Objectives: The prevalence of generalised anxiety disorder (GAD) is high. However, the underlying mechanisms remain elusive. Proteomics techniques can be employed to assess the pathological mechanisms involved in GAD.
Methods: Twenty-two drug-naive GAD patients were recruited, their serum samples were used for protein quantification and identified using Tandem Mass Tag and Multiple Reaction Monitoring (MRM). Machine learning models were employed to construct predictive models for disease occurrence by using clinical scores and target proteins as input variables.
Results: A total of 991 proteins were differentially expressed between GAD and healthy participants. Gene Ontology analysis revealed that these proteins were significantly associated with stress response and biological regulation, suggesting a significant implication in anxiety disorders. MRM validation revealed evident disparities in 12 specific proteins. The machine learning model found a set of five proteins accurately predicting the occurrence of the disease at a rate of 87.5%, such as alpha 1B-glycoprotein, complement component 4 A, transferrin, V3-3, and defensin alpha 1. These proteins had a functional association with immune inflammation.
Conclusions: The development of generalised anxiety disorder might be closely linked to the immune inflammatory stress response.
Keywords: Immune inflammation; drug-naive; machine learning; multiple reaction monitoring; tandem mass tags.