Evaluation of fMRI activation in hemiparetic stroke patients after rehabilitation with low-frequency repetitive transcranial magnetic stimulation and intensive occupational therapy

Abstract Purpose: To evaluate activity changes associated with the intervention of low-frequency repetitive transcranial magnetic stimulation (rTMS) and intensive occupational therapy (OT) after stroke using functional magnetic resonance (fMRI). Methods: Seventy stroke patients were scanned while performing finger tapping tasks twice, before and 12 days after the intervention. Recovery of motor functions assessed using Fugl–Meyer Assessment (FMA) and Wolf Motor Function Test-Functional Ability Scale (WMFT-FAS) for upper extremity at each time point. An fMRI analysis was performed, and a region of interest (ROI) analysis was conducted using percentage signal changes (% SC) to determine the magnitude of activation. Results: FMA and WMFT-FAS were significantly increased from pre-intervention to post-intervention. Intervention related activations were seen in the ipsilesional premotor cortex (PMC) and primary motor cortex (M1), thalamo-cortico regions with the paretic hand movements. With the unaffected hand movements, significant clusters in the contralesional primary somatosensory cortex (S1), superior parietal cortex, and bilateral cerebellum were observed. The ROI-based analysis revealed that ipsilesional M1, contralesional PMC, and supplementary motor area (SMA) showed significantly higher results with the paretic hand movements, a trend toward a significant decrease in the contralesional S1 with the unaffected hand movements from the pre-intervention to post-intervention. Conclusions: Our findings suggest that gains in motor functions produced by the intervention of rTMS and intensive OT in hemiparesis stroke patients may be associated with the ipsilesional hemisphere and contralesional hemisphere as well. Identifying rTMS and OT intervention based on cortical patterns may help to implement rTMS in motor rehabilitation after stroke. Supplementary data for this article is available online at https://doi.org/10.1080/00207454.2021.1968858 .


Introduction
Hemiparesis is the most commonly identified disability among patients with neurological disorders after stroke [1]. A variety of rehabilitation interventions have increasingly been applied to improve upper limb functions and decrease long-term disabilities by inducing the process of recovery [2]. Although recovery after stroke has been reported extensively for more than a century, the neural mechanism facilitating intensive rehabilitation-induced recovery is not clearly established [1]. Furthermore, it can be seen in the literature that the recovery after hemiparesis is associated with a convoluted pattern of the reorganization of the human brain [3].
The intervention of low-frequency repetitive transcranial magnetic stimulation (rTMS) and intensive occupational therapy (OT) has been shown to induce motor recovery in the affected upper extremity in hemiplegic stroke patients with mild to moderate hemiparesis [4][5][6]. Determining the neural changes that occur in functional recovery induced by rTMS and intensive OT intervention would help to illuminate the mechanism by which the intervention approach can Hemiparetic stroke; stroke recovery; low-frequency repetitive transcranial magnetic stimulations; intensive occupational therapy; finger movements; functional MRI promote post-stroke functional recovery. Also, it may guide the advancement of novel and more effective interventions.
A growing body of literature from brain imaging studies using magnetic resonance imaging (MRI) has demonstrated that intervention-induced recovery after stroke is associated with altered activity in brain networks [5,[7][8][9][10][11]. In addition, functional magnetic resonance imaging (fMRI) is a promising tool for evaluating neural mechanisms non-invasively, and therefore, extends a huge potential in studying the recovery process of the brain after stroke [7,10]. Task-related fMRI studies have been conducted to demonstrate that the movements with the affected hand after stroke are associated with massive recruitment of areas in the motor system located in the ipsilesional hemisphere [2,5,11]. While performing upper limb movements in patients recovering from a stroke, bilateral activations in the motor networks, additional sensory and motor structures have not commonly been found. Moreover, most of these studies have been conducted with selected patients who have gained complete recovery, as seen in the available literature. Therefore, it can be said that the non-recovered and partially recovered hemiplegic stroke patients have received less attention from the researchers so far [1].
A recent fMRI study following rTMS and intensive OT in stroke patients with moderate hemiparesis revealed that the improved motor functions after the rTMS and intensive OT were accompanied by increasing activations in the bilateral precentral gyri. Ueda et al. reported gains in motor functions following low-frequency rTMS and intensive OT intervention after stroke. Moreover, the authors clarified that improved functional connectivity between the ipsilesional precentral gyrus and contralesional hemisphere could be related [5]. Additionally, another study conducted on a sample of patients with moderate hemiparesis has also reported that rTMS and intensive OT intervention can induce cortical reorganization in the ipsilesional hemisphere [6].
The goal of the current study was to evaluate cortical changes associated with the recovery produced by low-frequency rTMS and intensive OT using task-based fMRI. Certain regions in the motor cortex have been previously shown to be related to reorganization after rTMS and OT in selected hemiparetic stroke patients [5]. However, outcomes can be varied, and some studies have reported increased brain activations, whereas some have shown decreased activations after rehabilitation.

