2.1.

Subjects

Twenty-nine patients with focal right-upper limb dystonia were recruited by neurologists at the participating movement disorders centers or by advertisement between the years 2017 and 2019. Twenty-seven healthy controls were recruited via advertisement and convenience. Inclusion criteria were age between 18 and 60 years and alphabetization in Portuguese with a minimum of 8 years of formal education. All subjects were right-handed; handedness was determined by self-report and confirmed by the Edinburgh Handedness Inventory. The diagnosis of idiopathic focal dystonia was established according to current recommendations,¹ including a normal brain image exam [magnetic resonance or computer tomography (CT) scan] before recruitment. All dystonia patients had focal right-upper limb dystonia, 11 of which were task specific (writer’s cramp). Exclusion criteria were neurological diseases (except for the dystonia group), history of neurosurgery, moderate to severe head or upper limb trauma, cancer, uncontrolled metabolic disorders, current use of anticholinergics, neuroleptics, antidepressants, benzodiazepines, vasodilators/constrictors, and history of substance abuse. Patients under botulinum toxin treatment had to wait for at least three months after receiving the last injection to participate in the study.

Two patients with dystonia were excluded due to MRI incidental findings (microangiopathy and a brain tumor). One control was excluded due to technical problems in the functional acquisition. In total, our fMRI sample had 27 patients (18 females) and 26 controls (14 females). The mean age of dystonia patients was 44.7 years (SD = 10.94 years), and the mean age of the control group was 42.7 years [standard deviation (SD) = 11.95 years]. Eleven patients had task-specific dystonia (writer’s cramp). The median disease duration was 10 years, ranging from 1 to 35 years. Clinical and demographic characteristics are available in Tables S1 and S2 in the Supplementary Material.

2.2.

Paradigm

The finger-tapping task consisted of 12 blocks, which contained one of three different conditions (four blocks each): right hand and left hand or both hands; each block lasted 30 s, followed by an interval of 30 s when subjects were instructed to remain still. Volunteers had to oppose the fingers to the thumb, starting from the index finger to the little finger, repeating the same sequence as fast and accurately as possible. Blocks were randomized using the E-Prime^® 2.0 software. The finger-tapping paradigm was the same for both fMRI and fNIRS acquisition, except for the fact that in fNIRS acquisition, participants remained seated on a chair with their hands resting on a desk. Participants were videotaped during the finger-tapping task and the investigator counted the number of finger taps to the thumb. For behavioral analysis concerning the number of finger taps for fNIRS and fMRI, we conducted a series of one-tailed independent samples $t$-test, with a significance level of $p<0.05$. Only correct finger taps were considered.

2.3.

fMRI

2.4.

fNIRS

2.4.1.

fNIRS acquisition

The hemodynamic signals were collected using a continuous-wave fNIRS system (NIRx Medical Technologies) with eight LED illumination sources emitting two wavelengths of near-infrared light (760 and 850 nm) and eight optical detectors. We recorded the signals at a sampling rate of 7.81 Hz. Sources and detectors were positioned bilaterally on the measuring cap with reference to the 10–5 international system.²⁷ The spatial distribution of the optodes on the cap was chosen to result in channels (i.e., source–detectors pairs) with standard interoptode distances of $\sim 30\mathrm{mm}$. A total of 23 channels covered regions of the middle frontal gyrus, motor, and sensorimotor cortices in addition to the supplementary motor area optodes set up are shown in Fig. 1 and corresponding anatomical landmarks are available in Table S9 in the Supplementary Material. Channels that displayed gain settings and coefficients of variation higher than 3% and 7.5%, respectively, were considered as bad channels and were excluded from the analyses (Table S10 in the Supplementary Material). Participants that had more than five bad channels were excluded. In total, 12 participants were excluded, and 21 patients and 20 controls were analyzed.

Fig. 1

fNIRS channels set up: sources and detectors are marked in red and blue, respectively. Each channel is represented in purple lines with its respective number.

