Neuroanatomical Characteristics Associated with Response to Dorsal Anterior Cingulotomy for Obsessive-Compulsive Disorder

Research

Original Investigation Neuroanatomical Characteristics Associated With Response to Dorsal Anterior Cingulotomy for Obsessive-Compulsive Disorder

Garrett P. Banks, BS; Charles B. Mikell, MD; Brett E. Youngerman, MD; Bryan Henriques, BS; Kathleen M. Kelly, BS; Andrew K. Chan, BS; Diana Herrera; Darin D. Dougherty, MD; Emad N. Eskandar, MD; Sameer A. Sheth, MD, PhD

Editorial page 108

IMPORTANCE Approximately 10% of patients with obsessive-compulsive disorder (OCD) have Author Audio Interview at symptoms that are refractory to pharmacologic and cognitive-behavioral therapies. jamapsychiatry.com Neurosurgical interventions can be effective therapeutic options in these patients, but not all Supplemental content at individuals respond. The mechanisms underlying this response variability are poorly jamapsychiatry.com understood. CME Quiz at jamanetworkcme.com and OBJECTIVE To identify neuroanatomical characteristics on preoperative imaging that CME Questions page 196 differentiate responders from nonresponders to dorsal anterior cingulotomy, a neurosurgical lesion procedure used to treat refractory OCD.

DESIGN, SETTING, AND PARTICIPANTS We retrospectively analyzed preoperative T1 and diffusion magnetic resonance imaging sequences from 15 patients (9 men and 6 women) who underwent dorsal anterior cingulotomy. Eight of the 15 patients (53%) responded to the procedure.

MAIN OUTCOMES AND MEASURES We used voxel-based morphometry (VBM) and diffusion tensor imaging to identify structural and connectivity variations that could differentiate eventual responders from nonresponders. The VBM and probabilistic tractography metrics were correlated with clinical response to the cingulotomy procedure as measured by changes in the Yale-Brown Obsessive Compulsive Scale score.

RESULTS Voxel-based morphometry analysis revealed a gray matter cluster in the right anterior cingulate cortex, anterior to the eventual lesion, for which signal strength correlated with poor response (P = .017). Decreased gray matter in this region of the dorsal anterior cingulate cortex predicted improved response (mean [SD] gray matter partial volume for responders vs nonresponders, 0.47 [0.03] vs 0.66 [0.03]; corresponding to mean Yale-Brown Obsessive Compulsive Scale score change, −60% [19] vs −11% [9], respectively). Hemispheric asymmetry in connectivity between the eventual lesion and the caudate (for responders vs nonresponders, mean [SD] group laterality for individual lesion seeds, −0.79 [0.18] vs −0.08 [0.65]; P = .04), putamen (−0.55 [0.35] vs 0.50 [0.33]; P = .001), thalamus (−0.82 [0.19] vs 0.41 [0.24]; P = .001), pallidum (−0.78 [0.18] vs 0.43 [0.48]; P = .001), and Author Affiliations: Department of hippocampus (−0.66 [0.33] vs 0.33 [0.18]; P = .001) also correlated significantly with clinical Neurological Surgery, Neurological response, with increased right-sided connectivity predicting greater response. Institute, Columbia University, New York, New York (Banks, Mikell, Youngerman, Kelly, Chan, Herrera, CONCLUSIONS AND RELEVANCE We identified features of anterior cingulate cortex structure Sheth); College of Physicians and and connectivity that predict clinical response to dorsal anterior cingulotomy for refractory Surgeons, Columbia University, New OCD. These results suggest that the variability seen in individual responses to a highly York, New York (Henriques); Department of Psychiatry, consistent, stereotyped procedure may be due to neuroanatomical variation in the patients. Massachusetts General Hospital, Furthermore, these variations may allow us to predict which patients are most likely to Boston (Dougherty); Department of respond to cingulotomy, thereby refining our ability to individualize this treatment for Neurosurgery, Massachusetts General Hospital, Boston (Eskandar). refractory psychiatric disorders. Corresponding Author: Sameer A. Sheth, MD, PhD, Department of Neurological Surgery, Neurological Institute, Room 427, Columbia University, 710 W 168th St, New York, JAMA Psychiatry. 2015;72(2):127-135. doi:10.1001/jamapsychiatry.2014.2216 NY 10032 ([email protected] Published online December 23, 2014. .edu).

(Reprinted) 127

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Research Original Investigation Response to Dorsal Anterior Cingulotomy

