Portal vein thrombosis in patients with cirrhosis.

BACKGROUND AND AIMS

Portal vein thrombosis (PVT) is common among patients with cirrhosis, but the independent impact on outcomes and management is uncertain. We aimed to determine whether the development of PVT is independently associated with mortality, bleeding, and hospitalization and whether anticoagulation improves these outcomes.

METHODS

Patients with cirrhosis and PVT were identified using billing codes from a large health system between 2016 and 2023 and compared to matched control cirrhosis patients without PVT. Among the cohort with PVT, those who received anticoagulation were compared to those who did not. Outcomes included mortality, gastrointestinal bleeding, and hospitalization. Adjustment for confounding was performed using propensity score analysis.

RESULTS

Among 48,596 patients with cirrhosis, 1332 formed the PVT cohort and 3440 formed the non-PVT matched cohort. On adjusted analysis, patients with PVT had higher mortality (hazard ratio [HR] 1.33,  < 0.001), bleeding (HR 1.41,  < 0.001), and hospitalization (incidence rate ratio [IRR] 1.25,  < 0.001). Among the 1161 PVT patients meeting inclusion criteria, 768 received no anticoagulation, 309 received anticoagulation for ≤90 days, and 84 received anticoagulation for >90 days. In the unadjusted analysis, anticoagulation was associated with lower mortality (log-rank  = 0.004), with a dose-response relationship. After propensity score adjustment, the association between anticoagulation and lower mortality persisted but no longer reached statistical significance (HR 0.8,  = 0.075). However, anticoagulation remained associated with higher bleeding (HR 1.67,  = 0.004) and hospitalization (IRR 1.43,  < 0.001).

CONCLUSIONS

Among patients with cirrhosis, PVT is independently associated with a higher risk of mortality, bleeding, and hospitalization. Anticoagulation may improve overall survival but is associated with a higher risk of bleeding and hospitalization.