3.1.

Behavioral Results of the Finger-Tapping Task

Dystonia patients and controls were compared in terms of the number of taps with the right hand, left hand, and both hands. Six participants showed dystonia posture during the task (evaluated and confirmed by a neurologist). In fMRI, a significant difference was found for the right hand $t(51)=1.89$, $p=0.032$ and for both hands $t(51)=1.73$, $p=0.045$, one-tailed. Dystonia patients had a mean number of 46.7 ($\mathrm{SD}=17.9$) taps per block, whereas controls had 56.2 ($\mathrm{SD}=18.9$). The degree of difference in means was moderate (Cohen’s $d=0.518$). Accordingly, a significant difference was found for the right-hand finger-tapping in fNIRS, $t(38)=2.26$, $p=0.015$ and for both hands $t(38)=3.02$, $p=0.004$, one-tailed. The degrees of freedom are different in both analyses because 12 participants were excluded from fNIRS data due to the mentioned criteria in Sec. 2.

3.2.

Group Comparison (fMRI)

Both dystonia patients and controls had bilateral cortical activation during right-hand finger tapping and, in both, the clusters peaked in the contralateral somatosensory cortex and ipsilateral cerebellum. Clusters are shown in Fig. 2 and Table 1. The contrasts regarding left-hand finger tapping are shown in Figs. S3 and S4, and Tables S5 and S6 in the Supplementary Material; and for both hands finger tapping in Figs. S5 and S6 and Tables S7 and S8 in the Supplementary Material.

Fig. 2

fMRI activation maps during the right-hand finger-tapping: clusters are marked in red for dystonia patients and in green for controls. Overlapping areas are shown in light brown. Coronal slices in MNI coordinates ($y=-84$, $-56$, $-30$, $-2$, 24, and 50).

Table 1

Activation clusters for right-hand finger tapping.

Differences between patients with upper limb dystonia and controls were assessed by voxel-wise analysis using a two-tailed independent sample $t$-test, a cluster-corrected significant threshold of $p<0.001$.¹⁸ No differences were detected between patients and controls using the $Z>3.09$ threshold in any condition. However, a threshold of $Z>2.3$ resulted in two clusters by contrasting $\text{dystonia}>\text{controls}$ during right-hand finger-tapping: one in the left cerebellum (crus I, crus II, and lobule VI: segment of cerebellar posterior lobe), peak-cluster MNI coordinates $-24$ $-68$ $-34$ ($Z\max =3.54$, $p=0.000732$), another in the left occipital lobe (occipital pole and lateral occipital cortex, inferior and superior division), peak-cluster MNI coordinates: $-28$ $-96$ $+12$ ($Z\max =3.74$, $p=0.000647$), shown in Fig. 3(a); boxplots regarding the mean of contrast parameter estimate are shown in Figs. 3(b) and 3(c). For both hands condition, a cluster in the frontal cortex (frontal medial cortex and paracingulate gyrus) peak-cluster MNI coordinates $+10$ $+46$ $-12$, resulted in contrasting $\text{controls}>\text{patients}$ ($Z\max =3.57$, $p=0.011$), as shown in Figs. 3(d) and 3(e). For the left-hand finger-tapping, a similar pattern was observed as for both hands condition, patients had diminished BOLD effect in the frontal cortex (frontal pole and frontal medial cortex), cluster-peak MNI coordinates $+10$ $+62$ $-8$ ($Z\max =4.07$, $p<0.0001$).

