Two people may be the same age and have similar family histories or risk factors, yet only one of them may develop thrombosis. To better understand why this occurs, researchers from the Complex Disease Genomics Unit at the Sant Pau Research Institute (IR Sant Pau) and the Biomedical Research Networking Center for Rare Diseases (CIBERER) have developed an artificial intelligence–based tool that integrates clinical, genetic, and transcriptomic information to identify signals associated with the disease. The results, published in the Journal of Thrombosis and Haemostasis , identify hundreds of molecular signals associated with thrombosis and improve the characterization of people with different risk profiles.
Venous thrombosis is one of the most common cardiovascular diseases and a major cause of morbidity and mortality. Although several factors are known to increase the risk of developing a thrombotic event, some cases occur without any clear triggering factors. This form of the disease, known as idiopathic venous thromboembolism, makes it more difficult to identify people with a greater predisposition and limits the ability to anticipate its onset. Various studies have shown that more than 60% of the individual variability in the risk of venous thrombosis may be influenced by genetic factors. However, known hereditary factors do not fully explain why some people develop the disease while others do not, prompting the search for new biomarkers and risk-stratification tools.
To identify signals that traditional risk factors cannot fully capture, the team analyzed information from 790 people belonging to families with a history of venous thromboembolic disease, including 70 who had previously experienced idiopathic venous thrombosis. The data came from the GAIT2 (Genetic Analysis of Idiopathic Thrombophilia) family cohort, one of the largest initiatives devoted to studying hereditary predisposition to thrombosis. The researchers integrated clinical and genetic variables with expression profiles derived from the activity of 12,981 genes to build models capable of identifying patterns associated with the disease and assessing whether they could improve risk stratification beyond conventional approaches.
"The study's main contribution is not only the identification of new genes associated with thrombosis, but also the demonstration that integrating thousands of biological variables makes it possible to describe risk profiles far more accurately than when traditional factors are analyzed in isolation," says Dr. José Manuel Soria, director of the Complex Disease Genomics Unit at IR Sant Pau and co-senior author of the study.
Beyond Traditional Risk Factors
Until now, thrombotic risk assessment has primarily been based on known clinical factors, such as age, obesity, certain hormone treatments, or specific genetic alterations. However, these factors do not fully explain why some people develop thrombosis in the absence of clear triggers.
To address this question, the researchers applied different machine-learning algorithms capable of simultaneously analyzing thousands of biological variables. This approach enabled them to identify a combination of clinical and molecular factors associated with a history of idiopathic venous thrombosis.
The most relevant predictors included previously known markers, such as von Willebrand factor levels, body mass index, age, and certain variants of the ABO system. In addition, the analysis identified 494 genes whose activity helped distinguish people who had experienced thrombosis from those with no history of the disease, including numerous long noncoding RNAs, a type of regulatory molecule that has received little study in the context of thrombosis.
"The incorporation of transcriptomic data allowed us to identify disease-associated signals that could not be detected using conventional approaches. This demonstrates the potential of combining artificial intelligence and gene expression to achieve a more precise characterization of patients," explains Dr. Pol Ezquerra, first author of the study and a researcher at IR Sant Pau.
A Molecular Signature Associated With Thrombosis
Beyond identifying specific biomarkers, the team developed a molecular signature associated with thrombosis based on the combination of thousands of clinical, genetic, and transcriptomic variables. Using this signature, they created a similarity score capable of measuring how closely a person's profile resembles that of individuals who had already experienced a thrombotic event. This made it possible to identify participants with no history of the disease whose profiles showed characteristics similar to those observed in patients with thrombosis.
The incorporation of transcriptomic data significantly refined the classification of participants. When the model used only clinical and genetic variables, 43% of people with no history of thrombosis were classified within the high-risk range. After gene-expression information was added, this proportion fell to 23%. At the same time, the identification of people with a history of thrombosis increased from 70% to 74%.
"Gene expression provides an additional layer of information that complements traditional risk factors. This allows us to better distinguish between individuals who share similar clinical characteristics but have different profiles when we analyze their molecular activity," explains Dr. Soria.
Toward More Personalized Prevention
The study also identified signals related to cardiovascular and renal processes previously associated with thrombotic risk, reinforcing the biological relevance of the findings. These included several molecular pathways related to cardiomyopathies and the function of the kidney's proximal tubules, two systems that previous research had linked to venous thromboembolic disease.
The authors emphasize that the tool still requires validation in independent cohorts before its direct clinical application can be considered. Nevertheless, they believe this approach represents an important step toward more accurate models for thrombotic risk stratification and increasingly personalized medicine.
"Our work demonstrates the value of integrating clinical, genetic, and transcriptomic information to obtain a more comprehensive view of the factors associated with thrombosis. In the future, these types of strategies could help identify people with a high-risk profile more accurately and facilitate the development of preventive measures tailored to each patient," Dr. Soria concludes.
The researchers also note that the genes and noncoding RNAs identified in this study represent a new source of potential biomarkers and research targets for gaining a more profound understanding of the mechanisms involved in the development of idiopathic venous thrombosis.
