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

Repetitive transcranial magnetic stimulation (TMS), when applied to the dorsolateral prefrontal cortex (dlPFC), treats depression. Therapeutic effects are hypothesized to arise from propagation of local dlPFC stimulation effects across distributed networks; however, the mechanisms of this remain unresolved. dlPFC contains representations of different networks. As such, dlPFC TMS may exert different effects depending on the network being stimulated. Here, to test this, we applied high-frequency TMS to two nearby dlPFC targets functionally embedded in distinct anti-correlated networks—the default and salience networks— in the same individuals in separate sessions. Local and distributed TMS effects were measured with combined 18fluorodeoxyglucose positron emission tomography and functional magnetic resonance imaging. Identical TMS patterns caused opposing effects on local glucose metabolism: metabolism increased at the salience target following salience TMS but decreased at the default target following default TMS. At the distributed level, both conditions increased functional connectivity between the default and salience networks, with this effect being dramatically larger following default TMS. Metabolic and haemodynamic effects were also linked: across subjects, the magnitude of local metabolic changes correlated with the degree of functional connectivity changes. These results suggest that TMS effects upon dlPFC are network specific. They also invoke putative antidepressant mechanisms of TMS: network de-coupling.


Figure Legends
Fig. S1.Confirmation of network identity at target sites.Surface-rendered group-averaged seed-based functional connectivity (FC) maps defined using each subject's salience network (SAL) targets (Top, black rings) and default network (DN) targets (Bottom, white rings) as seeds.The SAL targets showed positive correlation with regions of the salience network (black outlines) and negative correlations with regions of the default network (white outlines).In contrast, the DN targets showed positive correlation with the default network and negative correlation with the salience network.

Fig. S3
. Target locations as they relate to gyral morphology.Coronal T1 images of each subject are shown in standardized MNI space.Subjects were numbered in the order in which experimental procedures were performed.The red dots represent functionally derived target sites which were targeted using a neuronavigation system during administration of TMS.The targets in each subject were far apart, with a more dorsomedial site representing the default network (DN) target, while the more inferolateral site represents the salience network (SAL) target.We did not observe systematic differences between sites, such as in the depth or cortical orientation of the target in relation to the cranial surface, that might be relevant for understanding the different effects of stimulation.

Fig. S4. Comparison of TMS induced functional connectivity (FC) between different scanning sessions.
Surface rendered voxel-wise statistical T maps (p<0.05,FWE cluster corrected) showing effects of TMS on FC between scanning sessions (left, Post > Pre; right, Post > Baseline).The top three rows represent FC changes due to stimulation of the salience network (SAL-TMS) and the bottom three rows represent FC changes due to stimulation of the default network (DN-TMS).Each row displays FC changes estimated using a different seed.'SAL target FC' and 'DN target FC' displays regions showing differences in FC with the individually defined stimulation site in each subject, averaged across subjects.'SAL FC' and 'DN FC' displays regions showing differences in FC with the average signal calculated from ROIs encompassing the entire default (white borders) and salience networks (black borders).For the SAL-TMS condition, between-network FC changes were more prominent in the Post > Baseline subtraction.For DN-TMS, the two comparisons differed in the magnitude of changes, but the topography and direction of FC changes were comparable, e.g., FC decreases within network and FC increases (i.e., reduced anticorrelations) between networks were observed in both comparisons.

Fig. S5
. The magnitude of metabolic effects of salience network TMS correlate with FC changes across subjects.Surface rendered voxel-wise correlation (r) maps (p<0.05,uncorrected) depicting correlations between SAL target metabolism (WBn-suv) changes and the observed FC changes across subjects.Each row indicates FC changes from a specific seed/ROI: the SAL target itself, the sgACC and the entire SAL network as an ROI.The left panel shows regions where a positive correlation between metabolism and FC effects of SAL-TMS was observed (yellow/pink).The right panel shows regions where a negative correlation between metabolism and FC effects following SAL-TMS was observed (green/blue).Across subjects, increases in SAL target metabolism correlated with increased FC between the sgACC and the salience network, particularly frontoinsular regions of this network (middle row).SAL target metabolism also correlated positively with cross-network FC changes between the default and salience networks (bottom row).Fig. S6.The magnitude of metabolic effects of default network TMS correlate with FC changes across subjects.Surface rendered voxel-wise correlation (r) maps (p<0.05,uncorrected) depicting correlations between DN target metabolism (WBn-suv) changes and the observed FC changes across subjects.Each row indicates FC changes from a specific seed/ROI: the DN target itself, the sgACC and the entire DN network as an ROI.The left panel shows regions where a positive correlation between metabolism and FC effects of DN-TMS was observed (yellow/pink).The right panel shows regions where a negative correlation between metabolism and FC effects following DN-TMS was observed (green/blue).Across subjects, decreases in DN target metabolism correlated with increased FC between the DN target and the salience network (top row).DN target metabolism decreases also correlated with sgACC FC decreases with the default network (middle row) and FC decreases within the default network proper (bottom row).