Fig. 3

(a) Contrast dystonia > controls for right-hand finger tapping: clusters in left cerebellum ($x=-24$, $y=-68$, and $z=-34$; $Z\max =3.54$, cluster peak) and left occipital lobe ($x=-28$, $y=-96$, and $z=12$; $Z\max =3.74$, cluster peak), found in voxel-wise analyses contrasting patients and controls, axial slices in MNI coordinates ($z=-40$, $-24$, 12, and 20). (b) Boxplot of the cerebellar cluster: mean signal change relative to resting ($p=0.000732$). (c) Boxplot of the left occipital lobe: mean signal change relative to resting ($p=0.000647$). (d) $\text{Contrast controls}>\text{dystonia}$ for both hands finger tapping: cluster in frontal pole ($x=10$, $y=46$, and $z=-12$) found in voxel-wise analyses ($Z\max =3.57$, cluster peak) contrasting controls and patients, axial slices in MNI coordinates ($z=-15$, $-12$, $-8$, 3, and 1). (e) Box plot of the frontal pole cluster: mean signal change relative to resting ($p=0.011$).

3.3.

Group Comparison (fNIRS)

Patients showed a lower activation during the right hand and both hands finger-tapping tasks. For oxy-Hb contrasts, significant results comprised the left middle frontal gyrus, medial postcentral gyrus, and the right supramarginal gyrus ($p_{\text{thresholded}}<0.00217$) as shown in Fig. 4(a) (right). The contrast regarding total-Hb exhibited statistically significant differences ($p_{\text{thresholded}}<0.00217$) for the superior frontal gyrus and the right supramarginal gyrus Fig. 4(a) (left). For both hands finger tapping, a similar pattern was observed in the group comparison, but no difference was detected in the right supramarginal gyrus [Fig. 4(b)]. For the left hand finger-tapping, we did not find group differences in cortical activation for oxy-Hb and total-Hb contrasts. No difference was detected in terms of deoxy-Hb in any condition.

Fig. 4

(a) $\text{Controls}>\text{patients}$ contrast during the right-hand finger-tapping task: contrasts relative to finger tapping with the right hand for oxy-Hb (left)—left middle frontal gyrus, medial postcentral gyrus, and right supramarginal gyrus and total-Hb (right)—superior frontal gyrus and right supramarginal gyrus. (b) $\text{Controls}>\text{patients}$ contrast during both hands finger tapping: contrasts relative to finger tapping with both hands for oxy-Hb (left)—left middle frontal gyrus, medial postcentral gyrus, and total-Hb (right)—superior frontal gyrus. Statistical maps show lower cortical activation in upper limb dystonia patients, considering Bonferroni correction ($p_{\text{thresholded}}<0.00217$). Color bars based on statistical parametric map SPM $\{t\}$ image.

To ensure the results from the channel-wise analyses, we conducted a series of nonparametric two-sample $t$-tests (Mann–Whitney U) to compare differences in each channel and condition considering the oxy-Hb (Tables S11, S12, and S13 in the Supplementary Material), total-Hb (Tables S14, S15, and S16 in the Supplementary Material), and deoxy-Hb (Tables S17, S18, and S19 in the Supplementary Material) contrasts.

3.4.

fMRI and fNIRS Data Description

Brain areas exhibiting cortical activation during finger tapping with the right and left hands relative to rest are shown in Fig. S7 in the Supplementary Material. In terms of BOLD effect (fMRI), oxy-Hb and deoxy-Hb contrasts (fNIRS), both patients and controls exhibited activation of areas implicated in the motor activity, including the bilateral somatosensory cortex and supplementary motor cortex. For deoxy-Hb contrasts, the finger-tapping task was associated with the contralateral somatosensory cortex in patients and controls.

3.5.

Resting-State Functional Connectivity

The dystonia group exhibited an increased correlation between the seed of the entire cerebellar somatomotor network and a cerebellar cluster spanning the medial prefrontal cortex and paracingulate gyrus, peak-cluster MNI coordinates $+14$ $+48$ $+8$; cluster size: 1989; cluster $p$-value family-wise error: 0.001). When the other cerebellar networks were used, no other between-group differences were observed.

4.1.

