Abstract Number: OC 23.1
Meeting: ISTH 2021 Congress
Theme: Venous Thromboembolism » Cancer Associated Thrombosis
Background: Accurate risk assessment cancer associated thrombosis (CAT) can inform optimal patient selection for thromboprophylaxis. Unfortunately, a uniform biomarker does not exist and the current risk scoring methods exhibit modest accuracy. Machine learning can capture rich multi-dimensional patient information and identify patients at increased risk for CAT.
Aims: We aimed to identify a novel set of variables predictive of CAT using machine learning to analyze a comprehensive feature space combining proteomic and clinical features in a well-curated prospective cancer cohort.
Methods: A proximity extension assay was used to quantify 1,161 proteins in 183 gastric and lung cancer patients enrolled in the HYPERCAN Study (ClinicalTrials.gov #NCT02622815). Samples were collected at time of cancer diagnosis and patients were prospectively followed for the development of CAT. Machine learning predictive models were built using plasma protein concentrations and clinical/laboratory information from the accompanying patient metadata (>300 variables). The derived machine learning models were compared to Khorana Score to predict venous thromboembolism (VTE).
Results: Of the total cohort of 183 cancer patients, 103 had gastric cancer and 80 lung cancer, and 58% had metastasis at baseline. Median age was 64 years and 32% were female. Approximately one-third (32%) of patients in the cohort developed a thrombotic event on follow-up. A machine learning driven model identified 10 plasma proteins and 6 clinical features predictive of thrombosis. The algorithm predicted VTE with high accuracy (AUC 0.75+0.04)). By comparison, the Khorana Score was not predictive of VTE (AUC 0.52). (Figure 1).
Receiver Operating Characteristics analysis of machine learning driven model combining 10 plasma proteins and 6 clinical features to predict venous thrombosis in patients with active cancer.
Conclusions: This study supports utilizing machine learning in conjunction with novel biomarker discovery for the creation of a next generation high accuracy CAT predictive modeling. Validation of prediction model is ongoing.
To cite this abstract in AMA style:
Zwicker J, Elango E, Patell R, Marchetti M, Russo L, Verzeroli C, Giaccherini C, Gamba S, Flaumenhaft R, J Tartari C, Vlachos I, Falanga A. Personalized Cancer-associated Thrombosis Risk Assessment: Integration of Plasma Proteomics, Clinical Characteristics, and Machine Learning [abstract]. Res Pract Thromb Haemost. 2021; 5 (Suppl 2). https://abstracts.isth.org/abstract/personalized-cancer-associated-thrombosis-risk-assessment-integration-of-plasma-proteomics-clinical-characteristics-and-machine-learning/. Accessed September 22, 2023.« Back to ISTH 2021 Congress
ISTH Congress Abstracts - https://abstracts.isth.org/abstract/personalized-cancer-associated-thrombosis-risk-assessment-integration-of-plasma-proteomics-clinical-characteristics-and-machine-learning/