Introduction: Deep brain stimulation (DBS) is an established treatment in patients of various ages with pharmaco-resistant neurological disorders. Surgical targeting and postoperative programming of DBS depend on the spatial location of the stimulating electrodes in relation to the surrounding anatomical structures, and on electrode connectivity to a specific distribution pattern within brain networks. Such information is usually collected using group-level analysis, which relies on the availability of normative imaging resources (atlases and connectomes). Analysis of DBS data in children with debilitating neurological disorders such as dystonia would benefit from such resources, especially given the developmental differences in neuroimaging data between adults and children. We assembled pediatric normative neuroimaging resources from open-access datasets in order to comply with age-related anatomical and functional differences in pediatric DBS populations. We illustrated their utility in a cohort of children with dystonia treated with pallidal DBS. We aimed to derive a local pallidal sweetspot and explore a connectivity fingerprint associated with pallidal stimulation to exemplify the utility of the assembled imaging resources.
Methods: An average pediatric brain template (the MNI brain template 4.5-18.5 years) was implemented and used to localize the DBS electrodes in 20 patients from the GEPESTIM registry cohort. A pediatric subcortical atlas, analogous to the DISTAL atlas known in DBS research, was also employed to highlight the anatomical structures of interest. A local pallidal sweetspot was modeled, and its degree of overlap with stimulation volumes was calculated as a correlate of individual clinical outcomes. Additionally, a pediatric functional connectome of 100 neurotypical subjects from the Consortium for Reliability and Reproducibility was built to allow network-based analyses and decipher a connectivity fingerprint responsible for the clinical improvements in our cohort.
Results: We successfully implemented a pediatric neuroimaging dataset that will be made available for public use as a tool for DBS analyses. Overlap of stimulation volumes with the identified DBS-sweetspot model correlated significantly with improvement on a local spatial level (R = 0.46, permuted p = 0.019). The functional connectivity fingerprint of DBS outcomes was determined to be a network correlate of therapeutic pallidal stimulation in children with dystonia (R = 0.30, permuted p = 0.003).
Conclusions: Local sweetspot and distributed network models provide neuroanatomical substrates for DBS-associated clinical outcomes in dystonia using pediatric neuroimaging surrogate data. Implementation of this pediatric neuroimaging dataset might help to improve the practice and pave the road towards a personalized DBS-neuroimaging analyses in pediatric patients.