Advertisement

Internal capsule/nucleus accumbens deep brain stimulation increases impulsive decision-making in obsessive-compulsive disorder

  • Thomas Schüller
    Affiliations
    University of Cologne, Faculty of Medicine and University Hospital Cologne, Department of Psychiatry and Psychotherapy, Cologne, Germany
    Search for articles by this author
  • Sina Kohl
    Affiliations
    University of Cologne, Faculty of Medicine and University Hospital Cologne, Department of Psychiatry and Psychotherapy, Cologne, Germany
    Search for articles by this author
  • Till Dembek
    Affiliations
    University of Cologne, Faculty of Medicine and University Hospital Cologne, Department of Neurology, Cologne, Germany
    Search for articles by this author
  • Marc Tittgemeyer
    Affiliations
    Max-Planck-Institute for Metabolism Research, Cologne, Germany

    Cologne Cluster of Excellence in Cellular Stress and Aging associated Disease (CECAD), Cologne, Germany
    Search for articles by this author
  • Daniel Huys
    Affiliations
    University of Cologne, Faculty of Medicine and University Hospital Cologne, Department of Psychiatry and Psychotherapy, Cologne, Germany

    LVR Hospital Bonn, Department of Psychiatry and Psychotherapy III, Bonn, Germany
    Search for articles by this author
  • Veerle Visser-Vandewalle
    Affiliations
    LVR Hospital Bonn, Department of Psychiatry and Psychotherapy III, Bonn, Germany
    Search for articles by this author
  • Ningfei Li
    Affiliations
    Movement Disorders and Neuromodulation Unit, Department for Neurology, Charité – University Medicine Berlin, Germany
    Search for articles by this author
  • Laura Wehmeyer
    Affiliations
    University of Cologne, Faculty of Medicine and University Hospital Cologne, Department of Psychiatry and Psychotherapy, Cologne, Germany

    University of Cologne, Faculty of Medicine and University Hospital Cologne, Department of Stereotactic and Functional Neurosurgery, Cologne, Germany
    Search for articles by this author
  • Michael Barbe
    Affiliations
    University of Cologne, Faculty of Medicine and University Hospital Cologne, Department of Neurology, Cologne, Germany
    Search for articles by this author
  • Jens Kuhn
    Affiliations
    University of Cologne, Faculty of Medicine and University Hospital Cologne, Department of Psychiatry and Psychotherapy, Cologne, Germany

    Department of Psychiatry, Psychotherapy, and Psychosomatics, Johanniter Hospital Oberhausen, Oberhausen, Germany
    Search for articles by this author
  • Juan Carlos Baldermann
    Correspondence
    Corresponding Author: Dr. Juan Carlos Baldermann, University of Cologne, Faculty of Medicine and University Hospital Cologne, Department of Neurology, Kerpener Straße 62, 50937 Cologne, GermanyPhone: +49 221 478 98844
    Affiliations
    University of Cologne, Faculty of Medicine and University Hospital Cologne, Department of Psychiatry and Psychotherapy, Cologne, Germany

    University of Cologne, Faculty of Medicine and University Hospital Cologne, Department of Neurology, Cologne, Germany
    Search for articles by this author
Open AccessPublished:October 27, 2022DOI:https://doi.org/10.1016/j.bpsc.2022.10.005

      Abstract

      Background

      Deep brain stimulation of the anterior limb of the internal capsule/nucleus accumbens (ALIC/NAc) is an effective treatment in patients with obsessive-compulsive disorder but may increase impulsive behavior. We aimed to investigate how active stimulation alters subdomains of impulsive decision-making and whether respective effects depend on the location of stimulation sites.

      Methods

      We assessed fifteen participants with obsessive-compulsive disorder performing the Cambridge Gambling Task during active and inactive ALIC/NAc deep brain stimulation. Specifically, we determined stimulation-induced changes in risk adjustment and delay aversion. To characterize underlying neural pathways, we computed probabilistic stimulation maps and applied fiber filtering based on normative structural connectivity data to identify “hot” and “cold” spots/fibers related to changes in impulsive decision-making.

      Results

      Active stimulation significantly reduced risk adjustment while increasing delay aversion, both implying increased impulsive decision-making. Changes in risk adjustment were robustly associated with stimulation sites located in the central ALIC and fibers connecting the thalamus and subthalamic nucleus with the medial and lateral prefrontal cortex. Both hot spots and fibers for changes in risk adjustment were robust to leave-one-out cross-validation. Changes in delay aversion were similarly associated with central ALIC stimulation but validation hereof was non-significant.

      Conclusion

      Our findings provide experimental evidence that ALIC/NAc stimulation increases impulsive decision-making in obsessive compulsive-disorder. We show that changes in risk adjustment depend on the location of stimulation volumes and affected fiber bundles. The relationship between impulsive decision-making and clinical long-term outcomes requires further investigations.

      Key Words

      Introduction

      Deep brain stimulation (DBS) targeting the anterior limb of the internal capsule (ALIC) and the nucleus accumbens (NAc) is an effective and generally well tolerated treatment option for severely affected patients with obsessive-compulsive disorder (OCD) (
      • Huys D.
      • Kohl S.
      • Baldermann J.C.
      • Timmermann L.
      • Sturm V.
      • Visser-Vandewalle V.
      • Kuhn J.
      Open-label trial of anterior limb of internal capsule–nucleus accumbens deep brain stimulation for obsessive-compulsive disorder: insights gained.
      ,
      • Denys D.
      • Graat I.
      • Mocking R.
      • de Koning P.
      • Vulink N.
      • Figee M.
      • et al.
      Efficacy of Deep Brain Stimulation of the Ventral Anterior Limb of the Internal Capsule for Refractory Obsessive-Compulsive Disorder: A Clinical Cohort of 70 Patients.
      ). Increase in impulsive behavior has been repeatedly reported as a side effect of ALIC/NAc-DBS (
      • Denys D.
      • Graat I.
      • Mocking R.
      • de Koning P.
      • Vulink N.
      • Figee M.
      • et al.
      Efficacy of Deep Brain Stimulation of the Ventral Anterior Limb of the Internal Capsule for Refractory Obsessive-Compulsive Disorder: A Clinical Cohort of 70 Patients.
      ,
      • Alonso P.
      • Cuadras D.
      • Gabriëls L.
      • Denys D.
      • Goodman W.
      • Greenberg B.D.
      • et al.
      Deep brain stimulation for obsessive-compulsive disorder: A meta-analysis of treatment outcome and predictors of response.
      ). Despite its clinical relevance, the neuropsychological and -physiological underpinnings of DBS-induced impulsivity are poorly understood.
      Impulsivity is a psychopathological feature in a multitude of neuropsychiatric disorders including addiction, attention deficit hyperactivity disorder, mania and impulse control disorders. Moreover, impulsivity is a multidimensional construct and alterations in different components (e.g., interference control, decision-making and response inhibition) may lead to increased impulsive behavior (
      • Basar K.
      • Sesia T.
      • Groenewegen H.
      • Steinbusch H.W.M.
      • Visser-Vandewalle V.
      • Temel Y.
      Nucleus accumbens and impulsivity.
      ). Among those facets of impulsivity, impulsive decision-making is characterized by insufficient information accumulation and/or premature responding. The Cambridge Gambling Task (
      • Rogers R.D.
      • Owen a M.
      • Middleton H.C.
      • Williams E.J.
      • Pickard J.D.
      • Sahakian B.J.
      • Robbins T.W.
      Choosing between small, likely rewards and large, unlikely rewards activates inferior and orbital prefrontal cortex.
      ) is a well-established decision-making task that allows researchers to assess these features of impulsive decision-making, here termed risk adjustment and delay aversion. Risk adjustment corresponds to the ability to adapt gambling choices by taking into account the win probabilities (
      • Czapla M.
      • Simon J.J.
      • Richter B.
      • Kluge M.
      • Friederich H.-C.
      • Herpertz S.
      • et al.
      The impact of cognitive impairment and impulsivity on relapse of alcohol-dependent patients: implications for psychotherapeutic treatment.
      ). Hence, a person that gambles a lot of money regardless of the given odds shows decreased risk adjustment, corresponding to increased impulsive behavior. Delay aversion, in turn, is reflected by choosing prematurely and thereby neglecting the possibility to earn greater rewards or to avoid greater losses (
      • Czapla M.
      • Simon J.J.
      • Richter B.
      • Kluge M.
      • Friederich H.-C.
      • Herpertz S.
      • et al.
      The impact of cognitive impairment and impulsivity on relapse of alcohol-dependent patients: implications for psychotherapeutic treatment.
      ). Therefore, increased delay aversion reflects increased impulsive behavior.
      Impulsive decision-making has been linked to specific prefrontal cortex - basal ganglia networks (
      • Kim B.S.
      • Im H.I.
      The role of the dorsal striatum in choice impulsivity.
      ,
      • Basar K.
      • Sesia T.
      • Groenewegen H.
      • Steinbusch H.W.M.
      • Visser-Vandewalle V.
      • Temel Y.
      Nucleus accumbens and impulsivity.
      ). A study by Newcombe et al. utilized diffusion tensor imaging in patients after traumatic brain injury to disentangle distinctive networks associated with subcomponents of the Cambridge Gambling Task (
      • Newcombe V.F.J.
      • Outtrim J.G.
      • Chatfield D.A.
      • Manktelow A.
      • Hutchinson P.J.
      • Coles J.P.
      • et al.
      Parcellating the neuroanatomical basis of impaired decision-making in traumatic brain injury.
      ). While changes in risk adjustment were associated with alterations in the dorsal striatum and thalamus, changes in delay aversion correlated with alterations in the orbitofrontal/ventromedial prefrontal cortex and caudate nucleus. Fujihara et al. further demonstrated that both risk adjustment and delay aversion were negatively correlated with glutamate/glutamine and γ-aminobutyric acid concentration in the anterior cingulate cortex (
      • Fujihara K.
      • Narita K.
      • Suzuki Y.
      • Takei Y.
      • Suda M.
      • Tagawa M.
      • et al.
      Relationship of γ-aminobutyric acid and glutamate + glutamine concentrations in the perigenual anterior cingulate cortex with performance of Cambridge Gambling Task.
      ). These findings highlight the functional importance of specific prefrontal cortex–basal ganglia networks for regulating impulsive behavior (
      • Ullsperger M.
      • Danielmeier C.
      • Jocham G.
      Neurophysiology of Performance Monitoring and Adaptive Behavior.
      ). Thus, DBS of the ALIC/NAc may interfere with impulsive decision-making via modulation of aforementioned networks.
      To test how ALIC/NAc DBS influences impulsive decision-making and whether such changes depend on the individual location of stimulation sites, we employed the Cambridge Gambling Task in OCD patients under active and inactive DBS. We expected that active ALIC/NA DBS increases impulsive decision-making and that behavioral changes can be linked to modulation of specific neural pathways.

