Abstract
INTRODUCTION
Individuals with Parkinson’s disease (PD) are known to have reduced movement speed and amplitude as well as learning difficulty in performing self-initiated movements [1]. It has been proposed that the defective striato-cortical circuit in individuals with PD is compensated by increased activation in the cerebello-cortical pathway during movement execution [2, 3]. Consistent with this hypothesis, increased cerebellar activity in individuals with PD has been reported [4–8]. However, in some other studies, reduced cerebellar activity was revealed in PD patients during movement execution [9–12].
The functional role of the cerebellum has been traditionally attributed only to motor control until studies found that cerebellar activities could also be associated with higher cognitive functions [3, 13]. Using resting state functional connectivity, O’Reilly et al. [14] found that there are at least two functional areas within the cerebellum, i.e., the sensorimotor and supramodal zones. The sensorimotor zone, including cerebellar lobules VI, VIIb and VIII, has functional connectivity with the motor, premotor cortex, somatosensory, visual and auditory cortex, and is involved in sensory guidance of movements [15]. The supramodal (executive) zone, comprising Crus I, II of the cerebellar lobule VIIa, has connection with the dorsolateral prefrontal area, and is possibly involved in the control of executive functions [16]. Due to impaired basal ganglia functions, individuals with PD have been reported to have increased reliance on attention for controlling movement that can be performed automatically by healthy individuals [17].
The present study aimed to compare the brain activation patterns of individuals with PD and healthy control subjects in a simple finger tapping task. Most previous studies employed demanding finger or hand tasks, i.e., complex and/or fast speed, and conducted the test while PD individuals were off-medicated [4, 19]. The present study used a simple finger tapping task at a relatively slow speed during the patients’ on-medicated state, i.e., the patients’ peak medication cycle. Subjects were required to perform the task in self-initiated and cue-initiated conditions. We hypothesized that the PD and control subjects would have similar activation patterns in the motor cortex because patients performed the task during their best medication state. We further hypothesized that individuals with PD would exhibit higher activation in the executive zone but lower activation in the sensorimotor zone of the cerebellum during theself-initiated finger tapping task. The use of visual cues has been shown to enhance motor output of individuals with PD [20]; we therefore hypothesized that there would be an overall increase in the activation of the cerebellum during cue-initiated finger tapping task in PD subjects.
MATERIALS AND METHODS
Subjects
Twenty-seven individuals with PD and 22 age-matched healthy subjects were recruited for the study. Patients were recruited from the Hong Kong PD Association, a patient self-help group. Healthy subjects were recruited from local health centres. Idiopathic PD was diagnosed by a neurologist according to UK PD Brain Bank Criteria [21]. All PD subjects had PD for at least one year, were at mild to moderate disease severity as measured by the Hoehn and Yahr Staging scale [22], and were stable with anti-Parkinsonian medications. In addition, they were able to move their index finger freely with full range of motion, and had sufficient communication and cognitive abilities to allow full participation in the study. Those who had orthopaedic or arthritic diseases that affected thumb and finger movements, neurological disorders other than PD, and previous neurosurgical treatment(s) were excluded from the study. Patients with atypical Parkinsonism syndromes including autonomic failure, pyramidal signs, cerebellar signs, gaze palsy, severe dysarthria or myoclonus, and any conditions that were not suitable for fMRI scans were also excluded. Healthy subjects had the same inclusion and exclusion criteria except that they did not have PD. Ethics approval and written consent were obtained prior to data collection. All tests were performed during “on” time, i.e., the peak medication cycle of individuals with PD. One patient was excluded because he could not complete the task, and one control was excluded due to excessive head movement. Twenty-six PD and 21 healthy subjects were included in the final analysis. All subjects were right handed.
