Anticoagulation Practice and Risk of Portal Vein Thrombosis Following Pancreaticoduodenectomy or Total Pancreatectomy with Venous Resection

Objective: Assess anticoagulation practice and portal vein thrombosis (PVT) risk following pancreaticoduodenectomy (PD) or total pancreatectomy (TP) with venous resection (VR). Background: Retrospective studies suggest an increased risk of PVT following PD/TP with VR. However, anticoagulation practice is variable and its efficacy at preventing PVT is unknown. Methods: This multicentre cohort study (Europe, USA, Mexico, Turkey, Australia, New Zealand) included consecutive patients undergoing PD/TP with VR between 2018-2022. A 1:1 age and sex matched cohort undergoing PD/TP without VR was also collected to assess PVT risk without VR. Results: Among 972 patients who underwent PD/TP with VR, 259 (26.6%) received inpatient therapeutic anticoagulation and 242 (25.0%) were discharged on therapeutic anticoagulation. Thirty-day, 90-day and one-year PVT risk following VR was 5.1%, 7.3%, and 11.6%, versus 1.0%, 1.3% and 2.6% in patients without VR (P<0.001). Predictors of 90-day PVT included prior history of venous thromboembolism (odds ratio [OR] 2.67), VR type (OR 2.29, 6.28, 6.90 and 23.75 for type 1-4 VR, P<0.001) and graft material (OR 0.78, 0.94, 5.28, 4.90 and 5.99 for peritoneal, autologous vein, cadaveric vein, bovine and synthetic grafts, P<0.001). Postoperative therapeutic anticoagulation reduced 30-day PVT risk (OR 0.06, P<0.001), but not 90-day (OR 0.06, P=0.075) or >90-day PVT risk (OR 1.23, P=0.466). The strongest predictor of >90-day PVT was cancer recurrence (OR 3.96, P<0.001). Conclusions: VR increases PVT risk following PD/TP, with technical factors influencing <90-day PVT and cancer-related factors influencing >90-day PVT. The benefits of early postoperative anticoagulation in preventing PVT post-VR remain unclear.