Can Thrombosis Be Predicted in Patients with Gastroparesis?

Abstract

Purpose: Recent reports have shown that nearly 90% of patients with gastroparesis have some sort of detectable hypercoagulable state. Also, a recent series of gastroparesis patients has noted an unusually high incidence of deep venous thrombosis, pulmonary embolism and/or line related thrombosis. The purpose of our study was to retrospectively examine a group of patients with gastroparesis to identify risk factors for hypercoagulablility. Methods: We retrospectively examined a series of 191 patients (30 were male, 161 were female; 161 were white, 28 African American, and 2 were Hispanic with the average age of 44 years) with gastroparesis. We determined if these patients had history of DVT, PE and/or line-related thrombosis (CLOT, N = 50) or no history of thrombosis (NOCLOT, N = 141). Thirteen factors were compiled on these patients: activated Protein C resistance, fibrinogen levels, Protein C and S levels, antiphospholipid IgA, IgM and IgG levels, anti-b2-glycoprotein IgA, IgM, and IgG levels, Factor VIII levels, Antithrombin III levels, as well as the presence of the C677t methyl-tetrahydrofolate reductase mutation. We used Fisher's discriminant analysis to determine a linear function of these 13 variables that classified the 191 patients into either the CLOT or NOCLOT group. Because true clotting history was known, we were able to assess the ability of the function to classify each patient as if that patient's clotting status was unknown. Results: Of our 191 patients, 50 of whom had history of abnormal clotting and 141 who did not, the derived function correctly classified 74% of the patients into either the CLOT or NOCLOT group. The classification was correct for 74% of the CLOT group (sensitivity) and 74% of the NOCLOT group (specificity). Conclusion: Our results suggest that clotting risk can be accurately predicted with 74% certainty in these patients on the basis of his/her outcomes on the 13 factors listed above. Clinically, this could be used to risk stratify these patients, the majority of whom have some hypercoagulable propensity, for life-threatening thrombosis.