Introduction
The striatum is a component of the basal ganglia and is composed of the caudate nucleus, putamen, and nucleus accumbens. These regions are involved in sensorimotor function, cognition, and emotional information processing (1–3). Previous studies have demonstrated that the striatum interacts with other cortical regions to plan and execute goal-directed behavior (4–6), as well as motor activity (7). The striatum also receives projections from specific, remote cortical areas and projects them to the thalamus and brainstem via the pallidum and then back to the cortex. This pathway is known as the corticostriatal network (5).
Multiple sclerosis (MS) is characterized by multiple lesions in the brain and/or spinal cord, resulting in various nervous system impairments (8). In recent years, corticostriatal resting-state functional connectivity (rsFC) has been implicated in the pathophysiology of MS (9, 10). For example, using voxelwise analysis, Sbardella and colleagues found that MS patients showed reduced basal ganglia-related rsFC when compared with healthy controls (HCs) (11). In other studies, investigators found that MS patients had significantly reduced functional connectivity within the corticostriatal motor loop (12) and revealed a spatial expansion of motor resting-state connectivity in subcortical nuclei (13, 14). Furthermore, Finke et al. found that MS patients were associated with altered functional connectivity of the striatum with the frontal gyrus, prefrontal cortex, motor area, and parietal cortex (15).
Although abnormal corticostriatal functional connectivity has been detected in MS patients, the precise role of the striatal subdivisions in MS remains unclear. Individual striatal subregions are believed to connect to specific functional cortical networks (9, 16, 17). More specifically, the inferior ventral striatum is positively connected with the orbital frontal cortex and the arterial cingulate cortex, the dorsal caudate (DC) is positively connected with the superior and middle frontal gyri, and the dorsal caudal putamen is positively connected with the precentral gyrus.
Previous studies have also found that portions of the striatal subregions, such as the superior ventral striatum (VSs) and DC, demonstrate abnormal rsFC with reward-related brain regions in depression patients (18, 19). In children with autism, nearly all striatal subregions showed increased rsFC with the limbic cortex (20), and in patients with obsessive-compulsive disorder, there is a clear functional distinction in the corticostriatal axis between the dorsal and ventral striatal regions (21). Similarly, exploring rsFC changes in striatal subregions in MS patients may shed light on neural plasticity changes in MS patients.
In this study, we used six subdivisions of the striatum as seeds to investigate the rsFC of patients with relapsing–remitting multiple sclerosis (RRMS), as well as the association between rsFC and patients’ symptoms, as indicated by the Expanded Disability Status Scale (EDSS). We hypothesized that compared to HCs, RRMS patients would have altered corticostriatal functional connectivity of striatal subregions, particularly subregions of the putamen due to its important role in the corticostriatal motor loop. Additionally, in order to test the reliability of the corticostriatal rsFC changes in RRMS patients, we scanned a subcohort of the RRMS patients a second time in the fMRI scanner.
Materials and Methods
Participants
Twenty patients with a primary diagnosis of RRMS were recruited for this study at Dongzhimen Hospital, Beijing, China. All patients met the revised McDonald criteria (22) and classification standards for MS. Fifteen age and sex-matched right-handed healthy subjects were used as a control group. Healthy subjects had no history of neurological or psychiatric disease.
The inclusion criteria for RRMS patients were: (1) patients were in the remission stage of RRMS with no acute attack or exacerbation of MS during the last month; (2) patients were not taking any glucocorticoid medications; (3) patients’ medication and treatment had no significant recent adjustments; (4) patients had no history of serious psychiatric illness or neurological disease other than MS; (5) Chinese was their primary language; (6) patients were right-handed according to the modified Edinburgh Handedness Questionnaire (23); and (7) patients did not have any contraindications to MRI. Patients who had contraindications to MRI, poor quality MRI images, or showed one or more gadolinium-enhancing lesions during their baseline MRI were excluded.
A neurologist assessed patients’ disability using the EDSS (24) on the day of the neuropsychological assessment.
fMRI Scan
Twenty RRMS patients (RRMS1 group) and 15 HCs received the first fMRI scan. In order to investigate the reliability of our findings, 12 RRMS patients received a second fMRI scan (RRMS2 group) approximately 7 months apart.
fMRI Data Acquisition
In this study, all fMRI data and structural data were acquired using two-pulse sequences on a 3-T scanner (Siemens AG, Erlangen, Germany) with an 8-channel head coil. The functional data were collected using EPI sequences oriented parallel to the AC-PC line (repetition time: 2,000 ms, echo time: 30 ms, flip angle: 90°, matrix: 64 mm × 64 mm, field of view: 225 mm × 225 mm, slice thickness: 3.5 mm, 36 slices, no gap, 306 time points). A T1-weighted 3D magnetization-prepared rapid acquisition gradient echo sequence (repetition time: 2,700 ms, echo time: 2.97 ms, flip angle: 7°, matrix: 256 mm × 256 mm, field of view: 250 mm × 250 mm, slice thickness: 1 mm, 176 sagittal slices covering the whole brain, no gap, acquisition voxel size: 1 mm3) was also applied. Participants were instructed to lie still with their eyes closed while staying awake.
