Cortical Thickness and White Matter Surface Morphology in Tourette Syndrome: A Cohort Study Authors: Sahar Delavari, Adam C Frank, Ravi Bansal, Flora M Vaccarino, Robert A King, Paul Lombroso, Lawrence Scahill, James F Leckman, Bradley S Peterson Published: 2026 ABSTRACT AbstractObjectiveTo examine cortical thickness and white matter surface morphology in a large sample of individuals with Tourette syndrome (TS) and neurotypical controls across the lifespan, and to assess associations with symptom severity, comorbidities, and medication use.MethodsCross‐sectional neuroimaging cohort study utilizing high‐resolution, T1‐weighted magnetic resonance imaging scans obtained in 264 participants (151 TS, 113 neurotypical controls), aged 6–63 years. Image processing included rigorous quality control and deformation‐based morphometry. Voxel‐wise measures of cortical thickness and white matter surface displacement from a template brain were calculated. General linear models applied voxel‐wise tested group differences adjusted for age, sex, and socioeconomic status. Additional covariates in sensitivity analyses included full‐scale IQ, comorbid illnesses, and medication use. Interactions of group with age and sex, as well as associations with symptom severity, were also evaluated.ResultsCompared with controls, TS participants showed extensive cortical thinning in precentral and postcentral gyri, inferior and middle frontal gyri, and inferior parietal lobule, as well as white matter surface displacement in overlapping regions. These effects remained significant after covarying for IQ, comorbidity, and medication use. Neither age nor sex significantly moderated these group differences. Attention deficit hyperactivity disorder symptom severity was associated with additional frontoparietal cortical thinning; obsessive–compulsive disorder severity showed no effect. Tic symptom severity did not associate significantly with brain measures anywhere.InterpretationThese findings suggest that cortical thinning and associated white matter reductions in TS reflect stable neurodevelopmental traits rather than transient symptom states. These cortical portions of cortico‐striato‐thalamo‐cortical circuits may serve as targets for future therapeutic interventions. ANN NEUROL 2026;100:1104–1114 The clinical presentation of Tourette syndrome (TS) varies widely in terms of age‐specific symptom expression, differences in prevalence across sexes, and patterns of comorbidity. 1 , 2 , 3 , 4 , 5 , 6 , 7 , 8 , 9 Identifying the neuroanatomical correlates of these clinical features is important for understanding the pathogenesis of TS, and potentially for better informing diagnosis, prognosis, and treatment. Investigating this variability requires representative samples that span a broad age range and that include individuals with diverse symptom profiles. Most prior neuroimaging studies have focused on pediatric samples in efforts to minimize sample heterogeneity and to improve internal validity. 10 Those sampling strategies, however, limit the generalizability and reproducibility of findings, especially underrepresenting adults who have persistent symptoms and functional impairment. 11 Additionally, most anatomical magnetic resonance imaging (MRI) studies in TS patients have used coarse region‐of‐interest analyses or volumetric methods, which may obscure localized anatomical differences. 12 , 13 Although some studies have reported volume reductions in the basal ganglia, frontal cortex, or sensorimotor areas, 14 , 15 few have used voxel‐wise measures of cortical thickness or white matter across the cerebral surface. These higher‐resolution methods may be better able to detect more subtle, regionally specific alterations expected in TS and other neurodevelopmental disorders. 13 , 16 Furthermore, the extent to which structural differences are associated with clinical symptoms, such as attention deficit hyperactivity disorder (ADHD) or obsessive–compulsive disorder (OCD), or with psychotropic medication use, remains unclear due to small sample sizes and limited control for these covariates in previous studies. We aimed to address these knowledge gaps by analyzing high‐resolution anatomical MRI data from a large, well‐characterized cohort of children and adults with TS and age‐matched neurotypical controls, which enabled us to examine age‐related effects across the lifespan. In this same sample, we previously reported cerebral volume reduction in sensorimotor and frontoparietal regions, 17 which suggested the presence of underlying disruptions in cortico‐striato‐thalamo‐cortical (CSTC) circuits that support motor control, inhibition, and multisensory integration—functional domains commonly impaired in TS. 15 , 16 , 18 , 19 , 20 , 21 Those volume parcellations were coarse, however, and they included both gray and white matter volumes, limiting our ability to localize disturbances in TS to gray or white matter and to specific cortical subregions. In the present study, we therefore applied deformation‐based morphometry to quantify fine‐grained alterations at millimeter‐level spatial resolution in cortical thickness and white matter surface morphology. We also assessed the moderating effects of age and sex of these group differences, as well as the effects of comorbid ADHD and OCD symptom severity and medication use on cortical thickness and white matter morphology. This study provides one of the most detailed characterizations to date of cortical morphology in TS that may inform pathophysiological models and treatment interventions across the lifespan. Materials and Methods Participant Characteristics Participants with a lifetime diagnosis of TS (n = 151) were recruited from the Tic and Obsessive‐Compulsive Disorders Specialty Clinic at the Yale Child Study Center, New Haven, CT, US. Control participants (n = 113) were randomly selected from a database of 10,000 individuals provided by a telemarketing company, then recruited and group‐matched to the TS participants by age and ZIP code (a 5‐digit number used by the US postal service to code a local geographic neighborhood, which generally