Introduction

In cases of medically intractable psychiatric disease, neurosurgical treatment is an option for reducing symptom burden and increasing quality of life. Forty to sixty percent of patients with obsessive-compulsive disorder (OCD) do not have satisfactory response to optimal management (Pallanti and Quercioli 2006). As a result, neurosurgical techniques, which include stereotactic lesioning of specific areas of the brain involved in psychopathology, have been used for therapeutic benefit. Procedures currently used include anterior cingulotomy, subcaudate tractotomy, and a combination of the two, called the limbic leucotomy (Kelly et al. 1973; Mashour et al. 2005).

Outcomes for these procedures have been largely positive. The most recent studies in OCD patients have shown that approximately 45 % improve after cingulotomy (Dougherty et al. 2002; Jung et al. 2006) and 70 % improve after subcaudate tractotomy (Bridges et al. 1994; Hodgkiss et al. 1995). At Massachusetts General Hospital, the limbic leucotomy is completed with a subcaudate tractotomy after non-response to anterior cingulotomy, and a recent study showed that 73 % of these patients demonstrate symptom improvement (Bourne et al. 2013).

The target of the anterior cingulotomy is more anatomically homogenous, with the lesion directly targeting the cingulum bundle (Leiphart and Valone 2010). In contrast, the effects of subcaudate tractotomy, developed by Knight in 1960, are less clear; the lesion targets the white matter ventral to the head of the caudate nucleus to disrupt corticostriatothalamic pathways (Feldman et al. 2001; Knight 1969; Mashour et al. 2005). Previous work by our group studied the direct gray and white matter structures that are lesioned, which include the uncinate fasciculus, orbitofrontal cortex, and nucleus accumbens, but the disrupted connections between specific structures were not delineated (Yang et al. 2014).

Given the notable efficacy of SCT in cases of medically intractable OCD, an understanding of the disrupted connectivity may provide further insight into the neural substrates of OCD and provide guidance in making the SCT lesion more effective. In this study, we used multi-tensor based tractography in 60 normal subjects to delineate and to identify the anatomical connections between the medial, central, and lateral orbitofrontal cortex (OFC) and thalamus or brainstem that pass through the region affected by SCT.

Materials and methods

Subjects

The initial dataset for this study involved healthy subjects with an average age of 34 and without diagnosed neurological disorder or drug dependency, with demographics described in a prior study (Makris et al. 2013). Informed consent was obtained from all subjects, and the study was IRB-approved.

MRI acquisition

Description of MRI parameters is described in further detail in a prior study (Makris et al. 2013). A 3T GE Echo speed system (General Electric Medical Systems, Milwaukee, WI) was used to obtain T1-weighted and T2-weighted images, and diffusion data were collected with an echo planar imaging (EPI) HARDI pulse sequence.

Briefly, parameters are as follows: T1-weighted sequence: TR = 7.4 ms, TE = 3 ms, TI = 600, 100 flip angle, 25.6 cm2 field of view, matrix = 256 × 256, and voxel dimensions of 1 × 1 × 1 mm3. T2-weighted sequence: TR = 2,500 ms, TE = 80 ms, 25.6 cm2 field of view, and voxel dimensions of 1 × 1 × 1 mm3. EPI HARDI sequence: TR = 17,000 ms, TE = 78 ms, FOV = 24 cm, 144 × 144 encoding steps with slice thickness = 1.7 mm, resulting in isotropic 1.7 × 1.7 × 1.7 mm3 voxels. A double echo sequence with an 8-channel coil and ASSET (Array Spatial Sensitivity Encoding Techniques, GE) was used. Ultimately, 51 non collinear diffusion directions with b = 900 s/mm2 and eight baseline scans with b = 0 s/mm2 were acquired.

Diffusion image processing and tractography

Diffusion images were corrected for motion and eddy currents by using FSL-based software (The Oxford Centre for Functional MRI of the Brain, University of Oxford, Oxford, UK).

