephalopathy (36% vs 18% at 24 months, P = 0.011). Eight patients died in each group during the follow-up period.


Bleeding from esophageal varices is one of the most troublesome, dissatisfying conditions the intensivist encounters. It draws heavily on health care resources, is a major use for donated blood and blood products, and all too often has an unfavorable outcome. Mortality and morbidity are high, due in large part to the poor medical condition of many of the patients, but also because all available therapies fall short of the ideal. The efficacy of noninvasive therapies such as octreotide infusion has been difficult to establish, and mechanical intervention with Sengstaken-Blakemore and similar multiple-balloon tubes is cumbersome and often ineffective. Surgical shunts have high morbidity and substantial mortality in the acute setting. For these and other reasons, the TIPS procedure has gained widespread favor in the management of variceal hemorrhage.

This study provides substantial support for the use of TIPS in patients who present with variceal bleeding. It is not, however, without problems in terms of study design and potential generalizability. The patients included in the study may or may not compare to those encountered in the reader’s practice. In general, they were less severely ill than those seen in the ICUs of many teaching hospitals; patients with significant hepatic encephalopathy on admission were excluded, as were those with serum bilirubin levels greater than 5 mg/dL, patients in whom propranolol was considered contraindicated, and those with "bleeding emergency." One-third of the patients considered for the study were not randomized; rebleeding during the observation period included Mallory-Weiss tears, peptic ulcers, and other etiologies that may not have pertained to the disorder being studied.

Nonetheless, this study provides considerable support for TIPS as an alternative for surgical shunt—and perhaps even for endoscopic therapy—in patients with acute variceal bleeding. Although a number of serious complications have been observed following this procedure, in a disease with no wholly satisfactory treatments TIPS may prove to be the least of available evils. Until a randomized control trial of TIPS vs. surgical shunting is available, it would seem reasonable to consider the former as a less morbid alternative to surgery for patients with bleeding varices, one that may also be preferable to endoscopic therapies in terms of rebleeding and the need for medical resources.