Factors related to early mortality after transjugular intrahepatic portosystemic shunt for failed endoscopic therapy in acute variceal bleeding

Background: Uncontrolled variceal haemorrhage is the main indication for transjugular intrahepatic portosystemic shunt. However, mortality is 50% for this high-risk group. We have evaluated clinical and laboratory variables prior to transjugular intrahepatic portosystemic shunt in order to establish predictors of mortality, validated prospectively. Method: Over a 4-year period, 367 patients were admitted with variceal bleeding. In 54 patient endoscopic therapy for acute variceal bleeding failed and they had emergency transjugular intrahepatic portosystemic shunt. Failure of therapy was defined as continued bleeding after 2 endoscopy sessions (n=39) or vasoconstrictor-resistant bleeding from gastric/ectopic varices (n=15). Thirty-three variables were analysed from data available immediately prior to transjugular intrahepatic portosystemic shunt. Results: Twenty-six patients died within 6 weeks. In a multivariate analysis, 6 factors had independent prognostic value: moderate/severe ascites, requirement for ventilation, white cell blood count (WBC), platelet count (PLT), partial thromboplastin time with kaolin (PTTK) and creatinine. A prognostic index (PI) score was derived, in which presence of moderate/severe ascites, or need for ventilation, scored 1: PI=1.54 (Ascites)+1.27 (Ventilation)+1.38 Ln (WBC)+2.48 In (PTTK)+1.55 Ln (Creat)-1.05 Ln (PLT). Using this equation, 42% (n=10) of deaths occurred in the fifth quintile (PI≥18.52), where the mortality was 100%. The score was prospectively validated in a further 31 patients, giving 100% positive predictive value. Eleven further patients died, including all seven with a PI> 18.5. No survivors had a PI> 18.3. Conclusions: Despite immediate control of bleeding by transjugular intrahepatic portosystemic shunt, patients with uncontrolled variceal haemorrhage have a high mortality, particularly when associated with markers of advanced liver disease, sepsis and multiorgan failure. The use of transjugular intrahepatic portosystemic shunt is probably not justified in this subgroup. Our prognostic index can help identify such patients, and, if validated elsewhere, will help in deciding when to use transjugular intrahepatic portosystemic shunt.