Announcement
Thank you for attending THSNA 2026. The virtual meeting is now closed.
Thank you for attending THSNA 2026. The virtual meeting is now closed.
Presentation Details
| Prediction of Anticoagulant-related Bleeding in Patients with Cancer on Primary Thromboprophylaxis Kristen M Sanfilippo1, 2, Yan Yan1, Brian F Gage1. 1Washington University in St.Louis School of Medicine, Saint Louis, MO, USA.2St.Louis Veterans Administration Medical Center, Saint Louis, MO, USA |
Abstract
Introduction: Low-dose direct oral anticoagulant (DOAC) therapy reduces the risk of venous thromboembolism (VTE) in patients with cancer at high-risk of VTE based on validated prediction models. However, patients with cancer have double the risk of anticoagulant-related major bleeding (MB). As such, there has been limited implementation of prescription of primary thromboprophylaxis in clinical practice. This apprehension has been shown to be, in part, due to concerns about the risk of bleeding. Thus, we aimed to determine if available risk prediction scores could discriminate risk of anticoagulant-related bleeding in cancer patients prescribed low-dose DOAC while enrolled in the AVERT and CASSINI randomized trials. Methods: Using the modified intent-to-treat populations in the AVERT and CASSINI randomized trial, we conducted a post-hoc analysis to assess the discrimination of five risk prediction scores developed for anticoagulant-related bleeding. We selected scores based on availability of candidate variables, modifying when necessary: ATRIA, Kuijer et al, ORBIT, RIETE, and VTE Bleed scores. The outcome of interest was development of a MB or clinically relevant non-major bleed (CRNMB). Each bleeding risk score was applied to the cohort with clinical point assignments as in the original papers. Using a Fine and Gray competing risk model, we tested the association between each score and development of first bleed following anticoagulant prescription within 180 days of anticoagulant start. The discrimination of each model was evaluated using time-dependent IPCW concordance Index (c-index) as well as by using the area under the curve (AUC) as suggested by Blanche et al. All analyses were conducted using R (4.2.3) and SAS (9.4) statistical software. Results: The total cohort size was 693 patients, 288 prescribed low-dose apixaban 2.5mg twice daily in the AVERT trial and 405 prescribed low-dose rivaroxaban 10mg daily in the CASSINI trial. The median age was 61 years in AVERT and 63 years in CASSINI. Frequent cancers included: gynecologic (25.8%), lymphoma (25.3%), and pancreatic cancer (13.6%) in AVERT and pancreatic (32.4%), upper gastrointestinal (21.2%), and lung cancer (14.8%) in CASSINI. Adherence rates of 83.6% and 98.3% were noted with apixaban and rivaroxaban, respectively. The primary outcome occurred in 19 (6.6%) patients receiving apixaban and 28 (6.9%) patients receiving rivaroxaban. The c-index at 180 days was 0.60 for ATRIA, 0.53 for Kuijer et al., 0.63 for ORBIT, 0.60 for RIETE, and 0.52 for VTE BLEED. Using AUC, results were: 0.63 for ATRIA, 0.54 for Kuijer et al., 0.66 for ORBIT, 0.63 for RIETE, and 0.56 for VTE BLEED. Conclusions: Of the scores analyzed, the ORBIT score performed best with moderate discrimination. The ability to discriminate between those at high-risk of anticoagulant-related bleeding in patients with cancer who are candidates for primary thromboprophylaxis could inform providers and patients to make informed decisions about primary thromboprophylaxis based on the competing risk of VTE versus risk of anticoagulant-related bleeding.
No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, including photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the author.
No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, including photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the author.