If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Neurology, Faculty of Medicine and University Hospital Cologne, University of Cologne, Cologne, GermanyDepartment of Psychiatry and Psychotherapy, Faculty of Medicine and University Hospital Cologne, University of Cologne, Cologne, Germany
Department of Neurology, Movement Disorders and Neuromodulation Section, Charité - University Medicine Berlin, Berlin, GermanyEinstein Center for Neurosciences, Charité - University Medicine Berlin, Berlin, GermanyFaculty of Philosophy, Humboldt University of Berlin, Berlin School of Mind and Brain, Berlin, Germany
Department of Psychiatry and Psychotherapy, Faculty of Medicine and University Hospital Cologne, University of Cologne, Cologne, GermanyDepartment of Psychiatry, Psychotherapy and Psychosomatic, Johanniter Hospital Oberhausen, Oberhausen, Germany
Obsessive-compulsive disorder is among the most disabling psychiatric disorders. Although deep brain stimulation is considered an effective treatment, its use in clinical practice is not fully established. This is, at least in part, due to ambiguity about the best suited target and insufficient knowledge about underlying mechanisms. Recent advances suggest that changes in broader brain networks are responsible for improvement of obsessions and compulsions, rather than local impact at the stimulation site. These findings were fueled by innovative methodological approaches using brain connectivity analyses in combination with neuromodulatory interventions. Such a connectomic approach for neuromodulation constitutes an integrative account that aims to characterize optimal target networks. In this critical review, we integrate findings from connectomic studies and deep brain stimulation interventions to characterize a neural network presumably effective in reducing obsessions and compulsions. To this end, we scrutinize methodologies and seemingly conflicting findings with the aim to merge observations to identify common and diverse pathways for treating obsessive-compulsive disorder. Ultimately, we propose a unified network that—when modulated by means of cortical or subcortical interventions—alleviates obsessive-compulsive symptoms.
). This is at least in part due to uncertainty about the precise brain networks to modulate for optimal treatment response. The anatomical and functional characterization of circuits that, when stimulated, reduce obsessions and compulsions could improve the risk-benefit profile of DBS and provide testable hypotheses for neuromodulation of OCD in general.
Remarkably, different DBS targets have shown comparable response rates in OCD (
) has motivated the concept of a broader, potentially shared neural network responsible for improvement of obsessions and compulsions. In parallel, OCD is a heterogeneous disorder, with evidence suggesting that a varying set of multiple networks may be affected in each patient, and thus relevant for neuromodulatory treatment (
). Connectomic DBS is a rapidly developing neuroscientific concept that can help to understand how different target regions contribute to clinical improvement via linked networks. In this critical review, we aim to scrutinize methodologies and findings from connectomic studies and DBS interventions for OCD to identify common and diverse pathways likely to be effective for reducing obsessions and compulsions. We focus on structural connectivity for the sake of conciseness; relevant functional magnetic resonance imaging (MRI) studies in OCD DBS (
The idea of modulating a network (instead of a focal brain region) with surgery is not new. Around 1950, Jean Talairach and Lars Leksell independently began lesioning the anterior limb of the internal capsule (ALIC), with the aim of disrupting a network between limbic and prefrontal regions (
). In the following years, different nuclei adjacent to the ALIC, including the ventral striatum (VS) containing the nucleus accumbens (NAc) as well as the bed nucleus of the stria terminalis (BNST), have been proposed as key regions for successful DBS (
). Through empirical evidence from DBS in movement disorders, other brain targets such as the subthalamic nucleus (STN), the inferior thalamic peduncle, and the superolateral branch of the medial forebrain bundle (MFB) [later referred to as ventral tegmental area (VTA) projection pathway (
). Remarkably, modulation of these distinct subcortical targets (Figure 1) all show the potential to improve obsessions and compulsions, pointing toward a common network responsible for clinical efficacy. Using modern MRI technology such as diffusion-weighted imaging based tractography (diffusion MRI [dMRI]), we are now poised to create realistic in silico models of how these different sites of intervention may form nodes that assemble a common network (Figure 2). Specifically, by mapping clinical effects onto modulated neural pathways, researchers have begun to identify connectivity profiles associated with clinical efficacy (
). Undeniably, classic analysis of optimal spots can further complement such network analysis to characterize or validate specific hubs (i.e., for surgical targeting) within a given network. Box S1 outlines different methodological approaches that have been used in OCD DBS so far (see Supplement for a more detailed discussion).
