Ninety-four children with TS were recruited from outpatient clinics of the Second Affiliated Hospital, Zhejiang University School of Medicine. All patients met the Diagnostic and Statistical Manual of Mental Disorders, 4th edition, text revision (DSM-IV-TR) criteria. Sixty-three age-, sex-, and handedness-matched healthy children without any neurological or psychiatric disorders were recruited as HC. Participants were excluded if they had (i) any neurological (other than tics) or psychiatric diseases (other than ADHD and OCD), (ii) structural abnormalities on visual inspection of structural imaging, and (iii) head motion exceeding 3 mm in translation or 3° rotation in any direction. After head-motion control, fifteen patients were excluded. Seventy-nine TS patients were eligible to take part in the current study. All clinical evaluations were performed on the day of acquisition of MRI scans by an experienced pediatric neurologist (Table 1). The Yale Global Tic Severity Scale (YGTSS) was used to assess current tic severity. The Swanson, Nolan, and Pelham IV Scale (SNAP-IV) was used to assess ADHD symptoms, and the Children's Yale–Brown Obsessive Compulsive Scale (CY-BOCS) was used to quantify OCD symptoms. Twenty-five children with TS in this study were using psychoactive medication during the MRI scan. Eighteen children with TS were on monotherapy (Tiapride 11/18, Topamax 5/18, Haloperidol 2/18), while seven subjects were taking combination drug therapy. Informed written consent from parents and assent by children for TS and HC were obtained before participation. The study was approved by the Medical Ethics Committee of the Center for Cognition and Brain Disorders, Hangzhou Normal University, China.

MRI images were acquired on a 3.0-Tesla MRI scanner (GE Discovery 750 MRI, General Electric, Milwaukee, WI, USA) at the Center for Cognition and Brain Disorders, Hangzhou Normal University. Foam pads were used to minimize head motion for all subjects. A gradient-recalled echo planar imaging sequence (repetition time = 2,000 ms, echo time = 30 ms, and flip angle = 90°) was used to obtain functional images. Forty-three axial slices (field of view = 220 × 220 mm, matrix = 64 × 64, slice thickness/gap = 3.2/0 mm, and 240 volumes) were acquired. Participants were instructed to rest with their eyes closed, not to think of anything in particular, and not to fall asleep. Three-dimensional T1-weighted images were obtained in the sagittal orientation using a magnetization-prepared rapid acquisition gradient-echo sequence (repetition time = 8.068 ms, echo time = 3.136 ms, flip angle = 8°, field of view = 250 × 250 mm, matrix = 256 × 256, slice thickness/gap = 1/0 mm, and 176 slices). After each scanning session, the responsiveness of the subjects was determined by asking whether they had fallen asleep during the scan.

Statistical Parametric Map (SPM12, http://www.fil.ion.ucl.ac.uk/spm) software and the Data Processing & Analysis for Brain Imaging (DPABI) toolbox (http://rfmri.org/dpabi) were used for preprocessing (16). The first 10 image volumes of fMRI scans were discarded for scanner calibration and the subject's adaptation to the scanning noise. After slice time correction, functional images were spatially realigned to the first image of each session for head motion correction. Head movements assessed by the realignment parameters were tolerated up to ±3 mm and ±3°. Then, several nuisance variables including a linear trend, head-motion parameter (the Friston 24-parameter model) (17), white matter, cerebrospinal fluid, and global mean signals were regressed out (16). Individual T1 images were coregistered to functional images, then segmented (gray matter, white matter, and cerebrospinal fluid) and normalized to Montreal Neurological Institute (MNI) space. The transformation matrix obtained from T1 segmentation was applied to the functional images. The normalized functional images were resampled to 3 × 3 × 3 mm. Finally, we performed bandpass filtering (0.01–0.08 Hz, 0–0.01 Hz, 0.01–0.027 Hz, 0.027–0.073 Hz, 0.073–0.198 Hz, 0.198–0.25 Hz). ReHowas calculated (27 voxels) and ReHo maps from each frequency band for each individual were divided by the global mean ReHo value and then spatially smoothed with a Gaussian kernel (full width at half maximum = 6 mm). Ultimately, smReHo maps were used for statistical analysis.

For ICA, fMRI data were spatially smoothed with 6 mm full width half maximum Gaussian kernel after normalization. Subsequently, Group ICA of the fMRI Toolbox v3.0b (GIFT, http://icatb.sourceforge.net) was used to calculate spatially independent components (ICs). Dimension estimation of data from both groups was conducted using the minimum description length (MDL) criterion to determine the number of ICs (18). Then, individual maps in each group were conjoined and the temporal dimension of the convergent data was reduced via principal component analysis, followed by IC estimation. For all subjects, the spatial component maps were converted into z-score maps.

The number of ICs in the two groups (TS and HC) was 43. These components were selected based on the largest spatial correlation with a specific resting-state network template (http://findlab.stanford.edu/functional_ROIs.html) for further analysis. The brain networks included basal ganglia, sensorimotor, and the left and right executive control networks (one sample t-test, multiple-comparison correction based on Gaussian random field theory, single voxel p < 0.001, cluster level p < 0.05, Figure 1).