Anatomic structures close to the OFC, such as bone and air-filled sinuses, result in magnetic field inhomogeneity that often causes distortion in diffusion data. As a result, using Advanced Normalization Tools (ANTs) (Avants et al. 2008), EPI distortion correction was performed via 1D deformable registration with mutual information metric of the b = 0 s/mm2 image to a rigidly aligned reference T2-weighted image (Ardekani and Sinha 2005). This method restricts the registration of the diffusion image onto the T2-weighted image such that the diffusion data may only be warped in the same direction as the phase encoding direction. Ultimately, 60 subjects from the original dataset were fully processed in preparation for tractography.

Two-tensor whole brain tractography was subsequently implemented with MATLAB using an algorithm that uses an unscented Kalman filter (Malcolm et al. 2010). This form of tractography involves a recursive process that both fits local parameters at each step and propagates the fiber in the most stable direction, allowing simultaneous tractography and model estimation. This model also uses a covariance matrix to measure confidence, which reduces false positives, and implements two eigenvalues per voxel, which improves resolution of and accounts for branching and crossing fibers. Whole-brain tractography was used to best account for all potential tracts among the ROIs, which may be otherwise limited in solely ROI-based tractography.

Region of interest (ROI) delineation

The representative SCT lesion was delineated by using data from 11 subjects who had undergone a limbic leucotomy for medically intractable psychiatric diagnoses between 2001 and 2008. At our institution, the limbic leucotomy is carried out by adding the subcaudate tractotomy lesion after a prior anterior cingulotomy. Details of the subcaudate tractotomy lesion procedure can be found in a prior study (Yang et al. 2014).

T1-weighted images of subjects with SCT lesions were segmented in Free view, which is part of the Free Surfer imaging software package (Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Boston, MA, USA), and the resulting output was a volume mask (per subject). Since these lesion masks were delineated on clinical MR images (in native acquisition space), a linear, manual, landmark-based (anterior commissure, posterior commissure, and inter hemispheric plane (Filipek et al. 1994; Talairach and Tournoux 1988) registration with 12 degrees of freedom was used to register each subject to the Montreal Neurological Institute (MNI) MNI152 brain. As clinical MR images containing large lesions may cause automated methods to fail, and because clinical images may not be of the entire brain due to patient considerations, this approach was considered the most accurate. One author (JCY) performed the lesion segmentations, which were corroborated by a neuroanatomist (NM). To determine intra-rater reliability, re-segmentation on a subset of lesions was performed and Cronbach’s alpha was calculated to identify volumetric and spatial reliability.

The representative SCT lesion was subsequently created by using FSLMaths, part of the FMRIB Software Library package. All registered lesion masks were added using FSLMaths and voxels that overlapped in 5 or more subjects were extracted to form the representative lesion. The average voxel size of all subjects was 13320 ± 686 voxels (standard error), and the representative SCT lesion was 13584 voxels.

ROI identification for medial OFC (mOFC), central OFC (cOFC), lateral OFC (lOFC), thalamus, and brainstem were isolated from the MNI152-based Massachusetts General Hospital-Center for Morphometric Analysis (CMA) structural atlas (Fig. 1) (Caviness et al. 1996; Makris et al. 1999; Meyer et al. 1999; Poellinger et al. 2001; Rademacher et al. 1992). Of note, the cOFC ROI was derived from combining the ROIs known as the anterior OFC (aOFC) and posterior OFC (pOFC) (Fig. 1). As many studies that have divided the OFC into different regions use the cOFC ROI rather than the aOFC and pOFC ROIs, this cOFC ROI definition allowed the resultant data to be best comparable to prior studies (Jbabdi et al. 2013; Lehman et al. 2011). Lehman et al. (2011) used a tracer approach to understanding connections from homologous ventral medial prefrontal cortex (vmPFC), mOFC, cOFC, and lOFC regions of the macaque, and Jbabdi et al. (2013) built upon this work by using probabilistic tractography with human and macaque diffusion imaging that was seeded in discrete areas of the vmPFC, cOFC, and lOFC. Notable differences in our study’s ROI delineation include the investigation of the mOFC region as well as utilization of the entire ROI for extracting fiber tracts from whole brain tractography analysis, rather than from seed regions.

