This study significantly improves the prediction of cancer-associated venous thromboembolism (VTE) by integrating high-throughput plasma proteomics with machine learning models, and reveals the central role of IL17A in thrombotic inflammation, providing a new paradigm for personalized thrombosis risk assessment and anti-inflammatory intervention strategies in related fields.
Literature Overview
The article titled 'Plasma proteomics improves thrombosis prediction in cancer and implicates a targetable IL-17-driven endothelial activation pathway,' published in Science Translational Medicine, systematically explores how plasma proteomics combined with machine learning can be used to build more accurate models for predicting cancer-associated thrombosis (CAT), and deeply analyzes the mechanistic link between CD200R1 and IL17A. The study not only enhances clinical risk stratification but also uncovers a targetable thrombo-inflammatory regulatory pathway, offering theoretical support for developing non-anticoagulant antithrombotic strategies.Background Knowledge
Cancer-associated thrombosis (CAT) is the second leading cause of death in cancer patients, significantly affecting treatment safety and quality of life. Current clinical reliance on the Khorana score system has limited predictive performance, with a positive predictive value of only about 10%, resulting in an unfavorable benefit-risk ratio for anticoagulant prophylaxis. Although tissue factor and neutrophil extracellular traps (NETs) are known to participate in CAT, stable and reliable biomarkers remain lacking. In recent years, the concept of thrombo-inflammation has emerged, emphasizing the interaction between the immune and coagulation systems. Immune mediators such as IL17A and Th17 cells have been linked to various thrombotic diseases, but their mechanisms in CAT remain unclear. This study's approach lies in using a high-sensitivity Olink platform to perform unbiased screening of 1,105 plasma proteins, integrating machine learning to construct a novel predictive model, and validating the causal roles of key proteins through animal models and Mendelian randomization analysis—thereby overcoming the limitations of traditional coagulation biomarkers and exploring the role of immune regulatory axes in driving CAT.
Research Methods and Experiments
The study adopted a prospective cohort design. The discovery cohort (HYPERCAN study) included 163 newly diagnosed lung and gastric cancer patients, with baseline plasma analyzed using the Olink plasma proteomics platform. Through machine learning-based feature selection, a Bayesian prediction model (TOP model) incorporating 11 proteins and 5 clinical variables was developed. The model was externally validated in an independent cohort (placebo arm of the AVERT trial, n=72), demonstrating superior discriminative ability compared to the Khorana score (c-statistic 0.71 vs. 0.36).
To investigate the mechanistic role of the key protein CD200R1, phenotypic analysis was performed using Cd200r1−/− mice. Results showed significantly elevated thrombin-antithrombin complex (TAT) levels in knockout mice, indicating a prothrombotic state. Further RNA-seq and mouse Olink proteomic analyses revealed significant upregulation of IL17A, along with increased endothelial activation markers (e.g., sE-selectin, sICAM-1). In vitro experiments demonstrated that IL17A directly induces endothelial cells to express adhesion molecules and promotes monocyte adhesion.
To validate the causal role of IL17A, anti-IL17A antibody was administered to Cd200r1−/− mice, successfully reversing the elevated TAT levels, thus proving that IL17A mediates the hypercoagulable state caused by CD200R1 deficiency. Additionally, Mendelian randomization analysis using human genetic data confirmed that genetically predicted low CD200R1 expression is associated with increased risk of venous thrombosis (OR=1.04).Key Conclusions and Perspectives
Research Significance and Prospects
This study marks a paradigm shift from traditional coagulation biomarkers toward systemic analysis of thrombo-inflammatory networks. The development of the TOP model paves the way for precise thromboprophylaxis in clinical practice, with future integration of liquid biopsy and immune protein biomarkers enabling dynamic risk scoring systems.
Mechanistically, the discovery of the CD200R1–IL17A axis provides a new target for non-anticoagulant antithrombotic strategies. Particularly in the context of widespread immune checkpoint inhibitor use, such therapies may break immune tolerance and activate the Th17/IL17A pathway, increasing thrombosis risk. Therefore, monitoring CD200R1 or IL17A levels could help identify high-risk patients and potentially enable combination therapy with anti-IL17A agents to reduce thrombotic complications.
Furthermore, this study highlights the powerful value of integrating multi-omics data with animal model validation in translational medicine. Future studies could extend to more cancer types, exploring the role of CD200R1 in the tumor microenvironment and its potential as a biomarker for immune-related adverse events (irAEs).
Conclusion
This study, by integrating plasma proteomics, machine learning, and mechanistic validation, not only constructs a superior model for predicting cancer-associated thrombosis compared to existing standards but, more importantly, reveals the CD200R1–IL17A–endothelial activation pathway as a targetable thrombo-inflammatory axis. This discovery extends thrombosis risk assessment from the coagulation system to immune regulatory networks, offering more precise thromboprophylaxis strategies for cancer patients. From bench to bedside, this research provides solid evidence for developing antithrombotic interventions based on IL17A inhibition, especially for patient populations in high-inflammatory states. In the future, combining CD200R1 genetic background with dynamic protein monitoring may enable personalized thrombosis risk management, significantly improving survival quality and treatment safety for cancer patients. This work sets a new benchmark for mechanistic research and translational applications in thrombo-inflammatory diseases, advancing the field from 'anticoagulation' toward 'immune modulation' as an antithrombotic strategy.
