Venous thromboembolic events (VTEs) are significant complications in patients with systemic malignancies. Thrombosis risk is poorly defined for patients with brain metastasis, and available risk calculation scores are not validated for these patients.
We identified 811 patients with brain metastasis followed at our institution and reviewed electronic charts retrospectively for the occurrence of VTEs, along with candidate risk factors. Risk factors were tested in univariate and multivariate analyses and finally integrated in a score model for risk estimation. An independent cohort of 346 patients with brain metastasis was available for validation.
VTEs were documented in 97 of 811 patients (12.0%). Primary tumours with high thrombogenicity (p = 0.02, hazard ratio 1.7, 95% confidence interval (CI) = 1.1-2.8), dexamethasone (p = 0.011, hazard ratio 2.27, 95% CI = 1.5-4.5), chemotherapy (p = 0.005, hazard ratio 3.4, 95% CI = 1.6-7.5), body mass index > 35 kg/m (p = 0.002, hazard ratio 3.4, 95% CI = 1.6-7.5) and immobilisation (p = 0.003, hazard ratio 2.4, 95% CI = 1.3-4.3) were confirmed to be independently associated with VTEs. We derived a score model for VTE risk estimation, the thrombogenic primary, immobilization, chemotherapy, obesity, steroid (PICOS) score (0-7 points). Receiver-operating characteristic curve analysis demonstrated its prognostic accuracy (area under the curve [AUC] = 0.71, 95% CI = 0.64-0.77), and its value for the evaluation of VTE risk was superior to that of other scores such as the Khorana (AUC = 0.51) or CONKO (AUC = 0.52) scores. The potential value of the PICOS score was confirmed in the validation cohort (AUC = 0.72, 95% CI = 0.63-0.82).
The PICOS score may become a helpful tool for the identification of patients with brain metastasis at high risk for VTEs and for stratification in controlled studies.

Copyright © 2020 Elsevier Ltd. All rights reserved.