Fig. 1
figure 1

Region of Interest (ROI) Definition (a) Sagittal T1-weighted MNI152 brain with medial orbitofrontal cortex (mOFC, blue), subcaudate tractotomy lesion (SCT, yellow), thalamus (green), and brainstem (orange). Red vertical lines indicate the coronal sections in D, E, and F. (b) 3D anterior projection of ROIs on T1-weighed MNI152 brain. In addition to (a), lateral orbitofrontal cortex (lOFC, red) and central orbitofrontal cortex (cOFC, purple) are shown. (c) 3D projection of ventral surface of MNI12 brain, with mOFC (blue), cOFC (purple), and lOFC (red). For comparison, aOFC (pink) and pOFC (cyan) additionally shown. (d) Coronal T1-weighted MNI152 brain with mOFC (blue), cOFC (purple), and lOFC (red) (e) Coronal T1-weighted MNI152 brain with SCT (yellow) (f) Coronal T1-weighted MNI152 brain with thalamus (green) and brainstem (orange)

The SCT lesion was subsequently added to the CMA atlas, creating a complete MNI152-based label map. Finally, the T1-weighted MNI152 brain was registered to the baseline DWI of each subject using ANTs, and the label map underwent the same deformation. These labels were subsequently used to filter out fibers from the whole-brain tractography that traversed the following ROI combinations: (1) mOFC-SCT-thalamus, (2) mOFC-SCT-brainstem, (3) cOFC-SCT-thalamus, (4) cOFC-SCT-brainstem, (5) lOFC-SCT-thalamus, and (6) lOFC-SCT-brainstem.

Fiber warping and probability maps

To create probability maps for tracts affected by the SCT lesion, the baseline DWI data for each subject was ultimately registered to the MNI152 template using ANTs. These deformations were then applied to the fiber tracts to place all fibers in MNI152 space. Fiber tract masks were created and averaged to obtain a probabilistic map, and maps were thresholded at 1.5 % for clarity. For each map, center-of-mass coordinates were subsequently obtained in each intersecting coronal plane, and coordinates were transformed to Talairach space.

Results

Data from a total of 60 subjects were corrected using a registration-based method, on which whole-brain tractography and fiber isolation were performed to delineate fibers that traversed the mOFC or cOFC or lOFC, SCT lesion, and thalamus or brainstem. As described previously (Yang et al. 2014), creation of the SCT lesion, based on clinical MRI scans, had high intra-rater reliability, with the percent of overlapping voxels calculated to be 91 % and with a Cronbach’s alpha of 0.98 based on lesion volume.

Fibers connecting the mOFC, SCT, and thalamus were most commonly identified in this dataset, with mOFC-SCT-thalamus fibers identified on the right side in 86.7 % of subjects and on the left side in 95 % of subjects. lOFC-SCT-thalamus fibers were isolated on the right side in 56.7 % and on the left side in 53.3 % of subjects. Right mOFC-SCT-brainstem fibers were found in 55 %, left mOFC-SCT-brainstem fibers in 63.3 %, right lOFC-SCT-brainstem fibers in 53.3 %, and left lOFC-SCT-brainstem fibers in 46.7 % of subjects. For the relatively larger cOFC ROI, cOFC-SCT-thalamus fibers were found in 100 % of subjects, bilaterally. The cOFC-SCT-brainstem fiber was isolated in 88.3 % of subjects on the right side and 95 % of subjects on the left side.

Fibers coursing between the OFC and brainstem or thalamus, that were interrupted by the SCT lesion, were found to run in the internal capsule with a specific orientation (Figs. 2 and 3). Tracts running from the lOFC travel dorsally to those from the cOFC, which are similarly dorsal to the mOFC. While mOFC and cOFC tracts to the thalamus additionally involved a white matter tract that traveled ventrolateral to the putamen in the external capsule, these fibers were not seen in mOFC, cOFC, or lOFC tracts to the brainstem (Fig. 3). In examining the mOFC, cOFC, and lOFC, fibers targeting the thalamus versus the brainstem diverged posterior to the anterior commissure (Figs. 2 and 3).