Patients
The Ethics Committee of Tokyo Metropolitan University, Tokyo, Japan, approved the study. A total of seventy post-stroke patients who had been diagnosed with their first-ever unilateral stroke with upper limb hemiparesis were recruited in this study. Informed consent was obtained from all the participants. Patients were admitted to the Rehabilitation Unit of the Shimizu Hospital to receive an inpatient 12-days combined intervention. Inclusion criteria for stroke patients were; (1) hemiparetic stroke patients with Brunnstrom Recovery Stages (BRS) 3-6 for upper extremity and hand and finger; The degree of upper limb hemiparesis was categorized as (BRS) 2-6 considering the ability to flex the fingers with or without extension [12], (2) time-lapse after stroke >12 months caused by infarction or haemorrhage, (3) exhibits a plateau in clinical improvement as indicated by the Fugl-Meyer Assessment (FMA), wherein patient's score did not differ for more than 5 points for three months before enrolment in this study, and (4) has lesions located in the putamen, thalamus, and corona radiate. Individuals with lesions located at the brainstem and cortical strokes and those with neurological or psychiatric diseases were excluded from this study. The clinical characteristics and demographic data of participants are listed in Table 1.

Behavioural evaluation
All patients were assessed using the following functional outcome measures: 1. FMA 2. Wolf Motor Function Test-Functional Ability Scale (WMFT-FAS) at the admission and discharge from the hospital. An expert occupational therapist evaluated the two assessments.

Application of rTMS and occupational therapy
Low-frequency rTMS was applied over the primary motor cortex in the contralesional hemisphere in each patient. Thus, interhemispheric inhibition is decreased towards the ipsilesional hemisphere, leading to indirect activation of cortical excitability [6]. Each rTMS session consisted of either 1200 pulses for 20 min or 2400 pulses for 40 min. A 70 mm figure coil and the Magpro R30 stimulator (magVenture Company, Farum, Denmark) with a focal size of 1 Hz were used to apply stimulations in this study. The optimal site to place the stimulation on the skull was defined as the location where the largest motor-evoked potentials (MEPs) in the first dorsal interosseous (FDI) muscles of the unaffected upper limb using electromyography. The motor threshold (MT) of the FDI muscles of the unaffected upper limb was outlined as the lowest intensity of stimulation that can activate MEPs in that muscles. The intensity of stimulation was set to 90% of MT of the FDI muscle. All the patients were carefully monitored throughout the application of rTMS by the same physician who was providing the stimulations.
The rehabilitative program consisted of intensive OT with 60 min one-to-one training and 60 min of self-training after the rTMS sessions. One-to-one training included shaping techniques such as writing letters, drawing pictures, and repetitive task practise techniques, including pinching small coins. These one-to-one training activities were focused mainly on the motor functions of the affected upper limb in individuals and their lifestyles, such as occupation, interest, and household works. The self-training activities were also similar to one-to-one training sessions. An expert occupational therapist evaluated the performance of training activities.

The motor task during fMRI
Verbal instructions were given to patients using headphones at the time of imaging. The patients were trained with finger movement tasks before the fMRI scanning to ensure stable performance of the intended tasks. These tasks included repetitive movements of flexion and extension of fingers alternately between right hand and left hand. The design of the experiment was planned to have 30 s of rest and then 30 s of movement task alternately. This pattern of hand movements was repeated three times for 6.5 min. Patients were instructed to perform 1 movement per 3 s. However, in case a patient was unable to perform movements at the instructed rate, the patient was then asked to perform movements at their fastest possible rate.

