Prior History of Thrombosis, Meld, BMI and the Presence of Esophageal Varices Predict Thrombus Progression in Patients with Untreated Splanchnic Vein Thrombosis

Background: Splanchnic vein thrombosis (SVT) occurs in a heterogenous group of patients secondary to a variety of risk factors including acute and chronic liver disease, malignancy, and myeloproliferative neoplasms. There is equipoise on the utility of anticoagulation in many patients with SVT given the perceived risks of bleeding and unclear benefits. We sought to determine which clinical factors predict new or progressive thrombosis in a cohort of patients with SVT.

Methods: We undertook a retrospective cohort study of patients over 18 years of age identified to have an SVT at the Oregon Health & Science University from 01/01/2015 - 12/31/2020, including only patients who were not initially treated with anticoagulation at the time of their initial VTE diagnosis. Relevant clinical variables were selected apriori. The primary study endpoint was imaging-confirmed progression of SVT, development of cavernous transformation, intestinal ischemia, or new venous or arterial thrombosis. Chart demographics, patient history, and relevant lab values at the time of initial SVT were extracted for the analysis. Descriptive analysis, univariate logistic regression, and multivariable logistic regression were performed in STATA version 12.1 and R (R core team 2019).

Results: Seventy-eight patients were included in the analysis (mean age 60 years old, 74% male). The most common SVT in the cohort was isolated portal vein thrombosis (N=60) followed by thrombosis of multiple splanchnic veins (N=14). The most prevalent causes of liver disease in the cohort were viral hepatitis (N=33) alcoholic cirrhosis (N=20) and non-alcoholic steatohepatitis (N=12). 66% of patients had known varices at diagnosis. 33% had thrombosis directly associated with a tumor. The mean platelet count and INR were 105 x 109/L and 1.55 respectively. Twenty-two patients (28%) developed the primary endpoint of thrombus progression. Univariate logistic regression found that prior history of thrombosis (OR 6, P= 0.04) and the presence of varices at diagnosis (OR 4.4, P= 0.02) were associated with progression. We then created a multivariable logistic regression model and observed that total bilirubin (ORadj = 0.34, p = 0.03), MELD score (ORadj = 1.33, p = 0.04), the presence of varices (ORadj = 11.7, 0.03), and BMI (ORadj = 1.14, p = 0.047) were significant predictors of our composite outcome while controlling for age, glomerular filtration rate, INR, and prior history of VTE.

Discussion: In our heterogenous cohort of patients with SVT not treated with anticoagulation, one in four patients developed the composite endpoint of SVT progression, development of cavernous transformation, intestinal ischemia, or new venous or arterial thrombosis. Several common clinical variables appear to be predictive for thrombus progression, suggesting that predictive models may be feasible to determine which patients with SVT are likely to benefit from anticoagulation.