      Methods and Materials

      Participants

      Fifteen patients with treatment-refractory OCD (age: 44.4 ±14.21, nine female) who participated in a clinical trial to assess the clinical effectiveness of ALIC/NAc DBS in OCD (
      • Huys D.
      • Kohl S.
      • Baldermann J.C.
      • Timmermann L.
      • Sturm V.
      • Visser-Vandewalle V.
      • Kuhn J.
      Open-label trial of anterior limb of internal capsule–nucleus accumbens deep brain stimulation for obsessive-compulsive disorder: insights gained.
      ) were included in this study. All subjects underwent bilateral stereotactic implantation of quadripolar leads (Model 3387 or 3389 DBS Lead; Medtronic; Minneapolis, MN, USA) under local or general anesthesia guided by preoperative MRI and stereotactic cerebral CT. Targets were identified using Atlas for Stereotaxy of the Human Brain by Schaltenbrand, Wahren and Hassler (

      Schaltenbrand G, Wahren W, Hassler RG (1977): Atlas for Stereotaxy of the Human Brain, 2nd ed. Stuttgart: Thieme.

      ). Stereotactic planning was performed using the STP 3.5 software (Howmedica Leibinger, Freiburg, Germany). The two most distal contacts (i.e. contact 0 and 1 on the left side, contact 8 and 9 on the right side) were placed bilaterally in the posterior border of the Nac. The two most dorsal contacts (contacts 2 and 3 on the left side, 10 and 11 on the right side) were placed in the ventral part of the ALIC (Figure 1 and supplemental Figure 1). All participants were informed about the study goals and gave oral and written consent to participation. The study was approved by the local research ethics committee of the University of Cologne (No. 09-155) and was conducted in accordance with the Declaration of Helsinki. Demographic data are presented in Table 1.
      Figure thumbnail gr1
      Figure 1Reconstruction of electrodes, task design and behavior. (A) Displayed are reconstructed electrodes in standard space for group analysis. Given is the front view overlaid on a 7 Tesla ultra-high resolution MRI coronal slice (Edlow et al., 2019) at y = 5. Nucleus accumbens is shown in yellow, taken from (
      • Pauli W.M.
      • Nili A.N.
      • Michael Tyszka J.
      Data Descriptor: A high-resolution probabilistic in vivo atlas of human subcortical brain nuclei.
      ). (B) Individual trial of the Cambridge Gambling Task. First, the participant observes the given odds (e.g. 80% chance to win for red) and subsequently chooses a color and the amount of points to gamble. Then, the outcome of points won/lost is presented. For further explanation please refer to the behavioral assessment paragraph. (C) Behavioral changes in risk adjustment (left) and delay aversion (right) for the DBS-OFF and DBS-ON. Asterisks indicate significant changes (p < 0.05). (D) Explorative depiction of the outcomes of the Cambridge Gambling Task. Displayed are the averaged percentages during DBS-ON (red) and DBS-OFF (blue) of points gambled in dependence of the indicated odds (e.g. 9:1 = 0.9, 8:2 = 0.8 etc.). Error bars indicate standard errors.
      Table 1Demographic and clinical characteristics of patient cohort. Yale-Brown Obsessive Compulsive Scale (YBOCS).
      IDAge (years)GenderYBOCS before surgeryYBOCS at time of studyStimulation settings at time of study
      132F28321-,2-,c+/9-,10-,c+; 5.5V; 150μs: 130Hz
      232F34302-,3-,c+/10-,11-,c+; 2.5V; 180μs: 90Hz
      336F33342-,3-,c+/10-,11-,c+; 4.2V; 130μs: 120Hz
      444F34202-,3-,c+/10-,11-,c+; 4V; 150μs; 130Hz
      529F33220-1,1-,c+/8-,9-,c+; 5.5V; 150μs; 130Hz
      660F25202-,3-,c+/10-,11-,c+; 4.5V; 90μs; 130Hz
      725M35242-,3-,c+/10-,11-,c+; 2.6V; 120μs; 150Hz
      849F32272-,3-,c+/10-,11-,c+; 4.7V; 150μs; 130Hz
      934F37332-,3-,c+/10-,11-,c+; 4.3V; 120μs; 150Hz
      1056M35252-,3-,c+/10-,11-,c+; 5.2V; 120μs; 130Hz
      1159M36280-,1-,2-,c+/8-,9-,10-; 5.1V; 150μs; 130Hz
      1227M26160-,1-,c+/8-,9-,; 6.5V; 150μs; 130Hz
      1364M25172-,3-,c+/10-,11-,c+; 4V; 120μs; 130Hz
      1447M28152-, 3 -,c+/10-,11-,c+; 3.3V; 130 Hz; 120μs
      1546M26291-, 2-, 3 -,c+/9-, 10-,11-,c+; 4.5V; 130 Hz; 120μs
      YBOCS = Yale-Brown Obsessive Compulsive Scale.

      Behavioral assessment

      Participants performed the Cambridge Gambling Task (Figure 1 B) as implemented in the Cambridge Neuropsychological Automated Test Battery (CANTAB, Cambridge Cognition Ltd.). Participants were presented with 10 boxes, with a token hidden in one of them. The boxes were colored in either red or blue (e.g., 7 blue and 3 red), which represents the outcome probabilities of the current trial. First, participants had to choose a color (Figure 1 B). Subsequently, they had to choose the amount of points to wage on the chosen color. The amount of points to wage started either low from 5 % and increased (ascending condition) or started high from 95 % and decreased (descending condition) over time (Figure 1 B). If the token showed up in a box of the chosen color, the waged points were added to the total score. Otherwise, the amount of points was deducted from the total score. We assessed the two main behavioral parameters of the Cambridge Gambling Task related to impulsive decision-making: risk adjustment and delay aversion as calculated in the CANTAB software. In detail, to assess the risk adjustment score, the proportion of current points that the participant bets are calculated, which are then merged across the different trials in dependence of the ratio of boxes (2x for 9:1 + 1x for 8:1 – 1x for 7:3 – 2x for 6:4) (
      • Fujihara K.
      • Narita K.
      • Suzuki Y.
      • Takei Y.
      • Suda M.
      • Tagawa M.
      • et al.
      Relationship of γ-aminobutyric acid and glutamate + glutamine concentrations in the perigenual anterior cingulate cortex with performance of Cambridge Gambling Task.
      ). A higher risk adjustment score indicates that the participant adapted the wager more to the probability of outcomes. The delay aversion score is calculated as the difference of risk taking between the ascending and the descending conditions. Risk taking is calculated as the mean proportion (0 – 1) of current points gambled over all trials in which the ratio between colors was 9:1, 8:2, 7:3 or 6:4 and the participant chose the majority box color.
      Assessments during active (DBS-ON) and inactive (DBS-OFF) stimulation were performed on two consecutive days in an open-label counterbalanced pseudo-randomized design (the order of recording sessions was switched for each successive participant). Prior to DBS-OFF measurements stimulation was switched off for at least 12 hours. DBS was activated again immediately after DBS-OFF recordings. When DBS-OFF recordings were first, this ensured 24 h of stimulation before DBS-ON recordings. Patients’ data were collected at 6 (n = 9) or 12 months (n = 6) post-surgery. One patient did not tolerate DBS-OFF due to anxiety and was removed from the sample. Statistical analyses were performed with SPSS 25 (IBM Corp., New York, NY, USA). Behavioral results were analyzed using a repeated measures multivariate general linear model with main effect stimulation (DBS-ON, DBS-OFF) and risk adjustment and delay aversion as dependent variables. Of note, increased impulsive decision-making is reflected by lower risk adjustment values and higher delay aversion values. Thus, for the general linear model the effect of risk adjustment was inversed. Planned post-hoc comparisons for DBS-ON and DBS-OFF were performed using two-tailed paired t-tests. As a potential confounder, we compared DBS-induced changes between participants who were tested first in DBS-OFF and second DBS-ON (n = 7) with those tested vice-versa (n = 7) using two-tailed independent-sample t-tests. Similar, to explore whether the time to follow-up influences DBS induced changes, we compared participants assessed at six months follow-up with those assessed at twelve months follow-up using two-sample t-tests. To test the overall ability to perform the paradigm, we performed post-hoc analysis of the quality of decision-making and deliberation time. We calculated the quality of decision-making as the proportion (0 – 1) of all trials where the subject chose the majority box color and the number of boxes of each color differed. Deliberation time is the decision latency time in ms between the presentation of the boxes and the bet choice.

