Deep Brain Stimulation for Obsessive-Compulsive Disorder: Optimal Stimulation Sites

Abstract

Background: Deep brain stimulation (DBS) is a promising treatment option for treatment-refractory obsessive-compulsive disorder (OCD). Several stimulation targets have been used, mostly in and around the anterior limb of the internal capsule and ventral striatum. However, the precise target within this region remains a matter of debate.

Methods: Here, we retrospectively studied a multicenter cohort of 82 patients with OCD who underwent DBS of the ventral capsule/ventral striatum and mapped optimal stimulation sites in this region.

Results: DBS sweet-spot mapping performed on a discovery set of 58 patients revealed 2 optimal stimulation sites associated with improvements on the Yale-Brown Obsessive Compulsive Scale, one in the anterior limb of the internal capsule that overlapped with a previously identified OCD-DBS response tract and one in the region of the inferior thalamic peduncle and bed nucleus of the stria terminalis. Critically, the nucleus accumbens proper and anterior commissure were associated with beneficial but suboptimal clinical improvements. Moreover, overlap with the resulting sweet- and sour-spots significantly estimated variance in outcomes in an independent cohort of 22 patients from 2 additional DBS centers. Finally, beyond obsessive-compulsive symptoms, stimulation of the anterior site was associated with optimal outcomes for both depression and anxiety, while the posterior site was only associated with improvements in depression.

Conclusions: Our results suggest how to refine targeting of DBS in OCD and may be helpful in guiding DBS programming in existing patients.

Keywords: Anterior limb of the internal capsule; Bed nucleus of the stria terminalis; Deep brain stimulation; Obsessive-compulsive disorder; Sweet-spot; Ventral striatum.

Fig. 1:. Electrode and active contact locations (view from anterosuperior).

Electrode locations in each cohort (A) as well as active contact locations (B) are shown relative to relevant atlas structures and overlaid on an ultrahigh-resolution MRI template (42). The nucleus accumbens (NAc; purple, obtained from the California Institute of Technology reinforcement learning atlas, CIT168, 43), ventral pallidum (VeP; pink, obtained from the CIT168) and bed nucleus of the stria terminalis (BNST; orange, obtained from the atlas of the human hypothalamus, 44) are shown as 3-D structures.

Fig. 2:. Y-BOCS sweet-spot mapping results.

(A, B, C) Axial, coronal and sagittal views of the Y-BOCS sweet-spot. (D) Schematic representation of the sweet-spot mapping method. The sweet-spot was determined by calculating, at each voxel, the Spearman correlation coefficient (ρ) between E-field magnitude and clinical improvement. Only voxels receiving significant positive or negative ρ values were retained as part of the sweet- or sour-spot (respectively represented in red and blue). Examples are shown for representative voxels (E, F). The Sweet-spot Score – an estimate of how well a patient’s E-fields match the sweet-spot – significantly correlated with Y-BOCS improvement (G). Note that, for visualization only, and owing to the symmetry of the initial sweet-spot, the E-fields were non-linearly warped to the contralateral hemisphere such that the sweet-spot was mirrored across hemispheres, allowing to display only the left hemisphere. Views of the non-mirrored sweet-spot are available as supplementary material. AC: Anterior commissure; Ca: Caudate nucleus; GP: Globus Pallidus; IC: Internal Capsule; NAc: Nucleus Accumbens; Pu: Putamen; STN: Subthalamic Nucleus; VeP: Ventral Pallidum.

Fig. 3:. Depression and anxiety sweet-spot mapping results.

(A, B) Axial, coronal and sagittal views of the depression and anxiety sweet-spots. The Sweet-spot Score significantly correlated with clinical improvement (C, D). As above, E-fields were non-linearly warped to the contralateral hemisphere for visualization. Views of the non-mirrored sweet-spots and direct comparison with the Y-BOCS sweet-spot are available as supplementary figures. AC: Anterior commissure; Ca: Caudate nucleus; GP: Globus Pallidus; IC: Internal Capsule; NAc: Nucleus Accumbens; Pu: Putamen.

Fig. 4:. Out-of-sample validation of the Y-BOCS sweet-spot.

(A) Electrode locations for the independent cohort of 22 OCD patients receiving VC/VS-DBS used for validation. (B) The Sweet-spot Score significantly correlated with clinical improvement, meaning that the degree of agreement between E-fields and sweet-spots explained significant amounts of variance in clinical outcome in this unseen cohort. BNST: Bed Nucleus of the Stria Terminalis; NAc: Nucleus Accumbens; VeP: Ventral Pallidum.

Fig. 5:. 3-D views of the sweet- and sour-spots and relevant anatomy.

(A) Two optimal VC/VS stimulation sites were identified (dark red) for Y-BOCS, as well as one suboptimal stimulation site (dark blue). (B) The most anterior sweet-spot partially overlaps with the OCD response tract within the anterior limb of the internal capsule (ALIC) whose stimulation was found to be positively associated with outcome in a previous study (red fibers, 18; note that 36 out of the 58 patients included in the present study were also included in the study by Li et al.; Table 1). The sour-spot overlaps with the nucleus accumbens (NAc; light blue) and with the anterior commissure, the latter of which has previously been identified as negatively associated with outcome by Li et al. (blue fibers). (C) The postero-ventral sweet-spot lies at the intersection between various limbic structures and pathways, including the bed nucleus of the stria terminalis (BNST; light green), inferior thalamic peduncle (ITP; yellow), ventral amygdalofugal pathway (purple, reconstructed using DSI studio; 49) and medial forebrain bundle (orange, obtained from the DBS tractography atlas; 50).

Fig. 6:. Proximity of the sweet-spots to previously reported optimal targets.

(A) Overall, the sweet-spots are consistent with previous experience and evolution of targeting in two large international cohorts (as indicated by the arrows). (B) Denys et al. (6) recently reported on 70 consecutive patients receiving DBS for OCD. While they initially targeted the two lower electrode contacts to the NAc (10), they later switched to targeting the three upper contacts to the ALIC after observing that contacts selected for chronic stimulation were often located within capsular white matter rather than NAc (6,66). Their suggested optimal target (orange sphere; as defined in 66) is hence more dorsal than their initial one, and better aligns with our antero-dorsal sweet-spot. (C) Similarly, Greenberg et al. (9) initially targeted the NAc but moved towards a more posterior target over time (purple spheres). This refined target, which led to improved outcomes (referred to as “Group C” in the original publication) and was described by the authors as lying at the at the junction of the ALIC, anterior commissure, and posterior ventral striatum, precisely matches our second, postero-ventral sweet-spot. In contrast, the initial target (“Group A”, which had poorer outcomes) resided close to the sour-spot identified here. The arrows represent the direction of the evolution of the respective targets (i.e., shift in the corresponding plane).

Fig. 7:. Overlap between the postero-ventral Y-BOCS sweet-spot and previously reported hotspots for acute stimulation-induced side-effects in the hypothalamic region.

The hotspots were identified by voxel-wise mapping of 627 acute clinical responses to stimulation in 58 Alzheimer’s disease patients receiving DBS of the fornix region (48). Three-dimensional dorsal view (A) as well as 2-D axial, coronal and sagittal views (B) are provided. The lateral hypothalamus (white; obtained from the atlas of the human hypothalamus, 44) is displayed as a reference anatomical structure. Our Y-BOCS sweet-spot overlaps with the fear and tachycardia/hypertension hotspots and is close to the flushing/sweating hotspot.