contain residences with similar incomes and home values). They were contacted via introductory letter and follow‐up phone screening; approximately 10% of eligible controls participated. The protocol was approved by the Yale University Human Investigation Committee. Written informed consent (with parental permission and child assent for minors) was obtained from all participants. Participants ranged in age from 6 to 63 years and were predominantly right‐handed. Lifetime psychiatric disorders excluded from both groups were DSM‐IV diagnoses of pervasive developmental disorder, intellectual disability, attachment disorders, gender identity disorder, post‐traumatic stress disorder, psychosis, mania, and substance dependence. Additional exclusions in the control participants were a lifetime history of tics, OCD, or ADHD, and any current DSM‐IV Axis I disorders. Both groups excluded individuals with seizures, head trauma with loss of consciousness, current substance abuse or past dependence, full‐scale IQ <75, or other movement disorders (Sydenham's chorea, Parkinson's disease, tardive dyskinesia, dystonia). OCD and ADHD were not excluded in the TS group, for two reasons: (1) they may be inherent expressions of the pathophysiology of TS, and (2) they are such common co‐occurring conditions in TS that excluding them would distort the representativeness of the sample and impair the generalizability of findings. Diagnoses were confirmed through clinical interviews and the clinician‐administered Schedule for Tourette and Other Behavioral Syndromes, 22 which integrates the Schedule for Affective Disorders and Schizophrenia–Present and Lifetime Version (child or adult forms) 23 , 24 with TS‐ and OCD‐specific modules. Diagnoses were assigned by consensus of 2 child psychiatrists reviewing all available data. Severity of current and worst‐ever tics, OCD, and ADHD symptoms was assessed using the Yale Global Tic Severity Scale, 25 Yale‐Brown Obsessive Compulsive Scale, 26 , 27 and DuPaul‐Barkley ADHD Rating Scale. 28 Full‐scale IQ was estimated using the Kaufman Brief Intelligence Test 29 , 30 for children and selected subtests of the Wechsler Adult Intelligence Scale–Revised for adults. 31 Socioeconomic status (SES) was assessed using the Hollingshead Index. 32 MRI Scanning T1‐weighted anatomical MRI data were acquired on single 1.5‐T scanner with a quadrature head coil (GE Signa, Milwaukee, WI, US) using a 3D spoiled‐gradient echo sequence. To minimize noise discomfort and motion, participants wore earplugs and received relaxation training before the scan. They were monitored by a staff member positioned at the magnet bore who provided real‐time encouragement to remain still (pulse sequence details in Supplemental Methods). Image Processing All processing was performed with operators blinded to participant characteristics. We implemented a standardized pipeline for identifying and excluding scans with moderate or severe motion artifact. Rates of exclusion were similar across diagnostic groups (TS group, 6.1%; neurotypical controls group, 6.5%). We implemented a standardized preprocessing pipeline that included segmentation, image alignment, scalp, dura, and cerebellum removal, and tissue classification (details of quality control and preprocessing in Supplemental Methods). Cortical Thickness and White Matter Measures We used previously validated tools for deformation‐based morphometry, developed in‐house, to perform fine‐grained analyses of localized morphological features across the cerebral surface. 33 We first used a rigid‐body transformation followed by nonlinear transformation to spatially scale and coregister each brain to a template brain. This registration permitted us to identify corresponding points between the surfaces of each brain and the template, 34 , 35 which were used in statistical analyses to compare tissue measures point‐wise across groups. The nonlinear transformation was then reversed. Cortical thickness was calculated as the smallest distance of each point on the cerebral surface to the outermost surface of the white matter. White matter measures were calculated as the Euclidean distance from each point on the white matter surface of each brain to the corresponding point on the white matter surface of a template brain. 34 These distances represent shape features of the white matter surface, either outward or inward deformations (protrusions and indentations) relative to the surface of the template. For simplicity, we refer to these effects as increases or decreases in “local white matter volumes.” Statistical Analyses We applied general linear models voxel‐wise at the surface of the template brain, testing the association of TS diagnosis with each imaging measure. All models covaried for participant age at the time of scan, sex assigned at birth, and SES: The dependent variable, “imaging measure,” was either cortical thickness or white matter distance. Overall brain volume was not included as a covariate for anatomical measures because spatial coregistration of participant brains to a template brain included an affine transformation that inherently adjusted anatomical measures for overall brain size (Supplemental Methods). We assessed whether including age2 as an additional covariate influenced our findings, and it did not. We also used scatterplots of age versus either cortical thickness or white matter surface measures to assess whether we could visually discern any nonlinear effects in our data, and we could not. Therefore, age2 was not included in the model. Statistical maps for each regression model were corrected for multiple comparisons using a cluster‐based family‐wise error rate procedure, 36 , 37 with a cluster‐defining threshold of 2.0. Clusters that survived family‐wise error rate correction were displayed. The statistical significance (family‐wise error rate‐corrected p values) was visualized on the template surface using a color‐coded scale. Sensitivity Analyses To assess robustness of group differences, we conducted a series of sensitivity analyses, including in separate models the additional covariates of: (1) full‐scale