Image analysis
Image analysis was carried out using the Functional Magnetic Resonance Imaging of the Brain centre Software Library, FSL (www.fmrib.ox.ac.uk/fsl).

First-level analysis
Data from each patient was initially analysed separately. Pre-processing and analysis were performed using fMRI Expert Analysis Tool (FEAT) version 6.0.4, a part of FSL. The following pre-statistics were applied: (1) Motion correction using the Linear Image Registration Tool (MCFLIRT), (2) Removing non-brain using Brain Extraction Tool (BET), (3) Spatial smoothing using a Gaussian kernel of 6 mm full width half maximum, (4) High-pass filtering (Gaussian-weighted least-squares straight-line fitting with sigma 90 s), (5) Slice time correction, (6) The first two volumes of fMRI  (7) continuous variables are presented as mean ± SD and, categorical data are presented as numbers (n) and percentage (%).
images were deleted to remove unstable BOLD signals. (7) Co-registration of functional images, first to the high-resolution 3D T1 images and then to the standard template in the Montreal Neurological Institute (MNI) using Boundary-Based Registration (BBR) and 12 DOF robust affine. At the first level analysis, a task-specific effect was estimated with voxel-wise analysis to obtain contrast images for each condition (Affected hand movement, unaffected hand movement, and affected hand versus unaffected hand movements) using the FMRIB's Improved Linear Modelling (FILM).

Higher-level analysis
The higher-level analysis was performed using FMRIB's Local Analysis of Mixed-effects. For the group analysis, the images of patients with right hemisphere lesions were flipped in the mid-sagittal plane. Therefore, all subjects were assumed to have left-sided lesions virtually. To estimate group baseline activation patterns associated with hand movements, we produce Z statistic images for pre-intervention and post-intervention analysis for movement versus rest. In the post-statistics processing, Z statistic images were thresholded using clusters determined by Z > 2.7 with the corrected cluster significance threshold of p < .05. The group-level analysis for pre-intervention versus post-intervention was performed using paired two group differences in FEAT. Clusters were evaluated with Z > 2.3 with the significance threshold of corrected p < .05.

Region of interest analysis
For the region of interest (ROI)-based analyses, we used the most studied regions found in the literature in motor training considering functional findings and theoretical background [8,9]. Thus, 10 ROIs were computed in this study, including the primary motor cortex (M1), the primary somatosensory cortex (S1), and supplementary motor area (SMA), premotor cortex (PMC) in both hemispheres and bilateral cerebellum using FEATQUERY in FSL. The ROI-based parameters were converted to BOLD Percentage Signal Changes (% SC), which is used as a magnitude of activation [13].

Statistical analysis
Statistical analysis was performed using IBM SPSS statistics 25. The normality distribution was assessed by the Shapiro-Wilk test. Since the data was not normally distributed, we have used a nonparametric test for paired measures (Wilcoxon Signed-Rank test) to determine the pre-post differences in motor performance and signal changes of 10 ROIs. In each statistical test for ROI, correction for multiple comparisons were performed separately using a false discovery rate (FDR) [14]. All p-values reported for the ROI analysis are uncorrected p-value which is also known as q-value. A p-value of less than .05 was considered to have a statistically significant difference, and a p-value of less than .07 was considered to have a trend towards significance. After FDR correction, any significant value was reported with an asterisk in the Supplementary Table 1.

Motor performance
The motor functional measures demonstrated statistically significant improvement in the performance of all patients. The Wilcoxon Signed-Rank test indicated that WMFT-FAS in post-intervention significantly increased in contrast to the pre-intervention (Z = −5.992, p < .001). FMA was also found to increase significantly (Z = -6.399, p < .001). The WMFT-FAS changed from 41.77 ± 14.37 at pre-intervention to 46.06 ± 14.98 at post-intervention, and FMA changed from 40.86 ± 13.04 at pre-intervention to 46.00 ± 12.65 at post-intervention.

