Objective: To define the long-term characteristics, prognostic factors, and outcomes of patients undergoing selective splenorenal shunting procedures for portal hypertension-induced recurrent upper gastrointestinal bleeding.
Materials and methods: A retrospective evaluation of a prospectively collected data set.
Results: From June 1971 through May 2005, 507 Warren-Zeppa shunts were performed at a single institution. Indications included: alcoholic cirrhosis, 52.6%; viral cirrhosis, 21.8%; cryptogenic cirrhosis, 8.4%; autoimmune cirrhosis, 5.8%; and other causes, 6.3%. Median survival was 81 months (5-year survival, 58.9%; 10-year survival, 34.4%; 20-year survival, 12.5%). patients with portal vein thrombosis and biliary cirrhosis demonstrated better survival than others (P = 0.03), while patients with alcoholic cirrhosis trended toward worse survival than those with nonalcoholic causes (P = 0.11). Multivariate analysis of preoperative risk factors found body hair loss (hazard ratio, 17.3; P > 0.005), preoperative encephalopathy (hazard ratio, 1.93; P > 0.003), diuretic use (hazard ratio, 1.43; P > 0.003), and age (hazard ratio, 1.02 per year of age; P > 0.051) were independent predictors of poor long-term survival. Multivariate analysis of operative factors demonstrated blood loss <500 mL was predictive of up to a 4-fold improved long-term survival (hazard ratio, 3.95; P < 0.013). Postoperative complications included: recurrent bleeding, 12%; ascites, 17.5%; and encephalopathy, 13.9%. Multivariate analysis of postoperative factors prospectively collected in 130 patients found that alcoholic recidivism (hazard ratio, 2.66; P > 0.001) was the only independent predictor of poor prognosis.
Conclusions: The Warren-Zeppa shunt provides long-term survival and control of bleeding in most patients with portal hypertension. Excellent long-term survival can be obtained in properly selected patients with portal hypertension and relatively spared hepatic function.