Two-way repeated-measures analysis of variance (ANOVA) was performed to examine the effects of group and frequency band on the ReHo maps. Group (TS and HC) served as a between-subject factor and frequency band (0–0.01 Hz, 0.01–0.027 Hz, 0.027–0.073 Hz, 0.073–0.198 Hz, and 0.198–0.25 Hz) served as a within-subject factor. An F-map of the “frequency by group” interaction effect and an F-map of the group main effect were obtained. Multiple-comparison correction was conducted based on Gaussian random field (GRF) theory (single voxel p < 0.001, cluster level p < 0.05). The peak ReHo values of the clusters surviving from the group main effect were extracted and then entered SPSS software for two-sample t-tests in each frequency band

The peak ReHo values of TS patients were correlated against the clinical measurements, including YGTSS scores and SNAP-IV scores.

To explore connectivity differences between the two groups, two sample t-tests were performed within each network. Multiple-comparison correction was conducted based on GRF theory (single voxel p < 0.001, cluster level p < 0.05).

ANOVA showed a significant main effect of group and a significant interaction effect (Figure 2, Table 2). For the main effect of group, the brain regions included the left precentral gyrus and the right operculum (frontal and temporal lobes). For the interaction effect, the brain regions included the right anterior cingulate gyrus (ACC)/supplementary motor cortex (SMA), right superior frontal gyrus, right Rolandic operculum, right putamen, right superior parietal gyrus, left superior temporal gyrus, and left white matter. Generally, in TS patients, the right superior frontal, and superior parietal gyrus showed an increased ReHo in lower bands and turned to normal in the higher bands, while the right Rolandic operculum and ACC/SMA showed a decreased ReHo in lower bands and turned to normal in higher bands. The right putamen and the left superior temporal gyrus showed a trend of decreased ReHo in lower frequency bands and increased ReHo in higher frequency bands (Figure 3, Table 3).

The two-sample t-tests showed significantly increased ReHo in full bands of the left precentral gyrus and significantly decreased ReHo in full bands of the right operculum (frontal and temporal lobes) (Table 4).

The ReHo value of the right operculum (frontal and temporal lobes) showed significant negative correlations with vocal scores of YGTSS in the highest frequency bands (0.198–0.25 Hz) (Bonferroni correction, 0.05/15 = 0.0033; Figure 4, Table 5). No significant correlation was found between the ReHo value and any rating score in the left precentral gyrus.

Compared with HC, TS patients showed increased connectivity of the right superior frontal gyrus (BA 6, MNI coordinates X = 21, Y = 9, Z = 63, T = 4.132, Cluster size = 945 mm³) within the left executive control network (Figure 5). No significant group difference was found in the basal ganglia, sensorimotor, and right executive control network. The peak Z-value of the right superior frontal gyrus was extracted to perform correlation analysis against clinical assessment scores. Only the motor score of YGTSS showed a positive correlation with abnormal connectivity (Figure 6).

The present study found decreased ReHo in the right putamen and increased ReHo in the left precentral gyrus of TS children. The putamen, a key node of the basal ganglia, is involved in motor loop circuits (19). The precentral gyrus responds to visual input and codes information during movement preparation (20), which involves cognitive processing for movements such as motor preparation and execution (21). The putamen and precentral gyrus are both parts of the CSTC circuit.

Functional deficits in the CSTC circuit may contribute to the motor symptoms of TS (7). The putamen, the center of the motor portion of the CSTC in the genesis of tics, receives abundant projections from the motor and somatosensory cortices (22). Defects in the modulation of activity in the structures of the motor CSTC network may underlie the failure to inhibit unwanted impulses in patients with TS (7). Previous studies demonstrated the aberrant structure, metabolism, and activity of the putamen and primary motor cortices in TS patients (23–25). Microstructural changes in the white matter of the left precentral gyrus and putamen in unmedicated and pure TS, and the regional apparent diffusion coefficient values were positively correlated with tic severity (25). A proton magnetic resonance spectroscopy study of children with TS showed significant reductions of N-acetylaspartate and choline in the putamen and reduced N-acetylaspartate in the frontal cortical (24). TS children were shown to have increased spontaneous local brain activity in the putamen and bilateral thalamus during RS-fMRI scans (26). Two event-related functional neuroimaging studies support the idea that the primary motor cortex processes preparatory signals related to motor tic behavior in TS (23, 27). One study showed significant fMRI activity in the primary motor cortex before tic onset (23). Another study found that cortical structures, including the primary motor cortex, preceded subcortical activation prior to tic onset (27). Decreased ReHo in the putamen and increased ReHo in the precentral gyrus in the current study indicated reduced synchronization of spontaneous neural activity in the putamen might inhibit the output nuclei of the basal ganglia, which in turn excite the precentral cortex to produce tics.