IQ (FSIQ) and minority status, because groups differed significantly on these measures, (2) ADHD and OCD diagnoses concurrently to account for the most common comorbidities, (3) current use of any psychotropic medication. Moderation Effects Because age, sex, SES, and IQ differed significantly between groups, we assessed whether any of these variables moderated the association of diagnostic group with brain measures. In separate regression models, we included interaction terms for each variable with diagnosis, while also including main effects of age, sex, SES, and diagnosis as covariates (and FSIQ for that analysis). Associations with Symptom Severity and OCD/ADHD Diagnosis Within the TS group, we used linear regressions to examine associations of symptom severity and diagnoses of OCD or ADHD with brain measures. Separate regressions were performed for the severity of current and worst‐ever tic symptoms (motor, phonic, and total Yale Global Tic Severity Scale separately) while covarying for ADHD and OCD diagnosis. Separate models examined the association of current ADHD and OCD symptoms or lifetime diagnoses with brain measures. All models covaried for age, sex, and SES: where “diagnosis” was either lifetime OCD or ADHD diagnosis. Effects of Medication Use We examined effects of current medication use on imaging measures using voxel‐wise general linear regressions within the TS group for: (1) any psychotropic medication use, and (2) individual medication classes entered in separate models, including: (a) antipsychotics (n = 23), (b) stimulants (n = 4), (c) alpha‐agonists (n = 28), and (d) tricyclic antidepressants (n = 11). Results Demographics Demographics are shown in Table 1. The TS group had a higher proportion of males, lower IQ, and slightly lower socioeconomic status than controls. The TS adults all had persistent, functionally impairing tics. TABLE 1Sociodemographic and Clinical Characteristics of the Study SampleCharacteristicTourette syndrome n = 151Controls n = 113Comparisonn (%) or mean (SD) [range]Test statisticd.f. p valueAge at MRI scan (years)18.80 (13.50) [5.7–62.3]18.08 (11.62) [7.3–57.2] t = 0.452620.65Age bin (years)151 (100%)113 (100%)5.7–1289 (58.9%)59 (52.2%)13–2123 (15.2%)22 (19.5%) χ 2 = 5.9430.1122–4021 (13.9%)25 (22.1%)>4018 (11.9%)7 (6.2%)SexMale111 (73.5%)66 (58.4%) χ 2 = 6.6710.01Female40 (21.5%)47 (41.6%)Race/ethnicityWhite144 (95.4%)101 (89.4%) χ 2 = 3.5410.09Minority7 (4.6%)12 (10.6%)HandednessRight138 (91%)107 (95%) χ 2 = 1.0510.35Left13 (9%)6 (5%)SES45.49 (11.56) [17–66]48.61 (10.85) [15–66] t = −2.23262<0.03FSIQ112.62 (16.28) [75–151] (n = 129)120.22 (16.17) [75–154] (n = 104) t = −3.55231<0.001Comorbid diagnosisN/AADHD40 (26.5%)N/AOCD45 (29.8%)N/AAge of onset of TS (years)5.7–125.84 (1.97) [2–9] (n = 45)N/A13–217.38 (2.58) [3–12] (n = 21)N/AMedication use85 (56.3%)N/AAlpha agonist28 (18.5%)N/AAntipsychotic23 (15.2%)N/ASSRI19 (12.6%)N/AStimulant4 (2.6%)N/ATCA11 (7.3%)N/ASymptom severityYGTSS‐C20.49 (8.47) [0–46] (n = 142)N/ACurrent phonic tic8.05 (5.47) [0–24] (N = 142)N/ACurrent motor tic12.45 (4.70) [0–23] (n = 142)N/AYGTSS‐WE28.30 (7.06) [10–44] (n = 137)N/AWorst phonic tic12.25 (4.51) [0–22] (n = 137)N/AWorst motor tic16.05 (4.08) [0–25] (n = 137)N/AYBOCS4.27 (7.16) [0–28] (n = 140)N/AADHD score17.09 (11.73) [0–45] (n = 102)6.55 (0.7) [0–34] (n = 86) t = −7.13186< 0.001Clinical and sociodemographic possible scale ranges: YGTSS total tic score: 0–50; phonic tic: 0–25; motor tic: 0–25; YBOCS: 0–40; ADHD: 0–54. A total of 66 Tourette syndrome participants (56.3% missingness) provided age of tic onset; others were unable to provide it from retrospective recall, and most were from parent reports in youth. The missing data for age of Tourette syndrome onset were primarily for older participants, especially adults, who were unable to offer an estimate for when their tic symptoms began.Abbreviations: ADHD = attention deficit/hyperactivity disorder; FSIQ = full scale intelligence quotient; MRI = magnetic resonance imaging; OCD = obsessive–compulsive disorder; SES = socioeconomic status (Hollingshead); SSRI = selective serotonin reuptake inhibitor; TCA = tricyclic antidepressant; TS = Tourette syndrome; YBOCS = Yale‐Brown Obsessive Compulsive Scale; YGTSS = Yale Global Tic Severity Scale‐current; YGTSS‐WE = Yale Global Tic Severity Scale‐worst ever. Sociodemographic and Clinical Characteristics of the Study Sample Clinical and sociodemographic possible scale ranges: YGTSS total tic score: 0–50; phonic tic: 0–25; motor tic: 0–25; YBOCS: 0–40; ADHD: 0–54. A total of 66 Tourette syndrome participants (56.3% missingness) provided age of tic onset; others were unable to provide it from retrospective recall, and most were from parent reports in youth. The missing data for age of Tourette syndrome onset were primarily for older participants, especially adults, who were unable to offer an estimate for when their tic symptoms began. Abbreviations: ADHD = attention deficit/hyperactivity disorder; FSIQ = full scale intelligence quotient; MRI = magnetic resonance imaging; OCD = obsessive–compulsive disorder; SES = socioeconomic status (Hollingshead); SSRI = selective serotonin reuptake inhibitor; TCA = tricyclic antidepressant; TS = Tourette syndrome; YBOCS = Yale‐Brown Obsessive Compulsive Scale; YGTSS = Yale Global Tic Severity Scale‐current; YGTSS‐WE = Yale Global Tic Severity Scale‐worst ever. Group Differences Cortical Thickness Compared with controls, participants with TS showed widespread cortical thinning throughout the lateral cerebral surfaces bilaterally, in the inferior frontal (IFG), middle frontal (MFG), precentral (Pr), postcentral (Po) gyri, supramarginal (SMG), superior temporal (ST), middle temporal (MTG), and inferior temporal (IT) gyri, and inferior parietal lobule (IPL), with a greater spatial expanse in the right compared with left hemisphere (Fig. 1; Table S1). Sampling of representative voxels within areas of significant group differences revealed effect sizes of 0.42–0.52 (Fig 1). Statistical maps of Tourette syndrome (TS) associations with anatomical magnetic resonance imaging measures. A regression