135 Electrophysiological effects of deep brain stimulation in anorexia nervosa for coherence in right and left anterior channels (p<0.038) and left-posterior channels for wPLI (p=0.028). None of these differences survived correction for multiple testing. Whole-brain connectivity and graph measures Graph measures describe the organization of the brain connectivity as a property of the whole network. For this, we only used the wPLI-based connectivity matrices, as this algorithm is designed to suffer less from spurious connectivity due to volume conduction effects. The effects for the graph measures are shown in figure 8. The level for HC is indicated with the dashed line. Graph measures showed no significant differences with HC after multiple testing correction. Average connectivity, path length, diameter, rich-club metric showed a U curve (inverted for path length and rich-club). FDR-corrected p-values showed significant change at T2 compared to T1 for beta band average connectivity, path length, and the rich-club metric. Diameter showed a significant change at T2 compared to T1 for the alpha band (p=0.032 for all). Assortativity did not significantly change over time. DISCUSSION We hypothesized that DBS causes changes in electrophysiology of the brain of AN patients beyond the local influence of DBS at the stimulation target. We found evidence for such a widespread effect: strong and significant within-subject changes in oscillatory power were observed for all patients. These power changes indicated that DBS affected EEG across the scalp (patients 2-4), and for both frequency bands. In patient 1 significant changes were also found, albeit less widespread. The effects of DBS on brain activity were specific for individual patients: Patients 1, 3, and 4 showed a decrease in both alpha and beta power after the first DBS period (T2), and a subsequent increase at T4. Patient 2, however, showed an increase in power in many regions of the brain, mainly over the right hemisphere. Overall, network activity showed a V-shaped curve, with initial disorganization and later reorganization. Averaging individual data resulted in heterogeneity, suggesting that the effects of DBS on network activity in AN are strong and widespread yet highly personalized. Our results are in line with the findings of the recent MEG-study on DBS in AN that indicate that DBS modulates neuronal process in regions far outside the stimulation target site and at latencies possibly reflecting task specific processing (18). We observed a V-shaped curve over time for average connectivity. In particular, beta connectivity showed significant changes from T1 to T2. At T4, beta connectivity appears to increase again and shows no significant difference with baseline. A possible explanation for this curve is that at T1, there is a relatively highly structured but pathological network organisation. This could be consistent with the findings of a study that showed an altered global network architecture with local degradations in patients with AN (16). It could be hypothesized that DBS disrupts this high
RkJQdWJsaXNoZXIy MjY0ODMw