Connectomic Studies of DBS for OCD
In a first connectomic approach toward DBS for OCD, Hartmann et al. (
) investigated 6 patients who underwent ALIC/NAc-DBS employing tract activation modeling using a normative dMRI brain atlas (Approach A in Box S1). In 2 responding patients, stimulation particularly affected fibers reaching the right anterior middle frontal gyrus/dorsolateral prefrontal cortex (dlPFC), while nonresponders stimulated the lateral orbitofrontal cortex (OFC). The authors concluded that targeting right dlPFC fibers leads to optimal response, while negative outcomes resulted from widespread activation of nontherapeutically relevant fibers. While the latter conclusion (i.e., negative outcome associated with widespread activation) was not directly replicated in further studies discussed below, the study also suggested that modulation of a more centrally rather than ventrally located white matter pathway within the ALIC was associated with optimal treatment response (
) investigated 12 patients with ALIC/NAc-DBS using individual preoperative dMRI and calculated the distance of active electrode contacts to specific fibertracts, the anterior thalamic radiation, and the superolateral branch of the MFB (Approach B in Box S1). Both fiber bundles were reconstructed using probabilistic tractography with the anterior thalamic nucleus or the VTA as seed regions, respectively, and the ventral ALIC as waypoint based on previous studies (
). The authors observed a significant positive correlation between clinical improvements and the proximity ratio in favor of the MFB compared with the anterior thalamic radiation. We note that the nomenclature and conceptualization of this fiber tract identified as MFB has since evolved (
). Optimal voxelwise structural connectivity profiles were calculated based on individual dMRI data in one group and based on normative dMRI data in another. The resulting maps constituted models of optimal connectivity capable to explain significant amounts of variance in outcomes in patients of the other subsample, indirectly highlighting the utility of both individual and normative dMRI for connectomic DBS. A final model of optimal connectivity using the normative connectome data across the whole group revealed that connectivity between stimulation sites and both lateral and medial prefrontal cortices could be cross-validated with significant correlations in a leave-one-patient-out design. Finally, a fiber-centric analysis (Approach C2 in Box S1) was introduced to further determine the subcortical representations of this beneficial connectivity profile. This analysis revealed a fiber bundle that connected the lateral and medial prefrontal cortex with the thalamus and STN, which traversed the ALIC centrally.
Further developing this novel approach (C2)—which has since been termed DBS fiber filtering or discriminative tractography—Li et al. published the largest study (N = 50) to date to determine connectivity associated with response to DBS for OCD (
) was included (ALIC/NAc target). Second, a cohort of 14 subjects who received DBS of the STN was added. Results were validated by calculating the tract model on the first cohort and overlaying stimulation volumes of the second cohort with it to generate coefficients termed fiber T scores. High scores would suggest optimal clinical outcomes, while low scores would suggest poor clinical outcomes. The calculated scores significantly correlated with clinical improvements in both directions, i.e., after calculating the tract model on the first to explain variance in the second cohort, and vice versa. Finally, two smaller samples from independent centers [N = 8 targeting the NAc and N = 6 targeting the ALIC/NAc region and STN in a crossover trial (
)] were used to validate results. By doing so, the hypothesized pathway for OCD DBS was refined, showing again that streamlines connecting the lateral and medial prefrontal cortex with the anteromedial STN and medial dorsal nucleus of the thalamus were associated with successful DBS. A more conservative analysis of the data using a tractography atlas of basal ganglia pathways (
) (which is less prone to false positives) confirmed a fiber bundle connecting the dorsal anterior cingulate cortex (dACC) with the STN via the ALIC as the strongest candidate tract represented by the atlas. After consultation with 4 anatomists, the tract was classified to represent a central subsection of the ALIC that involves hyperdirect input to the nonmotor STN.