      Reconstruction of electrodes and volumes of activated tissue

      We reconstructed electrode locations and calculated patient-specific volumes of activated tissue (VTAs) based on individual stimulation parameters using the Lead-DBS toolbox (
      • Horn A.
      • Li N.
      • Dembek T.A.
      • Kappel A.
      • Boulay C.
      • Ewert S.
      • et al.
      Lead-DBS v2: Towards a comprehensive pipeline for deep brain stimulation imaging.
      ) following the pipeline described in (
      • Horn A.
      • Reich M.
      • Vorwerk J.
      • Li N.
      • Wenzel G.
      • Fang Q.
      • et al.
      Connectivity Predicts deep brain stimulation outcome in Parkinson disease.
      ). Briefly, we co-registered postoperative computer tomography scans on preoperative magnetic resonance imaging using advanced normalization tools (ANTs; (
      • Avants B.B.
      • Tustison N.
      • Song G.
      Advanced Normalization Tools (ANTS).
      ). In all patients, an additional subcortical refinement for potential brain shift was added using the “coarse mask” setting within Lead-DBS. Images were then nonlinearly normalized into standard space (ICBM 2009b NLIN, Asym), again using the ANTs tool. Of note, the process of normalization is essential to allow subsequent group analysis. To reduce potential bias due to inaccurate normalization, which may result in inaccurate anatomical alignment, we applied the SyN approach within the ANTs algorithm with nonlinear subcortical refinement which showed best performance when comparing different subcortical image registrations algorithms (
      • Klein A.
      • Andersson J.
      • Ardekani B.A.
      • Ashburner J.
      • Avants B.
      • Chiang M.C.
      • et al.
      Evaluation of 14 nonlinear deformation algorithms applied to human brain MRI registration.
      ), and all steps were visually inspected by experienced scientists in the field of imaging (JCB and TS).
      We reconstructed DBS electrodes with the PaCER-algorithm (
      • Husch A.
      • Petersen M V.
      • Gemmar P.
      • Goncalves J.
      • Hertel F.
      PaCER - A fully automated method for electrode trajectory and contact reconstruction in deep brain stimulation.
      ). If required, electrodes were manually refined as suggested in the Lead-DBS software. We estimated VTAs by employing a Finite Element Method model that includes gray and white matter as well as metal and insulation parts of the electrode (
      • Horn A.
      • Reich M.
      • Vorwerk J.
      • Li N.
      • Wenzel G.
      • Fang Q.
      • et al.
      Connectivity Predicts deep brain stimulation outcome in Parkinson disease.
      ). Gray and white matter areas were determined using the Atlas of Human Subcortical Brain Nuclei including the caudate nucleus, NAc and globus pallidus (
      • Pauli W.M.
      • Nili A.N.
      • Michael Tyszka J.
      Data Descriptor: A high-resolution probabilistic in vivo atlas of human subcortical brain nuclei.
      ). The distribution of electric fields was simulated using an adaptation of the FieldTrip – SimBio pipeline (
      • Vorwerk J.
      • Oostenveld R.
      • Piastra M.C.
      • Magyari L.
      • Wolters C.H.
      The FieldTrip-SimBio pipeline for EEG forward solutions.
      ) integrated in Lead-DBS. To this end, a volume conductor model field is computed by derivation of the finite element method on a four-compartment mesh describing local gray and white matter (defined by a subcortical atlas of the basal ganglia (
      • Pauli W.M.
      • Nili A.N.
      • Michael Tyszka J.
      Data Descriptor: A high-resolution probabilistic in vivo atlas of human subcortical brain nuclei.
      ) as well as electrode contact and insulating material. The electric field is then estimated as the derivative of the voltage distribution across the tissue surrounding the electrode. The electric field gradients were thresholded in dependence of the individually applied pulse width using a heuristic approximation suggested by Proverbio & Husch (

      Proverbio D, Husch A (n.d.): ApproXON: Heuristic Approximation to the E-Field-Threshold for Deep Brain Stimulation Volume-of-Tissue-Activated Estimation. https://doi.org/10.1101/863613

      ) based on data by Åström et al. (
      • Åström M.
      • Diczfalusy E.
      • Martens H.
      • Wårdell K.
      Relationship between neural activation and electric field distribution during deep brain stimulation.
      ). The resulting VTAs in patient’s native space were then transformed into standard space for group analyses.

      Probabilistic stimulation maps and hot/cold spots

      Modelling of probabilistic stimulation maps followed the methodology described in previous publications (
      • Dembek T.A.
      • Roediger J.
      • Horn A.
      • Reker P.
      • Oehrn C.
      • Dafsari H.S.
      • et al.
      Probabilistic sweet spots predict motor outcome for deep brain stimulation in Parkinson disease.
      ,
      • Reich M.M.
      • Horn A.
      • Lange F.
      • Roothans J.
      • Paschen S.
      • Runge J.
      • et al.
      Probabilistic mapping of the antidystonic effect of pallidal neurostimulation: a multicentre imaging study.
      ). First, each pair of VTA (left and right) was weighted by the respective behavioral outcome, i.e., the difference in risk adjustment/delay aversion between DBS-ON and DBS-OFF. For each voxel, we determined the mean behavioral outcome by averaging the weighted VTAs covering the respective voxel. To control for outliers, only voxels that were overlapped by at least 3 VTAs (minimum 20 % of VTAs) were used for subsequent analyses. Then, for each voxel, we performed a voxel-wise t-test, comparing the average outcome of the VTAs covering the respective voxel with those VTAs not covering it, resulting in a t-map that included voxel-wise t-statistics. Further following the approach by Reich et al. (
      • Reich M.M.
      • Horn A.
      • Lange F.
      • Roothans J.
      • Paschen S.
      • Runge J.
      • et al.
      Probabilistic mapping of the antidystonic effect of pallidal neurostimulation: a multicentre imaging study.
      ), t-maps were thresholded to only include voxels with a p-value < 0.05. The remaining voxels represent “hot” and “cold” spots indicating the location associated most with DBS-induced behavioral changes. Of note, we chose not to correct for the size of the VTAs or stimulation amplitude as it has been proposed in previous publications (
      • Elias G.J.B.
      • Boutet A.
      • Joel S.E.
      • Germann J.
      • Gwun D.
      • Neudorfer C.
      • et al.
      Probabilistic Mapping of Deep Brain Stimulation: Insights from 15 Years of Therapy.
      ,
      • Dembek T.A.
      • Roediger J.
      • Horn A.
      • Reker P.
      • Oehrn C.
      • Dafsari H.S.
      • et al.
      Probabilistic sweet spots predict motor outcome for deep brain stimulation in Parkinson disease.
      ), since larger and less focal VTAs may be especially relevant for stimulation-induced impulsivity.
      We validated the obtained probabilistic hot spots in a leave-one-out cross-validation. To this end, we repeated the aforementioned procedure 14 times, each time leaving out one subject’s pair of VTAs leading to 14 probabilistic hot spots with each missing one subject. These hot/cold spots were then compared with the VTAs of the left-out participant by averaging the t-values of all overlapping voxels, resulting in a single hot spot score. If a subject’s VTA covered parts of the probabilistic hot spot with high values, this resulted in a higher score. If a subject’s VTA did not encompass the hot spot, or only parts with lower values, this resulted in a lower hot spot score. Finally, the individual hot spot scores were Spearman rank-correlated with the respective behavioral outcome to determine significance (p < 0.05) of the validation process.

