Abstracts | HAMO Thieme 65th Annual Meeting of the Society of Thrombosis and Haemostasis Research Datum/Ort: Kongresspräsident: 22.–26. February 2021, simply online Prof. Lorenzo Alberio Lausanne, CH visualize the best recommended model, the D-Dimer LR model, which largely Oral Communication narrowed the gap between model development and clinical application. The web calculator can be found at https://webcalculatorofcancerassociateddvt. Cancer-associated Thrombophilia shinyapps.io/dynnomapp/. In the future, the performance and clinical applica- tion of ML models for cancer-associated DVT might be improved further. OC01-01 Development and validation of machine OC01-02 Modelization the impact of learning predictive models applying for cancer- antithrombotic agents on pancreatic tumoral micro- associated deep vein thrombosis: A 1035-sample environment. retrospective cohort study Authors Tran H1, Amrane R1, Mbemba E1, Sabbah M1, Vandreden P2,1, 1 1 2 1 1 1 Authors Jin S , Qin D , Liang BS , Zhang LC , Wei XX , Wang YJ , Gerotziafas G1 1 3 4 1 1 1 5 Zhuang B , Zhang T , Yang ZP , Cao YW , Jin SL , Yang P , Jiang B , Institutes 1 INSERM, UMR_S 938, Centre de Recherche Saint-Antoine- Team 4 4 1 Rao BQ , Shi HP , Lu Q Cancer Biology and Therapeutics, Group « Cancer-Hemostasis-Angiogenesis Institutes 1 Division of Medical & Surgical Nursing, School of Nursing, Peking », Institut Universitaire de Cancérologie, F-75012, Sorbonne Université, 2 University, Beijing; Department of Biostatistics, School of Public Health, Paris; 2 Clinical Research, Diagnostica Stago, Gennevilliers Peking University, Beijing; 3 Division of Medical & Surgical Nursing, School of DOI 10.1055/s-0041-1728080 Public Health, Peking University, Beijing; 4 Department of Gastrointestinal Objective Interactions between cancer cells and their micro-environment Surgery, Beijing Shijitan Hospital, Capital Medical University/The 9th Clinical with antithrombotic agents is an emerging field of research. In the present Medical College, Peking University, Beijing; 5 Department of Medical study we investigated the impact of apixaban, fondaparinux, enoxaparin and Oncology, Beijing Shijitan Hospital, Capital Medical University/The 9th tinzaparin on the procoagulant properties of pancreatic cancer cells BXPC3. Clinical Medical College, Peking University, Beijing Reciprocally, we investigated the impact of BXPC3 on the potency of these DOI 10.1055/s-0041-1728079 antithrombotic agents. Objective This study aims to develop machine learning(ML) models for Material and Methods BXPC3 (400 cells/ml) were exposed to apixaban (2 cancer-associated deep vein thrombosis (DVT), and compare the perfor- microgram/ml), fondaparinux (2 microgram/ml), enoxaparin, or tinzaparin (2 mance of these models with the currently widely-applied predictive model, anti-Xa IU/ml) for 48h. Then, cells and supernatants were separated and the Khorana Score, with or without using D-Dimer. added in normal platelet poor plasma (PPP) for thrombin generation (TG) Material and Methods We consecutively and retrospectively extracted data experiments. Viability of cancer cells (assessed with the MTT assay), gene of 1035 cancer patients from a tertiary hospital. Both uni-variable analysis expression for TF, VEGF, THSB1 (assessed with RT-qPCR) and expression of TF and the Lasso regression were applied to select the important predictors. protein and activity were also examined. Microparticles (MP) were tested for Model training (training set, 652/725) and hyper-parameter tuning (testing TF with specific ELISA. Residual anti-Xa activity was measured in the superna- set, 73/725) were implemented on 70 % (725/1035) of the data using a ten- tant using specific amidolytic assays. fold cross-validation method. The remaining 30 % (310/1035) data was used Results BXPC3 enhanced TG. Incubation of the BXPC3 cells with all antithrom- to compare the performance with six indicators, the area under the receiver botic agents did not significantly modify their TG capacity. Apixaban resulted operating characteristic curve(AUC), sensitivity, specificity, accuracy, the in significant TF mRNA expression decrease by BXPC3 cells. None of the anti- Brier Score, and the calibration curve, among all five ML models, linear discri- thrombotic agents significantly modified the amount of BXPC3 cells -TF pro- minant analysis (LDA), logistic regression (LR), classification tree (CT), tein. Fondaparinux and enoxaparin significantly decreased VEGF mRNA random forest (RF, an assemble algorithm), and support vector machine expression and apixaban significantly increased the expression of THBS1. The (SVM), and the Khorana Score, with or without using D-Dimer. viability of BXPC3 cells was significantly reduced following exposure to apixa- Results The