bsessive-compulsive disorder (OCD) is a debilitating these structures, and diffusion tensor imaging (DTI) and chronic disorder with a lifetime prevalence of 2% studies26,27 have demonstrated hemispheric asymmetries in O to 3%.1 The disorder is characterized by intrusive white matter bundles connecting limbic structures, including thoughts and repetitive intentional behaviors that persist de- the anterior limb of the internal capsule and cingulum. Sig- spite a desire to suppress them, often accompanied by marked nificant heterogeneities in these metrics exist across studies, anxiety. Although most patients attain adequate sympto- and the population of patients with intractable OCD who are matic relief with medication and cognitive-behavioral therapy, surgical candidates could have different organizational pat- it is estimated that OCD may be refractory to these treat- terns from those reported in the previous studies. Neverthe- ments in 10% to 20% of patients.1,2 Unfortunately, alternative less, existing data suggest that these metrics are reasonable treatment options are limited. candidate predictors of response. A subset of patients with refractory OCD may be candi- We therefore hypothesized that preoperative structural or dates for surgical treatment. Given the risk of morbidity with connectivity variations in and between these regions may un- surgery, candidate patients must be carefully evaluated by a derlie the variability in patients’ response to cingulotomy. We multidisciplinary team consisting of psychiatrists, neurosur- tested this hypothesis by applying VBM and DTI analyses to geons, psychologists, and neurologists in consultation with preoperative imaging data in patients who received this ste- ethicists.3-6 Typical inclusion criteria for surgical consider- reotyped surgical lesion to determine whether neuroanatomi- ation are a diagnosis of persistent severe OCD that is refrac- cal factors can predict response. tory to several adequate pharmacologic trials and cognitive- behavioral therapy, the ability to follow instructions and provide consent, and demonstration of realistic expecta- Methods tions. Typical exclusion criteria are severe medical comorbidi- ties, imminent suicidal intent, comorbid severe psychiatric dis- Ethics Statement orders, and evidence of neurocognitive disorders. Given these All study procedures were approved by the Columbia Univer- necessarily stringent requirements for surgical consider- sity Medical Center and Massachusetts General Hospital in- ation, the number of appropriate surgical candidates is rela- stitutional review boards. The requirement for informed con- tively small. sent was waived by both institutions. Stereotactic surgical lesions, originally developed in the mid-20th century, have been successfully used for decades to Participants treat severe refractory OCD and other psychiatric disorders. The We retrospectively reviewed the records of all patients receiv- necessity for safe, accurate, and reproducible surgical treating cingulotomy for severe, intractable OCD between Janu- ment options for psychiatric conditions was a major motiva- ary 1, 2000, and December 31, 2010, at Massachusetts General tion for the development of stereotactic neurosurgery in the Hospital. Patients who received preoperative magnetic reso- late 1940s.7 The dorsal anterior cingulotomy, one such stereo- nance imaging (MRI) with either a high-resolution T1 or DTI tactic procedure, involves lesioning the dorsal anterior cingu- sequence were included in the study. Patients with prior neu- late cortex (dACC), a region believed to play a role in the patho- rosurgical interventions to treat OCD were excluded. Details genesis of the neural network that causes OCD. Clinical series regarding the clinical outcome of these patients have been with long-term follow-up have demonstrated a durable re- described.5 Surgical technique is described in the eAppendix sponse rate of 45% to 70% following cingulotomy.8-11 This re- in the Supplement. sponse rate is significant considering that these patients had been refractory to conventional therapy for years or even de- Patient Evaluation cades. Response rates of an alternative lesion procedure, the Surgical candidacy was determined by a multidisciplinary anterior capsulotomy, were fairly similar in open-label team as described previously.3,4,9 Patients underwent a psy- series,12-15 but data from controlled blinded trials are limited.16,17 chiatric evaluation prior to cingulotomy and during follow-up Nevertheless, these are invasive procedures with a complica- visits. The severity of OCD was assessed by the treating psy- tion rate of 5% to 10%. Improving our ability to predict which chiatrist (D.D.D.) using the Yale-Brown Obsessive Compulsive patients will respond to surgical treatment would represent a Scale (Y-BOCS).28 The Y-BOCS score obtained during the major advance in our management of OCD. follow-up visit closest to 1 year after intervention was used as Whereas our understanding of the role of the dACC in the postoperative score. For patients who underwent another both normal cognitive processes18-20 and OCD21,22 is steadily neurosurgical procedure within the first 12 months, the score improving, few studies have investigated features that pre- obtained immediately before the second procedure was dict outcome after cingulotomy or other neurosurgical used. Patients with a 35% or greater reduction in the Y-BOCS procedures.23 Structural differences have been described24-27 score were considered responders. between patients with OCD and individuals serving as con- High-resolution (in-plane resolution, ≤1 mm; out-of- trols in components of the limbic corticobasal ganglia- plane spacing, ≤2.5 mm) T1-weighted and diffusion (6 or 25 di- thalamo-cortical (CBTC) network, including the orbitofron- rections) sequences were acquired on a 1.5T scanner (Signa tal, parahippocampal, and cingulate cortices; the striatum; HDxt; GE Healthcare). Mann-Whitney analyses were per- and the medial thalamus. Voxel-based morphometry (VBM) formed to assess for scan parameter biases between respond- studies24,25 have reported gray matter volume differences in ers and nonresponders (eTable 1 in the Supplement).

128 JAMA Psychiatry February 2015 Volume 72, Number 2 (Reprinted) jamapsychiatry.com

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Response to Dorsal Anterior Cingulotomy Original Investigation Research