FMRI paradigm
Subjects performed a simple finger tapping task in cue-initiated and self-initiated conditions during the scanning session. The two conditions were tested in a block design along with the resting condition in separate runs. Subjects were instructed to press a button by flexing their index finger. A force transducer was used to detect finger taps generated by the index finger inside the scanner. During the self-initiated condition, subjects were instructed to generate a finger tap at irregular intervals between 3 s to 5 s. In the cue-initiated task condition, subjects were instructed to generate a finger tap in response to a visual cue (“+”) on the computer screen. The rate of the movement was yoked to that generated by the subject in the self-initiated condition. This was achieved by saving the inter-tap time intervals produced by the subjects during the self-initiated finger tapping task. During the rest condition, subjects were asked to relax and view the screen in front of them. Each run consisted of 6 task and 6 resting blocks. Because the rate of the visual cue was yoked to that of the self-initiated movements, subjects performed the self-initiated task prior to the cue-initiated task. The design and schematic structure of the fMRI paradigm are shown in Fig. 1.
All patients performed the task using the hand of their more affected side, 13 of them performed the task with the left hand and the other 13 patients performed with the right hand. All healthy subjects performed the task with their non-dominant, left hand.
Imaging acquisition
A 3T Philips (Philips Achieva TX 3.0T, Philips Healthcare, Netherland) scanner with an 8-channel SENSE head coil was used for data collection. Functional data were collected using a T2-weighted gradient, echo-planar sequence that is sensitive to the BOLD contrast [echo time (TE) = 25 ms, flip angle = 90°, 3*3*3-mm isotropic voxels, volume repetition time (TR) = 2.5 s]. Whole brain coverage was obtained with 48 contiguous slices. First, all subjects underwent resting fMRI recording with visual fixation on a crosshair and eyes open in low-level illumination (without fixation). Then task fMRI data were acquired in cue-initiated, self-initiated, and rest conditions using a block design. Structural data, including a high-resolution (1.04*1.04*0.6-mm) sagittal, T1-weighted magnetization-prepared rapid gradient-echo scan (TR = 7.54 s, TE = 3.52 ms, flip angle = 8°) and a T2-weighted fast spin echo scan, were acquired in the end of the session. Head motion artifacts were minimized using a built-in helmet with foam pads. Residual head movement artifacts were corrected by spatially realigning each functional image to the first image in the time series using a six-parameter rigid body transformation.
Imaging data analysis
Pre-processing
Functional MR imaging data were preprocessed using Statistical Parametric Mapping software (SPM8, http://www.fil.ion.ucl.ac.uk/spm/). The first five volumes of each run were discarded to allow for T1 equilibration. The remaining 175 functional images per run were corrected for timing offsets between different slices and then realigned to the first volume of each run for rigid-body head motion correction. The T1-weighted anatomical image of each participant was then co-registered to the mean realigned functional images and subsequently segmented into gray matter, white matter and cerebrospinal fluid using a unified segmentation method.
A study-specific template was created using the Diffeomorphic Anatomical Registration Exponentiated Lie algebra (DARTEL) toolbox. The template was generated based on structural data of subjects who performed the task with their left hand. The corrected functional images were spatially normalized to the standard MNI space using the nonlinear normalization parameters estimated by the DARTEL toolbox. Functional images were resampled to 3×3×3 mm3 and spatially smoothed with an 8 mm full-width half maximum (FWHM) Gaussian kernel. Participants who had head movement larger than 3 mm translation or 3° of rotation in any direction were considered as having excessive head movement during scanning and were excluded.
Before all the pre-processing procedure, functional and structural images of the thirteen patients who performed the task with their right hand were flipped to the left to allow subsequent group analysis. Previous studies used a similar approach in patients with stroke [23, 24] and patients with PD [25]. We observed some degree of symmetry in the activation pattern of the finger tapping tasks performed by left hand and right hand groups of the PD individuals. Furthert-tests showed no significant brain activation difference between the right and left hand groups.