percentage of cancer-associated DVT in this study was 22.3 % (231/ ban (25 %), fondaparinux (12 %), enoxaparin (14 %) or tinzaparin (11 %). Expo- 1035). The top five important predictors were D-Dimer, age, Charlson Comorbi- sure of BXPC3 to antithrombotic agents did not significantly modify the dity Index (CCI), length of stay (LOS), and previously VTE history. Five ML models release of MP. Apixaban, fondaparinux, enoxaparin and tinzaparin decreased were developed and validated. In validation set, LDA (AUC: 0.756 & 0.773, none- TG induced by the supernatant by 70 %, 30 %, 40 %, 90 % respectively. After D-Dimer model & D-Dimer model, respectively), and LR (AUC: 0.752 & 0.772) exposure to BXPC3 cells the concentration of fondaparinux, enoxaparin and performed best, followed by RF (AUC: 0.638 & 0.660), Khorana Score (AUC: tinzaparin in the supernatant reduced by 27 %, 48 % and 26 % respectively. In 0.604 & 0.642), CT (AUC: 0.604 & 0.638), and SVM (AUC: 0.593 & 0.665). contrast the concentration of apixaban did not significantly change. Conclusion This study developed and validated ML predictive models for Conclusion Antithrombotic agents do not alter cancer cells’ TF expression or cancer-related DVT. The combination with D-Dimer showed improved perfor- procoagulant MP release, but inhibit the procoagulant potency of microenvi- mance of all models. LDA and LR out-performed Khorana Score, but CT, RF, ronment. Nevertheless, a LMWHs and fondaparinux degradation occurs follo- and SVM did not surpass it. A nomogram and a web calculator were used to wing two days of exposure to cancer cells. Antithrombotic agents reduced Hämostaseologie 2021; 41: S1–S61 | © 2021. Thieme. All rights reserved. S1 Abstracts |HAMO Thieme fi ▶Tab 1. Impact of antithrombotic agents on BXPC3 cells, on their Flow cytometry analysis of TF expression showed a signi cantly higher expres- MP, and on the capacity of trigger thrombin generation of their sion on the membrane of MCF7-sh-WISP2 and BT-20. Western blot analysis culture supernatants. Cells were treated with 2 Ul/ml of Fondapari- showed that TF was present in eluted proteins after immunoprecipitation of nux/Apixaban, or 2 micro gram/ml of Tinzaparin/Enoxaparin for cell lysates with anti-TF and in cell lysates. Two different major bands were 48h. mRNA expression were normalized using the 2^(-delta(del- observed at levels of 47 kDa and 25 kDa. In MCF7 a significant 47 KDa band taCq)) method. * p < 0.05 as compare to Control experiment. was observed though its intensity was lower compared to the corresponding band of the lysate from MCF7-sh-WISP2; MDAMB-231; BT-20 cells. Conclusion We demonstrate that procoagulant phenotype of pancreatic and breast cancer, is related to the expression of functional TF. The procoagulant fin- gerprint of breast cancer cells varies in function of the degree of aggressiveness. MCF7-ShWisp2 and BT-20 were the more aggressive cells and present higher procoagulant potential which is correlated with TF expression on the cell membrane. The present experimental model will allow the characterization of the procoagulant fingerprint of cell lines from the same or different histological types of cancer. It will also allow toevaluate the efficiency of antithrombotic agents to downregulate hypercoagulability induced by cancer cells. OC01-04 LMWH or DOACs for cancer associated thrombosis (CAT) in daily clinical practice ? – Insights from the GECAT registry Authors Klamroth R1, Pollich C1, Riess H2, Sinn M3 Institutes 1 Internal medicine, vascular medicine and coagulation disorders, Vivantes Klinikum Friedrichshain, Berlin; 2 Haematology and Oncology, Charite University Hospital, Berlin; 3 Haematology and Oncology, University Hospital Hamburg-Eppendorf, Hamburg DOI 10.1055/s-0041-1728082 Objective National and international guidelines for the diagnosis and treat- ment of venous thromboembolism (VTE) recommend anticoagulation treat- ment for 3 to 6 months with preference for direct oral anticoagulants DOACs and a re-evaluation for resumption depending on the individual risk of every patient. In cancer associated VTE (CAT) however, low molecular weight heparin (LMWH) was the guideline recommended treatment until recently. The prospective German Evaluation of CAT (GECAT) registry investigated the anticoagulation practice within the Berlin region. tumour cells’ viability and impaired mRNA expression of pro- and antiangio- Material and Methods The GECAT registry was set up for Berlin´s two main genic factors. hospital companies Charité-Universitaetsmedizin Berlin and Vivantes (cove- ring about 50 % of the hospital beds in Berlin) to document prospectively in- OC01-03 Aggressiveness of breast cancer cells is hospital newly diagnosed VTE in patients (pts) with active cancer. A follow-up related