Gray Matter Analysis The tractography analysis was performed in patient diffu- Analyses were performed at Columbia University Medical Cen- sion space using the Diffusion Toolbox of the FMRIB Software ter. All patients with a preoperative high-resolution T1 MRI se- Library.33-35 For each patient we sought to estimate the strength quence were included. We used VBM to identify differences of connectivity between the origin seed (lesion mask) and the in gray matter between the brains of responders and nonre- atlas-defined target structures. To do so, we calculated prob- sponders, along with a model-free algorithm to avoid precon- ability distributions based on 2 fiber directions using a previ- ceived biases regarding regions of interest. Strict statistical ously described33-35 multiple fiber extension algorithm. We used thresholding was used to account for multiple comparisons. these calculated streamline distributions to estimate fiber tracts An optimized VBM protocol was performed using the from the origin seeds to ipsilateral target regions. The strength FMRIB Software Library, version 5.0 (FMRIB; University of of connectivity from an origin seed to any other area in the brain Oxford), which included brain extraction, segmentation, and correlates with the number of seed-originating streamline traces linear/nonlinear transformation.29 A left-right, symmetric, passing through the target.36 To suppress spuriously gener- study-specific gray matter template was designed from ated tracts passing through areas unlikely to support white mat- native-space T1 images after affine transformation to the gray ter connectivity, a patient-specific cerebrospinal fluid termi- matter ICBM 152, version 2009c (McConnell Brain Imaging nation mask was designed by thresholding the nondirectionally Center) 2-mm standard template. Native-space gray matter vol- weighted diffusion image. Streamline tracking originated in the umes were nonlinearly normalized to the study-specific tem- seed and stopped after leaving brain space or encountering the plate. A modulation algorithm, part of the FMRIB Software cerebrospinal fluid mask. To quantify connections from the ori- Library protocol, was used to compensate for distortion from gin seed to ipsilateral targets, we calculated the percentage of nonlinear transformation. Resultant gray matter volumes were seed-originating streamlines reaching the targets.36 Left- and smoothed with an isotropic gaussian kernel of sigma 3.5. right-hemisphere connections were calculated separately. For The FMRIB Software Library randomize function was used the primary analysis, we only considered ipsilateral tracts, ig- to perform permutation-based nonparametric inference using noring any hemisphere-crossing streamlines.36 An analysis al- a generalized linear model to compare responders and lowing hemispheric crossings through the corpus callosum is nonresponders.30 In addition to changes in the Y-BOCS score, included in eTable 2 in the Supplement. the generalized linear model included age and sex as nui- To select appropriate seed-target pairs, only pairs wherein sance variables. A threshold-free cluster enhancement analy- at least 1 streamline connecting the origin seed to the ipsilat- sis was used to compare modulated gray matter maps be- eral target in all patients were used. This requirement pre- tween groups on a voxelwise basis (5000 permutations). The vented unnecessary analysis on spurious seed-target pairs that results were subjected to a voxelwise Bonferroni correction to had no anatomical basis. In addition, dorsal cingulate and para- account for familywise errors. An after-correction value of cingulate targets were excluded because they partially over- P < .05 was considered significant. lapped with the origin seed in standard space, and tracts would A spherical region-of-interest diameter of 10 mm was con- therefore automatically connect to their target and create spu- structed using FSLView in standard space, centered on the rious results. Target pairs meeting the above criteria were con- voxel with the smallest postcorrection P value. For each pa- sidered in network and were included in the analysis. tient’s modulated gray matter volume map, the mean gray mat- We quantified hemispheric asymmetry in seed-target con- ter fraction in the area of the spherical region of interest was nectivity by calculating a laterality metric (LM). For each seed- calculated using fslstats and regressed against patient- target pair, we subtracted the right-sided streamline percent- specific Y-BOCS score changes. age from the left-sided streamline percentage and divided by the sum. The LM is thus a ratio between −1 and 1 that de- White Matter Analysis scribes whether the right- or left-sided target was more highly All patients who underwent preoperative DTI were included. connected to the seed area, with negative values indicating We used DTI data to estimate the connectivity between the greater right-sided connectivity. A Mann-Whitney test was per- eventual lesion site and predefined target structures. To de- formed to compare the LM between responders and nonre- lineate the lesioned area, we created patient-specific lesion sponders for in-network targets. All results were subjected to masks using postoperative T2-weighted images.31 This delin- a Benjamini-Hochberg false discovery rate correction to ac- eation was performed by 2 investigators (B.H. and D.H.) blinded count for multiple comparisons. to patient information. We assessed interrater reliability using The primary analysis used each patient’s lesion mask as the Cohen κ test. The lesion masks were determined for pa- the seed. As a secondary analysis, we created a standard le- tients individually and then transformed to the diffusion space sion mask by aggregating the individual lesion masks. All in- defined by each patient’s preoperative DTI scan. These lesion dividual lesions were transformed to standard space, and the masks were the seeds for the subsequent DTI analysis. threshold for the resultant lesion probability distribution was To objectively define the target structures, we used the set at P < .05 after threshold-free cluster enhancement and 2-mm cortical and subcortical Harvard-Oxford structural prob- familywise error correction to create the standard lesion mask. ability atlases, with a threshold set a priori at 25% probability.32 This standard lesion mask was transformed linearly and non- By using these atlases, there were 56 possible seed-target pairs. linearly to patient-specific diffusion space and used as the seed The atlases were transformed to patient diffusion space using for this secondary analysis. In-network targets were deter- linear/nonlinear transformation. mined separately for the standard lesion mask analysis.

jamapsychiatry.com (Reprinted) JAMA Psychiatry February 2015 Volume 72, Number 2 129

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Research Original Investigation Response to Dorsal Anterior Cingulotomy

Table 1. Demographic Information and Clinical Severity Scores of Participants

Y-BOCS Scorea Postoperative Patient Sex/Age, y Year Interval, mob Preoperativec 1 Yeard Deltae Delta (%) Imaging Responders M/30 2001 12 32 2 −30 −93.8 T1 F/56 2009 12 40 11 −29 −72.5 T1, DTI F/28 2008 13 35 10 −25 −71.4 T1, DTI M/29 2003 7 34 15 −19 −55.9 T1, DTI M/27 2003 12 36 16 −20 −55.6 T1, DTI F/43 2007 7 40 22 −18 −45.0 T1, DTI F/49 2008 12 38 21 −17 −44.7 T1, DTI M/46 2008 10 34 21 −13 −38.2 T1, DTI Nonresponders M/22 2008 13 34 27 −7 −20.6 T1, DTI M/44 2000 9 35 29 −6 −17.1 T1 F/49 2003 14 38 32 −6 −15.8 T1, DTI M/33 2007 14 39 35 −4 −10.3 T1, DTI F/38 2003 7 28 27 −1 −3.6 T1, DTI M/29 2006 11 35 36 +1 +2.9 DTI M/33 2007 13 35 36 +1 +2.9 T1, DTI Group comparison, mean (SD) All patients NA/37.1 (10.1) 2005.4 (2.9) 11.1 (2.6) 35.5 (3.18) 22.7 (10.4) −15 (9.7) −35.9 (30.0) NA Responders NA/38.5 (11.3) 2005.9 (3.0) 10.6 (2.4) 36.1 (2.9) 14.8 (6.9) −21.4 (6.0) −59.6 (18.5) NA Nonresponders NA/35.4 (9.1) 2004.9 (2.9) 11.6 (2.7) 34.9 (3.5) 31.7 (4.1) −3.1 (3.4) −8.8 (9.7) NA P Valuef .75/.84 .34 .33 .54 <.001 <.001 <.001

Abbreviations: DTI, diffusion tensor imaging; NA, not applicable; Y-BOCS, during the month prior to the operation. Yale-Brown Obsessive Compulsive scale. d The postoperative Y-BOCS score was obtained as part of psychiatric evaluation a The Y-BOCS score was calculated by a staff psychiatrist at the preoperative during the follow-up visit closest to 1 year after surgery but before additional and follow-up visits. neurosurgical procedures. b Number of months after the operation that the 1-year postoperative e Delta value was calculated by subtracting the preoperative Y-BOCS score from evaluation was performed. the Y-BOCS score at 1 year. c The preoperative Y-BOCS score was obtained as part of psychiatric evaluation f P value was calculated using a Mann-Whitney test.