fMRI Data Analysis
We examined six subregions of the striatum. These six subregions were widely used in previous studies (18, 20, 21). The six seeds included the inferior [Montreal Neurological Institute (MNI) peak coordinate x y z: ±9 9 −8] and superior (±10 15 0) ventral striata (inferior ventral striatum (VSi) and VSs), DC (±13 15 9), dorsal caudal putamen (DCP) (±28 1 3), dorsal rostral putamen (DRP) (±25 8 6), and ventral rostral putamen (VRP) (±20 12 −3) (16). Each striatal subregion was extracted as a 3 mm radial sphere using WFU-Pick Atlas software (25). The rsFC measures were computed between each seed and every voxel in the brain.
The preprocessing of fMRI data was performed with statistical parametric mapping (Wellcome Department of Cognitive Neurology, University College, London, UK) in MATLAB 8.2 (Mathworks, Inc., Natick, MA, USA). The preprocessing steps included realignment and coregistration of fMRI data to the structural images, normalization to the MNI standard template, and finally, smoothing of the data with an 8-mm full width at half maximum (FWHM) kernel. Band-pass filtering was performed with a frequency window of 0.008–0.09 Hz. In addition to our signals of interest, we employed segmentation of white matter and cerebrospinal fluid areas to remove confounding factors (26). Finally, time points were marked as outliers if the global mean intensity exceeded three SDs or if composite movement from a preceding image exceeded a 0.5 mm deviation using the artifact detection toolbox (http://www.nitrc.org/projects/artifact_detect/). Outlier time points were also included as regressors in the first-level general linear model along with motion parameters.
Functional connectivity analysis was carried out using a seed-based approach in the CONN toolbox v15.p (26). First-level correlation maps were produced by extracting the residual BOLD time course from each striatal seed region and by computing Pearson’s correlation coefficients between that time course and the time courses of whole-brain voxels. Correlation coefficients were Fisher transformed into “Z” scores, which increased normality and allowed for improved second-level general linear model analyses.
The second-level group analysis was applied using two-sample t-tests to compare the functional connectivity changes between the first scans of MS patients and HCs. Voxelwise linear regression analysis in CONN was also performed to investigate the association between the potential relationships between patients’ disability symptoms (EDSS) and the rsFC of each striatal subregion in the first and second scans of the MS patients. Age and gender were included in the analysis as covariates of non-interest.
A threshold of voxelwise p < 0.005 uncorrected and cluster-level p < 0.05 false discovery rate correction was applied for within and between-group fMRI data analyses.
Behavioral Statistical Analysis
Behavioral analysis was performed using SPSS 18.0 Software (SPSS Inc., Chicago, IL, USA). Two-sample t-tests were applied to compare the baseline characteristics between the two RRMS patient cohorts and HCs.
Discussion
In this study, we used seed-based resting-state fMRI to explore the corticostriatal circuitry in RRMS patients and the association between symptoms and functional connectivity. We found enhanced rsFC between the left DCP and the left premotor cortex, left DLPFC, insula, and parietal areas, and between the right VSs and the right posterior cingulate cortex in RRMS patients compared to HCs. We also found a positive association between corticostriatal rsFC and EDSS scores in RRMS patients. Specifically, after comparing RRMS1 patients with HCs and performing a regression between clinical outcomes and connectivity, we found overlapping results at the DLPFC and parietal areas when using the DCP as a seed. This result highlights the importance of the DCP in MS. In addition, a subcohort of RRMS patients was scanned twice in order to test the reliability of our findings. We found similar results after comparing both scans. Our results demonstrate the involvement of the striatum, particularly the DCP, in the pathophysiology of RRMS.
Previous studies have indicated that extensive lesions (8) might be the cause of the dysregulation of movement and the reduced connectivity of the motor network in the corticalsubcortical network in RRMS patients (13, 14, 27, 28). The basal ganglia interacts with other cortical regions to play a key role in contextually based motor decision making (7). More specifically, the striatum is thought to play a critical role in the storage and retention of motor programs (29). Within the striatum, the dorsal lateral putamen receives projections from the motor and motor association cortices (5, 30, 31), the main domain connected to the putamen seeds in healthy subjects (9, 16). Previous studies have found that the putamen is also connected to the DLPFC (17, 32–34) and the cingulate cortex (35), which implies that it is also involved in cognition and emotion.
