State anxiety may mediate the association between striato-cortical circuitry and anxiety symptom severity in generalized anxiety disorder: A resting-state fMRI study and support vector machine analysis.

BACKGROUND: Patients with generalized anxiety disorder (GAD) show structural and functional striatal abnormalities. While state and trait anxiety are known to modulate neural circuits influencing anxiety progression, the specific role of striatal-cortical circuitry in relation to anxiety dimensions in GAD remains unclear. METHODS: We included 43 GAD patients and 36 healthy controls (HCs), assessing trait/state anxiety and collecting resting-state fMRI data. The striatal seed-based functional connectivity (FC) was compared between groups. Correlation analyses evaluated links between striatal FC, clinical symptoms, and anxiety measures. Mediation analysis tested whether state anxiety mediates FC-symptom relationships. A linear support vector machine (SVM) model assessed striatal FC's ability to classify GAD vs. HCs. RESULTS: GAD patients exhibited increased FC between the right ventral superior striatum and medial prefrontal cortex, left dorsal caudal putamen (DCP) and left middle temporal gyrus, and right DCP and right MTG/fusiform gyrus, but decreased FC between the left ventral rostral putamen and left inferior parietal lobule/supramarginal gyrus. The FC between the right DCP and right MTG/fusiform exploratorily negatively correlated with HAMA and State Anxiety Inventory scores. State anxiety statistically mediate the relationship between striatal FC and anxiety severity. Linear and Gaussian SVM classifiers achieved accuracies of 81.07% and 79.82%, respectively. CONCLUSIONS: GAD involves disrupted striato-cortical connectivity. State anxiety statistically mediate the relationship between striatal FC and clinical anxiety in an exploratory cross-sectional mediation framework, which may highlight its central role. Altered striatal-cortical FC may show potential candidate features for distinguish GAD patients from HC in exploratory proof-of-concept analyses.