heterogeneity in acquisition parameters, several additional Results analyses were performed (eTables 3-5 and eFigures 1-3 in the Supplement). No analysis suggested that the distinction be- Patients tween responders and nonresponders was driven by this Fifteen patients (9 men [60%]; mean age, 37 years) met the heterogeneity. Representative images of the postoperative T2 study inclusion criteria. Of these, 14 individuals (93%) had pre- sequences along with the segmented lesions are shown in eFig- operative high-resolution MRI T1 sequences and 13 (87%) had ure 4 in the Supplement. preoperative DTI sequences. Eight of the 14 patients (57%) with high-resolution T1 and 7 of the 13 patients (54%) with DTI data Gray Matter Results were responders (Table 1). There were no significant differ- The VBM analysis of whole-brain gray matter topography dem- ences between responders and nonresponders in sex, age, sur- onstrated a gray matter cluster associated with the outcome. gical year, or preoperative Y-BOCS score as determined by a The cluster, centered in the right anterior cingulate cortex Mann-Whitney test. (x = 2, y = 42, and z = 20) several millimeters anterior to the standard lesion area, demonstrated greater gray matter vol- Imaging Parameters ume in nonresponders (Figure 1A). Postcorrection signifi- We found no significant differences between responders and cance of the cluster was P = .017. No clusters of gray matter sur- nonresponders in voxel volume on T1 images, slice spacing on viving familywise error correction correlated with age or sex. T1 images, or DTI gradient direction number (eTable 1 in the Given the heterogeneity of the scans, the scan parameters were Supplement). Nonetheless, to address the possible effect of included as nuisance repressors in the generalized linear model;

130 JAMA Psychiatry February 2015 Volume 72, Number 2 (Reprinted) jamapsychiatry.com

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Response to Dorsal Anterior Cingulotomy Original Investigation Research

subsequently, correlations in the same area remained statis- Figure 1. Differences in Gray Matter Between Responders tically significant (eFigure 1 in the Supplement). and Nonresponders We quantified the relationship for illustrative purposes by placing a 10-mm diameter seed in the right dACC at the clus- A Gray matter cluster ter center and plotting gray matter volume against the Y- BOCS score change (Figure 1B). Decreased gray matter in this region of the dACC predicted improved response (mean [SD] gray matter partial volume for responders vs nonresponders, 0.47 [0.03] vs 0.66 [0.03]; corresponding to mean Y-BOCS score change, −60% [19] vs −11% [9], respectively).

White Matter Analysis B Mean ROI gray matter volume Two investigators (B.H. and D.H.) blinded to response status 20 delineated the lesions on postoperative T2 MRIs of individual Responders Nonresponders patients. The Cohen interrater κ statistic was 0.92, indicating 0 a consistent segmentation. Of the 56 potential seed-target pairs, 18 pairs (32%) were –20 considered in network for the lesion study and included in the –40 analysis (Table 2). After false discovery rate correction, the LM for 4 of the 18 structures (22%) differed significantly between –60 Y-BOCS Change, % Y-BOCS the groups. Connectivity to the thalamus, putamen, pallidum, and hippocampus readily distinguished responders from –80 nonresponders (Table 2). The LMs of these 4 targets were plot- –100 ted against patient-specific Y-BOCS score changes (Figure 2). 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 Greater right-sided connectivity between the lesioned areas Mean Gray Matter Volume in the dACC and these 4 target areas correlated with better response to cingulotomy. A, Sagittal (left), coronal (center), and axial (right) views centered on the dorsal anterior cingulate. Shown is the cluster of gray matter that was greater in To permit preoperative evaluation of patients in whom the nonresponders, with a threshold after correction of P < .05 and overlaid on a lesion was not yet created, the analysis was repeated using a 1-mm standard brain. Orange/yellow voxels indicate areas that were standardized lesion area (eFigure 5 in the Supplement). In this significantly different between responders and nonresponders. B, Yale-Brown Obsessive Compulsive Scale (Y-BOCS) score changes regressed against the analysis, 23 structures were considered in network (Table 2). mean partial volume gray matter fraction (ie, number between 0 and 1 without Again, 4 seed-target LM pairs differed significantly between units) in the cluster center region of interest (ROI). responders and nonresponders. However, in this standardized analysis, the caudate nucleus instead of the putamen LM was significant. Targets of the significant pairs included the These structures are part of the limbic CBTC loop impli- thalamus, pallidum, hippocampus, and caudate nucleus. Ana- cated in the neural network dysfunction thought to be tomical targets and tracts are described in eFigures 6 and 7, re- responsible for OCD. Other than defining the DTI seed based spectively, in the Supplement. Significant LMs for the stan- on the location of the lesion, our analysis was agnostic to dard areas were plotted against Y-BOCS score changes for candidate brain structures. Even without an a priori region illustrative purposes (Figure 3), and greater right-sided con- of interest definition, limbic CBTC structures were the only nectivity predicted better response. Right-sided dominance of areas to survive stringent significance testing. Previous at least 2 of the 3 consistent structures (thalamus, pallidum, studies25,37,38 have identified structural differences in com- and hippocampus) had strong positive and negative predic- ponents of this circuit between patients with OCD and tive values for treatment response in this small cohort (eAp- matched controls. Our results extend these findings by dem- pendix in the Supplement). Allowing for streamline crossing onstrating that anatomical characteristics of these struc- through the corpus callosum did not significantly alter the re- tures also predict which patients are likely to respond to sults (eTable 2 in the Supplement). cingulotomy. Our findings were strongly lateralized: less gray matter in the right dACC and greater right-sided connectivity in the lim- Discussion bic CBTC circuit predicted improved response. This notion of lateralized anatomical differences in OCD is consistent with the We identified several neuroanatomical features associated with findings of previous studies demonstrating unilateral differ- clinical outcome following a stereotactic surgical lesion pro- ences between patients and healthy controls throughout the cedure for severe intractable OCD. We observed differences in CBTC circuit,39,40 although bilateral differences have also been gray matter signal in the anterior cingulate gyrus as well as in identified.25,37 Several pharmacologic and cognitive-beha- connectivity between the dorsal cingulate, caudate, puta- vioral therapy trials have demonstrated unilateral right-sided men, pallidum, thalamus, and hippocampus that readily dis- imaging changes across metabolic41-44 and perfusion45-47 do- tinguished responders from nonresponders to cingulotomy. mains, all of which correlated with response to treatment. Fewer

jamapsychiatry.com (Reprinted) JAMA Psychiatry February 2015 Volume 72, Number 2 131

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Research Original Investigation Response to Dorsal Anterior Cingulotomy