model tested group effects at each point on the brain or white matter surface: Imaging Measure = β 0 + β1 × Age + β 2 × Sex + β 3 × SES + β 4 × Dx + ε, where the “imaging measure” represents either cortical thickness or the signed Euclidean distance (“local volume”) of the white matter surface from a template brain. The analysis included 264 participants (TS: 151, neurotypical controls [NC]: 113). The statistical significance (cluster size family‐wise error rate‐corrected p values) of the associations between TS and brain structure at each point on the surface of the brain is visualized using color coding. Warm colors (yellow, orange, red) indicate significant positive associations, whereas cooler colors (blue, purple) indicate significant inverse associations. Only p values that survived cluster size family‐wise error rate correction are shown, with color coding according to the scale shown in the color bar. Lower panel shows bar graphs for TS diagnosis with cortical thickness shown in (A–C) and white matter surface distances in (D, E). The point in the brain where the data were sampled is shown in a small yellow circle, and a line leads from that point to the bar graph. Imaging measures were adjusted for (1) age at the time of scan, (2) sex, and (3) socioeconomic status of the participants. Error bars represent standard error. Effect size was calculated using Cohen's d = (M 2 − M 1) / SDpooled. Means and 95% CIs: (A) TS 2.27 [2.16, 2.38]; NC 2.85 [2.51, 3.19], (B) TS 2.36 [2.26, 2.45]; NC 2.71[2.57, 2.85], (C) TS 1.65 [1.55, 1.75]; NC 2.16 [1.76, 2.56] (D) TS −1.81 [−1.98, −1.64]; NC −1.3 [−1.51, −1.09], (E) TS −0.49 [−0.78, −0.19]; NC 0.4 [−0.03, 0.83]. CG = cingulate gyrus; GR = gyrus rectus; IFG = inferior frontal gyrus; IPL = inferior parietal lobule; IT = inferior temporal; MFG = middle frontal gyrus; MTG = middle temporal gyrus; Po = postcentral gyrus; Pr = precentral gyrus; SFG = superior frontal gyrus; SMG = supramarginal gyrus; ST = superior temporal gyrus. White Matter Surface Significant reductions in local white matter volumes were detected in the TS group along the mesial surface of the left hemisphere in the Pr, Po, and superior frontal (SFG), cingulate (CG), and rectus gyri (GR) (Fig 1; Table S1). Sensitivity Analyses Group differences persisted when covarying for FSIQ and minority status, ADHD and OCD comorbidity, or any current psychotropic medication use (Figs S1–S3), although cortical thinning in the right hemisphere was no longer significant when covarying for any medication use (Fig S3). Attenuation of group differences was expected, however, as controls did not have ADHD, OCD, or medication use, and these covariates would covary away a portion of diagnosis effects. Moderation Effects No significant age‐by‐diagnosis interactions were detected for cortical thickness (Fig S4). Significant interactions were detected in white matter only in the right hemisphere Pr, Po, SMG, IFG, MFG, and IPL (Fig S4). Maps for age effects in each group and scatterplots for the interaction revealed that local white matter volumes varied little with age in controls, but declined significantly with age in the TS group, in right hemisphere frontal and parietal regions (Fig S4). No significant interactions of sex, SES, or FSIQ with diagnosis were detected for cortical thickness or white matter measures. Symptom Severity and OCD/ADHD Associations Within the TS group, neither current tic symptom severity nor worst‐ever tic severity (either motor, vocal, or total tic scores), OCD symptom severity, nor lifetime OCD or ADHD diagnosis associated significantly with imaging measures anywhere. ADHD symptom severity, however, associated inversely with right hemisphere cortical thickness in the IPL, MTG, IFG, MFG, Pr, Po, SMG, SFG, and precuneus (Fig 2) in parietal and frontal regions adjacent to those where significant main effects for group differences were detected. Statistical maps of association of current attention deficit hyperactivity disorder (ADHD) symptom severity with anatomical magnetic resonance imaging measures. A regression model tested group effects at each point on the brain or white matter surface: Imaging Measure = β 0 + β 1 × Age + β 2 × Sex + β 3 × SES + β 4 × Current ADHD Symptom Severity + ε where the “imaging measure” represents either cortical thickness or the signed Euclidean distance (“local volume”) of the white matter surface from a template brain. The analysis included 101 Tourette syndrome participants who had ADHD symptom severity scores. The statistical significance (cluster size family‐wise error rate‐corrected p values) of the associations between current ADHD symptom severity and brain structure at each point on the surface of the brain is visualized using color coding. Warm colors (yellow, orange, red) indicate significant positive associations, whereas cooler colors (blue, purple) indicate significant inverse associations. Only p values that survived cluster size family‐wise error rate correction are shown, with color coding according to the scale shown in the color bar. The lower panel shows scatterplots for the associations of current ADHD symptom severity with cortical thickness. The point in the brain where the data were sampled is shown in a small yellow circle, and a line leads from that point to the scatterplot. Cortical thickness was adjusted for: (1) age at the time of scan, (2) sex, and (3) socioeconomic status of the participants. IFG = inferior frontal gyrus; IPL = inferior parietal lobule; IT = inferior temporal; MFG = middle frontal gyrus; PCu = pre‐cuneus; Po = postcentral gyrus; Pr = precentral gyrus; SFG = superior frontal gyrus; SMG = supramarginal gyrus; SPG = superior parietal gyrus; ST = superior temporal. Medication Effects Associations of current medication use in TS participants were small in spatial extent and varied by drug class (Fig S5), with little to no overlap with diagnosis effects. Diagnosis effects were largely unchanged when covarying for any medication use, although right lateral cortical thickness findings and left medial white matter surface findings were no longer significant (Fig S3). Discussion Group Differences Participants with TS showed prominent cortical thinning in primary sensorimotor regions (Po, Pr), and adjacent posterior frontal (IFG, MFG, SFG) and parietal (SMG, IPL) regions, while covarying for age, sex, and SES. These regions respectively support, among other processes, motor planning and execution (Pr), somatosensory processing (Po), higher‐order motor, premotor, and inhibitory control (posterior frontal regions), and multisensory integration, sensorimotor transformation, and attentional control (posterior parietal regions), all of which have been implicated frequently in the pathophysiology of TS and the generation of tics. 38 , 39 , 40 Findings persisted when covarying for comorbid ADHD and OCD diagnoses, FSIQ, and psychotropic medication use at the time of MRI scan. These findings align with abundant clinical and preclinical evidence implicating dysfunction in primary sensorimotor, motor planning, and attentional cortex portions of CSTC circuits in TS. 15 , 16 , 19 , 20 , 21 These group differences generally did not vary by sex, age, SES, or FSIQ, although significant age‐by‐diagnosis interactions in white matter indicated that older participants with TS had progressively greater reductions in right hemisphere frontal and parietal white matter volumes, suggesting that reduced white matter volumes in these regions may be a biomarker for the persistence of tic symptoms throughout adulthood. Symptom Severity The severity of tic and OCD symptoms (either at the time of scanning or worst‐ever) generally did not associate significantly with cortical thickness or white matter measures. Together with the stability of the group differences across the lifespan, these findings suggest that the group differences in these measures—especially the large expanse of cortical thinning involving bilateral sensorimotor, posterior frontal, and posterior parietal cortices—represent stable traits in the brains of persons with TS. This interpretation is consistent with findings from studies of induced pluripotent stem cells, which suggest a disturbance in basal ganglia inhibitory neurogenesis during fetal development in TS individuals who have persistent, life‐long symptoms. 41 In contrast, more severe ADHD symptoms associated significantly with cortical thinning in right hemisphere parietal and frontal regions, adjacent to those where significant main effects for group differences were detected. The locations of these associations are consistent with the cortical networks that have long been implicated in attentional control and executive functioning problems in ADHD. 42 Prior imaging studies have reported that co‐occurring TS and ADHD is associated with a larger magnitude of disturbance than either condition alone. 43 , 44 , 45 The present study suggests that comorbid TS and ADHD involves a spatial extension of the cortical thinning that defines the likely trait disturbance in TS alone, and that expansion is into frontoparietal cortices, thereby disrupting attention and executive functioning in those who have TS. Relation to Previous Studies Cortical thinning has been previously reported in individuals with TS, although thickening and increased gray matter volumes have been found in some regions as well. 45 , 46 , 47 Thinning of the precentral and postcentral gyri, frontal regions including inferior and middle frontal gyri, and parietal lobe have been reported in both pediatric and adult cohorts. 16 , 20 , 39 , 48 , 49 , 50 , 51 The additional thinning in temporal cortices that we observed has also been reported previously, 46 , 48 as has the thinning that we observed in limbic and paralimbic regions, including cingulate and rectus gyri. 19 , 39 , 45 , 52 These anatomical abnormalities appear robust across image processing and analytic platforms, both region‐of‐interest and whole‐brain approaches. 11 , 47 , 53 We have shown that this thinning is present throughout the lifespan and likely represents a primary (trait) rather than a secondary (state) characteristic of TS. A few studies report null findings regarding structural differences in TS, suggesting that observed anomalies in prior studies may reflect secondary effects of comorbidity or treatment, 54 , 55 but those studies were limited by modest sample sizes. Although prior studies reported inverse correlations of symptom severity with cortical thickness in CSTC regions, 20 , 39 , 52 we found no significant associations. These discrepancies may reflect sample or methodological differences. Overall, our findings support the view that CSTC structural anomalies are core features of TS, independent of symptom severity. Model of Pathogenesis Recent research has provided insights into the cellular mechanisms underlying inhibitory deficits in TS. Reduced GABAergic inhibition has been consistently observed in the primary sensorimotor cortices of individuals with TS and is thought to reflect either diminished numbers or compromised functionality of inhibitory cortical interneurons. 56 , 57 , 58 , 59 , 60 These interneurons play a critical role in regulating the balance of cortical excitation and inhibition. Disturbances affecting their integrity or function could lead to cortical disinhibition, directly facilitating the expression of tics. 61 , 62 Reduced GABAergic inhibition in TS likely extends to the basal ganglia. Human postmortem studies have reported significantly lower densities of parvalbumin‐positive GABAergic and cholinergic interneurons in the striatum, 57 , 63 and recent single‐nucleus transcriptomic data show an approximately 50% reduction in GABAergic and cholinergic interneuron populations in the caudate/putamen. 58 These studies have also demonstrated downregulation of synaptic adhesion and receptor pathways in interneurons, suggesting that those neurons have diminished arborization and connectivity. Finally, basal ganglia organoids derived from patients with TS show disrupted differentiation of interneurons, consistent with a model of altered embryological development and migration of interneurons in TS. 41 Taken together, these findings suggest a model of pathogenesis in which impaired inhibitory control over excitatory medium spiny neurons elevates striatal excitability and destabilizes motor output. 