Network-specific metabolic and hemodynamic effects elicited by non-invasive brain stimulation

Abstract Repetitive transcranial magnetic stimulation (TMS), when applied to dorsolateral prefrontal cortex (dlPFC), is an effective treatment for depression. Therapeutic effects are believed to arise from propagation of dlPFC stimulation effects across distributed corticolimbic networks. However, the mechanisms of the local and distributed effects of TMS are unresolved, and studies have yielded variable results. One possible explanation for this variability is that dlPFC contains representations of different, closely juxtaposed, large-scale networks. As such, TMS may exert different effects depending on the network being stimulated. If true, targeting networks non-specifically may account for the large number of TMS treatment failures in depression. To test this, we used TMS to stimulate two nearby dlPFC targets in the same individuals. The two targets were functionally embedded in distinct anticorrelated networks: the default network and the salience network. Local and distributed effects of TMS were measured with combined whole-brain 18fluorodeoxyglucose positron emission tomography (FDG-PET) and functional magnetic resonance imaging (fMRI). Identical TMS patterns caused opposing effects on local glucose metabolism, with metabolism increasing at the salience target and decreasing at the default target. At the distributed level, both conditions reduced FC anticorrelation between the default and salience networks. However, FC effects were dramatically larger following default TMS. Metabolic and hemodynamic effects were also linked: across subjects, the magnitude of local metabolic changes correlated with the degree of FC changes. These results suggest that TMS effects upon dlPFC are network-specific. They also invoke putative mechanisms of TMS in depression, specifically the de-coupling of large-scale networks.