Table 2. Statistical Comparison of LM Between Responders and Nonrespondersa

LM Values for Individual Lesion Seeds LM Values for Standardized Lesion Seeds Group Laterality, Mean (SD) Group Laterality, Mean (SD) Uncorrected Uncorrected Network Target Responder NonresponderP Valueb Responder Nonresponder P Valueb Anterior frontal pole 0.21 (0.37) 0.01 (0.53) .84 0.19 (0.43) 0.21 (0.34) >.99 Insula −0.27 (0.64) 0.58 (0.44) .05 0.09 (0.63) 0.57 (0.42) .14 Superior frontal gyrus 0.05 (0.25) 0.18 (0.38) .63 0.01 (0.16) 0.18 (0.23) .37 Middle frontal gyrus −0.35 (0.58) 0.02 (0.61) .37 −0.04 (0.16) 0.07 (0.49) .73 Pars opercularis 0.41 (0.62) 0.40 (0.79) .84 0.59 (0.48) 0.42 (0.67) >.99 Pars triangularis NA NA −0.16 (0.56) 0.35 (0.49) .18 Primary motor cortex −0.08 (0.62) 0.05 (0.60) .63 −0.20 (0.55) 0.07 (0.57) .37 Primary sensory cortex −0.28 (0.73) −0.13 (0.68) .73 −0.24 (0.79) −0.10 (0.56) .45 Lateral occipital cortex 0.09 (0.78) 0.24 (0.51) .73 −0.32 (0.66) 0.20 (0.53) .23 Ventromedial PFC NA NA −0.24 (0.81) 0.41 (0.31) .23 Supplementary motor area 0.48 (0.32) 0.18 (0.26) .23 0.54 (0.30) 0.36 (0.28) .37 Subgenual PFC NA NA −0.10 (0.53) −0.02 (0.48) .84 Precuneus NA NA 0.02 (0.69) 0.06 (0.61) >.99 Lateral orbital frontal −0.01 (0.73) 0.35 (0.68) .30 0.33 (0.64) 0.36 (0.58) .95 Anterior parahippocampal NA NA −0.54 (0.42) 0.11 (0.64) .10 Posterior parahippocampal −0.72 (0.22) 0.07 (0.62) .01 −0.59 (0.27) 0.18 (0.60) .02 Frontal operculum −0.31 (0.73) 0.46 (0.56) .07 Parietal operculum NA NA −0.59 (0.68) 0.29 (0.43) .04 Thalamusc −0.82 (0.19) 0.41 (0.24) .001 −0.73 (0.35) 0.37 (0.24) .001 Caudate nucleusd −0.79 (0.18) −0.08 (0.65) .04 −0.70 (0.27) 0.08 (0.45) .008 Putamene −0.55 (0.35) 0.50 (0.33) .001 −0.29 (0.57) 0.50 (0.36) .01 Pallidumc −0.78 (0.18) 0.43 (0.48) .001 −0.64 (0.28) 0.42 (0.50) .002 Hippocampusc −0.66 (0.33) 0.33 (0.18) .001 −0.64 (0.27) 0.33 (0.34) .002 Amygdala −0.67 (0.23) −0.10 (0.69) .14 −0.61 (0.51) 0.05 (0.63) .05

Abbreviations: LM, laterality metric; NA, not applicable (not in network); were .02 for individual lesion seeds and .03 for standardized lesion seeds. PFC, prefrontal cortex. d P value determined using Benjamini-Hochberg false discovery rate correction a The top portion of the table compares the responder and nonresponder was .046 for standardized lesion seeds. groups. e P value determined using Benjamini-Hochberg false discovery rate correction b P values were determined by a Mann-Whitney test. was .02 for individual lesion seeds. c P values determined using Benjamini-Hochberg false discovery rate correction

examples exist of left-sided47 and bilateral48,49 imaging- nal in the right (but not left) nucleus accumbens during a response correlations. reward anticipation task increased to levels near those of Targeted surgical interventions provide further support controls during stimulation and exhibited decreased signal for the laterality hypothesis. Although most clinical series during the off state. This intriguing finding suggests that DBS and trials of deep brain stimulation (DBS) for OCD have used normalizes nucleus accumbens activity, potentially in a later- bilateral stimulation,50 2 reports2,51 described results with alized fashion. The confluence of these results therefore per- unilateral right-sided stimulation of the ventral striatum, mits consideration of whether most of the clinical response with similar response to bilateral stimulation.52 A similar to these surgical procedures is derived from the right-sided situation holds for the results from ventral capsulotomy, a intervention. stereotactic lesion targeting the ventral portion of the ante- The significance of the potential unilateral (right predomi- rior limb of the internal capsule, a region very close to the nant) aspect of the pathogenesis of OCD remains unclear. It is DBS target. In a study by Lippitz et al53 including 29 patients, possible that, just as language and verbal memory are heavily all 16 patients with a good outcome had lesions that over- lateralized functions, the planning, decision making, and re- lapped in the right, but not left, anterior limb of the internal ward circuitry dysfunction that underlies OCD could also be capsule. Again, most capsulotomy studies use bilateral significantly lateralized. Further work in this area is neces- lesions and have reported similar outcomes.12-14,54 However, sary to explore this possibility. image evaluators in the Lippitz et al study were not blinded An important implication of our results pertains to the to response, and those findings were not reproduced in a presurgical evaluation of patients with severe refractory later capsulotomy study55 in patients with non-OCD anxiety OCD who are being considered for neurosurgical treatment. disorders from the same group. Finally, a recent DBS study56 Although the complication rate of these procedures is rela- found that blood oxygen level–dependent functional MRI sig- tively low, the procedures are nonetheless invasive and

132 JAMA Psychiatry February 2015 Volume 72, Number 2 (Reprinted) jamapsychiatry.com

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Response to Dorsal Anterior Cingulotomy Original Investigation Research

Figure 2. Regression of Yale-Brown Obsessive Compulsive Scale (Y-BOCS) Score Changes Against Laterality Metric Values for Each of the Significant Target Pairs in the Patient-Specific Lesion Study

A Lesion to pallidum connectivity B Lesion to hippocampus connectivity 10 10

0 0

–10 –10

–20 –20

–30 –30

–40 –40

–50 –50 Y-BOCS Change, % Y-BOCS Change, % Y-BOCS –60 –60

–70 –70

–80 –80 –1.0 –0.8 –0.6 –0.4 –0.2 0 0.2 0.4 0.6 0.8 1.0 –1.0 –0.8 –0.6 –0.4 –0.2 0 0.2 0.4 0.6 0.8 1.0 Laterality Metric Laterality Metric