GLM Analysis
Data for each subject were normalized separately by dividing the BOLD time series with its grand mean in each session. Then a General Linear Model (GLM) was used to estimate the activity of the brain during the self-initiated and cued runs. The onsets of the self-initiated and cued movements, as well as the instruction to start finger tapping and rest were entered as different regressors. For the cued task, the onsets of the cues with no response and repeated response were modeled as separate regressors, and these incorrect trials were removed from the final within- and between-group contrasts. The regressors were convolved with the canonical hemodynamic response function (HRF). Additionally, six realignment parameters and ventricle and white matter signals were included as nuisance covariates. A high-pass filter (128 s cutoff period) was used to remove low-frequency confounds, and a first-order autoregressive model was used to correct for serial correlations.
We then modeled two contrasts for each individual, representing the two task conditions (cue-initiated or self-initiated) compared to the control condition (rest) and these contrast images were entered into the second level analyses. Conjunction analysis was conducted to identify the areas that were commonly activated during self- and cue-initiated movement respectively in both subject groups. The beta-weight contrast images obtained from GLM analysis were entered into a repeated measure ANOVA with group (PD & healthy controls) as between-group factor and condition (self- & cue-initiated movement) as within-group factor. We used a statistical threshold of p < 0.05 at the peak level corrected by Family wise error (FWE), along with a voxel number >100 for the conjunction analysis [26]. Within-group comparisons were conducted by comparing the beta contrast images between the cue- and self-initiated movement using paired t-tests. Between-group differences in each movement condition were evaluated using independent t-tests. For both within- and between-group comparisons, statistical threshold with p < 0.001 at the peak and cluster-extent based threshold with p < 0.05 for multiple comparison correction was applied [27, 28]. This thresholding method was commonly used to reduce the possibility of obtaining large activation clusters and to improve the degree of confidence on inference about specific locations [28]. All statistical tests were performed using the SPSS 20.0 software (SPSS Inc., Chicago, USA). A significance level of 0.05 was set for the statistical analysis of the behavioral data.
RESULTS
Twenty-six PD and 21 healthy subjects were included in the final analysis. All subjects were right handed. Table 1 presents the demographic and clinical characteristics of the subjects.
Task Performance
Individuals with PD performed the tasks at a comparable level as the healthy controls in terms of the tapping intervals and the number of finger taps in both conditions. Patients made finger taps following the visual cues at an accuracy level similar to that of the healthy subjects during cue-initiated movement. However, they took a significantly longer time than healthy subjects to respond to the visual cue (t (34,74) = 3.12, p = 0.004). A summary of task performance measures are shown in Table 2.
FMRI results
Conjunction analysis
Brain activations during self-initiated movement were evaluated in the control and PD groups separately, then a conjunction analysis of these two contrasts was conducted to reveal regions that were involved during self-initiated movement in both groups. Results revealed that there were activations in the bilateral cerebellum and SMA, and left (L.) motor areas including the precentral gyrus and postcentral gyrus (Table 3, Fig. 2a). Likewise, brain regions that were activated during cue-initiated movements in both groups included bilateral regions in the cerebellum, right (R.) SMA, and L. postcentral gyrus (Table 3, Fig. 2b).
Within-group comparisons
Healthy subjects had higher brain activity in the R. precentral gyrus during the self- than cue-initiated task; whilst they had higher brain activity in theL. middle occipital gyrus and the R. middle temporal gyrus during the cue- than self-initiated task. In contrast, PD patients had higher brain activity in the cerebellum Crus I (R,L) and lobules VI (R) during the self- than cue-initiated task and higher activation in the R. middle frontal gyrus (BA 25) during the cue-initiated as compared to self-initiated task (Table 4, Fig. 3).
Between-group comparisons
When compared with the healthy controls, the PD patients had lower brain activity in the R. inferior/superior parietal lobule during the self-initiated task, and lower brain activity in the L. cerebellum lobule VIII, lobule VIIB and vermis VIII, and thalamus during the cue-initiated task. There was a trend for PD patients to exhibit higher brain activity in the L. lingual gyrus and R. cerebellum Crus II during the self-initiated movements (Table 5, Fig. 4).
