INTEGRATED PREDICTION OF VENOUS THROMBOEMBOLISM IN GLIOMA PATIENTS

Abstract BACKGROUND Venous thromboembolism (VTE) is a debilitating and life-threating condition that is common in patients with cancer, especially glioma. There are no widely-accepted guidelines for antithrombotic prophylaxis in glioma patients, in part because of the risk of iatrogenic intracerebral hemorrhage. Effective VTE prediction models exist for patients with other cancers, but not glioma. Our prior publications suggested a protective role of IDH mutation (IDHmut) against VTE, as well as a positive association between elevated circulating Tissue Factor (TF), a procoagulant secreted by cancers, and VTE in glioma patients. Others have recently found a correlation between podoplanin, a platelet activator, and cancer-associated VTE, but this has not yet been proven in gliomas. Our objective was to develop a multidimensional VTE prediction tool to improve glioma patient care, incorporating clinical, blood-based, histologic, and molecular markers. METHODS The Northwestern University Nervous System Tumor Bank provided preoperative arterial blood, tumor tissue, and clinical-pathologic data from 254 grade II-IV glioma patients over a 4-year interval. Molecular profiling was done by GlioSeq NGS. RESULTS Forty-six out of 254 grade II-IV glioma patients (18%) experienced VTE during their disease course. VTE was associated with worse median overall survival (18 vs. 54 months, P< 0.001). Positive risk factors for VTE included advanced patient age, high body mass index, high glioma grade, prior history of VTE, coronary artery disease, hyperlipidemia, hypertension, and elevated circulating TF activity, among other variables. While tumor expression of podoplanin positively correlated with VTE, circulating podoplanin did not. Among molecular markers, the strongest was IDHmut, being associated with greatly reduced VTE risk (OR=0.28; 95% CI=0.12-0.64, P=0.001). CONCLUSIONS This study identified multiple risk factors for glioma-associated VTE. Using these data, we have developed a web-based multivariable integrated prediction calculator to identify glioma patients who would benefit the most from prophylactic anticoagulation.