      Stimulation dependent connectivity analysis

      To complement the hot spot analysis, we computed a tractographic model for each behavioral parameter following a previously introduced fiber-filtering approach (
      • Irmen F.
      • Horn A.
      • Mosley P.
      • Perry A.
      • Petry-Schmelzer J.N.
      • Dafsari H.S.
      • et al.
      Left prefrontal impact links subthalamic stimulation with depressive symptoms.
      ,
      • Baldermann J.C.
      • Melzer C.
      • Zapf A.
      • Kohl S.
      • Timmermann L.
      • Tittgemeyer M.
      • et al.
      Connectivity profile predictive of effective deep brain stimulation in obsessive compulsive disorder.
      ,
      • Li N.
      • Baldermann J.C.
      • Kibleur A.
      • Treu S.
      • Akram H.
      • Elias G.J.B.
      • et al.
      A unified connectomic target for deep brain stimulation in obsessive-compulsive disorder.
      ) using the Fiber Filtering Explorer within the Lead-DBS toolbox (V. 2.5; www.lead-dbs.org) (
      • Horn A.
      • Li N.
      • Dembek T.A.
      • Kappel A.
      • Boulay C.
      • Ewert S.
      • et al.
      Lead-DBS v2: Towards a comprehensive pipeline for deep brain stimulation imaging.
      ). To this end, we employed a normative whole-brain structural connectome derived from multi-shell diffusion-weighted data collected in 985 healthy subjects from the Human Connectome Project (
      • Van Essen D.C.
      • Ugurbil K.
      • Auerbach E.
      • Barch D.
      • Behrens T.E.J.
      • Bucholz R.
      • et al.
      The Human Connectome Project: A data acquisition perspective.
      ) which was previously reconstructed within the Lead-DBS toolbox (for details of the processing pipeline see (
      • Li N.
      • Baldermann J.C.
      • Kibleur A.
      • Treu S.
      • Akram H.
      • Elias G.J.B.
      • et al.
      A unified connectomic target for deep brain stimulation in obsessive-compulsive disorder.
      )). This resulted in an aggregated dataset of 6,000,000 fibers in standard space. For each participant, those fibers travelling through his or her VTAs were filtered and labelled as “connected”. Thus, for each participant and each fiber of the connectome, we determined whether the VTAs were connected to the fiber or not. Like the voxel-wise analysis, only fibers that were connected to at least 20 % of VTAs were included to control for outliers. For each fiber, we then performed a t-test between the outcome (i.e. changes in risk adjustment or delay version) of connected VTAs and unconnected VTAs. Each fiber thus contained a t-statistic, here referred to as fiber score. A higher fiber score implies a stronger relationship between changes in behavior and modulation of this particular fiber. Similar to the voxel-wise analysis, we included only fibers with p < 0.05 for subsequent validation processes.
      To validate the resulting tractographic model, we again performed leave-one-out cross-validation. Resembling the voxel-wise approach, each model was re-calculated without the data of one participant at a time. We then calculated the mean fiber score of this subject’s fibers based on the model derived from the remaining subjects. A higher mean fiber score indicates similarity between the left-out subject’s modulated fibers and the leave-one-out model. This procedure was repeated for all participants and resulting averaged fiber scores were Spearman rank-correlated with the empirical behavioral change to assess significance (p < 0.05).

      Results

      Choices were significantly modulated by DBS indicated by a main effect of stimulation (F1, 13 = 14.905; ηp2 = .534, p = 0.002), a main effect of the dependent variables risk adjustment and delay aversion (F1, 13 = 25.944, ηp2 = .666, p < 0.001) and a significant interaction effect (F1, 13 = 9.005, ηp2 = .409, p = 0.010). Planned post-hoc comparisons revealed significantly decreased risk adjustment (DBS-ON: -0.80 (mean) ± .18 (standard error); DBS-OFF: -1.30 ± 0.17; t13 = 3.625, d = .969, p = 0.003) and increased delay aversion (DBS-ON 0.19 ± 0.06; DBS-OFF: 0.08 ± 0.05; t13 = 2.500, d = .668, p = 0.027), both pointing towards increased impulsivity (Figure 1). There was no difference in DBS-induced changes between patients tested first in DBS-ON, then -OFF or patients tested first in DBS-OFF, then -ON regarding risk adjustment (t12 = 0.006, p = 0.995) and delay aversion adjustment (t12 = -0.797, p = 0.441). Comparing patients that were assessed at six months follow-up with those assessed at twelve months follow-up revealed no significant differences for changes in risk adjustment (t12 = -1.025, p = 0.326) or delay aversion t12 = 0.720, p = 0.485). An exploratory post-hoc depiction of the outcomes of the Cambridge Gambling Task revealed that risk adjustment was altered in the sense that participants showed less reduction of points gambled in the DBS-ON compared to DBS-OFF condition when odds became more even (Figure 1). Risk adjustment and delay aversion scores did not correlate significantly with each other during DBS-ON (r = - .134, p = 0.477) or DBS-OFF (r = - .011, p = 0.956). We found no significant correlation between changes in clinical outcomes (defined as Yale-Brown Obsessive-Compulsive Scale (YBOCS) changes between assessments before surgery and at time of the experiment) with changes in risk adjustment (r = - .114, p = 0.581) or delay aversion (r = - .0.023, p = 0.912). Likewise, changes in depression symptoms, assessed with the Beck’s Depression Inventory, did not significantly correlate with changes in risk adjustment (r = .250, p = 0.409) or delay aversion (r = - .014, p = 0.974). In a post-hoc analysis, we tested whether participants were generally impaired to handle the paradigm. We found that both quality of decision making (t13 = 0.388, p = 0.704) and deliberation times (t13 = 0.271, p = 0.791) were similar between DBS-ON and DBS-OFF (see suppl. Figure 2).
      Figure thumbnail gr2
      Figure 2Probabilistic stimulation maps and discriminative fibers for risk adjustment. (A) Displayed are the t-maps that indicate decreased risk adjustment with active stimulation (warm colors) or increased risk adjustment (cold colors). Outlined in red are hot and cold spots, i.e. voxels that discriminated between decreased and increased risk adjustment (p < 0.05). Maps are overlaid on coronal (top row), axial (bottom left) and sagittal (bottom right) slices of a 7 Tesla ultra-high resolution MRI dataset (
      • Edlow B.L.
      • Mareyam A.
      • Horn A.
      • Polimeni J.R.
      • Witzel T.
      • Tisdall M.D.
      • et al.
      7 Tesla MRI of the ex vivo human brain at 100 micron resolution.
      ). MNI coordinates and directions are given in white. The outline of the nucleus accumbens is shown as a dashed line. (B) Displayed are fiber tracts discriminative for changes in risk adjustment colored by respective t-values (p < 0.05). Red fibers indicate decreased risk adjustment, while blue fibers indicate increased risk adjustment. Displayed are the caudate nucleus (blue), nucleus accumbens (yellow), subthalamic nucleus (purple) and thalamus (green). Fibers are overlaid on sagittal (left column), sagittal/axial (top right) and axial (bottom right) slices of a 7 Tesla ultra-high resolution MRI dataset (
      • Edlow B.L.
      • Mareyam A.
      • Horn A.
      • Polimeni J.R.
      • Witzel T.
      • Tisdall M.D.
      • et al.
      7 Tesla MRI of the ex vivo human brain at 100 micron resolution.
      ). MNI coordinates and directions are given in white. Scatterplots depict the respective leave-one-out cross-validation. The black line represents a linear fit.
      For risk adjustment, probabilistic stimulation maps revealed a ventral-dorsal gradient for DBS-induced changes, indicating that more dorsally located stimulation sites were associated with decreased risk adjustment (Figure 2A). The resulting risk adjustment hot spot was located within the ventral/central ALIC and encompassed the anterior globus pallidus internus. The cold spots - suggesting increases in risk adjustment - were located more ventrally and medially partly overlapping the ventral striatum. Validation of the risk adjustment hot and cold spots using leave-one-out cross-validation was significant (r = 0.578; p = 0.028).
      Congruent with the probabilistic stimulation maps, the fiber filtering approach primarily revealed discriminative fibers within the ALIC associated with decreased risk adjustment (Figure 2B). These fibers were mainly located in the right hemisphere and connected the medial and lateral prefrontal cortex with the thalamus and subthalamic nucleus. From here, additional fibers ascended to cortical motor/premotor areas. Furthermore, fibers outside the ALIC that were associated with increased risk adjustment were located anterior and ventral to the NAc. The tractographic model of hot and cold fibers for changes in risk adjustment was robust to leave-one-out cross-validation (r = 0.565; p = 0.016). Changing the applied threshold of VTAs connected to each fiber to 10 % or 30 % did not alter the results, both resulting in a significant leave-one-out cross-validation (see suppl. Figure 3).
      Figure thumbnail gr3
      Figure 3Probabilistic stimulation maps and discriminative fibers for delay aversion. (A) Displayed are the t-maps that indicate increased delay aversion with active stimulation (warm colors) or decreased delay aversion (cold colors). Outlined in red are hot and cold spots, i.e., voxels that discriminated between decreased and increased delay aversion (p < 0.05). Maps are overlaid on coronal (top row), axial (bottom left) and sagittal (bottom right) slices of a 7 Tesla ultra-high resolution MRI dataset (
      • Edlow B.L.
      • Mareyam A.
      • Horn A.
      • Polimeni J.R.
      • Witzel T.
      • Tisdall M.D.
      • et al.
      7 Tesla MRI of the ex vivo human brain at 100 micron resolution.
      ). MNI coordinates and directions are given in white. The outline of the nucleus accumbens is shown as a dashed line. (B) Displayed are fiber tracts discriminative for changes in delay aversion colored by respective t-values (p < 0.05). Red fibers indicate increased delay aversion, while no fibers associated with decreased delay aversion were apparent. Displayed are the caudate nucleus (blue), nucleus accumbens (yellow), subthalamic nucleus (purple) and the thalamus (green). Fibers are overlaid on sagittal (top left, bottom left), sagittal/axial (top right) and axial (bottom right) slices of a 7 Tesla ultra-high resolution MRI dataset (
      • Edlow B.L.
      • Mareyam A.
      • Horn A.
      • Polimeni J.R.
      • Witzel T.
      • Tisdall M.D.
      • et al.
      7 Tesla MRI of the ex vivo human brain at 100 micron resolution.
      ). MNI coordinates and directions are given in white. Scatterplots depict the respective leave-one-out cross-validation. The black line represents a linear fit.
      Probabilistic stimulation maps of delay aversion showed a ventral-to-dorsal gradient for increased delay aversion, comparable to the findings for risk adjustment (Figure 3A). The hot spot was similarly located in the dorsal stimulation volumes but was less widespread than the risk adjustment hot spot. Corresponding voxels were also located within the ventral/central ALIC, with a more medial manifestation in proximity to the bed nucleus of the stria terminalis (Figure 3). Validation of hot and cold spots for delay aversion was statistically non-significant (r = 0.432; p = 0.059). The complementary fiber filtering approach revealed fibers within the left ALIC to be associated with increase in delay aversion (Figure 3B). Validation of the tractographic model for delay aversion fibers was non-significant (r = 0.170, p = 0.301). Changing the applied threshold of VTAs connected to each fiber to 10 % or 30 % did not alter the results, both resulting in a non-significant leave-one-out cross-validation (see suppl. Figure 3).