64 As a portion of GABAergic interneurons born in the basal ganglia migrate to the cortex in the prenatal and early perinatal period, 65 , 66 , 67 their reduction would likely similarly affect cortical portions of CSTC circuits, and hence the cortical thinning that we observed could represent diminished arborization and connectivity, together with reduced GABAergic interneurons in sensory, motor, and adjacent association cortices. An additional possible mechanism for trait‐level cortical thinning in CSTC circuits in TS, and which is not mutually exclusive with deficits in interneuron populations, is chronic neuroinflammation from microglial activation. Microglia, the brain's resident immune cells, respond to inflammatory triggers with the release of cytokines, chemokines, and reactive oxygen species. 68 , 69 , 70 Although adaptive in acute settings, chronic activation during development can impair neuronal growth and synapse formation. 71 Animal models of TS have reported microglial activation and increased expression of inflammatory markers, including interleukin‐6 and tumor necrosis factor‐α throughout CSTC circuits. 72 , 73 Consistent with these animal models, a positron emission tomography brain imaging study used (11)C‐[R]‐PK11195 to measure microglial activity and found excess activity in the basal ganglia and thalamic portions of CTSC circuits in patients with TS. 74 Transcriptome analysis of the striatum in individuals with TS connects these potential pathogenic mechanisms, having found both downregulated interneuron transcripts and upregulated immune‐related genes within microglia. 75 Sustained microglial activation within CSTC circuits may eliminate synaptic connections through phagocytosis, reduce dendritic spine density, and release neurotoxic mediators that impair local neuronal and glial function. 76 , 77 These effects could diminish cortical structural complexity and cortical thickness, as we observed in TS participants in regions vital for sensorimotor integration and inhibitory control. Limitations This study has several limitations. First, its cross‐sectional design precludes inferences about developmental trajectories or directions of causality in the observed associations. Second, MRI data were acquired using a 1.5‐T scanner. We note, however, that 1.5 T compared with 3T systems provide excellent signal‐to‐noise ratios (albeit at longer scan times), fewer magnetic field inhomogeneities, and reduced susceptibility artifacts, which together facilitate accurate and valid measurements that are comparable or, in several important ways, preferable to measures obtained at higher field strengths. 78 , 79 , 80 , 81 Second, the TS and control groups differed in their sex composition. The absence of a significant sex × diagnosis interaction in our sample, however, indicated that group differences in our brain measures did not differ across sexes, suggesting that the imbalance in sex composition across groups did not unduly influence our findings of group differences. Third, this was a cohort study in a clinical sample, which is subject to ascertainment bias. An MRI study within an epidemiological sampling frame would be required to minimize that bias. 82 Finally, the sample was predominantly white, which may limit the generalizability of findings to more diverse populations. Conclusions The anatomical alterations we observed in persons with TS across the lifespan likely represent stable, trait‐like neurodevelopmental disturbances that are present regardless of symptom fluctuations and that therefore may serve as enduring biological markers of vulnerability. Future multigenerational imaging studies can assess cortical thickness in offspring before, during, and after developing tic symptoms to determine if this putative trait disturbance is also an endophenotypic marker for vulnerability to developing TS‐related disorders. 83 Future studies should also assess the associations of cortical thinning with measures of GABAergic inhibition and immune activation, both of which could have contributed to the bilateral thinning that we observed in persons with TS. Author Contributions B.P., F.V., R.K., P.L., L.S., and J.L. contributed to the conception and design of the study; B.P., F.V., R.K., P.L., L.S., J.L., R.B., S.D., and A.F. contributed to the acquisition and analysis of data; S.D., A.F., B.P., R.B., F.V., R.K., P.L., L.S., and J.L. contributed to drafting the text or preparing the figures. Potential Conflicts of Interest Nothing to report. Supporting information Figure S1. Statistical maps of Tourette Syndrome associations with anatomical MRI measures while covarying for FSIQ and minority status. A regression model tested group effects at each point on the brain or white matter surface: Imaging Measure = β 0 + β 1 × Age + β 2 × Sex + β 3 × SES + β 4 × Dx + β 5 × FSIQ + β 6 × Minority + ε where the “imaging measure” represents either cortical thickness or the signed Euclidean distance (“local volume”) of the white matter surface from a template brain. The analysis included 264 participants (Tourette syndrome: 151, Neurotypical controls: 113). The statistical significance (cluster size FWER‐corrected p‐values) of the associations between Tourette Syndrome and brain structure at each point on the surface of the brain is visualized using color‐coding. Warm colors (yellow, orange, red) indicate significant positive associations, whereas cooler colors (blue, purple) indicate significant inverse associations. Only p‐values that survived cluster size FWER correction are displayed, with color coding according to the scale shown in the color bar. Figure S2. Statistical maps of Tourette Syndrome associations with anatomical MRI measures while covarying for OCD and ADHD diagnosis. A regression model tested group effects at each point on the brain or white matter surface: Imaging Measure = β 0 + β 1 × Age + β 2 × Sex + β 3 × SES + β 4 × Dx + β 5 × OCD + β 6 × ADHD + ε where the “imaging measure” represents