Dynamic ALFF and dynamic Reho in bipolar disorder: Genetic links and predictive value for diagnosis and therapy

Alterations in resting-state brain activity in bipolar disorder (BD) and their pharmacological response remain unclear. This study investigated dynamic local brain activity, assessed the discriminative ability of dALFF and dReHo, and explored related transcriptomic correlates.

Resting-state fMRI and clinical data were obtained from 77 BD patients (with 38 completing the 3-month follow-up) and 83 healthy controls (HC). Dynamic local metrics (dALFF and dReho) were computed to quantify brain activity variability. Support vector machine (SVM) models were used for group classification and treatment-response prediction. Neuroimaging-transcriptomic associations were evaluated using gene expression data from the Allen Human Brain Atlas.

Baseline, BD patients showed decreased dReho in the right calcarine gyrus and decreased dALFF in the left superior and middle frontal gyri, alongside increased dALFF in the right hippocampus, corrected by Gaussian Random Field theory, with voxel-level significance set at p < 0.001 and cluster-level significance at p < 0.05. However, no significant alterations were observed following pharmacological treatment. The SVM models demonstrated moderate discriminative performance (AUC_dALFF = 0.83; AUC_dReho = 0.71). Cross-sample transcriptomic analyses identified 171 and 240 genes, after Bonferroni correction, that were potentially correlated with dReHo and dALFF variability, respectively, and were enriched in broad synaptic and immune-related pathways.

In summary, altered dALFF and dReHo patterns may represent neuroimaging markers of BD, reflecting disrupted dynamic brain activity. Neuroimaging-transcriptomic associations offer molecular insights into BD pathophysiology. However, findings should be interpreted cautiously due to the modest sample size, follow-up attrition, and lack of external validation.