Identifying poor responders to repetitive transcranial magnetic stimulation using structural brain networks.

Background

The efficacy of repetitive transcranial magnetic stimulation (rTMS) for depression varies widely. We examined whether pretreatment structural brain networks, specifically rich-club connection strength (RCS) and modularity, were associated with this variability and could help predict therapeutic outcomes.

Methods

In a prospective cohort of 204 depressed patients receiving 4-week left prefrontal rTMS, baseline diffusion MRI was used to quantify structural network rich-club organization and modularity. We assessed their associations with clinical outcomes (improvement, response, remission) and compared the predictive accuracy of clinical, network, and combined models.

Results

Patients with non-response and non-remission showed significantly lower baseline RCS and modularity than their respective comparison groups (all P < 0.05). Patients with concurrent reductions in both measures (low RCS/low modularity group) had the lowest response and remission rates among the four network groups (both P < 0.05), and this group was independently associated with non-response and non-remission in adjusted logistic models (both P < 0.05). The combined clinical-imaging model showed better discrimination than the clinical model alone for both response and remission. In the repeated-scan subsample, the low RCS/low modularity group showed less treatment-related network reorganization.

Conclusions

Lower baseline RCS and modularity were associated with poorer antidepressant outcome after rTMS in depression. Their combination identified a subgroup with particularly poor clinical improvement and limited network reorganization, suggesting that network-based markers may have prognostic value.