C Lesion to putamen connectivity D Lesion to thalamus connectivity 10 10

0 0

–10 –10

–20 –20

–30 –30

–40 –40

–50 –50 Y-BOCS Change, % Y-BOCS Change, % Y-BOCS –60 –60 Responders Nonresponders –70 –70

–80 –80 –1.0 –0.8 –0.6 –0.4 –0.2 0 0.2 0.4 0.6 0.8 1.0 –1.0 –0.8 –0.6 –0.4 –0.2 0 0.2 0.4 0.6 0.8 1.0 Laterality Metric Laterality Metric

Lesion connectivity to the pallidum (A), hippocampus (B), putamen (C), and thalamus (D).

should be offered only to appropriately selected patients. ever, because there were no significant differences between Previous studies57 of cingulotomy lesions suggest that lesion responders and nonresponders when imaging parameters size is not a strong predictor of response. Given this finding were compared, we believe this limitation was unlikely to as well as the reproducibility of stereotactic lesion proce- have significantly affected the results of our study. More- dures, it may be that response variability is more dependent over, our tractography analysis used only within-subject on patients’ neuroanatomical variation58-60 than on lesion comparisons to avoid the difficulty of intersubject compari- heterogeneity.9,57 The ability to preoperatively, noninva- sons with differing scan parameters. Although the sample sively identify patients who are more likely to respond to size was not large, these procedures are rare, and this study specific neurosurgical interventions would therefore repre- represents, to our knowledge, the largest investigating present a significant advance in our treatment algorithm. The operative imaging predictors of response to a stereotactic imaging metrics we identified could be applied prospectively lesion. to patients undergoing targeted interventions, such as stereotactic lesions or DBS. Better preoperative evaluation techniques should ide- Conclusions ally lead to a more individualized view of each patient’s disorder. Improved understanding of the neuroanatomical In this group of patients undergoing dorsal anterior cingu- basis of OCD will inevitably lead to more refined and indi- lotomy for severe intractable OCD, we found structural and con- vidualized targeting for stereotactic neurosurgical proce- nectivity differences in the limbic CBTC circuit that distin- dures. A more comprehensive appreciation of the underly- guished responders from nonresponders. Decreasing right ing circuit will not only help predict which patients are anterior cingulate gray matter volume and increasing connec- likely to respond but also which aspects of their disorder are tivity between the right cingulate and caudate, putamen, pal- likely to improve. lidum, thalamus, and hippocampus correlated with im- Important limitations of our study are its retrospective proved clinical outcomes. These results emphasize the design and consequent scan parameter heterogeneity. How- importance of hemispheric asymmetry in the pathophysiol-

jamapsychiatry.com (Reprinted) JAMA Psychiatry February 2015 Volume 72, Number 2 133

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Research Original Investigation Response to Dorsal Anterior Cingulotomy

Figure 3. Regression of Yale-Brown Obsessive Compulsive Scale (Y-BOCS) Score Changes Against Laterality Metric Values for Each of the Significant Target Pairs in the Standard Area Study

A Standard area to pallidum connectivity B Standard area to hippocampus connectivity 10 10

0 0

–10 –10

–20 –20

–30 –30

–40 –40

–50 –50 Y-BOCS Change, % Y-BOCS Change, % Y-BOCS –60 –60

–70 –70

–80 –80 –1.0 –0.8 –0.6 –0.4 –0.2 0 0.2 0.4 0.6 0.8 1.0 –1.0 –0.8 –0.6 –0.4 –0.2 0 0.2 0.4 0.6 0.8 1.0 Laterality Metric Laterality Metric

C Standard area to caudate connectivity D Standard area to thalamus connectivity 10 10

0 0

–10 –10

–20 –20

–30 –30

–40 –40

–50 –50 Y-BOCS Change, % Y-BOCS Change, % Y-BOCS –60 –60 Responders Nonresponders –70 –70

–80 –80 –1.0 –0.8 –0.6 –0.4 –0.2 0 0.2 0.4 0.6 0.8 1.0 –1.0 –0.8 –0.6 –0.4 –0.2 0 0.2 0.4 0.6 0.8 1.0 Laterality Metric Laterality Metric

Standard area connectivity to the pallidum (A), hippocampus (B), caudate nucleus (C), and thalamus (D).

ogy of OCD as well as the possibility of using neuroanatomi- determine the generalizability of these results as well as cal markers to inform treatment decisions and predict out- whether these findings should be considered for clinical de- comes in an individualized manner. Prospective work will help cision making.

ARTICLE INFORMATION Conflict of Interest Disclosures: None reported. 6. Garnaat SL, Greenberg BD, Sibrava NJ, et al. Submitted for Publication: February 18, 2014; final Who qualifies for deep brain stimulation for OCD? revision received August 16, 2014; accepted August REFERENCES J Neuropsychiatry Clin Neurosci. 2014;26(1):81-86. 25, 2014. 1. Greenberg BD, Rauch SL, Haber SN. Invasive 7. Diering SL, Bell WO. Functional neurosurgery for circuitry-based neurotherapeutics: stereotactic psychiatric disorders: a historical perspective. Published Online: December 23, 2014. ablation and deep brain stimulation for OCD. Stereotact Funct Neurosurg. 1991;57(4):175-194. doi:10.1001/jamapsychiatry.2014.2216. Neuropsychopharmacology. 2010;35(1):317-336. 8. Dougherty DD, Baer L, Cosgrove GR, et al. Author Contributions: Mr Banks and Dr Sheth had 2. Huff W, Lenartz D, Schormann M, et al. Unilateral Prospective long-term follow-up of 44 patients full access to all the data in the study and take deep brain stimulation of the nucleus accumbens in who received cingulotomy for treatment-refractory responsibility for the integrity of the data and the patients with treatment-resistant obsessive-compulsive disorder. Am J Psychiatry. accuracy of the data analysis. obsessive-compulsive disorder. Clin Neurol Neurosurg. 2002;159(2):269-275. Study concept and design: Banks, Mikell, Eskandar, 2010;112(2):137-143. Sheth. 9. Sheth SA, Neal J, Tangherlini F, et al. Limbic Acquisition, analysis, or interpretation of data: All 3. Mian MK, Campos M, Sheth SA, Eskandar EN. system surgery for treatment-refractory authors. Deep brain stimulation for obsessive-compulsive obsessive-compulsive disorder. J Neurosurg.2013; Drafting of the manuscript: Banks, Mikell, disorder: past, present, and future. Neurosurg Focus. 118(3):491-497. Henriques, Kelly, Herrera, Dougherty, Sheth. 2010;29(2):10. doi:10.3171/2010.4.FOCUS10107. 10. Jung HH, Kim C-H, Chang JH, Park YG, Chung Critical revision of the manuscript for important 4. Cosgrove GR, Rauch SL. Stereotactic SS, Chang JW. Bilateral anterior cingulotomy for intellectual content: Banks, Mikell, Youngerman, cingulotomy. Neurosurg Clin N Am. 2003;14(2):225- refractory obsessive-compulsive disorder. Chan, Eskandar, Sheth. 235. Stereotact Funct Neurosurg. 2006;84(4):184-189. Statistical analysis: Banks, Mikell, Henriques, Herrera. 5. Gabriëls L, Nuttin B, Cosyns P. Applicants for 11. Bourne SK, Sheth SA, Neal J, et al. Beneficial Obtained funding: Sheth. stereotactic neurosurgery for psychiatric disorders. effect of subsequent lesion procedures after Administrative, technical, or material support: Acta Psychiatr Scand. 2008;117(5):381-389. nonresponse to initial cingulotomy for severe, Banks, Mikell, Youngerman, Chan, Eskandar, Sheth. Study supervision: Sheth.