either cortical thickness or the signed Euclidean distance (“local volume”) of the white matter surface from a template brain. The analysis included 264 participants (Tourette syndrome: 151, Neurotypical controls: 113). The statistical significance (cluster size FWER‐corrected p‐values) of the associations between Tourette Syndrome and brain structure at each point on the surface of the brain is visualized using color‐coding. Warm colors (yellow, orange, red) indicate significant positive associations, whereas cooler colors (blue, purple) indicate significant inverse associations. Only p‐values that survived cluster size FWER correction are displayed, with color coding according to the scale shown in the color bar. Figure S3. Statistical maps of Tourette Syndrome associations with anatomical MRI measures while covarying for active medication use. A regression model tested group effects at each point on the brain or white matter surface: Imaging Measure = β 0 + β 1 × Age + β 2 × Sex + β 3 × SES + β 4 × Dx + β 5 × Medication Use + ε where the “imaging measure” represents either cortical thickness or the signed Euclidean distance (“local volume”) of the white matter surface from a template brain. The analysis included 264 participants (Tourette syndrome: 151, Neurotypical controls: 113), and psychotropic medication use as follows (N = 69): antipsychotic use = 23, stimulant use = 4, alpha agonist use = 28, SSRI Use = 19, TCA use = 11. The statistical significance (cluster size FWER‐corrected p‐values) of the associations between Tourette Syndrome and brain structure at each point on the surface of the brain is visualized using color‐coding. Warm colors (yellow, orange, red) indicate significant positive associations, whereas cooler colors (blue, purple) indicate significant inverse associations. Only p‐values that survived cluster size FWER correction are displayed, with color coding according to the scale shown in the color bar. Figure S4. Statistical maps of association of age moderation of the diagnosis effect on anatomical MRI measures. We assessed whether age significantly moderated the effects of diagnosis on Imaging Measures by adding the term (β × [age × diagnosis]) to the regression model (Imaging Measure = β 0 + β 1 × Age + β 2 × Sex + β 3 × SES + β 4 × Dx + β 5 × Dx × Age + ε) and then testing the coefficient for the age × diagnosis interaction at each point on the surface of the brain. Given well‐documented effects of age on cortical thickness and white matter volume, separate regression models (Imaging Measure = β 0 + β 1 × Age + β 2 × Sex + β 3 × SES + ε) were used for TS (N = 151) and NC (N = 113) to assess for changes in cortical thickness and white matter volume across the lifespan. “Imaging Measures” represent either cortical thickness or the signed Euclidean distance (“local volume”) of the white matter surface from a template brain. The statistical significance (cluster size FWER‐corrected p‐values) of the associations between age modification of Tourette syndrome and brain structure at each point on the surface of the brain is visualized using color‐coding. Warm colors (yellow, orange, red) indicate significant positive associations, whereas cooler colors (blue, purple) indicate significant inverse associations. Only p‐values that survived cluster size FWER correction are displayed, with color coding according to the scale shown in the color bar. No significant age × diagnosis interaction effects were found for cortical thickness (not shown). For white matter volume, significant age × diagnosis interactions occur in right lateral structures including IPL, SMF, Po, Pr, MFG, and IFG. Figure S5. Statistical maps of medication use associations with anatomical MRI measures. A regression model tested medication use effects at each point on the brain or white matter surface: Imaging Measure = β 0 + β 1 × Age + β 2 × Sex + β 3 × SES + β 4 × Antipsychotic Use + β 5 × Stimulant Use + β 6 × Alpha agonist Use + β 7 × SSRI Use + β 8 × TCA Use + ε, where the “imaging measure” represents either cortical thickness or the signed Euclidean distance (“local volume”) of the white matter surface from a template brain. The analysis included 151 Tourette participants. The statistical significance (cluster size FWER‐corrected p‐values) of the associations between antipsychotic use (N = 23), stimulant use (N = 4), alpha agonist use (N = 28), TCA use (N = 11) and brain structure at each point on the surface of the brain is visualized using color‐coding. SSRI use (N = 19) did not have any significant effects and is not displayed. Warm colors (yellow, orange, red) indicate significant positive associations, whereas cooler colors (blue, purple) indicate significant inverse associations. Only p‐values that survived cluster size FWER correction are displayed, with color coding according to the scale shown in the color bar. Table S1. MNI coordinates of peak group differences. Acknowledgments This work was supported in part by grants MH01232, MH59139, MH068318‐01, MH‐K02‐74677, and MH36197 from the National Institute of Mental Health, by grants from the Tourette Syndrome Association, and by grant KL2TR000131 from the Southern California Clinical and Translational Science Institute (SC CTSI). Data Availability Deidentified participant data and a data dictionary are available on request from the corresponding author (bpeterson@chla.usc.edu) for researchers whose proposed use of the data has been approved. The data are not publicly available due to privacy restrictions. References Associated Data Supplementary Materials Figure S1. Statistical maps of Tourette Syndrome associations with anatomical MRI measures while covarying for FSIQ and minority status. A regression model tested group effects at each point on the brain or white matter surface: Imaging Measure = β 0 + β 1 × Age + β 2 × Sex + β 3 × SES + β 4 × Dx + β 5 × FSIQ + β 6 × Minority + ε where the “imaging measure” represents either cortical thickness or the signed Euclidean distance (“local volume”) of the white matter surface from a template brain. The analysis included 264 participants (Tourette