BACKGROUND & AIMS: The outcome of portal vein thrombosis in relation to associated prothrombotic states has not been evaluated. We assessed current outcome and predictors of bleeding and thrombotic events in a cohort of 136 adults with nonmalignant, noncirrhotic portal vein thrombosis, of whom 84 received anticoagulant therapy. METHODS: Multivariate Cox model analysis for event-free survival and analysis taking into account multiple events were used. RESULTS: Median follow-up was 46 months. The incidence rate of gastrointestinal bleeding was 12.5 (95% confidence interval [CI], 10-15) per 100 patient-years. Large varices were an independent predictor for bleeding. Anticoagulant therapy did not increase the risk or the severity of bleeding. The incidence rate of thrombotic events was 5.5 (95% CI, 3.8-7.2) per 100 patient-years. Underlying prothrombotic state and absence of anticoagulant therapy were independent predictors for thrombosis. In patients with underlying prothrombotic state, the incidence rates of splanchnic venous infarction were 0.82 and 5.2 per 100 patient-years in periods with and without anticoagulant therapy, respectively (P = 0.01). Two nonanticoagulated patients died of bleeding and thrombosis, respectively. CONCLUSIONS: In patients with portal vein thrombosis, the risk of thrombosis is currently as clinically significant as the risk of bleeding. The benefit-risk ratio favors anticoagulant therapy.
BACKGROUND & AIMS: The outcome of portal vein thrombosis in relation to associated prothrombotic states has not been evaluated. We assessed current outcome and predictors of bleeding and thrombotic events in a cohort of 136 adults with nonmalignant, noncirrhotic portal vein thrombosis, of whom 84 received anticoagulant therapy. METHODS: Multivariate Cox model analysis for event-free survival and analysis taking into account multiple events were used. RESULTS: Median follow-up was 46 months. The incidence rate of gastrointestinal bleeding was 12.5 (95% confidence interval [CI], 10-15) per 100 patient-years. Large varices were an independent predictor for bleeding. Anticoagulant therapy did not increase the risk or the severity of bleeding. The incidence rate of thrombotic events was 5.5 (95% CI, 3.8-7.2) per 100 patient-years. Underlying prothrombotic state and absence of anticoagulant therapy were independent predictors for thrombosis. In patients with underlying prothrombotic state, the incidence rates of splanchnic venous infarction were 0.82 and 5.2 per 100 patient-years in periods with and without anticoagulant therapy, respectively (P = 0.01). Two nonanticoagulated patients died of bleeding and thrombosis, respectively. CONCLUSIONS: In patients with portal vein thrombosis, the risk of thrombosis is currently as clinically significant as the risk of bleeding. The benefit-risk ratio favors anticoagulant therapy.