      Discussion

      In this study we found experimental evidence that DBS of the ALIC/NAc increases impulsive decision-making as indicated by reduced risk adjustment and increased delay aversion in patients with severe OCD. We further showed that DBS-induced changes in risk adjustment depend on the localization of ALIC stimulation sites.

      DBS-induced impulsivity in Obsessive-Compulsive Disorder

      In our study, DBS of the ALIC/NAc region resulted in increased impulsive decision-making across differentiable subdomains, while overall task performance (i.e., quality of decision making and deliberation times) were unaltered. Our results extend previous findings showing a shift towards risky decision-making in four patients with stimulation of the NAc (
      • Nachev P.
      • Lopez-Sosa F.
      • Gonzalez-Rosa J.J.
      • Galarza A.
      • Avecillas J.
      • Pineda-Pardo J.A.
      • et al.
      Dynamic risk control by human nucleus accumbens.
      ) and increased decision impulsivity in patients with ALIC/NAc DBS compared to healthy controls (
      • Grassi G.
      • Figee M.
      • Ooms P.
      • Righi L.
      • Nakamae T.
      • Pallanti S.
      • et al.
      Impulsivity and decision-making in obsessive-compulsive disorder after effective deep brain stimulation or treatment as usual.
      ). Our results further are in line with clinical studies reporting increased impulsivity as a common side effect of DBS in OCD (
      • Huys D.
      • Kohl S.
      • Baldermann J.C.
      • Timmermann L.
      • Sturm V.
      • Visser-Vandewalle V.
      • Kuhn J.
      Open-label trial of anterior limb of internal capsule–nucleus accumbens deep brain stimulation for obsessive-compulsive disorder: insights gained.
      ,
      • Denys D.
      • Graat I.
      • Mocking R.
      • de Koning P.
      • Vulink N.
      • Figee M.
      • et al.
      Efficacy of Deep Brain Stimulation of the Ventral Anterior Limb of the Internal Capsule for Refractory Obsessive-Compulsive Disorder: A Clinical Cohort of 70 Patients.
      ,
      • Luigjes J.
      • Mantione M.
      • van den Brink W.
      • Schuurman P.R.
      • van den Munckhof P.
      • Denys D.
      Deep brain stimulation increases impulsivity in two patients with obsessive-compulsive disorder.
      ). Specifically, impulsivity related adverse events (i.e., disinhibition, impulsivity and hypomanic symptoms) were present in over 25 % of OCD patients reported in a meta-analysis, constituting the most common stimulation-induced side effect (
      • Alonso P.
      • Cuadras D.
      • Gabriëls L.
      • Denys D.
      • Goodman W.
      • Greenberg B.D.
      • et al.
      Deep brain stimulation for obsessive-compulsive disorder: A meta-analysis of treatment outcome and predictors of response.
      ). Remarkably, single cases of DBS-induced transient manic episodes have been reported (
      • Haq I.U.
      • Foote K.D.
      • Goodman W.K.
      • Ricciuti N.
      • Ward H.
      • Sudhyadhom A.
      • et al.
      A Case of Mania following Deep Brain Stimulation for Obsessive Compulsive Disorder.
      ,
      • Kuhn J.
      • Lenartz D.
      • Huff W.
      • Mai J.K.
      • Koulousakis A.
      • Maarouf M.
      • et al.
      Transient manic-like episode following bilateral deep brain stimulation of the nucleus accumbens and the internal capsule in a patient with Tourette syndrome.
      ) which may represent an extreme end on the impulsivity spectrum. Our results suggest that the clinically observed increase in impulsive behavior may be associated with changes in two separable dimensions of impulsive decision-making. Both decreased risk adjustment (indicating that participants waged points with less consideration of the apparent win probabilities, especially when odds became more conflicting) and increased delay aversion (indicating that participants placed their bets prematurely, neglecting the possibility to avoid losses by waiting) are characteristics of careless and hasty decisions that are reminiscent of the reported clinical side effects.
      Notably, it has been suggested that increased impulsive decision-making may not only be considered as a side effect of DBS in OCD, but conversely that may indicate decreased compulsive behavior. Specifically, DBS of the associative-limbic of the subthalamic nucleus decreased evidence accumulation in OCD and this was proposed to relate to clinical efficacy (
      • Voon V.
      • Droux F.
      • Morris L.
      • Chabardes S.
      • Bougerol T.
      • David O.
      • et al.
      Decisional impulsivity and the associative-limbic subthalamic nucleus in obsessive-compulsive disorder: Stimulation and connectivity.
      ). Based on clinical experience, acute stimulation-induced impulsive behavior has also been proposed as a potential predictor of DBS treatment response for OCD (
      • Denys D.
      • Graat I.
      • Mocking R.
      • de Koning P.
      • Vulink N.
      • Figee M.
      • et al.
      Efficacy of Deep Brain Stimulation of the Ventral Anterior Limb of the Internal Capsule for Refractory Obsessive-Compulsive Disorder: A Clinical Cohort of 70 Patients.
      ,
      • Tsai H.C.
      • Chang C.H.
      • Pan J.I.
      • Hsieh H.J.
      • Tsai S.T.
      • Hung H.Y.
      • Chen S.Y.
      Pilot study of deep brain stimulation in refractory obsessive-compulsive disorder ethnic Chinese patients.
      ). In our study, we found a compelling effect of DBS on risk adjustment, and it is conceivable that acting independently of previously indicated odds, especially in more conflicting conditions, may be related to a decrease of compulsive behavior when faced with an obsessive fear. However, we did not find a significant correlation between DBS-induced changes in risk adjustment YBOCS scores at the time of the experiment. Importantly, the observed changes in the Cambridge Gambling Task reflect rather acute changes (12 – 24 hours DBS-OFF) while clinical efficacy of DBS can take several months to reach its full effect. In our sample, improvement rates where rather modest which is, at least partly, explainable by ongoing parameter optimization in patients included six months after surgery. Whether the applied stimulation settings associated with decreased risk adjustment will later result in a reduction of obsessions and compulsions remains to be investigated. Another potential explanation for the lack of correlation is that impulsivity may also be a sign of overdosed stimulation within the clinically effective pathway, comparable to stimulation-induced hyperkinesia in patients with Parkinson’s disease (
      • Zheng Z.
      • Li Y.
      • Li J.
      • Zhang Y.
      • Zhang X.
      • Zhuang P.
      Stimulation-induced dyskinesia in the early stage after subthalamic deep brain stimulation.
      ), which may then result in a detrimental clinical outcome (
      • Volkmann J.
      • Daniels C.
      • Witt K.
      Neuropsychiatric effects of subthalamic neurostimulation in Parkinson disease.
      ).
      We propose that future studies may investigate more in-depth if acute DBS-induced changes in impulsive decision relate to improvement of obsessions and compulsions in the long-term. This could be an important clinical tool, since testing of stimulation parameters in OCD-DBS usually does not provide direct feedback (i.e., whether a specific setting is beneficial for obsessions and compulsions) since changes in OCD symptoms often take several weeks or months of stimulation.

