1. Browse Program
2. Poster Session
3. Virtual Posters
4. Presentation

Presentation Details

Diagnostic Performance of the Khorana, ONKOTEV, and PROTECHT Scores for Predicting Venous Thromboembolism in Ovarian Cancer Emilie Matheson1, Wesley Chorney1, Rahul Singh1, Susan O'Shea1, 2. 1University College Cork, Cork, Ireland.2Bons Secours Hospital, Cork, Ireland

Abstract

Background The risk of venous thromboembolism (VTE) is significantly higher in cancer patients than the general population. The Khorana score is the standard tool to assess VTE risk in the cancer however, its accuracy in predicting VTE for ovarian cancer patients receiving neoadjuvant chemotherapy is low. The PROTECHT and ONKOTEV scores have been applied to solid tumours and performed well for lung and pancreatic cancer, however, there ability to assess risk of VTE in ovarian cancer is not well understood. Our aim is to evaluate the Khorana, PROTECHT, and ONKOTEV scores predicting VTE risk in patients with ovarian cancer. Methods We conducted a retrospective cohort analysis using the Medical Information Mart for Intensive Care (MIMIC-IV) database. Women with a diagnosis of ovarian cancer (ICD-9/10) aged  18, who had not used anticoagulants within one month of the date they initiated chemotherapy, immunotherapy, targeted therapy or hormonal therapy were included. We extracted relevant variables to compute the Khorana, ONKOTEV, and PROTECHT scores. The index date used was the start of cancer therapy and the outcome, venous thromboembolism (VTE), was assessed within 6 months of the index date. The Khorana, ONKOTEV and PROTECHT scores were computed. Patients were classified as high or low risk based on our prespecified threshold. Thresholds were determined by maximizing the harmonic mean of sensitivity and specificity. The thresholds were 2, 1, and 4 for the Khorana score, ONKOTEV score, and PROTECHT score, respectively. For each score, we calculated the accuracy, sensitivity, specificity, negative pred value, positive predictive value, and area under the receiver operator characteristic (AUC-ROC) curve. Results A total of 85 patients were included, of whom 15.29% developed VTE. The diagnostic performance of the three VTE risk models varied. The Khorana score had a sensitivity, specificity, and accuracy of 69.2%, 76.4%, and 75.3%, respectively. The PPV was 34.6% and NPV was 93.2% (AUC 0.763). The ONKOTEV demonstrated values 61.5%, 80.5%, and 77.6% with a PPV of 36.4% and NPV of 92.1% (AUC 0.576). The PROTECHT score showed values of 53.8%, 95.8%, 89% with a PPV of 70% and NPV of 92% (AUC 0.733). The Khorana score demonstrated the highest sensitivity and negative predictive value, supporting its use as a tool for ruling out patients at low risk of VTE. The ONKOTEV score performed the poorest with limited discriminative ability (AUC=0.57). The PROTECHT score had the highest specificity and positive predictive value, indicating a strong ability to confirm high-risk patients once identified. Overall, the Khorana score performed best (with respect to AUC) and was also best at ruling out patients at risk of a VTE. The PROTECHT score performed best at confirming correctly identifying patients who went on to develop a VTE. Conclusion VTE is a significant challenge for patients with cancer, specifically high-risk cancers such as ovarian. These findings suggest a potential two step clinical approach. The Khorana score may be useful for initial risk exclusion, while the PROTECHT score may aid in confirming high risk patients who warrant closer monitoring or prophylaxis.

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.