Treating obsessive-compulsive disorder by invasively modulating thoughts, feelings, or both

The defining characteristic of pathological compulsions is that they are carried out to reduce momentary distress, even though performing the compulsive act or thought is rationally not in the individual’s true best interest. This inability to flexibly adapt is a prominent component of obsessive-compulsive disorder (OCD) and closely related illnesses such as Tourette Syndrome, and is observed in substance use disorders as well (1). Following the 20^th century experience with ablative procedures such as capsulotomy, deep brain stimulation (DBS) was applied to the treatment of OCD. Electrodes are most commonly placed within the anterior limb of the internal capsule (ALIC), modulating parallel cortico-basal ganglia-thalamo-cortical (CBGTC) white matter loops relevant to limbic, associative, and motor functions, respectively. ALIC stimulation has proven to be a highly effective treatment for severe OCD (2). DBS to a different anatomical location, the subthalamic nucleus (STN), was initially developed for Parkinson’s Disease (PD), but positive early experience in patients with co-morbid PD and OCD led to research establishing high-frequency stimulation straight to STN as clinically efficacious for OCD (3). In a similar manner, promising early pilot studies led to further exploration of other targets. These include the bed nucleus of the stria terminalis (BNST), the ‘extended amygdala,’ located between the ALIC and STN targets (4). More recently, evidence has emerged for effectiveness of stimulation delivered deeper in the midbrain than STN, adjacent to the ventral tegmental area (VTA), a source of dopaminergic projections to cortex (via the medial forebrain bundle (MFB)) and a recipient of proposed regulatory projections back from cortex (via the ‘supralateral’ MFB (slMFB)) (5).

DBS as a treatment for OCD thus faces a surfeit of options. Grouping them conceptually, one strategy is to focus on ‘limbic’ aspects of the illness. Attempts can be made to alleviate negative affect (i.e., target BSNT), or boost positive affect (i.e., target the slMFB). A different strategy is to focus on improving a patient’s ability to think adaptively. CBGTC loops can be divided into a faciliatory ‘direct’ pathway, and a suppressive ‘indirect’ pathway that runs through the STN. High-frequency stimulation to STN may disrupt the suppressive indirect pathway, decreasing inhibitory control and promoting behavioral flexibility in a way that benefits OCD patients, an ‘overcontrolled’ population stuck in rigid and repetitive compulsive patterns. The net effect would be to cognitively ‘release’ OCD patients, analogous to the motoric and behavioral ‘release’ seen in PD STN DBS patients. Very recently, the advent of ‘connectomic DBS’ has reinforced a move toward thinking of all these targets as ‘hubs’ in their respective emotional (limbic) or cognitive (associative) networks. Moreover, Li et al. took the lead in elucidating a robust (6) and replicable (7) single set of fiber pathways that can be physically accessed from any of these anti-OCD targets. On a population level, patients whose electrodes overlap with these fibers have better clinical outcomes, regardless of target. This ‘unifying’ approach does not mean that different targets don’t have distinct advantages vis a vis one another. In fact, the team that originally discovered this set of ‘OCD response fibers’ is now actively working to ‘break it apart’ into more refined and symptom specific sub-networks, which may ultimately coincide at least partially with the targetable ‘hubs’ reviewed above.

Haber et al. propose a new means of simultaneously influencing limbic and associative domains of mental function. In the current issue of Biological Psychiatry (8), they build on their long and distinguished track record integrating two different methods: injections of viral tracers into non-human primate (NHP) brains, and high-resolution diffusion magnetic resonance imaging (dMRI) scans performed on healthy adult human subject. Each method has advantages and disadvantages. Tracer injections avoid the false positive findings that are a risk in dMRI studies, whereas only dMRI studies can be conducted in humans. Haber et al. have previously used these techniques to define the spatial topography of cortical projections traveling down through the ALIC (9), and to map out how these cortical projections synapse onto the STN (10). Following on these foundational earlier works, Haber et al. now turn their attention to the Zona Incerta (ZI), an area located superior to the STN. They argue that relative to other locations in the brain, the ZI presents a unique opportunity to modulate limbic and associative phenomenology through a single intervention.