syndrome: 151, Neurotypical controls: 113). The statistical significance (cluster size FWER‐corrected p‐values) of the associations between Tourette Syndrome and brain structure at each point on the surface of the brain is visualized using color‐coding. Warm colors (yellow, orange, red) indicate significant positive associations, whereas cooler colors (blue, purple) indicate significant inverse associations. Only p‐values that survived cluster size FWER correction are displayed, with color coding according to the scale shown in the color bar. Figure S2. Statistical maps of Tourette Syndrome associations with anatomical MRI measures while covarying for OCD and ADHD diagnosis. A regression model tested group effects at each point on the brain or white matter surface: Imaging Measure = β 0 + β 1 × Age + β 2 × Sex + β 3 × SES + β 4 × Dx + β 5 × OCD + β 6 × ADHD + ε where the “imaging measure” represents either cortical thickness or the signed Euclidean distance (“local volume”) of the white matter surface from a template brain. The analysis included 264 participants (Tourette syndrome: 151, Neurotypical controls: 113). The statistical significance (cluster size FWER‐corrected p‐values) of the associations between Tourette Syndrome and brain structure at each point on the surface of the brain is visualized using color‐coding. Warm colors (yellow, orange, red) indicate significant positive associations, whereas cooler colors (blue, purple) indicate significant inverse associations. Only p‐values that survived cluster size FWER correction are displayed, with color coding according to the scale shown in the color bar. Figure S3. Statistical maps of Tourette Syndrome associations with anatomical MRI measures while covarying for active medication use. A regression model tested group effects at each point on the brain or white matter surface: Imaging Measure = β 0 + β 1 × Age + β 2 × Sex + β 3 × SES + β 4 × Dx + β 5 × Medication Use + ε where the “imaging measure” represents either cortical thickness or the signed Euclidean distance (“local volume”) of the white matter surface from a template brain. The analysis included 264 participants (Tourette syndrome: 151, Neurotypical controls: 113), and psychotropic medication use as follows (N = 69): antipsychotic use = 23, stimulant use = 4, alpha agonist use = 28, SSRI Use = 19, TCA use = 11. The statistical significance (cluster size FWER‐corrected p‐values) of the associations between Tourette Syndrome and brain structure at each point on the surface of the brain is visualized using color‐coding. Warm colors (yellow, orange, red) indicate significant positive associations, whereas cooler colors (blue, purple) indicate significant inverse associations. Only p‐values that survived cluster size FWER correction are displayed, with color coding according to the scale shown in the color bar. Figure S4. Statistical maps of association of age moderation of the diagnosis effect on anatomical MRI measures. We assessed whether age significantly moderated the effects of diagnosis on Imaging Measures by adding the term (β × [age × diagnosis]) to the regression model (Imaging Measure = β 0 + β 1 × Age + β 2 × Sex + β 3 × SES + β 4 × Dx + β 5 × Dx × Age + ε) and then testing the coefficient for the age × diagnosis interaction at each point on the surface of the brain. Given well‐documented effects of age on cortical thickness and white matter volume, separate regression models (Imaging Measure = β 0 + β 1 × Age + β 2 × Sex + β 3 × SES + ε) were used for TS (N = 151) and NC (N = 113) to assess for changes in cortical thickness and white matter volume across the lifespan. “Imaging Measures” represent either cortical thickness or the signed Euclidean distance (“local volume”) of the white matter surface from a template brain. The statistical significance (cluster size FWER‐corrected p‐values) of the associations between age modification of Tourette syndrome and brain structure at each point on the surface of the brain is visualized using color‐coding. Warm colors (yellow, orange, red) indicate significant positive associations, whereas cooler colors (blue, purple) indicate significant inverse associations. Only p‐values that survived cluster size FWER correction are displayed, with color coding according to the scale shown in the color bar. No significant age × diagnosis interaction effects were found for cortical thickness (not shown). For white matter volume, significant age × diagnosis interactions occur in right lateral structures including IPL, SMF, Po, Pr, MFG, and IFG. Figure S5. Statistical maps of medication use associations with anatomical MRI measures. A regression model tested medication use effects at each point on the brain or white matter surface: Imaging Measure = β 0 + β 1 × Age + β 2 × Sex + β 3 × SES + β 4 × Antipsychotic Use + β 5 × Stimulant Use + β 6 × Alpha agonist Use + β 7 × SSRI Use + β 8 × TCA Use + ε, where the “imaging measure” represents either cortical thickness or the signed Euclidean distance (“local volume”) of the white matter surface from a template brain. The analysis included 151 Tourette participants. The statistical significance (cluster size FWER‐corrected p‐values) of the associations between antipsychotic use (N = 23), stimulant use (N = 4), alpha agonist use (N = 28), TCA use (N = 11) and brain structure at each point on the surface of the brain is visualized using color‐coding. SSRI use (N = 19) did not have any significant effects and is not displayed. Warm colors (yellow, orange, red) indicate significant positive associations, whereas cooler colors (blue, purple) indicate significant inverse associations. Only p‐values that survived cluster size FWER correction are displayed, with color coding according to the scale shown in the color bar. Table S1. MNI coordinates of peak group differences. Data Availability Statement Deidentified participant data and a data dictionary are available on request from the corresponding author (bpeterson@chla.usc.edu) for researchers whose proposed use of the data has been approved. The data are not publicly available due to privacy restrictions.