) was made publicly available as a reference for scientific use (Figure 3, top right). An independent research group confirmed the predictive value of the pathway in 10 subjects with OCD and DBS of the ALIC/VS applying the same analysis procedure (
). In a voxel-by-voxel analysis, structural connectivity to the right ventrolateral PFC (vlPFC) and hippocampal regions and also to parietal and dorsomedial prefrontal areas significantly explained variance of response to DBS. A complementary fiber filtering analysis revealed, among others, white matter fibers within the ALIC that connected the stimulation site to the midbrain, traversing the BNST onward to the right vlPFC. This tract again graphically matched the pathway identified by Li et al. (
) (with a slightly more ventral course and overlap with the originally published tract) and correlated positively with clinical outcomes to a similar degree—albeit not significantly (Figure 3). Lastly, a recent investigation of 28 patients with Tourette syndrome treated with DBS of the anterior globus pallidus internus showed that modulation of the pathway published by Li et al. (
). This observation is remarkable because it suggests that DBS for OCD might act via symptom-specific rather than disease-specific networks. Specifically, it shows first evidence that the same network modulation approach could be effective transdiagnostically (a narrative that complies with the longstanding idea to study brain functions instead of nosological entities). Figure 3 summarizes the initial findings (
), a putative network associated with DBS response in OCD is likely not restricted to the already identified pathway but rather involves further connections yet to be uncovered, which is discussed in the later section, Further Pathways and Factors Relevant for Neuromodulation in OCD. Along with studies involving structural connectomics, there is also a growing body of literature on changes in metabolic, functional, or electrophysiological activity in the brain during DBS for OCD. An oxygen-15 positron emission tomography study by Dougherty et al. revealed an acute perfusion increase within the dACC and basal ganglia during ventral ALIC/NAc stimulation, which correlated with improvement of affective symptoms (
). In this study, capsulotomy for OCD resulted in an analog reduction in metabolic activity in the ACC. Figee et al. reported a DBS-induced reduction in hyperconnectivity with the ACC and lateral prefrontal cortex seeding from the NAc, which was associated with greater reduction of obsessions and compulsions (
). A later study performed a region of interest–based analysis of directional functional connectivity, showing that active DBS increased the impact of the ventromedial PFC (vmPFC) on an amygdala-insula network along with improvements in depression and anxiety symptoms (
). Moreover, there is a number of electrophysiological studies showing that active DBS of the ALIC/NAc interferes with low frequency oscillations within the mPFC/ACC, which can be linked to improvements in obsessions and compulsions (
). Taken together, these studies represent further evidence for the involvement of a central pathway encompassing the ACC and vlPFC in DBS for OCD beyond structural connectomics. Still, they also show that further circuits, especially involving the vmPFC, likely play a role in DBS for OCD. In summary, connectomic studies for OCD DBS provide growing evidence that a specific pathway within the ALIC carries out reductions in obsessions and compulsions, which is in part supported by studies using different modalities (functional MRI, positron emission tomography, electrophysiology). As outlined below, studies using different stimulation sites (ALIC, NAc, STN) are for the most part congruent in that modulation of fibers from the medial and lateral PFC, centrally traversing the ALIC and connecting the STN and thalamus, accounts for positive outcomes of OCD DBS.
While some of the aforementioned studies agreed on the critical role of the same fiber bundle published as a three-dimensional dataset (
), others revealed seeming heterogeneity about which pathway would be critical to modulate for successful DBS in OCD. Namely, the study by Liebrand et al. suggested that the MFB, connecting the PFC with the VTA, would be associated with a beneficial response (
)—and seen in this light, the studies would imply a conflicting finding. Reviewing the respective literature more in depth, however, suggests that this apparent discrepancy is in fact a matter of nomenclature. The fiber tract defined by Liebrand et al. (
) labeled as MFB was reconstructed using dMRI by placing a waypoint seed into the ALIC. This resulted in a bundle that passed the ALIC considerably dorsally to the NAc (Figure 4). The group was relying on former work by Coenen et al., who conceptualized the tract based on dMRI tractography (referred to as superolateral branch of the MFB) as a potential target for treating depression (
) confirm that the MFB is not part of the internal capsule (Figure 4). Thus, the streamlines referred to as superolateral MFB instead represent fibers of the internal capsule. Of note, there is uncertainty whether the functionally relevant connections are the ones from the VTA projecting through the internal capsule to the PFC and/or descending PFC-brainstem connections that send axon collaterals to regions such as the STN and VTA (
) could indeed be a cortico-midbrain projection traversing within the ALIC. These insights harmonize aforementioned findings with reports of the superolateral MFB/midbrain target/VTA projection pathway as an effective target for OCD (
Assembling all evidence, multiple studies from differing research groups with differing patient samples and targets converge on a highly similar effective stimulation site within the ALIC (Figure 5)—although authors had used different pathophysiological concepts to explain results (Figures 1 and 4). Evidence from nonhuman primate tract tracing suggests that this spot may best be described by the central portion of the ALIC with projections from the dACC and vlPFC (
). In a recent report, the same patients studied by Li et al. were reexamined based on functional connectivity, which has the advantage to include indirect connections. Again, a common network attributed a central role to the dACC (
). As shown in Figure 5, white matter tracts from the PFC/ACC travel through the ALIC in a topologically organized manner. By nature, dMRI-based tractography may not be able to distinguish these cortical representations with certainty. Subcortically, the connectomic evidence so far highlights the pivotal role of the anteromedial STN and the thalamus (
A Mechanistic Model of Connectomic Neuromodulation for OCD
Based on the evidence of connectomic DBS for OCD reviewed above, we propose a novel network model for an underlying mechanism of neuromodulation for OCD. Data indicate a central role for the dACC and that modulation of a hyperdirect connection of medial and lateral prefrontal cortices to the STN is associated with DBS response. Thus, the STN as an entry point for cortical information in terms of a hyperdirect pathway appears to be relevant for treatment of OCD, apart from the commonly accepted dysfunctional frontostriatal input related to the direct and indirect pathway (
). Secondly, projections between the anterior thalamus and PFC seem important (Figure 6). Considering the topological configuration of white matter tracts in the ALIC, the pathway can be described as a central ALIC pathway. Precise origination and termination points of this pathway remain unclear. However, some clues exist. As outlined, the dACC is a strong candidate derived from tractographic studies and is in line with alternative effective neuromodulation strategies for OCD, i.e., cingulotomy and transcranial magnetic stimulation, but methodological limitations prevent a definite conclusion regarding other cortical areas (i.e., vlPFC, dlPFC, and vmPFC) that may be involved (Table 1).
