Whole Blood Gene Expression Profiles Distinguish Patients with Single Prior Venous Thromboembolism from Patients with Multiple Prior Events.

The reccurence rate for venous thromboembolic events (VTE) following completion of a standard cause of anticoagulant therapy for an initial episode of deep vein thrombosis (DVT) or pulmonary embolism (PE) is as high as 30%. Despite the identification of genetic risk factors and several co-morbid conditions such as surgery, malignancy, trauma, and immobilization, all of which increase the risk for VTE, it remains unclear why VTE recurs in certain patients following an initial event. We have previously used a gene expression strategy to identify metagenes that distinguish patients with VTE and the antiphospholipid syndrome (APS), a disorder characterized by a high rate of recurrent thrombosis, from patients with VTE who did not have APS. In this study, we sought to extend this observation by comparing gene expression profiles in patients with a single prior VTE to patients with multiple prior thromboembolic events. We enrolled patients followed in the Duke Anticoagulation Clinic who were on chronic vitamin K-antagonist therapy for VTE. Medical records were reviewed to document each prior thromboembolic event, and patients were stratified according to whether they had had a single prior event or more than one prior event. The presence of a transient risk factor (e.g., post-operative state), or an inherited risk factor such as factor V Leiden, did not exclude the patient from the study. Patients with unclear documentation for their VTE, on an anticoagulant other than warfarin, or with known APS and/or malignancy were excluded from the study. Blood samples were collected in PAXgene blood RNA tubes, processed, and stored at −80°C until further analysis. Starting with 10 mL of whole blood, between 1.6 to 70.5 mcg of total RNA was isolated and checked for quality using the Agilent RNA Bioanalyzer. The RNA was amplified, labeled and hybridized to the oligonucleotide arrays printed at the Duke Microarray Facility using the Operon Human Genome Oligo Set Version 4.0 (Operon, Huntsville, AL). This report summarizes an initial screen of 19 patients -- 9 individuals with single and 10 individuals with multiple VTE. We first used PAM, a shrunken centroid analysis, to estimate the number of genes required to separate these two phenotypic groups and found that a total of 20–29 genes was optimal for the best prediction for single vs. multiple events. We further used a binary regression model to identify a 25-gene predictor with the ability to distinguish the single vs. multiple phenotypes with an accuracy of 70%. The genes comprising this 25-gene metagene include genes involved in regulating cellular apoptosis, transcription programs in responses to extracellular stimuli. There was no overlap with the genes comprising the metagene distinguishing APS patients from patients with non-APS VTE. We are extending our initial training set to include a total of 20 individuals with each clinical phenotype, and have also recruited a similar number of patients in each group for the validation analyses. These studies suggest the utility of genomic approach to identify VTE patients who are likely to benefit from long-term anti-coagulation therapy as well as provide insights into the differing pathogenesis of single vs. multiple VTE.