Characteristics, anticoagulation, and outcomes of portal vein thrombosis after intra-abdominal surgery

Background: Intra-abdominal surgery is a cause of portal vein thrombosis; however, postsurgical portal vein thrombosis has not been extensively described. Methods: This is a retrospective study of 107 patients with postsurgical portal vein thrombosis followed for a median 25 months (interquartile range 11–51). Outcomes were complete radiographic resolution of portal vein thrombosis and development of clinical portal hypertension. Results: Surgeries associated with portal vein thrombosis included colectomy (n = 42), bariatric surgery (n = 25), and splenectomy (n = 11). Presentations were nonspecific, typically characterized by abdominal pain. Sixty-three patients (59%) achieved complete radiographic resolution. On univariable analysis, provoking surgery, occlusivity of portal vein thrombosis, and anticoagulant used were associated with complete radiographic resolution. Colectomy was associated with a complete radiographic resolution rate of 30/42 (71%), bariatric 10/25 (40%), splenectomy 2/11 (18%), and other 21/29 (72%), (log rank P = .0033). Nonocclusive thrombus was associated with a complete radiographic resolution rate of 44/62 (71%), occlusive thrombus 19/45 (42%), (log rank P = .0101). Direct oral anticoagulants were associated with a complete radiographic resolution rate of 27/35 (77%), enoxaparin 20/29 (69%), warfarin 14/31 (45%), and no anticoagulant 2/12 (17%), (log rank P = .0002). On multivariable analysis, only anticoagulant choice was significantly associated with complete radiographic resolution. Using direct oral anticoagulants as reference, no anticoagulant yielded an adjusted hazard ratio of 0.10 for complete radiographic resolution (95% confidence interval 0.023–0.44), warfarin 0.40 (95% confidence interval 0.20–0.78), and enoxaparin 0.64 (95% confidence interval 0.49–1.60). Failure to achieve complete radiographic resolution was associated with greater risk of future clinical portal hypertension. Twenty-three patients (21%) went on to develop clinical portal hypertension; 20 who failed to achieve complete radiographic resolution (45%), and only 3 who achieved complete radiographic resolution (5%), (log rank P < .0001). Conclusion: The natural history of postsurgical portal vein thrombosis is variable and influenced by type of surgery, degree of occlusion, and, most notably, type of anticoagulant used. Failure to recanalize the portal vein carries considerable risk of future clinical portal hypertension.