Fig. 2
figure 2

Probability Maps of Fibers Traversing the SCT lesion (a) Coronal T1-weighted MNI152 brain with probability map for fibers that connect thalamus to subcaudate tractotomy lesion (SCT) and medial orbitofrontal cortex (mOFC) (blue) or lateral orbitofrontal cortex (lOFC) (orange) (b) Coronal T1-weighted MNI152 brain with probability map for fibers that connect brainstem to SCT and mOFC (blue) or lOFC (orange) (c) Sagittal T1-weighted MNI152 brain with probability map for fibers that connect mOFC to thalamus (green) or brainstem (brown) (d) Sagittal T1-weighted MNI152 brain with probability map for fibers that connect lOFC to thalamus (green) or brainstem (brown)

Fig. 3
figure 3

3D Projections of Fiber Tracts (a) 3D anterior projection on T1-weighted MNI152 brain. ROIs: medial orbitofrontal cortex (mOFC, dark blue), lateral orbitofrontal cortex (lOFC, dark red), subcaudate tractotomy lesion (SCT, yellow), and thalamus (green). Blue tract connects mOFC, SCT, and thalamus; red tract connects lOFC, SCT, and thalamus. (b) 3D right lateral projection of tracts on T1-weighted MNI152 brain. Dark blue tract connects mOFC, SCT, and thalamus; dark red tract connects lOFC, SCT, and thalamus. The central orbitofrontal cortex (cOFC) tract is not shown to facilitate visualization. (c) 3D anterior projection on T1-weighted MNI152 brain. ROIs: mOFC (dark blue), lOFC (dark red), SCT (yellow), and brainstem (orange). Blue tract connects mOFC, SCT, and brainstem; red tract connects lOFC, SCT, and brainstem. (d) 3D right lateral projection of tracts on T1-weighted MNI152 brain. Light blue tract connects mOFC, SCT, and brainstem; light red tract connects lOFC, SCT, and brainstem. The cOFC tract is not shown to facilitate visualization. (e) 3D right lateral projection of tracts connected to mOFC on T1-weighted MNI152 brain. Green tract connects mOFC, SCT, and thalamus; orange tract connects mOFC, SCT, and brainstem. (f) 3D right lateral projection of tracts connected to lOFC on T1-weighted MNI152 brain. Green tract connects lOFC, SCT, and thalamus; orange tract connects lOFC, SCT, and brainstem

The SCT lesions and each tract probability map were plotted by using a center-of-mass approach to generate Talairach coordinates (Makris et al. 2013) (Figs. 4, 5, and 6; Supplementary Table 1). These plots show similar findings to the probability maps, in which the fibers originating from the mOFC are ventral to those from the lOFC, with an area of tract divergence slightly anterior to the anterior commissure with fibers separating to target either the thalamus or brainstem (Fig. 4). At the level of the anterior commissure (the 0 mm position on the Talairach anterior-posterior axis), the fibers have a particular pattern in which those connected to the thalamus are dorsal to those connected to the brainstem. Probability maps of cOFC fibers demonstrated overlap with mOFC. However, using the center-of-mass approach, separate averaged tracts could be resolved and demonstrate that cOFC fibers to both the thalamus and brainstem follow similar organizing rules as the mOFC and lOFC, as cOFC fibers were found to travel between the mOFC and lOFC fibers (Figs. 4, 5, and 6).

