Connectivity-based Parcellation Of The Nucleus Accumbens Into Core And Shell Portions For Stereotactic Target Localization And Alterations In Each NAc Subdivision In MTLE Patients

Introduction:The nucleus accumbens(NAc),an integral part of basal forebrain,has become an important target of deep brain stimulation for some neuropsychiatric disorders when refractory to medical treatment.Within the nucleus there is a clear distinction between the shell and core subregions,defined by distinct cytoarchitectonics and connectivity patterns.Growing evidence supported that the NAc has been involved in epileptogenesis,especially its shell portion.Previous studies performed NAc stimulation in patients with intractable partial epilepsy,which provided initial evidence for its safety and feasibility.However little is known about the exact parcellation within the NAc in individuals,and structural alterations of NAc subdivisions in TLE patients remains unclear.The purpose of our study was to parcel nucleus accumbens into core and shell portions using DTI probabilistic tractography in individual mTLE patients for guiding NAc shell stereotactic target localization and investigate the structural alteration of each NAc subdivision in mTLE patients.Methods:Subjects and data acquisition:MR imaging was conducted on 40 mTLE patients and 25 controls age and gender matched using a 3.0T MR scanner(Philips Achieva TX)with 8-channel head coil.After screening from inclusion and exclusion criteria,a total of 34 patients with unilateral mTLE and 25 healthy controls participated in this study.The DTI acquisition was planned along the AC/PC line using a SE-EPI sequence with TR=7542ms,TE=86ms,b=1000s/mm2,gradient directions=32,FOV=224 × 224 mm,112 ×112 matrix,isotropic 2 mm resolution,70 axial slices with no interslice gap.For anatomical segmentation a 3D T1-weighted scan with was also performed.Data analysis:The DTI connectivity-based parcellation was performed using procedures described in a previous report.For each subject,T1-weighted scans were processed using FIRST(fMRIB Integrated Registration and Segmentation Tool)to localize the NAc.We transformed the NAc seed mask to the individual DTI space.Whole-brain probabilistic tractography was performed using the PROBTRACKX to estimate the connections between each NAc seed voxel and any target voxel in the whole brain.The connectivity matrix between NAc seed voxels and target voxels was derived,and used to generate a symmetric cross-correlation matrix.The cross-correlation matrix was fed into a k-means clustering algorithm to group together the seed voxels that share a similar connection profiles with the rest of the brain.The automated classification using k-means clustering produced individual parcellation of the NAc into 2 subdivisions.Core and shell portions were identified by the location and shape as described in previous anatomical researches.Then we measured diffusion parameters and volumes of each NAc subdivison for statistics.To investigate different anatomical connection patterns of each NAc subregion in the three groups,the subject-specific NAc subdivisions were then transformed into standard MNI space to generate group-averaged subdivisions.Group-averaged subdivisions were determined by overlapping all subjects’ subdivisions.The group-averaged NAc subdivisions were then used as ROIs to define probabilistic connections with predefined ipsilateral target ROIs.We evaluated the probabilistic connections between each group-averaged NAc subdivision and 16 predefined target regions in the standard space using the PROBTRACKX for every subject.These target regions were extracted from the Harvard-Juliet atlas and included the Amygdala(AMY),Caudate(CAU),anterior cingulate cortex(ACC),posterior cingulate cortex(PCC),orbital frontal cortex(OFC),frontal pole(FP),hippocampus(HIP),insula(INS),globus pallidus(GP),paracingulate gyrus(PCG),anterior-parahippocampal gyrus(Ante-PHG),posterior-parahippocampal gyrus(Post-PHG),putamen(PUT),subcallosal cortex(SCC),thalamus(THA)and temporal pole(TP).After tractography,the tracts from each NAc subdivision to any targets were then overlapped onto standard space across all subjects to generate common pathways.The common pathways were then mapped back to the diffusion space,from which the mean FA and MD values were extracted.Statistical analysis:Statistical analysis was performed with SPSS 20.0(IBM SPSS,Chicago,IL).FA,MD,AD,and RD values and three eigenvalues in left and right mTLE groups were compared with those in the control group for each NAc subdivision,using repeated-measures ANOVA to test the effects of hemispheres(2 levels),subregions(2 levels)and groups(2 levels).Post hoc pairwise comparisons with Bonferroni corrections for multiple comparisons were used to discriminate the diffusion parameters of each subdivision of each hemisphere in the three groups.We also assessed group differences of volumes in both the core and shell using two-sample t tests.For the control group,a three-way repeated-measures ANOVA was performed to test the main effects of hemispheres,NAc subdivisions and target regions on the normalized connection probabilities.Pairwise comparisons with Bonferroni corrections for multiple comparisons were used to discriminate the connectivity probabilities between each NAc subdivision to all targets.To quantify the differences in anatomical connections of each NAc subregion in the healthy controls and TLE groups,the connection probabilities with the sixteen target regions were entered into an NAc subregions(2 levels)× target regions(16 levels)×hemispheres(2 levels)repeated-measures ANOVA.Post hoc pairwise t-tests with Bonferroni corrections for multiple comparisons were performed to visualize the anatomical connection probabilities differences in each subdivision in the healthy control and TLE groups.Two-sample t tests were performed to assess the group differences on each group common tract.Results:We successfully segmented the nucleus accumbens into two different portions in 24 of 25 controls,14 of 16 left mTLE and 14 of 18 right mTLE patients,on the basis of their anatomical connection patterns.In each subject,we observed an approximate medial-lateral parcellation,dividing the NAc into two parts.The location and shape of the two portions approximately matched cytoarchitecturally identified NAc core and shell in postmortem human tissue.Based on this correspondence,we identified the medial-caudal portion as shell and the lateral-rostral portion as core of NAc.Our results revealed that both left and right mTLE patients exhibited decreased FA and increased radial diffusivity(RD)in the shell portion,while there were no significant alterations in the core.Moreover we demonstrated relatively distinct structural connectivity of each NAc subregion,and detected more extensive connection abnormalities in the NAc shell when comparing mTLE groups with controls.Specially,the combination of increased connectivity and less-fiber integrity in SCC-shell and OFC-shell tracts was particularly striking and has not been described in TLE patients before.Discussion and Conclusions:This is the first study to parcel nucleus accumbens into core and shell portions using DTI probabilistic tractography among individual mTLE patients.The exact stereotactic anatomy of the core and shell portion of NAc would help neurosurgeons to perform the NAc shell deep brain stimulation(DBS)in suppressing seizure propagation.Our results revealed that significant structural abnormalities and more extensive connection alterations mainly existed within the NAc shell portion,while absence of significant alterations in the core,which may provide anatomical evidence in support of the role of the NAc shell in epileptogenesis.The neuronal fiber loss and paradoxical increased connectivity in the SCC-shell and OFC-shell tracts in mTLE patients may suggest the regeneration of aberrant connections,a compensatory and repair process ascribed to recurrent seizures that constitutes a characteristic change in the epileptic structural network.