Executive control networks are thought to include the dorsolateral prefrontal, anterior cingulate cortices, and the orbitofrontal cortex (28). Many studies provided evidence of possible executive deficits in TS during various executive tasks such as working memory, fluency, shifting, and especially inhibition (29). TS patients could not suppress unwanted movements, and mainly comorbid with ADHD (3). The present study found that ReHo was decreased in the frontal and temporal operculum and ACC, but increased in the superior frontal and parietal cortexes. These regions were reported to be involved in two principal attention control networks. The cingulo-opercular network showed sustained activation across all tasks or nearly all tasks and was hypothesized to maintain task sets (30). The fronto-parietal network, which flexibly segregated or integrated in different aspects of control such as focused proactive control and episodic memory (31), has a crucial role in rapidly adaptive online control (32).

The present results showed reduced synchronization within the cingulo-opercular network (frontal operculum and ACC). Task-maintenance processes may be affected resulting in unwanted breakthroughs of normally suppressed behaviors, such as tics. However, the hyperactive frontal-parietal network (superior frontal and parietal cortexes) motivated initiate and adjust change in task control, again leading to unwanted movements and vocalizations (9). In previous fMRI studies of TS, activation of the operculum, ACC, insular, SMA, primary motor, and somatosensory cortex were found prior to tic onset (23), while decreased amplitude of low-frequency fluctuation (ALFF) and fractional ALFF in the ACC, frontal and parietal gyrus were detected during the resting state (26). In addition, fewer functional connections were found in the fronto-parietal and cingulo-opercular networks in TS children vs. adults and adolescents (9). Abnormalities of metabolism, anatomy, and function in the ACC, an important component of the cingulo-opercular network, have been reported in many previous TS studies (33–37). For instance, significantly decreased gray matter volumes were found in the ACC of TS patients (36). A single-photon emission tomography study reported the hypoperfusion of the ACC during the resting state, which was related to the severity of tics (35), while another study found increased ACC activation during voluntary tic suppression (34, 37). A recent proton magnetic resonance spectroscopy study of TS patients also found that concentrations of glutamine, an important excitatory neurotransmitter, negatively correlated with tic severity scores in ACC (33). Furthermore, we observed a negative correlation between tic severity and ReHoin the operculum. The greater the reduction of synchronization in the operculum, the more serious the disease is thought to become. Above all, altered ReHo of the cingulo-opercular and fronto-parietal networks, especially in the operculum and ACC in the current study, suggested that these brain areas may have an important role in the progress of tic generation mainly by breaking task set maintenance and enhancing adaptive tasks.

In addition to local synchronization, the present study used ICA to detect connectivity differences within the basal ganglia, sensorimotor, and executive control networks. We found increased connectivity of the right superior frontal gyrus within the left executive control network. The executive control network shares major nodes with the frontal-parietal network, including the dorsolateral prefrontal cortex and posterior parietal cortex. It is involved in the control of higher-order cognitive neural functions such as attention, planning, decision-making, and working memory (38). Recent years transcranial magnetic stimulation (TMS) studies attempted to indirectly impact deep or remote brain areas via functional connectivity. Some researchers have successfully observed TMS-induced functional connectivity alterations by stimulating the superficial cortex (39, 40). Our previous work also found increased ReHo in cerebellum after high frequency TMS on precentral gyrus (41). In addition, there was a significant positive correlation between functional connectivity of the right superior frontal gyrus and YGTSS motor score. The right superior frontal gyrus might be a direct or indirect target of TMS for TS treatment. This also suggested that Tics may be aggravated by enhanced functional connectivity within the executive control network, supporting the putative role of the control network in the pathogenesis of TS.

To the best of our knowledge, this is the first study to examine abnormalities of regional synchronization in a relatively large and typical pediatric TS population using the frequency specificity of ReHo. Several studies demonstrated the frequency specificity of ReHo changes in neurologic and psychiatric disorders, such as paroxysmal kinesigenic dyskinesia (42), schizophrenia (43), and major depression (44). In the present study, significantly increased ReHoin the superior frontal gyrus and superior parietal gyrus of TS children indicated compensatory functions of the frontal–parietal network in the lower frequency bands. These findings are consistent with previous reports that the lower frequency band (0.01–0.027 Hz) had a higher BOLD fluctuation in the cortical regions (13). We also found decreased ReHo in the putamen, ACC, and superior temporal gyrus in the lower-frequency bands, which was normal or reversed in the higher frequency bands. Previous RS-fMRI studies indicated that the higher band (0.027–0.073 Hz) exhibited increased ALFF compared with the lower band (0.01–0.027 Hz) in the subcortical regions (13, 45). The interaction of frequency band × group on ReHo indicated a significant difference in frequency-specific alterations (Figure 3, Table 3). The results of the current study suggested that the local intrinsic brain activity of TS was sensitive to specific frequency bands. More frequency-specific investigations are needed in the future.

The significant negative correlation between ReHo and YGTSS scores was found only in the highest frequency bands (0.198–0.25 Hz) (Figure 4, Table 5), suggesting that the ultra-high frequency band might be of clinical importance. These findings indicate that abnormal ReHoin TS is frequency-dependent and might be missed when using routine frequency band. Although the nature of these frequency-specific alterations of local neuronal homogeneity is still unclear, different frequency bands should be considered when measuring the ReHo of TS children in future studies to further understand the pathology of TS.