      Neural pathways of impulsive decision-making

      To obtain a better understanding of the underlying neural pathways of DBS-induced changes in impulsive decision-making we applied two complementary mapping methods. First, we created probabilistic maps linking the applied stimulation volumes to DBS induced changes in behavior. Second, we aimed to characterize the associated structural connectivity profiles. For risk adjustment, probabilistic stimulation maps revealed a clear gradient showing that decreases were associated with dorsal stimulation volumes. The hot spot was primarily located on the right hemisphere within the central/ventral ALIC and in parts extending to the globus pallidus internus. The tractography analysis revealed a robust relationship between changes in risk adjustment and right-hemispheric ALIC fibers. Of note, the clear laterality of this finding (as for delay aversion) should be interpreted with caution since the probabilistic stimulation maps showed mostly bilateral effects. Further, all patients were stimulated bilaterally with the same contacts and amplitudes which makes it difficult to assume definite laterality of findings. Anatomically, the fibers connect broad areas including the medial and lateral prefrontal cortex with the basal ganglia, in particular the subthalamic nucleus (STN) and thalamus. Because of the high number of false-positive connections in diffusion magnetic resonance imaging (MRI) based tractography, a precise inference of cortical projections is challenging. The localization of both the risk adjustment fibers and hot spot in relation to previously published segmentations of the ALIC (based on tracing experiments in primates and human diffusion MRI (
      • Safadi Z.
      • Grisot G.
      • Jbabdi S.
      • Behrens T.E.
      • Heilbronner S.R.
      • McLaughlin N.C.R.
      • et al.
      Functional Segmentation of the Anterior Limb of the Internal Capsule: Linking White Matter Abnormalities to Specific Connections.
      )(40)(
      • Safadi Z.
      • Grisot G.
      • Jbabdi S.
      • Behrens T.E.
      • Heilbronner S.R.
      • McLaughlin N.C.R.
      • et al.
      Functional Segmentation of the Anterior Limb of the Internal Capsule: Linking White Matter Abnormalities to Specific Connections.
      )(40)(
      • Safadi Z.
      • Grisot G.
      • Jbabdi S.
      • Behrens T.E.
      • Heilbronner S.R.
      • McLaughlin N.C.R.
      • et al.
      Functional Segmentation of the Anterior Limb of the Internal Capsule: Linking White Matter Abnormalities to Specific Connections.
      )(40)(
      • Safadi Z.
      • Grisot G.
      • Jbabdi S.
      • Behrens T.E.
      • Heilbronner S.R.
      • McLaughlin N.C.R.
      • et al.
      Functional Segmentation of the Anterior Limb of the Internal Capsule: Linking White Matter Abnormalities to Specific Connections.
      ) indicates that modulation of fibers connecting to the lateral and medial prefrontal cortex, including the anterior cingulate and ventromedial prefrontal cortex, are crucial for decreased risk adjustment. Interestingly, the hyperdirect pathway has been shown to convey a signal from medial prefrontal cortex to the STN causing an increased decision threshold (i.e., acting as a global brake to enable further evidence accumulation) in situations with conflicting response tendencies (
      • Frank M.J.
      • Samanta J.
      • Moustafa A.A.
      • Sherman3 S.J.
      Hold Your Horses: Impulsivity, Deep Brain Stimulation, and Medication in Parkinsonism.
      ,
      • Cavanagh J.F.
      • Wiecki T.V.
      • Cohen M.X.
      • Figueroa C.M.
      • Samanta J.
      • Sherman S.J.
      • Frank M.J.
      Subthalamic nucleus stimulation reverses mediofrontal influence over decision threshold.
      ,
      • Zavala B.A.
      • Tan H.
      • Little S.
      • Ashkan K.
      • Hariz M.
      • Foltynie T.
      • et al.
      Midline frontal cortex low-frequency activity drives subthalamic nucleus oscillations during conflict.
      ,
      • Zavala B.
      • Tan H.
      • Little S.
      • Ashkan K.
      • Green A.L.
      • Aziz T.
      • et al.
      Decisions Made with Less Evidence Involve Higher Levels of Corticosubthalamic Nucleus Theta Band Synchrony.
      ). Considering that DBS of the STN also results in changes in decision impulsivity, we speculate that hyperdirect fibers within the internal capsule, connecting the prefrontal cortex with the STN, may play a crucial role for mediating DBS-induced changes in risk adjustment, especially when confronted with higher conflict levels. Notably, both hot spot and hot fibers for risk adjustment were located in close proximity to a previously published fiber tract of a large cohort (N = 50) that, if stimulated, explained DBS treatment response in OCD (
      • Li N.
      • Baldermann J.C.
      • Kibleur A.
      • Treu S.
      • Akram H.
      • Elias G.J.B.
      • et al.
      A unified connectomic target for deep brain stimulation in obsessive-compulsive disorder.
      ). Thus, our study suggests that clinical improvement and changes in risk adjustment may be driven by stimulation of closely related pathways.
      Changes in delay aversion were also related to a ventral-dorsal gradient in the probabilistic stimulation maps, the subsequent validation process however only showed a trend-level effect. In contrast to risk adjustment, the fiber filtering approach revealed predominantly left-hemispheric fibers that were similarly located in the central ALIC but validation hereof was not significant. Thus, it remains open whether changes in delay aversion can be attributed to specific neural pathways.
      Several limitations of this study must be acknowledged. First, switching off DBS in patients with OCD can result in a sudden relapse of OCD symptoms but may also induce unspecific withdrawal effects including anxiety and acute mood deterioration. We cannot rule out that such effects may have influenced the behavioral performance in our study. However, switching between DBS-ON and -OFF states did not compromise general task performance, as shown by maintained quality of decision making and deliberation times. Thus, changes in risk adjustment and delay aversion were rather specific and cannot be assigned to an overall failure in task performance. Further, since a limited number of OCD patients qualify for DBS surgery the size of our sample is relatively small and therefore medium behavioral effects might have been not sufficient to facilitate findings in the imaging analysis that are robust to cross-validation. Furthermore, our imaging analyses relied on binarized VTAs to determine the modulation of fibers and brain tissue. This must be regarded as a simplified approximation of the actual impact of the electric field on neural tissue. More advanced models (
      • Noecker A.M.
      • Frankemolle-Gilbert A.M.
      • Howell B.
      • Petersen M.V.
      • Beylergil S.B.
      • Shaikh A.G.
      • McIntyre C.C.
      StimVision v2: Examples and Applications in Subthalamic Deep Brain Stimulation for Parkinson’s Disease.
      ,
      • Butenko K.
      • Bahls C.
      • Schröder M.
      • Köhling R.
      • Van Rienen U.
      OSS-DBS: Open-source simulation platform for deep brain stimulation with a comprehensive automated modeling.
      ) are currently being developed and may improve modelling of DBS effects in the near future. Also, the usage of normative connectomes disregards individual anatomy while providing high-quality streamlines with high signal-to-noise ratios that have facilitated the prediction of clinical outcome in several disorders (
      • Li N.
      • Baldermann J.C.
      • Kibleur A.
      • Treu S.
      • Akram H.
      • Elias G.J.B.
      • et al.
      A unified connectomic target for deep brain stimulation in obsessive-compulsive disorder.
      ,
      • Wang Q.
      • Akram H.
      • Muthuraman M.
      • Gonzalez-Escamilla G.
      • Sheth S.A.
      • Oxenford S.
      • et al.
      Normative vs. patient-specific brain connectivity in deep brain stimulation.
      ,

      Johnson KA, Duffley G, Foltynie T, Hariz M, Zrinzo L, Joyce EM, et al. (2020): Basal Ganglia Pathways Associated with Therapeutic Pallidal Deep Brain Stimulation for Tourette Syndrome. Biol Psychiatry Cogn Neurosci Neuroimaging. https://doi.org/10.1016/j.bpsc.2020.11.005

      ,
      • Petry-Schmelzer J.N.
      • Jergas H.
      • Thies T.
      • Steffen J.K.
      • Reker P.
      • Dafsari H.S.
      • et al.
      Network Fingerprint of Stimulation-Induced Speech Impairment in Essential Tremor.
      ,
      • Al-Fatly B.
      • Ewert S.
      • Kübler D.
      • Kroneberg D.
      • Horn A.
      • Kühn A.A.
      Connectivity profile of thalamic deep brain stimulation to effectively treat essential tremor.
      ).
      In conclusion, we believe our results provide important insights regarding DBS induced impulsive decision-making in OCD. Our study shows that DBS reduces risk adjustment and increases delay aversion. Mapping these DBS-related behavioral changes provides a better understanding of the neuropsychological changes associated with DBS. Future studies may further investigate whether acute changes in impulsivity constitute solely a preventable side-effect or if they may relate to subsequent attenuation of obsessions and compulsions.

