Biomarkers and Clinical Prediction Rules in the Diagnosis of Suspected Deep Vein Thrombosis: A Comparison of Cancer and Non-Cancer Patients

Abstract

Abstract Introduction: Despite efforts toward prevention, thousands of patients suffer from venous thromboembolic disease (VTE), consisting of deep vein thrombosis (DVT) and pulmonary embolism (PE), each year; this results in significant morbidity and mortality. Clinical prediction rules, such as the Wells score, in conjunction with biomarkers like the D-dimer, are suggested by national guidelines to aide in evaluating for DVT. Substantial interest exists in developing improved diagnostic testing strategies and algorithms to limit the need for more costly, time consuming, and/or invasive testing. Novel biomarkers, like the soluble P selectin (sP-Sel), a cell adhesion molecule that functions to mediate thrombus formation and amplification, have shown promise in this area. While cancer patients suffer from a high burden of VTE, the D-dimer and Wells score are less helpful in this group. This is due to non-specific D-dimer elevations in these patients, and cancer being part of the Wells risk stratification schema. We sought to identify a more specific combination of biomarkers and Wells score, for the diagnosis of DVT that would apply to cancer patients as well. Methods: We previously performed a prospective cohort study of adults presenting with symptoms of DVT afflicting the upper or lower extremities (Vandy et al. J Vasc Surg Venous Lymphat Disord., 2013). Patients were enrolled from December 2008 to July 2013. Those with isolated calf DVT, superficial venous thrombosis, pregnant or breastfeeding, on therapeutic anticoagulation, or with symptoms of simultaneous upper and lower DVT were excluded. After informed consent was obtained, clinical characteristics and biomarkers (D-dimer, C-reactive protein (CRP), Von Willebrand Factor activity (VWF), and sP-Sel) were collected, and duplex ultrasonography was performed to determine if DVT was present. In this subset analysis, cancer patients were compared to non-cancer patients with regards to thrombotic risk factors, biomarker values, and to assess the test characteristics of various combinations of biomarkers and Wells score. Results: A total of 373 patients were enrolled, 151 (40%) in the cancer group, compared to 222 (60%) in the non-cancer group. Cancer patients were more likely to have a DVT on confirmatory testing (58.9% vs. 43.2%). Cancer patients tended to be older, male, have a lower BMI, more recent acute illness, and more central lines, relative to the non-cancer group. Non-cancer patients were more likely to have other potential risk factors for thrombosis, i.e. recent surgery, oral contraceptive use, and a family history of thrombosis. Biomarker values for CRP and sP-Sel were similar, however VWF and D-dimer values were significantly higher in the cancer patients (see table). D-Dimer > 500 with Wells score ≥ 2 was less specific for the diagnosis of DVT among cancer patients compared to non-cancer patients (p=0.003). However, D-Dimer ≥ 500 with sP-Sel ≥ 90 showed not only high specificity, but that specificity was not different between the cancer and non-cancer populations (p=0.88). sP-Sel ≥ 90 and Wells ≥ 2 had similar performance characteristics in both groups as well (p=0.54 for specificity, p=0.14 for positive predictive value (PPV)). Conclusion: Co-morbid risk factors for thrombosis among cancer patients included an older age, male gender, and an increased use of central lines. This group was less likely to have other potential thrombotic risk factors like family history or recent surgery. Our results concur with the finding that the D-dimer, combined with a clinical prediction rule, is not as helpful for DVT in cancer patients. Moreover, this study further supports sP-Sel as a specific test for DVT, that can be combined with clinical information (Wells score) or other laboratory data (D-dimer) to reflect the presence of DVT and potentially rule in clot; the test seems equally useful for both cancer and non-cancer populations. This could potentially be used to guide initial patient management in scenarios where imaging or further testing is not immediately available, such as in an outpatient clinic or rural area. Further exploration of the optimal strategy for utilizing biomarkers and clinical prediction rules in the diagnosis of DVT in cancer is needed. Table Table. Disclosures Sood: Bayer: Research Funding.