fMRI findings
At baseline with paretic hand (right) movements, there were significant activations in ipsilesional M1 and PMC in post-intervention. In addition, significant clusters were found to emerge in contralesional S1 (Z > 2.7) for the unaffected hand movements. There were no significant clusters with movements of the paretic and unaffected hand in pre-intervention.
The analysis for the post-intervention versus pre-intervention (post > pre) comparison revealed significant differences of Z > 2.3. With movements in the paretic (right) hand, we observed ipsilesional activations in the PMC and the M1 (Figure 1(A)). Additionally, significant activations were found in the white matter brain regions including, ipsilesional optic radiation, hippocampus subiculum, anterior thalamic radiation, contralesional hippocampus subiculum, putamen, caudate, and forceps minor in the corpus callosum ( Figure 2). In contrast, unaffected (left) hand movements showed activations in contralesional S1 (Figure 1(B)), superior parietal cortex, and bilateral cerebellum (Figure 3). There were no significant clusters with the pre-intervention versus post-intervention (pre > post) with movements of paretic and unaffected hand.

ROI analysis
The % SC with the paretic hand movements was significantly increased in the contralesional S1 (Z = -3.899, p < .05, FDR), contralesional PMC (Z = -2.238, q < 0.05), and the ipsilesional M1 (Z = -2.039, q < 0.05). Furthermore, contralesional M1 and contralesional SMA were increased after the intervention with borderline values. The unaffected hand movements also showed a trend toward a significant decrease in % SC of the contralesional S1 (Z = -1.952, q < 0.07). However, there were no significant changes in % SC in the bilateral cerebellum. Supplementary Table 1 presents the results of the ROI analysis.

Discussion
In this present study, we aimed to assess the effect of rTMS and intensive OT on stroke patients with moderate hemiparesis using fMRI. In general, several studies have proven the intervention of repetitive rTMS and intensive OT is safe and effective on stroke patients with upper limb hemiparesis [15,16]. The major finding of this study was that the stroke patients who have participated in the 12 days intervention program showed improvement in gaining motor functions in the paretic hand is associated with a relative increase in activations of the ipsilesional hemisphere. Additionally, the contralesional hemisphere also might be involved in the recovery process.
Significant activations in the PMC and M1of the ipsilesional hemisphere with the paretic hand movement after the intervention were observed in our study. Also, ROI analysis showed increased % SC in the ipsilesional M1 after the intervention. These findings suggest that altered recruitment of PMC and M1 might have contributed to functional recovery after the intervention. Ipsilesional M1 plays a vital role in the reorganization process after stroke and is the main target of rehabilitative regimes [17]. In addition, studies have demonstrated ipsilesional increase activations in M1 are linked with the recovery of motor functions [18][19][20]. The findings described here, together with several other studies based on TMS therapy are consistent with the active role of ipsilesional motor areas in stroke recovery [5,21,22].
In recent years, the role of the contralesional hemisphere in motor recovery after stroke has been extensively studied [17,23]. There are considerable studies that have reported to show that contralesional S1 contributed to motor recovery after stroke [24]. Interestingly, our ROI findings showed an increased % SC in the contralesional motor regions. The novel finding of this study is stroke patients participating in the 12 days intervention of rTMS and OT reported improved motor functions in the paretic upper extremity that were associated with the relative increased fMRI signal changes in the contralesional motor cortices during the paretic hand movements. Our findings suggested that the intervention of stroke patients can promote motor recovery by shifting the balance of motor cortices toward the undamaged hemisphere. In general, although ipsilesional motor cortex inhibits the contralesional hemisphere during motor activities, neuroimaging studies have shown increased motor cortex excitability on the contralesional hemisphere with the paretic hand movements [25][26][27][28]. A previous fMRI study reported similar results in which improved motor functions in the affected upper extremity produced by constraint-induced movement therapy (CIMT) associated with increased activations contralesional hemisphere compared to ipsilesional hemisphere [10].
Brain activation correlates of rTMS and OT intervention in motor functions was not only related to the movement of the paretic hand. We also showed that functional recovery was associated with activity in the contralesional motor cortex with the movement of the unaffected hand. The present study found decreased activations in the contralesional S1 after the intervention with intact hand movements. This is consistent with one of the previous rehabilitative therapy studies in which they found a large cluster in the primary motor cortex where decrease activity correlated with therapy-related improvement in grip strength. This may be possible due to non-use of the unaffected limb is contributed directly to the recovery process by increasing plasticity for the paretic hand [7].
Our fMRI findings of bilateral cerebellum with the unaffected hand movement were also consistent with the previous study of Johansen-Berg et al. (2012). This study reported increased activity after therapy in the cerebellum bilaterally correlated with recovery scores [2]. However, our observations on fMRI activations in the bilateral cerebellum was not statistically confirmed Figure 2. significant activations in post-intervention versus pre-intervention with the paretic hand movements; ipsilesional white matter brain regions including optic radiation, hippocampus subiculum, anterior thalamic radiation, and in the contralesional structures, hippocampus subiculum, anterior thalamic radiation, putamen, caudate, and forceps minor in the corpus callosum. significant clusters are projected on the MnI template in fsl (p < .05, corrected on the cluster level).
through the magnitude of activations (% SC). This could have been due to considerable heterogeneity regarding lesion size, stroke site, and time since stroke. There have been suggestions in the literature that the cerebellum could be necessary for the recovery after stroke [11,13].
The current study found fMRI activations in contralesional and ipsilesional thalamo-cortical pathways with the paretic hand movements in addition to the primary motor regions. Several prior studies have discussed the reestablishment of cortical-subcortical pathways related to functional recovery after stroke [11,26,[29][30][31]. It is believed that activation of motor cortices triggers brain plasticity that might cause strengthen cortical-subcortical connectivity [30]. Furthermore, the current results are in general agreement with a prior study that reported activations in the corpus callosum with the affected hand movements after mental practice and physical exercise intervention [32]. Since the corpus callosum plays a crucial role in integrating information between hemispheres, it is more likely that the corpus callosum is involved in the functional recovery of the hands in post-stroke patients [23]. Therefore, our findings suggested that learning motor functions could be associated with thalamo-cortico activations.

