Patient-specific structural connectomic differences of deep brain stimulation targets in treatment-resistant obsessive-compulsive disorder patients

Abstract

Deep brain stimulation (DBS) is an established therapy for treatment-resistant obsessive-compulsive disorder (trOCD), yet clinical outcomes vary. This is partly due to differential cortico-striato-thalamo-cortical networks (CSTC) engagement across stimulation in relevant anatomical regions. This study compared the structural and functional connectivity profiles of three relevant anatomical-regions that have also served as target for DBS in different centers for trOCD: the nucleus accumbens/anterior-limb of the internal-capsule (NAc/ALIC), the medial forebrain bundle (MFB), and the anteromedial subthalamic nucleus (amSTN). We analysed in a retrospective, observational, single-center cohort the structural connectivity on patient-specific space, using structural and diffusion MRI data from 22 trOCD patients treated with NAc/ALIC DBS at the University Hospital Cologne. For this end, Whole-brain probabilistic-tractography using single-shell-three-tissue constrained-spherical-deconvolution was applied to reconstruct patient-specific structural connectivity matrices. From whole-brain tractograms, target-specific streamline subsets were selected using ROI-based inclusion criteria using 2 mm radius spheric stereotactic ROIs representing clinically relevant trOCD-DBS target regions. Afterwards, Structural connectivity matrices were extracted using Spherical-deconvolution informed filtering of tractograms (SIFT) streamline count, fractional anisotropy (FA), and mean diffusivity (MD) across 13 cortico-subcortical regions implicated in OCD pathophysiology. Thereafter, we characterized each target’s connectivity to key OCD functional networks derived from meta-analytic functional connectivity maps. NAc/ALIC demonstrated stronger structural fronto-limbic connectivity: medial-orbitofrontal cortex, anterior-cingulate, insula, and accumbens (p < 0.0001), associated with the affect (AN), salience (SN), default-mode (DMN), and reward-motivation (RMN) networks anatomical hubs. The MFB predominantly connected with reward-related regions: pallidum and rostral-middle-frontal cortex (p < 0.0001). The amSTN showed to be mainly connected to SN and the cognitive/motor control network (CMCN): precentral and paracentral gyri (p < 0.0001). These exploratory findings suggest that clinically relevant trOCD-DBS targets may engage distinct but partially overlapping tractography derived structural networks and meta-analytic functional networks. The observed connectivity profiles may help generate hypotheses about symptom-specific networks for future DBS planning. Overall, these results support the proof-of-concept value of patient-specific probabilistic tractography in trOCD. However, prospective validation in independent, multicentric trials remains necessary before clinical translation.