Table 1Structural Connectomic Studies of DBS for Obsessive-Compulsive Symptoms
Crucially, modulation of this circuit could take place at different nodes of the network: first, via DBS to the ALIC, STN, thalamus, and, potentially, globus pallidus internus; second, via ablative neurosurgery to dACC and ALIC; and third, via transcranial magnetic stimulation of the dACC. Importantly, the different targets within this loop are not necessarily interchangeable. Indeed, the fact that different targets are equally capable of modulating this specific network makes it even more important to understand what surmounting differences exist between them. For instance, a clinical trial including both the ALIC/NAc and STN targets in the same patients revealed different structural connectivity of these targets, although clinical improvement of obsessive-compulsive symptoms of both targets could be assigned to the same pathway (
). This suggests that each target additionally modulated different brain networks and, possibly, functions. Indeed, the authors distinguished that while ALIC/NAc-DBS had a greater effect on comorbid depression, STN DBS was associated with improved cognitive flexibility.
Finally, the concept of a common network for improving OCD symptoms may be independent of the disorder. As outlined, comorbid obsessions and compulsions in patients with Tourette syndrome improved when the central ALIC pathway was stimulated (
) (Figure 3). Thus, the proposed network may be effective in improving obsessions and compulsions, rather than OCD (as a categorical disease). Importantly, OCD is a highly heterogeneous disorder. Apart from specific OCD subtypes, e.g., washing, checking, and so on, the putative underlying neuropsychological mechanisms are also widespread, e.g., impaired habit versus goal-directed behavior, cognitive inflexibility, emotional vulnerability, or altered risk evaluation. These underpinning principles may in turn serve as transdiagnostic dimensions for other compulsivity-related disorders, such as behavioral addiction, substance use disorders, Tourette syndrome, and autism-related stereotypies (
). Thus, a next step toward a more effective and personalized neuromodulation for OCD will be to characterize these endophenotypes and identify through which networks each may be effectively modulated (
). In line with this notion, connectomic studies for OCD DBS provide evidence that modulating specific circuits relevant in OCD pathophysiology (i.e., a central ACC-ALIC-STN pathway and possibly a vmPFC-related pathway) can lead to clinical improvement. Furthermore, our review highlights the potential role of the STN in OCD therapy as an entry point for cortical information from the PFC in terms of a hyperdirect pathway.
Further Pathways and Factors Relevant for Neuromodulation in OCD
We must reiterate that this proposed mechanistic model forms one possible mechanism of action—and could represent part of a larger network. Modulation of additional loops (e.g., ventral and dorsal frontostriatal loops, fronto-midbrain connections) and respective changes in symptom dimensions will further contribute to specific therapeutic outcomes (
). This is reflected by the fact that although a common pathway could be derived from connectomic studies in OCD DBS, there are evidently subjects in whom this pathway was not modulated but who still profited from DBS (
), implying that additional circuits will be relevant to consider. As an example (Figure S1), DBS for OCD is capable of changing affective states (i.e., anxiety, mood) that are accompanied by altered activity in a network comprising the vmPFC, insula, and amygdala (
). Congruent to this, it was recently shown that transcranial alternating-current stimulation of the OFC improved obsessive-compulsive behavior in a cohort of healthy subjects by interfering with reward-related beta-gamma oscillations (
), the frontostriatal input may also play a decisive role for improving affective states in OCD. The importance of this circuit for OCD DBS is also supported by animal studies showing that optogenetic stimulation of the OFC-VS pathway decreases grooming in a rodent model of OCD (
). A later study in the same OCD mouse model revealed that both DBS of the VS and ALIC resulted in a significant reduction in grooming independently (although the ALIC target was more effective on average), suggesting that both pathways are contributing to therapeutic success (
). Further evidence supporting the involvement of an affect-related circuitry stems from the comprehensively discussed study by Mosley et al., where connectivity with the amygdala was also associated with DBS response, along with modulation of the central ALIC (
). These different therapeutic circuitries could correspond to improvement of different symptoms or neuropsychological dimensions of OCD. Thus, we emphasize that in the same manner as different basal ganglia cortical loops are implicated in the pathophysiology of OCD (
), neuromodulation of different circuits may contribute to therapeutic success.
