136 Chapter 6 level of organisation, causing a temporary network disorganisation, which later reorganises into a stronger signal at T4. This may indicate that the brain adapts to DBS treatment. The changes in graph parameters showed a similar V (or corresponding inverted-V) shape, and are therefore likely to reflect a uniform effect. Lower connectivity indicates that effective communication is reduced. The interesting observation is that DBS at the target area does not change the activity in the subcortical area directly, but indirectly affects communication between cortical areas. Decreased connectivity and rich-club metric, in combination with increased Path Length are typical for a less communicative, more segregated, and less integrated network. Although the effect was not significant, assortativity monotonically increased over time (i.e., not in a V-shape). If these results hold in future studies, it may indicate that the network changes qualitatively from T1 to T4: a linear increase of greater assortativity combined with V-shaped development of graph metrics that reflect integration. Future investigations may establish whether the combined network parameter assortativity is a useful biomarker of DBS induced brain network changes in AN. At the initial measurement T1 AN patients showed evidence of abnormalities in their EEG spectra compared to healthy controls. The power spectra showed clear peaks in the lower beta range where HC did not. Although the averaging into regional beta and alpha power did not reveal significant effects, more detailed inspection revealed that significant differences were found. AN patients showed increased lower and upper beta power in a wide central region. Although we have no source-localized data, the central location may suggest involvement of somatosensory or motor cortices. The prominent role of beta in sensorimotor cortices could explain associations with sensory and motor processing deficits in AN (35-37). Interestingly, after the initial stimulation, beta power normalized to levels close to HC, after which they showed an increase back to approximately baseline (T1) levels. Whether or not this is correlated with treatment success (e.g. BMI increases, reduction of eating disorder psychopathology) remains to be investigated. It is also unclear whether these initial differences are a cause or consequence of the disease state. However, these first observations suggest that central beta oscillations may be used to track the progression of DBS. In addition, alpha peak frequency showed evidence for slowing in AN. Nominal significance was observed for right parietal peak alpha frequency that was lower at T1 than in healthy controls. Although alpha peak slowing has mostly been related to aging and dementias, it also seems to be a general psychiatric phenomenon and useful for tracking treatment response in depression (38-40). However, these changes must be verified in follow-up studies, for example, controlling for medication use (40).
RkJQdWJsaXNoZXIy MjY0ODMw