Fig. 4
figure 4

Orbitofrontal Tracts Represented as Talairach Coordinates (a) Right lateral and axial projections of tracts with approximate Talairach brain outline. The top graph is a right sagittal projection, with the Talairach anterior-posterior coordinate on the x-axis and the Talairach inferior-superior coordinate on the y-axis. The bottom graph is an axial projection, with the Talairach anterior-posterior coordinate on the x-axis and the Talairach right-left coordinate on the y-axis. These tracts run through the medial orbitofrontal cortex (mOFC), central orbitofrontal cortex (cOFC), or lateral orbitofrontal cortex (lOFC); subcaudate tractotomy lesion (SCT); and thalamus or brainstem. Tracts: mOFC-SCT-thalamus (red: fibers traveling through the internal capsule), mOFC-SCT-brainstem (blue), cOFC-SCT-thalamus (orange: fibers traveling through the internal capsule), cOFC-SCT-brainstem (green), lOFC-SCT-thalamus (yellow), and lOFC-SCT-brainstem (cyan). (b) 3D oblique and C. coronal projections of tracts in Talairach space, using center-of-mass coordinates. Tracts: mOFC-SCT-thalamus (red: fibers traveling through the internal capsule), mOFC-SCT-brainstem (blue), cOFC-SCT-thalamus (orange: fibers traveling through the internal capsule), cOFC-SCT-brainstem (green), lOFC-SCT-thalamus (yellow), and lOFC-SCT-brainstem (cyan)

Fig. 5
figure 5

3D Projections of Talairach Coordinates for Orbitofrontal Cortex-Thalamus Tract (a) 3D sagittal, (b) coronal (level of the nucleus accumbens, Talairach y-coordinate = +8 mm), and (c) coronal (level of the anterior commissure, Talairach y-coordinate = 0 mm) projections of tracts in Talairach space, using center-of-mass coordinates. These tracts run through the medial orbitofrontal cortex (mOFC), central orbitofrontal cortex (cOFC), or lateral orbitofrontal cortex (lOFC); subcaudate tractotomy lesion (SCT); and thalamus. Tracts: mOFC-SCT-thalamus (red: fibers traveling through the internal capsule), cOFC-SCT-thalamus (orange: fibers traveling through the internal capsule), and lOFC-SCT-thalamus (yellow)

Fig. 6
figure 6

3D Projections of Talairach Coordinates for Orbitofrontal Cortex-Brainstem Tract (a) 3D sagittal, (b) coronal (level of the nucleus accumbens, Talairach y-coordinate = +8 mm), (c) coronal (level of the anterior commissure, Talairach y-coordinate = 0 mm), and (d) axial projections of tracts in Talairach space, using center-of-mass coordinates. These tracts run through the medial orbitofrontal cortex (mOFC), central orbitofrontal cortex (cOFC), or lateral orbitofrontal cortex (lOFC); subcaudate tractotomy lesion (SCT); and brainstem. Tracts: mOFC-SCT-brainstem (blue), cOFC-SCT-brainstem (green), and lOFC-SCT-brainstem (cyan). (e) Expanded region of brainstem of (d). The yellow region indicates the approximate area of intersection between the tracts and the axial slice. (f) 3D oblique projection of tracts in Talairach space, with a view from the dorsal aspect

In visualizing the probability map projections and the SCT lesion, an understanding of the primarily affected tracts is made clear. Given their ventral position, mOFC tracts appear to be more significantly involved in the SCT lesion, as compared with the more dorsal lOFC tracts. Correspondingly, mOFC tracts were most affected by the dorsolateral portions of the SCT lesions.

Discussion

In this study, we used multi-tensor whole-brain tractography to isolate white matter tracts from the orbitofrontal cortex that are affected by subcaudate tractotomy (SCT). We show that tracts connecting the mOFC, cOFC and lOFC with the thalamus or brainstem are all affected by SCT and that these tracts have a particular topographic organization, in which tracts connecting the mOFC travel ventral to those connecting the lOFC, with cOFC fibers coursing in between. At the level of the anterior commissure, these fibers are organized such that those connected to the thalamus travel dorsally to those connected to the brainstem.