DISCUSSION
We compared the neural pathways recruited by individuals with PD and healthy controls during the performance of a simple finger tapping task in self-initiated and cue-initiated conditions. Unlike most previous studies which employed complex and/or fast finger tasks in patients in off-medication state [4, 19], we used a simple self-initiated finger tapping task and tested PD individuals during their on-medicated state. The present study showed that individuals with PD exhibited lower activations in the sensorimotor network, i.e., the IPL during self-initiated movement and cerebellar lobules VIIb, VIII during cue-initiated tasks. On the other hand, they had higher activations in the cognitive network, including cerebellar Crus I and II, during the self-initiated finger tapping task.
Conjunction analysis
Although patients performed the task using their more affected hand, they had comparable performance with healthy subjects in terms of the tapping intervals and the number of finger taps. We therefore anticipated a lot of similarity in the activation patterns between the PD and healthy groups. Both groups showed activations in bilateral supplementary motor area and, cerebellum lobule VI, and L. IPL, sensory cortex and primary motor cortex. These areas are involved in motor preparation, initiation and execution to allow subjects to properly perform self-initiated movements [29]. In the cue-initiated condition, the two groups showed comparable activations in bilateral cerebellum lobule VI, R. supplementary motor area, and L. primary motor area. The pattern was very similar to that of the self-initiated movements but the activation levels seemed to be slightly reduced, a finding similar to that reported by [30]. When a movement was triggered by a visual signal, it might require less planning and preparation.
Self- versus cue-initiated movements within each subject group
Self-initiated finger tapping movements require internal planning of the time interval between taps. Healthy subjects showed higher activities in the R. precentral gyrus and frontal operculum regions during self-initiated movements than during cue-initiated movements. The frontal operculum regions are connected to the somatosensory areas [31] and involved in the formation of internal representations of finger tapping actions [32] before the movements are executed by the motor cortex. On the other hand, people with PD recruited more cerebellar Crus I (R,L) and lobule VI (R.) during self-initiated movements than during cue-initiated movements. This finding concurs with previous studies that individuals with PD may increase the activity of the cerebellum to compensate for the impaired basal ganglia in motor preparation [3]. Both animal and human studies reported that cerebellar lobule VI receives information from sensorimotor cortex whilst Crus I receives information from dorsolateral prefrontal cortex and posterior-parietal regions that are associated with the executive functions [14, 33]. These findings highlight that the cerebellum receive information from different cortical areas as well as form reciprocal functional and/or anatomical circuits to select the appropriate response for action preparation and execution [34, 35].
Cue versus self-initiated movements within each group
Cue-initiated movements require subjects to initiate and execute finger taps in response to visual signals. Healthy subjects showed higher activation in the L. middle occipital gyrus in the cue- than self-initiated movements. This finding is consistent with previous studies that examined visually-initiated movements [36–38]. Visual cortex, located in the occipital gyrus, is used to detect and process visual information for motor execution [39] and this explains the activation pattern of healthy subjects. Individuals with PD, however, showed more activation in the R. medial frontal gyrus during cue- than self-initiated movements. No study has hitherto reported this finding. The medial frontal gyrus has been found to be associated with high-level executive function and decision-related process [40]. It is possible that people with PD require executive control in anticipating their response to the visual signal.