Haber et al begin by injecting tracers into multiple sites throughout the prefrontal cortex (PFC) and anterior cingulate cortex (ACC), and following cortical projections down through the ALIC. As they have shown before, just past the anterior commissure a bundle of axons branches out from the now posterior internal capsule (IC) toward the thalamus. More caudally, they again confirm that another branch departs the IC for the STN and the VTA beyond it. New to the present study, they fully characterize an intermediate branch, between the thalamus and STN branches, that reaches the ZI, and confirm this pattern of sequential branchings with human dMRI. Of particular importance, PFC/ACC projection terminations were found in the NHP tracer experiments to cluster together within the rostral portion of ZI (ZIr), in contrast to projections from motor and pre-motor cortex that synapse more caudally in ZI. Within the ZIr, however, projections from the ventral-medial portion of cortex were sparse. This finding was supported by a second round of experiments in which Haber et al. injected bidirectional tracers into ZIr so that they could observe the results of retrograde transport from ZIr back to cortex. After these ZIr injections, they found ample labeled cell bodies in ventrolateral PFC (vlPFC), dorsolateral PFC (dlPFC), and ACC, but not ventromedial PFC (vmPFC) cortex. The present cortico-ZI findings thus mirror earlier findings from this group related to cortico-STN projections: motor projections synapsing onto the posterior STN, associative (i.e., lateral and dorsal) cortex projections synapsing onto the anterior STN, and projections from limbic (i.e, ventromedial) cortex only sparsely synapsing onto the STN itself (though some of these limbic cortex projections do appear to terminate medial to the conventional STN borders, while other travel on to VTA).

Haber et al. argue that the unique advantage of ZIr is its dense connections with a wide variety of other deep brain structures that mediate visceral and emotionally laden acute responses to environmental disturbances. Injecting bidirectional tracers into the ZIr and following tracers transported retrogradely, they identified cell bodies in the thalamus and also the hypothalamus, reticular formation, and VTA. Confirming the validity of this last set of connections, a follow up analysis looked at anterograde tracers injected into VTA, and indeed mapped direct projections from VTA to ZIr. The authors then shifted their focus from ZIr inputs to ZIr outputs, and found outgoing axons from ZIr to many of the same regions that sent inputs to ZIr, such as the thalamus, hypothalamus, reticular formation, and VTA. Of note, prominent projections were found from ZIr to the lateral habenula, a region known for generating negative affect, and one which had not been found to project to ZIr. These non-reciprocated projections from ZIr to lateral habenula were confirmed by injecting a retrograde tracer into the lateral habenula, which resulted in the labeling of many cell bodies in ZIr. Taken together, this series of sub-cortical injections, performed by the authors in ZIr, VTA, and lateral habenula, demonstrates that ZIr does have properties making it uniquely intriguing as a potential anti-OCD DBS target. Namely, it unites inputs from higher order associative cortex regions with connections to a variety of sub-cortical regions underlying critical survival-related functions, such as the reticular formation, hypothalamus, and lateral habenula. Haber et al. thus speculate that a DBS electrode inserted into ZIr could modulate the phenomenological intersection of ‘top down’ cognitive control processes with ‘bottom up’ felt sensations.

Prior to testing ZIr stimulation in humans, speculation will center around the relative value of specificity vs. comprehensiveness. Are there situations where it would be preferable to specifically target primarily limbic domains, or primarily cognitive domains? Taking things a step farther, could it be beneficial to make a surgical intervention even more narrow? ‘Cognition’ is a very broad domain, and one could envision advantages to selectively stimulating only those circuits involved in particular aspects of cognition. The set of ‘OCD response fibers’ discussed above travels through central ALIC, pointing to the clinical relevance of central PFC associative territory that is not so ventral that it crosses into purely limbic cortex, nor so dorsal that it encompasses motor regions. A new generation of segmented DBS electrodes are now available, which can be implanted in ALIC and used to selectively stimulate projections to sub-areas of interest within central PFC, e.g, ACC (relevant to uncertainty) vs. vlPFC (relevant to inhibitory control). Importantly, these new electrodes also have the ability to longitudinally capture local field potentials (LFPs) from adjacent grey matter nuclei over the course of treatment, further helping to define which circuits mediate successful treatment. Ultimately, the choice between zeroing in on a sub-component of cognitive or emotional circuity, versus opting to target an integrative zone such as ZIr, may depend on the needs of the individual patient being treated.