Mapping Thrombosis Serum Markers by ¹H-NMR Allied with Machine Learning Tools.

Area of Science:

• Biochemistry
• Metabolomics
• Artificial Intelligence

Background:

• Antiphospholipid syndrome (APS) and venous thromboembolism (VTE) are complex conditions with significant health implications.
• Understanding the metabolic underpinnings of these diseases is crucial for developing effective diagnostic and therapeutic strategies.

Purpose of the Study:

• To investigate the discriminatory potential of blood serum metabolites for identifying thromboembolism (VTE) and antiphospholipid syndrome (APS) using machine learning.
• To identify specific metabolic pathways and biomarkers associated with the pathogenesis of APS and VTE.

Main Methods:

• Utilized ¹H-NMR-based metabonomics on serum samples from patients with VTE (n=32), thrombotic primary APS (n=32), and healthy controls (HCs, n=32).
• Applied machine learning and artificial intelligence tools to analyze unique metabolic profiles.
• Identified significant alterations in glycolysis, TCA cycle, lipid, and branched-chain amino acid (BCAA) metabolism.

Main Results:

• Distinct metabolic profiles were observed between VTE and HCs, APS and HCs, and VTE and triple-positive APS groups.
• Histidine, 3-hydroxybutyrate, and threonine emerged as the top three metabolites influencing model predictions.
• These metabolites demonstrated a substantial impact in distinguishing between APS, VTE, and HCs.

Conclusions:

• Metabolic pathway alterations in glycolysis, TCA cycle, lipid, and BCAA metabolism are implicated in the pathogenesis of APS and VTE.
• Histidine, 3-hydroxybutyrate, and threonine are identified as potential key biomarkers for differentiating APS and VTE patients from healthy individuals.