Limitations and recommendations
Some limitations were to be acknowledged, the observed outcomes perhaps were due to the intervention. Our study included a relatively small stroke cohort. Hence, future studies with a larger sample size are needed to generalize our findings. The volume and site of anatomical damage caused by the haemorrhages and infarcts were heterogeneous in the study sample. Moreover, the underlying mechanism of functional recovery after stroke may depend on several factors, including the degree of recovery, the severity of the lesion, and the location of lesion sites [2,33]. Future work will need to incorporate lesion mapping to assess the effect of lesion volumes.
The present study did not include a control group. The inclusion of a control group would have facilitated meaningful interpretation to investigate the effect of rTMS and OT more elaborately. However, the lack of a control group is mitigated by a previous randomized study that included a control group with the same intervention [34]. The existing literature suggested that the time from stroke to the beginning of rehabilitation may influence the reorganization of the brain [8,17]. The time-lapse since stroke could be an issue inasmuch as spontaneous recovery is a leading factor after stroke. However, since our sample included stroke patients with time-lapse after stroke onset >1 year, the patients might have passed the period that spontaneous recovery could occur. Therefore, spontaneous recovery is less plausible in our study. Taken together, this makes the absence of a control group a lesser limitation in the present study [5]. It is recommended that future works be focused on comparing results with a control group to confirm the beneficial effect of our combined intervention. Also, there is still a need to include a stroke sample with time since stroke <1 year in a future study to investigate the potential effectiveness of our intervention on acute and subacute stroke patients. It would be interesting to distinguish findings of various categories apart from patients with BRS from 3 to 6 to investigate the effects of the intervention. Moreover, false positives in statistical results were reduced by applying the FDR correction and the results that survived the corrected p-values were reported. However, the results with a trend toward significance were also reported, although these results are not significant, can have practical implication.

Conclusions
In conclusion, our findings provided exploratory evidence that the motor function of the paretic upper extremity was increased in moderate hemiparetic stroke patients after the intervention of low-frequency rTMS and intensive OT. Furthermore, gains in motor function induced by the intervention might also be associated with motor cortices in the contralesional hemisphere. Additionally, it was evident that a few non-motor areas such as thalamo-cortical regions and cerebellum were involved in the recovery process in upper limb hemiparesis stroke patients after the intervention. Identifying stroke patients suitable for low-frequency rTMS and intensive OT based on the individual pattern of cortical activation may help to implement rTMS and OT in motor rehabilitation after stroke.