In order to reduce the possibility of having isolated false positives using this method of multi-tensor tractography, a neuroanatomical examination of these fibers, with comparison to atlases was conducted. Tracts that travel between the orbitofrontal cortex and thalamus are part of the anterior thalamic radiation and connect with the medial aspect of the thalamus (Fig. 5). Using approximate Talairach coordinates and an atlas-based approach (Talairach and Tournoux 1988), fibers from the lOFC were seen to connect to the mediodorsal nucleus (−20 mm A-P axis, ±5 mm M-L axis, 8 mm I-S axis) while fibers from the mOFC terminate in the mediodorsal and pulvinar nuclei (−30 mm A-P axis, ±15 mm M-L axis, 10 mm I-S axis). The OFC-SCT-brainstem pathways were localized to the dorsal aspect of the brainstem (Fig. 6), and these fibers are believed to be part of the medial forebrain bundle based on their location relative to the cerebral aqueduct and their trajectory through the brainstem and ventral diencephalon (Fig. 55 (Paxinos and Huang 1995); Fig. 191.13 (Nieuwenhuys et al. 1988); (Nieuwenhuys et al. 2008)) (Coenen et al. 2012; Coenen et al. 2011). The latter is a major pathway interconnecting the forebrain with the brainstem and is associated with limbic functions (Nieuwenhuys et al. 2008).

In addition, these fiber connections are further supported by primate tracer studies. The connection between the OFC and the pulvinar nucleus has been substantiated in a number of studies (Barbas et al. 1991; Bos and Benevento 1975; Cavada et al. 2000; Kievit and Kuypers 1977; Romanski et al. 1997; Trojanowski and Jacobson 1976; Yeterian and Pandya 1988). In addition, the connections between the OFC and the mediodorsal nucleus (Barbas et al. 1991; Ray and Price 1993; Yeterian and Pandya 1988) and the OFC and the brainstem (An et al. 1998; Frankle et al. 2006; Porrino and Goldman-Rakic 1982) have been demonstrated. Overall, these primate anatomical studies support the OFC-SCT-thalamus and OFC-SCT-brainstem fibers elucidated in this study.

With the support of these anatomical data, the subcaudate tractotomy ROI used in this study provides further insight into the neuropathology underlying OCD. The neuropathobiology of OCD has been described in terms of the corticostriatothalamocortical (CSTC) circuit, which has been implicated in OCD (Bourne et al. 2012; Haber and Brucker 2009). Multiple studies have demonstrated that areas of cortex connected by the CSTC circuit can be anatomically and functionally different in disease states. For example, meta-analyses have shown that patients with OCD have reduced gray matter density in the dorsolateral prefrontal cortex (DLPFC) and OFC and reduced volume of the anterior cingulate cortex (ACC) and OFC (Rotge et al. 2009; Rotge et al. 2010a; Szeszko et al. 1999). In addition, fMRI and PET studies of OCD patients have shown that stimulus presentation activates the OFC, DLPFC, ACC, and basal ganglia (Adler et al. 2000; den Braber et al. 2010; Breiter et al. 1996; McGuire et al. 1994; Rauch et al. 1994; Rotge et al. 2008). Notably, successful treatment of OCD by pharmacologic, behavioral, and neurosurgical methods reduces activity in this circuit (Baxter et al. 1992; Haber and Brucker 2009; Perani et al. 1995; Schwartz et al. 1996; Swedo et al. 1992).

The OFC has been of particular interest in the context of OCD given its activity under disease states that predict treatment response and because OFC morphometric abnormalities may offer insight into OCD pathophysiology (Greenberg et al. 2010; Piras et al. 2013). Studies have implicated aberrant OFC activity and connectivity in OCD (Beucke et al. 2013; Hou et al. 2012), and differential roles of the mOFC and lOFC are thought to play a role in OCD symptomatology, such as with delay discounting (Bari and Robbins 2013). While it remains unclear whether the mOFC or the lOFC plays a more significant role in disease, with both being implicated in recent studies (Hoexter et al. 2013; Rotge et al. 2010b), animal models have provided possible insight into disease processes in  vivo. Recent studies have shown that stimulation of the OFC-ventromedial striatum network can recapitulate an OCD phenotype in rodents, which further supports the role of CSTC dysregulation in OCD (Ahmari et al. 2013). Interestingly, in a separate study, stimulation of lOFC and its terminals in the striatum in rodents led to a reduced OCD-like phenotype (Burguière et al. 2013).