134 JAMA Psychiatry February 2015 Volume 72, Number 2 (Reprinted) jamapsychiatry.com

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Response to Dorsal Anterior Cingulotomy Original Investigation Research

treatment-refractory obsessive-compulsive development, use, and reliability. Arch Gen Psychiatry. obsessive-compulsive disorder patients responding disorder. Neurosurgery. 2013;72(2):196-202. 1989;46(11):1006-1011. to fluvoxamine. Psychiatry Res. 2005;138(2):89-97. 12. Rück C, Karlsson A, Steele JD, et al. 29. Douaud G, Smith S, Jenkinson M, et al. 46. Nakatani E, Nakgawa A, Ohara Y, et al. Effects Capsulotomy for obsessive-compulsive disorder. Anatomically related grey and white matter of behavior therapy on regional cerebral blood flow Arch Gen Psychiatry. 2008;65(8):914-921. abnormalities in adolescent-onset schizophrenia. in obsessive-compulsive disorder. Psychiatry Res. 13. Sheehan JP, Patterson G, Schlesinger D, Xu Z. γ Brain. 2007;130(pt 9):2375-2386. 2003;124(2):113-120. Knife surgery anterior capsulotomy for severe and 30. Nichols TE, Holmes AP. Nonparametric 47. Yamanishi T, Nakaaki S, Omori IM, et al. refractory obsessive-compulsive disorder. permutation tests for functional neuroimaging. Changes after behavior therapy among responsive J Neurosurg. 2013;119(5):1112-1118. Hum Brain Mapp. 2002;15(1):1-25. and nonresponsive patients with obsessive-compulsive disorder. Psychiatry Res. 14. Lopes AC, Greenberg BD, Norén G, et al. 31. Tomlinson FH, Jack CR Jr, Kelly PJ. Sequential 2009;172(3):242-250. Treatment of resistant obsessive-compulsive magnetic resonance imaging following stereotactic disorder with ventral capsular/ventral striatal radiofrequency ventralis lateralis thalamotomy. 48. Benkelfat C, Nordahl TE, Semple WE, King AC, gamma capsulotomy: a pilot prospective study. J Neurosurg. 1991;74(4):579-584. Murphy DL, Cohen RM. Local cerebral glucose J Neuropsychiatry Clin Neurosci. 2009;21(4):381-392. metabolic rates in obsessive-compulsive disorder: 32. Desikan RS, Ségonne F, Fischl B, et al. An patients treated with clomipramine. Arch Gen 15. D’Astous M, Cottin S, Roy M, Picard C, Cantin L. automated labeling system for subdividing the Psychiatry. 1990;47(9):840-848. Bilateral stereotactic anterior capsulotomy for human cerebral cortex on MRI scans into gyral obsessive-compulsive disorder. J Neurol Neurosurg based regions of interest. Neuroimage. 2006;31(3): 49. Perani D, Colombo C, Bressi S, et al. [18F]FDG Psychiatry. 2013;84(11):1208-1213. 968-980. PET study in obsessive-compulsive disorder. Br J Psychiatry. 1995;166(2):244-250. 16. Gouvea F, Lopes A, Greenberg B, et al. 33. Behrens TEJ, Johansen-Berg H, Woolrich MW, Response to sham and active gamma ventral et al. Non-invasive mapping of connections 50. Blomstedt P, Sjöberg RL, Hansson M, Bodlund capsulotomy in otherwise intractable between human thalamus and cortex using O, Hariz MI. Deep brain stimulation in the treatment obsessive-compulsive disorder. Stereotact Funct diffusion imaging. Nat Neurosci. 2003;6(7):750-757. of obsessive-compulsive disorder. World Neurosurg. Neurosurg. 2010;88(3):177-182. 2013;80(6):e245-e253. doi:10.1016/j.wneu.2012.10 34. Behrens TE, Woolrich MW, Jenkinson M, et al. .006. 17. Lopes AC, Greenberg BD, Canteras MM, et al. Characterization and propagation of uncertainty in Gamma ventral capsulotomy for obsessive- diffusion-weighted MR imaging. Magn Reson Med. 51. Sturm V, Lenartz D, Koulousakis A, et al. The compulsive disorder: a randomized clinical trial. 2003;50(5):1077-1088. nucleus accumbens: a target for deep brain JAMA Psychiatry. 2014;71(9):1066-1076. stimulation in obsessive-compulsive- and 35. Behrens TEJ, Berg HJ, Jbabdi S, Rushworth anxiety-disorders. J Chem Neuroanat. 2003;26(4): 18. Carter CS, van Veen V. Anterior cingulate cortex MFS, Woolrich MW. Probabilistic diffusion 293-299. and conflict detection. Cogn Affect Behav Neurosci. tractography with multiple fibre orientations: what 2007;7(4):367-379. can we gain? Neuroimage. 2007;34(1):144-155. 52. Greenberg BD, Gabriels LA, Malone DA Jr, et al. 19. Sheth SA, Mian MK, Patel SR, et al. Human Deep brain stimulation of the ventral internal 36. Bach DR, Behrens TE, Garrido L, Weiskopf N, capsule/ventral striatum for obsessive-compulsive dorsal anterior cingulate cortex neurons mediate Dolan RJ. Deep and superficial amygdala nuclei ongoing behavioural adaptation. Nature. 2012;488 disorder: worldwide experience. Mol Psychiatry. projections revealed in vivo by probabilistic 2010;15(1):64-79. (7410):218-221. tractography. J Neurosci. 2011;31(2):618-623. 