Between-group differences for self-initiated movements
When compared with control subjects, individuals with PD showed decreased activities in the R. IPL and SPL and marginally higher activations in the L. lingual gyrus and R. cerebellum Crus II during self-initiated movements. The IPL is involved in sensorimotor integration for movement planning [41] and in the awareness of intention to move [42]. Our finding of a high level of IPL activation in healthy subjects concurs with those reported previously [36]. Individuals with PD have been reported to have lower activations in the IPL and SPL when performing a simple fist open-close task [43] and a finger lift task [30] and higher activation when performing a more complex sequential finger task [44–46]. Electrophysiological and fMRI studies on PD patients have found reduced functional connectivity between the parietal cortex and primary motor cortex [47] and between the IPL and dentate nucleus [48]. In the present study, the reduced IPL activation in the individuals with PD might reflect a decreased ability to use somatosensory and kinesthetic inputs from the fingers for action preparation and execution [48–50]. In contrast to the under-activation of the IPL and SPL, there was marginally increased R. crus II activation in PD individuals. This finding reflects a tendency for patients to compensate by using their executive network to plan and prepare the self-initiated task. The higher activation in the lingual gyrus in PD individuals suggests increased attention to the visual signals, i.e., a “cross” which appeared after the finger tapping task. However, this signal was irrelevant to movement control and it is possible that PD individuals were unable to inhibit irrelevant information in movement planning [51].
Between-group difference for cue-initiated movement
Individuals with PD had lower activations than healthy subjects in L. cerebellar lobules VIIb and VIII. The under-activation in the ipsilateral cerebellum has been found in a hand grip task [10], a sequential finger tapping task [9] and visuo-manual tracking tasks [12]. Other studies reported over-activation of the cerebellum during a finger tapping task [4, 5], and a thumb press task [8]. We initially expected that the use of visual cues would enhance the cerebellar activation for motor execution. However, structural and functional pathological changes of the cerebellum have been found in PD individuals [3, 53]. The under-activation of the sensorimotor network could lead to impaired ability to scale the level of muscle activation and hence prolong the response times when compared with healthysubjects.
There is increasing evidence to suggest that the cerebellum may play a more complicated role in PD. Instead of two separate regions, the basal ganglia and cerebellum are anatomically connected and linked to form an integrated network [54]. There was cerebellar projection to the striatum via thalamus [55] and basal ganglia projection to the cerebellum via STN [56]. The cerebellum has a sensorimotor network and an executive network [14, 57]. The postulation that the defective striato-cortical circuit in individuals with PD was “compensated” by the increased activation of the cerebello-cortical network may be over-simplified. The present finding of reduced activation of the ipsilateral cerebellar lobules VIIb, VIII together with increased activation of bilateral Crus I and II in PD subjects suggest that the defective sensorimotor network (i.e., cortico-striato-cerebellar) could be compensated by the intact cognitive network (prefrontal-cerebellar). Recent studies reported that in PD, greater cortical-cerebellar connectivity was associated with better motor function, favoring the compensatory role of cerebellum [58, 59]. However, in several instances, increased cortico-cerebellar connectivity was related to worse performance, suggesting pathophysiological changes in the cerebellum and/or striatum [58]. Other studies found pathological changes in the cerebellum in patients with PD such as grey matter atrophy [52] and white matter damage [3, 60]. These mixed results suggest that the cerebellum could be part of the problem rather than a compensatory structure for the impaired basal ganglia. Future study is needed to elucidate the relationship between the cerebellum and thestriatum.
To conclude, individuals with PD showed higher activation of bilateral cerebellar Crus I during self-initiated movements than cue-initiated movements. When compared with healthy subjects, individuals with PD exhibited lower activation in the contralateral IPL and marginally higher activation of the contralateral cerebellar Crus II during self-initiated movements, and lower activations of the ipsilateral cerebellar VIIb and VIII during cue-initiated movements. When individuals with PD used their more impaired hand to perform a simple tapping task during their peak medication state, they may underactivate the sensorimotor network and overactivate the cognitive network within the cerebellum.
CONFLICT OF INTEREST
The authors have no conflict of interest to report.
Footnotes
ACKNOWLEDGMENTS
The work described in this paper was by a grant from the Research Grants Council of the Hong Kong Special Administrative Region, China (Project No. PolyU: 562212).