Abstract
Essential tremor is the most prevalent movement disorder and is often refractory to medical treatment. Deep brain stimulation offers a therapeutic approach that can efficiently control tremor symptoms. Several deep brain stimulation targets (ventral intermediate nucleus, zona incerta, posterior subthalamic area) have been discussed for tremor treatment. Effective deep brain stimulation therapy for tremor critically involves optimal targeting to modulate the tremor network. This could potentially become more robust and precise by using state-of-the-art brain connectivity measurements. In the current study, we utilized two normative brain connectomes (structural and functional) to show the pattern of effective deep brain stimulation electrode connectivity in 36 essential tremor patients. Our structural and functional connectivity models were significantly predictive of post-operative tremor improvement in out-of-sample data (p < 0.001 for both structural and functional leave-one-out cross-validation). Additionally, we segregated the somatotopic brain network based on head and hand tremor scores. These resulted in segregations that mapped onto the well-known somatotopic maps of both motor cortex and cerebellum. Crucially, this shows that slightly distinct networks need to be modulated to ameliorate head vs. hand tremor and that those networks could be identified based on somatotopic zones in motor cortex and cerebellum.
Finally, we propose a multi-modal connectomic deep brain stimulation sweet spot that may serve as a reference to enhance clinical care, in the future. This spot resided in the posterior subthalamic area, encroaching on the inferior borders of ventral intermediate nucleus and sensory thalamus. Our results underscore the importance of integrating brain connectivity in optimizing deep brain stimulation targeting for essential tremor.
Footnotes
We added two major analyses: First, we analysed connectivity profiles associated with the emergence of side effects (dysarthria and ataxia). Crucially, largely distinct regions were involved in electrodes associated with side effects. In addition, we preoperatively scanned a novel patient with diffusion MRI to successfully predict his improvement using patient specific fibertracts. Albeit this analysis merely adds anecdotal evidence (and we present the results in supplementary material), it may line out that from a methodological standpoint, our pipeline could be used in clinical practice. Finally, we replicated our main findings using a different statistical method that works on the level of fibertracts directly (instead of voxels) and was recently introduced by our group (Baldermann et al., Biological Psychiatry). In doing so, we further confirm the validity of our results using a different method.