20. Shenhav A, Botvinick MM, Cohen JD. The 53. Lippitz BE, Mindus P, Meyerson BA, Kihlström 37. Radua J, Mataix-Cols D. Voxel-wise expected value of control: an integrative theory of L, Lindquist C. Lesion topography and outcome meta-analysis of grey matter changes in anterior cingulate cortex function. Neuron. 2013;79 after thermocapsulotomy or gamma knife obsessive-compulsive disorder. Br J Psychiatry. (2):217-240. capsulotomy for obsessive-compulsive disorder. 2009;195(5):393-402. Neurosurgery. 1999;44(3):452-458. 21. Fitzgerald KD, Welsh RC, Gehring WJ, et al. 38. Bourne SK, Eckhardt CA, Sheth SA, Eskandar Error-related hyperactivity of the anterior cingulate 54. Kondziolka D, Flickinger JC, Hudak R. Results EN. Mechanisms of deep brain stimulation for cortex in obsessive-compulsive disorder. Biol following gamma knife radiosurgical anterior obsessive compulsive disorder: effects upon cells Psychiatry. 2005;57(3):287-294. capsulotomies for obsessive compulsive disorder. and circuits. Front Integr Neurosci. 2012;6:29. Neurosurgery. 2011;68(1):28-32. 22. Milad MR, Rauch SL. Obsessive-compulsive 39. Scarone S, Colombo C, Livian S, et al. Increased disorder: beyond segregated cortico-striatal 55. Rück C, Svanborg P, Meyerson BA. Lesion right caudate nucleus size in obsessive-compulsive pathways. Trends Cogn Sci. 2012;16(1):43-51. topography in capsulotomy for refractory disorder. Psychiatry Res. 1992;45(2):115-121. anxiety—is the right side the right side? Stereotact 23. Rauch SL, Dougherty DD, Cosgrove GR, et al. 40. Tang W, Li B, Huang X, et al. Morphometric Funct Neurosurg. 2005;83(4):172-179. Cerebral metabolic correlates as potential predictors brain characterization of refractory of response to anterior cingulotomy for obsessive 56. Figee M, Luigjes J, Smolders R, et al. Deep brain obsessive-compulsive disorder. Prog compulsive disorder. Biol Psychiatry. 2001;50(9): stimulation restores frontostriatal network activity Neuropsychopharmacol Biol Psychiatry. 2013;46: 659-667. in obsessive-compulsive disorder. Nat Neurosci. 126-131. 2013;16(4):386-387. 24. Valente AA Jr, Miguel EC, Castro CC, et al. Regional gray matter abnormalities in 41. Baxter LR Jr, Schwartz JM, Bergman KS, et al. 57. Vogt BA, Nimchinsky EA, Vogt LJ, Hof PR. obsessive-compulsive disorder. Biol Psychiatry. Caudate glucose metabolic rate changes with both Human cingulate cortex: surface features, flat 2005;58(6):479-487. drug and behavior therapy for obsessive-compulsive maps, and cytoarchitecture. J Comp Neurol. 1995; disorder. Arch Gen Psychiatry. 1992;49(9):681-689. 359(3):490-506. 25. Radua J, van den Heuvel OA, Surguladze S, Mataix-Cols D. Meta-analytical comparison of 42. Swedo SE, Pietrini P, Leonard HL, et al. Cerebral 58. Fornito A, Whittle S, Wood SJ, Velakoulis D, voxel-based morphometry studies in glucose metabolism in childhood-onset Pantelis C, Yücel M. The influence of sulcal obsessive-compulsive disorder vs other anxiety obsessive-compulsive disorder. Arch Gen Psychiatry. variability on morphometry of the human anterior disorders. Arch Gen Psychiatry. 2010;67(7):701-711. 1992;49(9):690-694. cingulate and paracingulate cortex. Neuroimage. 2006;33(3):843-854. 26. Cannistraro PA, Makris N, Howard JD, et al. 43. Hansen ES, Hasselbalch S, Law I, Bolwig TG. A diffusion tensor imaging study of white matter in The caudate nucleus in obsessive-compulsive 59. Yücel M, Stuart GW, Maruff P, et al. obsessive-compulsive disorder. Depress Anxiety. disorder. Int J Neuropsychopharmacol. 2002;5(1):1-10. Hemispheric and gender-related differences in the 2007;24(6):440-446. 44. Saxena S, Brody AL, Maidment KM, et al. gross morphology of the anterior cingulate/paracingulate cortex in normal 27. ChiuC-H,LoY-C,TangH-S,etal.Whitematter Localized orbitofrontal and subcortical metabolic volunteers. Cereb Cortex. 2001;11(1):17-25. abnormalities of fronto-striato-thalamic circuitry in changes and predictors of response to paroxetine obsessive-compulsive disorder. Psychiatry Res. treatment in obsessive-compulsive disorder. 60. Yang JC, Ginat DT, Dougherty DD, Makris N, 2011;192(3):176-182. Neuropsychopharmacology. 1999;21(6):683-693. Eskandar EN. Lesion analysis for cingulotomy and limbic leucotomy. J Neurosurg. 2014;120(1):152-163. 28. Goodman WK, Price LH, Rasmussen SA, et al. 45. Ho Pian KL, van Megen HJGM, Ramsey NF, The Yale-Brown Obsessive Compulsive Scale, I: et al. Decreased thalamic blood flow in

jamapsychiatry.com (Reprinted) JAMA Psychiatry February 2015 Volume 72, Number 2 135

Downloaded From: https://jamanetwork.com/ on 09/27/2021