In this study, we found increased rsFC between the DCP and the premotor cortex, insula, DLPFC, precuneous, and superior parietal lobule in both RRMS1 and RRMS2 patients when compared to HCs. We also found increased rsFC of the DCP with the DLPFC and parietal cortex. This increase was positively associated with EDSS scores in RRMS patients.
The test-retest findings indicate the important role of the rsFC of the DCP with the DLPFC and parietal gyrus in the pathophysiology of RRMS. The results are consistent with findings from previous studies indicating the involvement of the striatum in MS (13–15, 36). The DLPFC has diffuse connectivity with the rostral and caudal components of both the putamen and caudate, which project extensively to the central striatum and bridges the ventral and dorsal areas of the striatum (1). Studies suggest that the DLPFC plays an important role in motor and sensory information processing (37, 38), maintenance of sensory stimuli and motor execution (37), and pre-movement processes (39). Furthermore, the DLPFC is characterized as a part of the executive control network, which monitors motor performance (40) and integrates motor information and sensorimotor transformations (41).
The executive control network interacts with memory, motor, and sensory structures to direct thought and action (42). Dysfunction of the prefrontal cortex has been shown to cause deficits in executive function (43), and in studies of stroke patients, disruption of the executive control network has been shown to result in motor response deficits (44). We speculate that enhanced rsFC between the DCP and DLPFC may result in enhanced monitoring of motor performance (40) and enhanced integration of information and sensorimotor transformations (41).
Compared to HCs, RRMS patients showed increased rsFC of the left DCP with the parietal cortex, including the left precuneus, left superior parietal lobule, and right inferior parietal lobule. The superior parietal lobe is associated with the ability to produce purposeful and skilled movements and participates in complex motor programming. Studies have shown that motor execution requires coherent representations of neural composition and that the movement of a limb may cause activity changes in the left posterior parietal cortex (45). Previous studies found that, during the execution phase, patients with brain injuries showed enhanced activation in the superior and inferior parietal lobes, precuneus, and DLPFC compared with HCs. The superior parietal lobule and DLPFC, key regions in the frontoparietal control network, are known to have reciprocal circuit connectivity (46, 47) and play a role in changing spatial coordinates and body posture (45). It was also found that the functional connectivity of the parietal and frontal cortex can be altered by motor learning (48). Studies suggest that functional plasticity may occur following motor rehabilitation in patients with MS (49), and disrupted rsFC has been associated with disability progression in MS patients (50).
Previous studies showed that MS patients may need to activate more widespread sensorimotor networks compared to HCs during a sequential finger-to-thumb opposition task, as well as in other motor-related tasks (51, 52). Considering the dysfunction in movement, sensation, and cognition in RRMS (8), the increased rsFC of the dorsal putamen with the motor associated cortex and the positive association with clinical disability suggests that there is a compensatory modulation of the movement system in response to axonal injury (49). Taken together with these findings, our results suggest that the increased rsFC of the dorsal putamen with the DLPFC and parietal areas in RRMS patients may signal a compensatory connection for maintaining the function of the executive network or may adjust information resources for the motor processes involved in cognitive function.
We also found increased rsFC of the right VSs with the right PCC in RRMS1 and RRMS2 patients compared to HCs. The PCC is one of the primary nodes of the default mode network (DMN) (53, 54) and is implicated in memory processing and emotion generation (55). In a previous study, we found increased rsFC between the ventral striatum and the DMN in subthreshold depression patients compared with HCs, which may reflect a self-compensation mechanism to ameliorate the patient’s lower reward function (56). The PCC is shared between both the DMN and the sensorimotor network and, therefore, serves as an important interaction hub (57). The rsFC between the bilateral PCC and the ACC was increased following cognition and motor recovery after treatment in stroke patients (58). The PCC was also identified as the area involved in preparatory motor execution during rest (59) and directing the segregation of motor input (4). We thus speculate that the increased rsFC of the ventral striatum in RRMS patients may be due to the multiple neurological impairments of the motor and cognition system and may represent a complex modulatory interaction of the corticostriatal networks.
Finally, RRMS2 patients showed increased rsFC between the putamen and the primary motor cortex, thalamus, cerebellum, orbital frontal cortex, and parietal area compared with RRMS1 patients. The increased rsFC of the putamen confirmed the prior findings that displayed the functional coupling of the dorsal putamen and motor cerebral networks (9, 16). Previous rs-fMRI studies suggest the presence of functional plasticity following motor rehabilitation in MS (60). Based on this previous research, we speculate that the increased rsFC of the dorsal putamen with the motor cortex may be due to cortical reorganization accompanying the progression of RRMS.
One limitation of our study is the difference in lesion location among RRMS patients. MS lesions can be widespread across the brain and vary from patient to patient. We acknowledge that different lesion distributions may alter rsFC differently. Further studies with larger sample sizes are needed to further validate our findings.