      Acknowledgements

      This work is funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) (KFO-219, KU 2665/1-2; Project-ID 431549029 – SFB 1451). MT, VVV and JCB are funded by the DFG (Project-ID 431549029 – SFB 1451) as well as NL (Project-ID 424778371 – TRR 295).
      We would like to thank Isabelle Ripp for her assistance with data acquisition. Data were provided in part by the Human Connectome Project, WU-Minn Consortium (principal investigators: David Van Essen and Kamil Ugurbil; 1U54MH091657) funded by the 16 NIH Institutes and Centers that support the NIH Blueprint for Neuroscience Research; and by the McDonnell Center for Systems Neuroscience at Washington University.
      Disclosures
      The authors report no biomedical financial interests or potential conflicts of interest.

      Literature

        • Huys D.
        • Kohl S.
        • Baldermann J.C.
        • Timmermann L.
        • Sturm V.
        • Visser-Vandewalle V.
        • Kuhn J.
        Open-label trial of anterior limb of internal capsule–nucleus accumbens deep brain stimulation for obsessive-compulsive disorder: insights gained.
        J Neurol Neurosurg Psychiatry. 2019; 90 (jnnp-2018-318996)
        • Denys D.
        • Graat I.
        • Mocking R.
        • de Koning P.
        • Vulink N.
        • Figee M.
        • et al.
        Efficacy of Deep Brain Stimulation of the Ventral Anterior Limb of the Internal Capsule for Refractory Obsessive-Compulsive Disorder: A Clinical Cohort of 70 Patients.
        Am J Psychiatry. 2020; https://doi.org/10.1176/appi.ajp.2019.19060656
        • Alonso P.
        • Cuadras D.
        • Gabriëls L.
        • Denys D.
        • Goodman W.
        • Greenberg B.D.
        • et al.
        Deep brain stimulation for obsessive-compulsive disorder: A meta-analysis of treatment outcome and predictors of response.
        PLoS One. 2015; 10https://doi.org/10.1371/journal.pone.0133591
        • Basar K.
        • Sesia T.
        • Groenewegen H.
        • Steinbusch H.W.M.
        • Visser-Vandewalle V.
        • Temel Y.
        Nucleus accumbens and impulsivity.
        Prog Neurobiol. 2010; 92: 533-557
        • Rogers R.D.
        • Owen a M.
        • Middleton H.C.
        • Williams E.J.
        • Pickard J.D.
        • Sahakian B.J.
        • Robbins T.W.
        Choosing between small, likely rewards and large, unlikely rewards activates inferior and orbital prefrontal cortex.
        J Neurosci. 1999; 19: 9029-9038
        • Czapla M.
        • Simon J.J.
        • Richter B.
        • Kluge M.
        • Friederich H.-C.
        • Herpertz S.
        • et al.
        The impact of cognitive impairment and impulsivity on relapse of alcohol-dependent patients: implications for psychotherapeutic treatment.
        Addiction Biology. 2016; 21: 873-884
        • Kim B.S.
        • Im H.I.
        The role of the dorsal striatum in choice impulsivity.
        Ann N Y Acad Sci. 2018; 1–20
        • Basar K.
        • Sesia T.
        • Groenewegen H.
        • Steinbusch H.W.M.
        • Visser-Vandewalle V.
        • Temel Y.
        Nucleus accumbens and impulsivity.
        Prog Neurobiol. 2010; 92: 533-557
        • Newcombe V.F.J.
        • Outtrim J.G.
        • Chatfield D.A.
        • Manktelow A.
        • Hutchinson P.J.
        • Coles J.P.
        • et al.
        Parcellating the neuroanatomical basis of impaired decision-making in traumatic brain injury.
        Brain. 2011; 134: 759-768
        • Fujihara K.
        • Narita K.
        • Suzuki Y.
        • Takei Y.
        • Suda M.
        • Tagawa M.
        • et al.
        Relationship of γ-aminobutyric acid and glutamate + glutamine concentrations in the perigenual anterior cingulate cortex with performance of Cambridge Gambling Task.
        Neuroimage. 2015; 109: 102-108
        • Ullsperger M.
        • Danielmeier C.
        • Jocham G.
        Neurophysiology of Performance Monitoring and Adaptive Behavior.
        Physiol Rev. 2014; 94: 35-79
      1. Schaltenbrand G, Wahren W, Hassler RG (1977): Atlas for Stereotaxy of the Human Brain, 2nd ed. Stuttgart: Thieme.

        • Horn A.
        • Li N.
        • Dembek T.A.
        • Kappel A.
        • Boulay C.
        • Ewert S.
        • et al.
        Lead-DBS v2: Towards a comprehensive pipeline for deep brain stimulation imaging.
        Neuroimage. 2019; https://doi.org/10.1016/j.neuroimage.2018.08.068
        • Horn A.
        • Reich M.
        • Vorwerk J.
        • Li N.
        • Wenzel G.
        • Fang Q.
        • et al.
        Connectivity Predicts deep brain stimulation outcome in Parkinson disease.
        Ann Neurol. 2017; 82: 67-78
        • Avants B.B.
        • Tustison N.
        • Song G.
        Advanced Normalization Tools (ANTS).
        Insight Journal. 2009; : 1-35
        • Klein A.
        • Andersson J.
        • Ardekani B.A.
        • Ashburner J.
        • Avants B.
        • Chiang M.C.
        • et al.
        Evaluation of 14 nonlinear deformation algorithms applied to human brain MRI registration.
        Neuroimage. 2009; 46: 786-802
        • Husch A.
        • Petersen M V.
        • Gemmar P.
        • Goncalves J.
        • Hertel F.
        PaCER - A fully automated method for electrode trajectory and contact reconstruction in deep brain stimulation.
        Neuroimage Clin. 2018; https://doi.org/10.1016/j.nicl.2017.10.004
        • Pauli W.M.
        • Nili A.N.
        • Michael Tyszka J.
        Data Descriptor: A high-resolution probabilistic in vivo atlas of human subcortical brain nuclei.
        Sci Data. 2018; https://doi.org/10.1038/sdata.2018.63
        • Vorwerk J.
        • Oostenveld R.
        • Piastra M.C.
        • Magyari L.
        • Wolters C.H.
        The FieldTrip-SimBio pipeline for EEG forward solutions.
        BioMedical Engineering OnLine 2018. 2018; 17: 1-17
      2. Proverbio D, Husch A (n.d.): ApproXON: Heuristic Approximation to the E-Field-Threshold for Deep Brain Stimulation Volume-of-Tissue-Activated Estimation. https://doi.org/10.1101/863613

