Anticoagulation improves survival in patients with cirrhosis and portal vein thrombosis: The IMPORTAL competing-risk meta-analysis

•Anticoagulation appears to reduce all-cause mortality in patients with cirrhosis and portal vein thrombosis.•The beneficial effect of anticoagulation on survival is independent of thrombosis severity and recanalization.•Anticoagulation increases the rate of portal vein recanalization at the expense of increased non-portal hypertension-related bleeding. Background & AimsPrevious meta-analyses demonstrated the safety and efficacy of anticoagulation in the recanalization of portal vein thrombosis in patients with cirrhosis. Whether this benefit translates into improved survival is unknown. We conducted an individual patient data (IPD) meta-analysis to assess the effect of anticoagulation on all-cause mortality in patients with cirrhosis and portal vein thrombosis.MethodsIn this IPD meta-analysis, we selected studies comparing anticoagulation vs. no treatment in patients with cirrhosis and portal vein thrombosis from PubMed, Embase, and Cochrane databases (until June 2020) (PROSPERO no.: CRD42020140026). IPD were subsequently requested from authors. The primary outcome – the effect of anticoagulation on all-cause mortality – was assessed by a one-step meta-analysis based on a competing-risk model with liver transplantation as the competing event. The model was adjusted for clinically relevant confounders. A multilevel mixed-effects logistic regression model was used to determine the effect of anticoagulation on recanalization.ResultsIndividual data on 500 patients from five studies were included; 205 (41%) received anticoagulation and 295 did not. Anticoagulation reduced all-cause mortality (adjusted subdistribution hazard ratio 0.59; 95% CI 0.49–0.70), independently of thrombosis severity and recanalization. The effect of anticoagulation on all-cause mortality was consistent with a reduction in liver-related mortality. The recanalization rate was higher in the anticoagulation arm (adjusted odds ratio 3.45; 95% CI 2.22–5.36). The non-portal-hypertension-related bleeding rate was significantly greater in the anticoagulation group.ConclusionsAnticoagulation reduces all-cause mortality in patients with cirrhosis and portal vein thrombosis independently of recanalization, but at the expense of increasing non-portal hypertension-related bleeding.PROSPERO registration numberCRD42020140026.Impact and implicationsAnticoagulation is effective in promoting recanalization of portal vein thrombosis in patients with cirrhosis, but whether this benefit translates into improved survival is controversial. Our individual patient data meta-analysis based on a competing-risk model with liver transplantation as the competing event shows that anticoagulation reduces all-cause mortality in patients with cirrhosis and portal vein thrombosis independently of recanalization. According to our findings, portal vein thrombosis may identify a group of patients with cirrhosis that benefit from long-term anticoagulation. Previous meta-analyses demonstrated the safety and efficacy of anticoagulation in the recanalization of portal vein thrombosis in patients with cirrhosis. Whether this benefit translates into improved survival is unknown. We conducted an individual patient data (IPD) meta-analysis to assess the effect of anticoagulation on all-cause mortality in patients with cirrhosis and portal vein thrombosis. In this IPD meta-analysis, we selected studies comparing anticoagulation vs. no treatment in patients with cirrhosis and portal vein thrombosis from PubMed, Embase, and Cochrane databases (until June 2020) (PROSPERO no.: CRD42020140026). IPD were subsequently requested from authors. The primary outcome – the effect of anticoagulation on all-cause mortality – was assessed by a one-step meta-analysis based on a competing-risk model with liver transplantation as the competing event. The model was adjusted for clinically relevant confounders. A multilevel mixed-effects logistic regression model was used to determine the effect of anticoagulation on recanalization. Individual data on 500 patients from five studies were included; 205 (41%) received anticoagulation and 295 did not. Anticoagulation reduced all-cause mortality (adjusted subdistribution hazard ratio 0.59; 95% CI 0.49–0.70), independently of thrombosis severity and recanalization. The effect of anticoagulation on all-cause mortality was consistent with a reduction in liver-related mortality. The recanalization rate was higher in the anticoagulation arm (adjusted odds ratio 3.45; 95% CI 2.22–5.36). The non-portal-hypertension-related bleeding rate was significantly greater in the anticoagulation group. Anticoagulation reduces all-cause mortality in patients with cirrhosis and portal vein thrombosis independently of recanalization, but at the expense of increasing non-portal hypertension-related bleeding.