Prior Decompensation Identifies Patients at High Mortality Risk Despite Standard Therapy After Variceal Hemorrhage: An Individual Patient Data Meta-Analysis.

BACKGROUND & AIMS: In patients with variceal hemorrhage, nonselective beta-blockers plus endoscopic variceal ligation are recommended to prevent rebleeding when pre-emptive transjugular intrahepatic portosystemic shunt is not indicated or placed. The aim of this study was to determine whether decompensating events occurring before variceal hemorrhage are associated with worse outcomes in patients treated with nonselective beta-blockers plus endoscopic variceal ligation to prevent recurrent variceal hemorrhage. METHODS: This was a systematic review and individual participant data meta-analysis of studies including patients with cirrhosis and variceal hemorrhage not eligible for pre-emptive transjugular intrahepatic portosystemic shunt in patients with Child-Pugh class A-B and receiving nonselective beta-blockers plus endoscopic variceal ligation to prevent recurrent bleeding. All-cause mortality and new or worsening decompensation were analyzed using cause-specific Cox models and random-effects individual participant data meta-analysis. RESULTS: Thirteen studies (5 randomized controlled trials, 8 observational) contributed 1659 patients (Child-Pugh class A, 805; class B, 853); 606 had had prior decompensation, 121 had ascites, 154 had encephalopathy (alone or combined), 106 had ascites plus variceal hemorrhage, and 225 had variceal hemorrhage alone. The 2-year mortality was 25.1% with prior decompensation and 13.5% without (P < .0001): specific adjusted hazard ratios were 1.8 (95% confidence interval, 1.2-2.6) for prior ascites, 1.7 (95% confidence interval, 1.3-2.3) for encephalopathy, 0.83 (95% confidence interval, 0.6-1.2) for variceal hemorrhage, and 1.4 (95% confidence interval, 0.92-2.3) for variceal hemorrhage + ascites. Individual patient data meta-analysis showed a pooled adjusted hazard ratio for death of 1.4 (95% confidence interval, 1.1-1.7) with prior decompensation. New or worsening decompensation was also significantly higher with any prior decompensation: pooled adjusted hazard ratio, 1.7 (95% confidence interval, 1.15-2.44). CONCLUSIONS: Among patients with Child-Pugh class A-B cirrhosis who were not candidates for pre-emptive transjugular intrahepatic portosystemic shunt and were treated with nonselective beta-blockers plus endoscopic variceal ligation after variceal hemorrhage, prior decompensation identified a subgroup at increased risk of mortality in whom the potential benefit of transjugular intrahepatic portosystemic shunt warrants further investigation.