Cortical and Cerebellar Activation During Task

In this study, we provide evidence of altered brain activation in dystonia patients during finger tapping with the affected limb. Dystonia patients had a lower number of finger taps with the affected limb compared with controls. However, the majority did not show a dystonic posture during the task. Some studies suggest that brain imparities in tasks that do not induce dystonia might reflect primary changes or long-standing consequences of the disorders or an inherent impairment of the motor network.¹⁵

Regarding functional brain differences, both fMRI and fNIRS imaging detected lower activation in the frontal cortex in dystonia patients during finger-tapping of both hands simultaneously. We found a trend of increased left occipital and cerebellar activation but decreased activation in the frontal, sensorimotor, and right parietal cortex during finger-tapping with the affected limb. The latter result is well in line with other studies that reported lower activation in the somatosensory and premotor areas in patients with focal hand dystonia during the finger-tapping task.^30,31

Nonspecific activation in the left occipital lobe and cerebellum has been reported in patients with cervical dystonia during the finger-tapping task,³² suggesting unbalanced activation across a large network not limited to the somatomotor cortical network.

Cerebellar activation findings in dystonia vary across studies, possibly due to methodological differences and task-related characteristics. While some detected decreased activation with the finger-tapping task and writing,^30,33,34 others reported increased activation for patients with cervical and myoclonus dystonia and during the writing task.^35–38 Regarding our results, we hypothesize that the activation of the contralateral cerebellum in dystonia might reflect an additional engagement of alternative motor pathways, including the ipsilateral anterior corticospinal tract, to optimize the performance of a challenging task. This finding has been observed in healthy subjects and in parkinsonian patients under motor urgency situations, and a simple task, such as finger tapping with speed and accuracy, certainly presents an extra level of difficulty for patients with hand dystonia.^39,40

Although the role of the cerebellum is recognized, the mechanism underlying its functions in task-dependent environments is still not fully understood. As increased or decreased activation of the cerebellum varies in literature, we believe that it is currently unattainable to determine whether functional cerebellar disparities are compensatory or disease-causing. Ideally, screening patients at risk for developing dystonia, such as asymptomatic dystonia genes mutation carriers, with a long-term follow-up, could provide some answers. When looking at neurotransmission, Pizoli et al.⁴¹ found that increased glutamate activity through injections of kainic acid in cerebellum induces dystonia in mice. Further, recent findings of Gallea et al.² showed reduced gamma-Aminobutyric acid (GABA)-A receptors in the right cerebellum and left sensorimotor cortex in patients with focal hand dystonia, and reduced gray matter in the somatosensory cortex, which can be explained by the loss of GABA-A cells. In addition, reduced GABA levels in the sensorimotor cortex might enhance sensorimotor plasticity.⁴²

4.2.

Cerebellar Connectivity

The cerebellar cluster that was more activated in dystonia patients during task-fMRI lies mainly in posterior lobe crus I. Studies concerning the fronto-cerebellar network showed that crus I is connected to medial prefrontal cortex^26,43 and it might be associated with executive functions.⁴⁴

Cerebellar abnormalities in structural components, distinct activation, and alterations in functional connectivity have been associated with dystonia. Mantel et al.⁴⁵ found reduced right cerebellar gray matter volume and diminished connectivity function with the basal nuclei in dystonia patients. In individuals with writer’s cramp, Rothkirch et al.⁸ identified stronger functional connectivity between the primary motor cortex and the ipsilateral cerebellum, but diminished putamen-cerebellum effective connectivity. Our results point to a positive correlation of the cerebellar somatomotor network with the medial prefrontal cortex and paracingulate gyrus.

4.3.

Limitations

In this study, fMRI and fNIRS imaging were not acquired simultaneously. Subtle differences occurred during acquisition, for example, the position of the head and a different environment. Even though fNIRS optodes were placed in a measuring cap, optode positions were not digitalized for each participant, and therefore, there was no account for intersubject variability of channels’ positioning. In most cases, the finger-tapping task did not induce dystonia symptoms, which might explain why our results are not showing group differences in subcortical structures strongly related to dystonia, such as the basal nuclei.