Fig. S7. The magnitude of transcranial magnetic stimulation (TMS) effects on metabolism at the target stimulation site is associated with the magnitude of effects of TMS on functional connectivity (FC).
Surface rendered voxel-wise statistical T maps (p<0.05, uncorrected) showing Post > Pre TMS FC changes.The right column shows the group mean (all subjects, n=16), and the other columns show the 5 subjects with the least and most robust changes in target metabolism.The top row shows FC changes with respect to an ROI encompassing the entire salience network (SAL ROI) following SAL-TMS and the bottom row shows FC changes with respect to an ROI encompassing the entire default network (DN ROI) following DN-TMS.Fig. S8.E-field modeling for each subject.Subjects are numbered, e.g., S1, S2.For each subject, the modeled E-field is shown in pairs for the SAL-TMS condition (left) and for the DN-TMS condition (right).Modeling was done based on tracking data from the Nexstim neuronavigation system, which recorded coil position/orientation during TMS for each subject (except for Subject 12, as tracking data was corrupted on the neuronavigation computer) using a Magventure MagPro B-65 coil (or a Magventure MagPro MCF-65 coil in the case of subject 7, S7) tangential to the scalp.The stimulated targets are displayed on the surface meshes as black rings for the SAL-TMS condition and white rings for the DN-TMS condition.The non-stimulated targets (e.g., the subject's DN-TMS target during SAL-TMS) are displayed as triangles.Subject doses (% maximum stimulation output, MSO) are also shown.The approximate coil position is shown on a standard head model behind the surface meshes.Each parcellation scheme uses a different name for a similar set of regions, i.e., the salience network (SAL), the cingulo-opercular network (CO), and the "task-positive" network.Right columns display the overlap between the selected network (black outlines) and regions showing metabolic increases, FC increases, and increased correlation between target metabolism and FC changes following the two transcranial magnetic stimulation (TMS) conditions (SAL-TMS and DN-TMS).This demonstrates that these selected results engage the same network regardless of the parcellation scheme used.DN=default network, SAL=salience network, CO=cingulo-opercular network, FPN=frontoparietal network.

Fig. S2 .
Fig. S2.Changes in target metabolism in each subject.Subjects are numbered in the order that data were acquired.For each subject, unthresholded images are shown displaying glucose metabolism at the target site during Baseline (left), following Transcranial Magnetic Stimulation (TMS) (middle), and the subtraction of the two (right).Left panel shows effects of SAL-TMS, with black rings representing SAL targets.Right panel shows effects of DN-TMS, with white rings representing DN targets.

Fig. S9 .
Fig. S9.Replication of results using different cortical parcellation schemes.Three different parcellation schemes: Glasser et al. 2016, Gordon et al. 2016 and Power et al. 2011 are shown in the left column, in addition to the Yeo et al. 2011 parcellation used in the main analyses.DN and SAL stimulation targets are shown using black and white rings (left column, see Fig 2, main text).Each parcellation scheme uses a different name for a similar set of regions, i.e., the salience network (SAL), the cingulo-opercular network (CO), and the "task-positive" network.Right columns display the overlap between the selected network (black outlines) and regions showing metabolic increases, FC increases, and increased correlation between target metabolism and FC changes following the two transcranial magnetic stimulation (TMS) conditions (SAL-TMS and DN-TMS).This demonstrates that these selected results engage the same network regardless of the parcellation scheme used.DN=default network, SAL=salience network, CO=cingulo-opercular network, FPN=frontoparietal network.