Needless to say, other factors beyond targeting are likely to influence the outcome of OCD DBS as well, but so far, reliable response predictors are unknown. Larger volumes of the striatum seem to be associated with better outcomes (
)]. In all of these trials, a monopolar high-frequency stimulation (>80 Hz) was applied, and the pulse width was mostly selected above 60 μsec, although often considerably higher (up to 120–450 μsec) for the ALIC/NAc/BNST area (
). Furthermore, on average and across centers, ALIC stimulation volume centers were more distant to the central ALIC target than in the STN groups—which could again explain lower stimulation amplitudes applied in the STN target.
Limitations and Methodological Considerations
Connectomic DBS for OCD is a novel and emerging field that comes with relevant limitations. Primarily, the most studies relied on small cohorts (inherent to psychiatric DBS), which comes with a greater risk of false positive findings. Second, connectomic studies for DBS strongly depend on the validity of the modeled white matter pathways and how activation hereof is determined, which is again subject to relevant limitations. In case of OCD, many studies relied on a similar whole-brain normative connectome and fiber filtering approach based on isotropic electric field models (
). For a discussion on potential limitations of activation volume tractography (as performed in most OCD DBS studies, so far) versus tractography/pathway-activation models [as performed for instance in the study by Hartman et al. (
). Third, until now, there is no prospective validation of the identified pathways in OCD DBS. Critically, prospective tractography-based DBS can result in substantial differences across centers, putatively because of differences in tractographic analysis (
). Despite these limitations, the field of connectomic DBS for OCD has made tremendous progress in the past years, and the current evidence stems from multiple centers using different targets and has been partly cross-validated using different connectivity estimates (e.g., dMRI and histology-based atlases). To face the obstacle of connectomic DBS for OCD, we call for future studies that 1) pool data from different centers for larger sample sizes, 2) focus on adequately assessed individual symptom/neuropsychological dimensions of OCD, 3) employ and ideally compare different approaches of DBS connectivity models, and 4) combine different neuromodulatory approaches for OCD. Finally, following the pioneering example of connectomic DBS for depression (
), prospective studies are now necessary to validate observations in DBS for OCD to make a step toward a more tailored, precise, and thus safe and effective neuromodulation for OCD.
In summary, we review evidence for a unified network spanning between cortical (the dACC, vlPFC, and assumingly others) and subcortical (anteromedial STN, medial dorsal nucleus of the thalamus) regions that—when modulated by means of DBS, ablative surgery, or noninvasive neuromodulation—alleviates obsessive-compulsive symptoms. We conclude that despite different uses of nomenclature, there is a high concordance between studies—especially regarding a specific surgical target site within the ALIC. Finally, we provide a mechanistic model with the most salient addition to include a limbic/associative hyperdirect pathway that traverses within the central segment of the ALIC as a critical component for clinical efficacy.
Acknowledgments and Disclosures
This study is supported by the National Institutes of Health (Grant No. MH106435 [to SNH]); German Research Foundation (Deutsche Forschungsgemeinschaft, Project-ID 431549029 –SFB 1451 [to JCB] and Emmy Noether grants 410169619 and 424778381 – TRR 295 [to AH]); as well as Deutsches Zentrum für Luft- und Raumfahrt (DynaSti grant within the EU Joint Programme Neurodegenerative Disease Research [to AH]).
PEM has previously received an honorarium for lecturing from Boston Scientific and received an unrestricted educational grant from Medtronic . He is currently an investigator in clinical trials of deep brain stimulation for obsessive-compulsive disorder and anorexia nervosa. SAS reports consulting agreements with Boston Scientific, Abbott, Neuropace, Zimmier Biomet, and Koh Young. MB reports personal fees from Medtronic, Boston Scientific, Abbott (formerly St Jude), GE Medical, UCB, Bial, and IQWIG, and grants from Gondola, Felgenhauer-Stiftung, and Forschungspool Klinische Studien, outside the submitted work. All other authors report no biomedical financial interests or potential conflicts of interest.