In non-human primates, the medial-lateral organization of the OFC subdivisions use a particular route, aligned in a ventral-dorsal pattern, to reach their targets (Lehman et al. 2011). Ultimately, fibers from the medial portions of the OFC and PFC were found to travel ventral to fibers from the lateral portions of the OFC (Lehman et al. 2011). A subsequent study used probabilistic tractography to re-demonstrate these findings regarding the cOFC and lOFC in small numbers of both macaques and humans, though with some technical challenges in replicating vmPFC tracer data (Jbabdi et al. 2013).

Our study builds upon these data by using a different tractography algorithm and by including analysis of the mOFC in a large number of patients. Ultimately, our method allowed us to show these organizational rules are clearly maintained in the mOFC, cOFC, and lOFC. In addition, our center-of-mass analysis allowed further consolidation and refinement of these tracts to illustrate better the spatial properties of these tracts and to define them as belonging to the anterior thalamic radiations (OFC-SCT-thalamus) and to the medial forebrain bundle (OFC-SCT-brainstem). Deterministic tractography may be more conservative in visualizing connecting tracts while probabilistic tractography may leak into unexpected regions (Moldrich et al. 2010; Qazi et al. 2009). In addition, our correlation with anatomical primate data further strengthens the elucidated fiber connections. Importantly, our approach, which uses clinical data rather than simulation (Schoene-Bake et al. 2010), shows the clinical relevance of the anatomical organization of these fiber tracts involved in OCD.

Deep brain stimulation has been used with success for OCD, with recent studies indicating that it improves overall quality of life (Ooms et al. 2014). One of the commonly used targets includes the ventral capsule/ventral striatum (VC/VS), which affects an area that includes the nucleus accumbens (Blomstedt et al. 2013; de Koning et al. 2011). The topographic pattern by which different parts of the OFC reach their targets in the thalamus and brainstem has implications for the VC/VS DBS site, as the organization of traveling fibers implies that stimulation of specific tracts can be modulated by examining the spatial characteristics of each electrode contact (Lehman et al. 2011). This region is similar to that affected by SCT, and our data further support the idea that differential fiber tracts that are connected to subsections of the OFC could potentially be activated by selective DBS programming.

Based on the organization of the traveling mOFC and lOFC fibers, the use of SCT for treatment of medically resistant OCD reinforces the role of these structures in psychopathology. Furthermore, our previous pilot study (Yang et al. 2014), which sought to understand better how lesion placement and morphometry could be related to clinical outcome, suggested that SCT lesions placed more posteroventrally tended to have more positive clinical outcome. Applied to the current study, this would suggest that modulation of the more ventral tracts from the mOFC would lead to better outcome, which is in line with animal model studies. However, given that the prior study that sought to link lesion analysis and clinical outcome was a pilot investigation, these results should be considered tentative.

Limitations of this work include the use of healthy volunteers for understanding white matter tract involvement in the SCT lesion; one study has suggested diffuse white matter changes in patients with OCD (Jayarajan et al. 2012). Another limitation of this study is its basis in tractography, as diffusion data are representative only of extracellular diffusion patterns rather than definitive anatomic tracts. Nevertheless, we have undertaken a large study of 60 patients to support our findings and have utilized a technique that accounts for crossing fibers as well as branching patterns. Overall, this dataset is robust and provides strong evidence that tracts connecting the mOFC, cOFC, and lOFC with the thalamus or brainstem are affected in SCT.

Conclusion

Subcaudate tractotomy is an effective neurosurgical procedure for the treatment of medically intractable OCD. Using multi-tensor tractography, tracts that connect with the mOFC, cOFC, and lOFC were found to travel through the SCT lesion, providing more evidence of their involvement in OCD psychopathology. The patient-derived SCT lesion ROI provides a clinical context for the tractography data, which has not been done in prior studies. These tracts follow a particular orientation as they connect with either the thalamus or brainstem – via the anterior thalamic radiations and medial forebrain bundle, respectively – suggesting that differential modulation of these tracts is possible and may have implications in making the SCT procedure more efficacious and in developing unique DBS programming paradigms.