        • Åström M.
        • Diczfalusy E.
        • Martens H.
        • Wårdell K.
        Relationship between neural activation and electric field distribution during deep brain stimulation.
        IEEE Trans Biomed Eng. 2015; 62: 664-672
        • Dembek T.A.
        • Roediger J.
        • Horn A.
        • Reker P.
        • Oehrn C.
        • Dafsari H.S.
        • et al.
        Probabilistic sweet spots predict motor outcome for deep brain stimulation in Parkinson disease.
        Ann Neurol. 2019; https://doi.org/10.1002/ana.25567
        • Reich M.M.
        • Horn A.
        • Lange F.
        • Roothans J.
        • Paschen S.
        • Runge J.
        • et al.
        Probabilistic mapping of the antidystonic effect of pallidal neurostimulation: a multicentre imaging study.
        Brain. 2019; 142: 1386-1398
        • Reich M.M.
        • Horn A.
        • Lange F.
        • Roothans J.
        • Paschen S.
        • Runge J.
        • et al.
        Probabilistic mapping of the antidystonic effect of pallidal neurostimulation: a multicentre imaging study.
        Brain. 2019; 142: 1386-1398
        • Elias G.J.B.
        • Boutet A.
        • Joel S.E.
        • Germann J.
        • Gwun D.
        • Neudorfer C.
        • et al.
        Probabilistic Mapping of Deep Brain Stimulation: Insights from 15 Years of Therapy.
        Ann Neurol. 2021; 89: 426-443
        • Dembek T.A.
        • Roediger J.
        • Horn A.
        • Reker P.
        • Oehrn C.
        • Dafsari H.S.
        • et al.
        Probabilistic sweet spots predict motor outcome for deep brain stimulation in Parkinson disease.
        Ann Neurol. 2019; https://doi.org/10.1002/ana.25567
        • Irmen F.
        • Horn A.
        • Mosley P.
        • Perry A.
        • Petry-Schmelzer J.N.
        • Dafsari H.S.
        • et al.
        Left prefrontal impact links subthalamic stimulation with depressive symptoms.
        Ann Neurol. 2020; https://doi.org/10.1002/ana.25734
        • Baldermann J.C.
        • Melzer C.
        • Zapf A.
        • Kohl S.
        • Timmermann L.
        • Tittgemeyer M.
        • et al.
        Connectivity profile predictive of effective deep brain stimulation in obsessive compulsive disorder.
        Biol Psychiatry. 2019; https://doi.org/10.1016/j.biopsych.2018.12.019
        • Li N.
        • Baldermann J.C.
        • Kibleur A.
        • Treu S.
        • Akram H.
        • Elias G.J.B.
        • et al.
        A unified connectomic target for deep brain stimulation in obsessive-compulsive disorder.
        Nat Commun. 2020; https://doi.org/10.1038/s41467-020-16734-3
        • Van Essen D.C.
        • Ugurbil K.
        • Auerbach E.
        • Barch D.
        • Behrens T.E.J.
        • Bucholz R.
        • et al.
        The Human Connectome Project: A data acquisition perspective.
        NeuroImage. 2012; https://doi.org/10.1016/j.neuroimage.2012.02.018
        • Nachev P.
        • Lopez-Sosa F.
        • Gonzalez-Rosa J.J.
        • Galarza A.
        • Avecillas J.
        • Pineda-Pardo J.A.
        • et al.
        Dynamic risk control by human nucleus accumbens.
        Brain. 2015; awv285
        • Grassi G.
        • Figee M.
        • Ooms P.
        • Righi L.
        • Nakamae T.
        • Pallanti S.
        • et al.
        Impulsivity and decision-making in obsessive-compulsive disorder after effective deep brain stimulation or treatment as usual.
        CNS Spectr. 2018; : 1-7
        • Luigjes J.
        • Mantione M.
        • van den Brink W.
        • Schuurman P.R.
        • van den Munckhof P.
        • Denys D.
        Deep brain stimulation increases impulsivity in two patients with obsessive-compulsive disorder.
        Int Clin Psychopharmacol. 2011; 26: 338-340
        • Haq I.U.
        • Foote K.D.
        • Goodman W.K.
        • Ricciuti N.
        • Ward H.
        • Sudhyadhom A.
        • et al.
        A Case of Mania following Deep Brain Stimulation for Obsessive Compulsive Disorder.
        Stereotact Funct Neurosurg. 2010; 88: 322
        • Kuhn J.
        • Lenartz D.
        • Huff W.
        • Mai J.K.
        • Koulousakis A.
        • Maarouf M.
        • et al.
        Transient manic-like episode following bilateral deep brain stimulation of the nucleus accumbens and the internal capsule in a patient with Tourette syndrome.
        Neuromodulation. 2008; 11: 128-131
        • Voon V.
        • Droux F.
        • Morris L.
        • Chabardes S.
        • Bougerol T.
        • David O.
        • et al.
        Decisional impulsivity and the associative-limbic subthalamic nucleus in obsessive-compulsive disorder: Stimulation and connectivity.
        Brain. 2017; 140: 442-456
        • Tsai H.C.
        • Chang C.H.
        • Pan J.I.
        • Hsieh H.J.
        • Tsai S.T.
        • Hung H.Y.
        • Chen S.Y.
        Pilot study of deep brain stimulation in refractory obsessive-compulsive disorder ethnic Chinese patients.
        Psychiatry Clin Neurosci. 2012; 66: 303-312
        • Zheng Z.
        • Li Y.
        • Li J.
        • Zhang Y.
        • Zhang X.
        • Zhuang P.
        Stimulation-induced dyskinesia in the early stage after subthalamic deep brain stimulation.
        Stereotact Funct Neurosurg. 2010; 88: 29-34
        • Volkmann J.
        • Daniels C.
        • Witt K.
        Neuropsychiatric effects of subthalamic neurostimulation in Parkinson disease.
        Nature Reviews Neurology. 2010; https://doi.org/10.1038/nrneurol.2010.111
        • Safadi Z.
        • Grisot G.
        • Jbabdi S.
        • Behrens T.E.
        • Heilbronner S.R.
        • McLaughlin N.C.R.
        • et al.
        Functional Segmentation of the Anterior Limb of the Internal Capsule: Linking White Matter Abnormalities to Specific Connections.
        The Journal of Neuroscience. 2018; 38: 2106-2117
        • Frank M.J.
        • Samanta J.
        • Moustafa A.A.
        • Sherman3 S.J.
        Hold Your Horses: Impulsivity, Deep Brain Stimulation, and Medication in Parkinsonism.
        Science (1979). 2007; 1309: 1309-1312
        • Cavanagh J.F.
        • Wiecki T.V.
        • Cohen M.X.
        • Figueroa C.M.
        • Samanta J.
        • Sherman S.J.
        • Frank M.J.
        Subthalamic nucleus stimulation reverses mediofrontal influence over decision threshold.
        Nat Neurosci. 2011; 14: 1462-1467
        • Zavala B.A.
        • Tan H.
        • Little S.
        • Ashkan K.
        • Hariz M.
        • Foltynie T.
        • et al.
        Midline frontal cortex low-frequency activity drives subthalamic nucleus oscillations during conflict.
        Journal of Neuroscience. 2014; https://doi.org/10.1523/JNEUROSCI.1169-14.2014
        • Zavala B.
        • Tan H.
        • Little S.
        • Ashkan K.
        • Green A.L.
        • Aziz T.
        • et al.
        Decisions Made with Less Evidence Involve Higher Levels of Corticosubthalamic Nucleus Theta Band Synchrony.
        J Cogn Neurosci. 2016; 6
        • Noecker A.M.
        • Frankemolle-Gilbert A.M.
        • Howell B.
        • Petersen M.V.
        • Beylergil S.B.
        • Shaikh A.G.
        • McIntyre C.C.
        StimVision v2: Examples and Applications in Subthalamic Deep Brain Stimulation for Parkinson’s Disease.
        Neuromodulation. 2021; https://doi.org/10.1111/ner.13350
        • Butenko K.
        • Bahls C.
        • Schröder M.
        • Köhling R.
        • Van Rienen U.
        OSS-DBS: Open-source simulation platform for deep brain stimulation with a comprehensive automated modeling.
        PLoS Comput Biol. 2020; https://doi.org/10.1371/journal.pcbi.1008023
        • Wang Q.
        • Akram H.
        • Muthuraman M.
        • Gonzalez-Escamilla G.
        • Sheth S.A.
        • Oxenford S.
        • et al.
        Normative vs. patient-specific brain connectivity in deep brain stimulation.
        Neuroimage. 2021; https://doi.org/10.1016/j.neuroimage.2020.117307
      3. Johnson KA, Duffley G, Foltynie T, Hariz M, Zrinzo L, Joyce EM, et al. (2020): Basal Ganglia Pathways Associated with Therapeutic Pallidal Deep Brain Stimulation for Tourette Syndrome. Biol Psychiatry Cogn Neurosci Neuroimaging. https://doi.org/10.1016/j.bpsc.2020.11.005

        • Petry-Schmelzer J.N.
        • Jergas H.
        • Thies T.
        • Steffen J.K.
        • Reker P.
        • Dafsari H.S.
        • et al.
        Network Fingerprint of Stimulation-Induced Speech Impairment in Essential Tremor.
        Ann Neurol. 2020; https://doi.org/10.1002/ana.25958
        • Al-Fatly B.
        • Ewert S.
        • Kübler D.
        • Kroneberg D.
        • Horn A.
        • Kühn A.A.
        Connectivity profile of thalamic deep brain stimulation to effectively treat essential tremor.
        Brain. 2019; https://doi.org/10.1093/brain/awz236
        • Edlow B.L.
        • Mareyam A.
        • Horn A.
        • Polimeni J.R.
        • Witzel T.
        • Tisdall M.D.
        • et al.
        7 Tesla MRI of the ex vivo human brain at 100 micron resolution.
        Sci Data. 2019; https://doi.org/10.1038/s41597-019-0254-8
        • Edlow B.L.
        • Mareyam A.
        • Horn A.
        • Polimeni J.R.
        • Witzel T.
        • Tisdall M.D.
        • et al.
        7 Tesla MRI of the ex vivo human brain at 100 micron resolution.
        Sci Data. 2019; https://doi.org/10.1038/s41597-019-0254-8