Abnormal dynamics of brain networks in bipolar disorder: a dynamic degree centrality and imaging-transcriptomic investigation.

backgroundBipolar disorder (BD) is a chronic psychiatric illness with high recurrence and disability rates. Identifying neuroimaging biomarkers may improve diagnostic precision and treatment prediction.

methodsThe resting-state functional MRI (rs-fMRI) data were acquired from patients with BD and age-, sex-, and education-matched healthy controls (HCs), with a subset of patients completing a three-month pharmacological follow-up. Dynamic degree centrality (dDC) was calculated using a sliding-window approach to characterize time-varying whole-brain network integration. Group differences were assessed at baseline and post-treatment. Support vector machine (SVM) models were used to evaluate the discriminative ability of dDC features. Imaging–transcriptomic association analyses were conducted using gene expression data from the Allen Human Brain Atlas to explore potential molecular mechanisms.

resultsCompared with HCs, patients with BD at baseline showed significantly reduced dDC variability in the cerebellar Crus I and Crus II. After treatment, increased dDC variability emerged in the bilateral anterior cingulate and paracingulate gyri. The SVM classifier based on dDC features achieved moderate performance in distinguishing BD patients from HCs. Transcriptomic analyses indicated that genes associated with dDC alterations were mainly enriched in synaptic signaling and immune–inflammatory pathways, with several hub genes identified in protein–protein interaction networks.

conclusionBD is characterized by disrupted temporal dynamics of whole-brain network integration involving cerebellar and cingulate regions, which appear partially modulated by pharmacological treatment. Integrating dynamic network metrics with transcriptomic data provides complementary insights into the neurobiological and molecular substrates of BD and supports the potential of dDC as a neuroimaging biomarker.