OUP user menu

Increased sensorimotor network activity in DYT1 dystonia: a functional imaging study

Maren Carbon, Miklos Argyelan, Christian Habeck, M. Felice Ghilardi, Toni Fitzpatrick, Vijay Dhawan, Michael Pourfar, Susan B. Bressman, David Eidelberg
DOI: http://dx.doi.org/10.1093/brain/awq017 690-700 First published online: 5 March 2010


Neurophysiological studies have provided evidence of primary motor cortex hyperexcitability in primary dystonia, but several functional imaging studies suggest otherwise. To address this issue, we measured sensorimotor activation at both the regional and network levels in carriers of the DYT1 dystonia mutation and in control subjects. We used 15Oxygen-labelled water and positron emission tomography to scan nine manifesting DYT1 carriers, 10 non-manifesting DYT1 carriers and 12 age-matched controls while they performed a kinematically controlled motor task; they were also scanned in a non-motor audio-visual control condition. Within- and between-group contrasts were analysed with statistical parametric mapping. For network analysis, we first identified a normal motor-related activation pattern in a set of 39 motor and audio-visual scans acquired in an independent cohort of 18 healthy volunteer subjects. The expression of this pattern was prospectively quantified in the motor and control scans acquired in each of the gene carriers and controls. Network values for the three groups were compared with ANOVA and post hoc contrasts. Voxel-wise comparison of DYT1 carriers and controls revealed abnormally increased motor activation responses in the former group (P < 0.05, corrected; statistical parametric mapping), localized to the sensorimotor cortex, dorsal premotor cortex, supplementary motor area and the inferior parietal cortex. Network analysis of the normative derivation cohort revealed a significant normal motor-related activation pattern topography (P < 0.0001) characterized by covarying neural activity in the sensorimotor cortex, dorsal premotor cortex, supplementary motor area and cerebellum. In the study cohort, normal motor-related activation pattern expression measured during movement was abnormally elevated in the manifesting gene carriers (P < 0.001) but not in their non-manifesting counterparts. In contrast, in the non-motor control condition, abnormal increases in network activity were present in both groups of gene carriers (P < 0.001). In this condition, normal motor-related activation pattern expression in non-manifesting carriers was greater than in controls, but lower than in affected carriers. In the latter group, measures of normal motor-related activation pattern expression in the audio-visual condition correlated with independent dystonia clinical ratings (r = 0.70, P = 0.04). These findings confirm that overexcitability of the sensorimotor system is a robust feature of dystonia. The presence of elevated normal motor-related activation pattern expression in the non-motor condition suggests that abnormal integration of audio-visual input with sensorimotor network activity is an important trait feature of this disorder. Lastly, quantification of normal motor-related activation pattern expression in individual cases may have utility as an objective descriptor of therapeutic response in trials of new treatments for dystonia and related disorders.

  • DYT1 dystonia
  • imaging marker
  • positron emission tomography
  • motor activation


Dystonia is a neurological syndrome manifested clinically by focal or generalized sustained muscle contractions, postures and/or involuntary movements, ranging from action-induced dystonic symptoms to disabling, generalized dystonia. The most frequent genetic form of primary torsion dystonia relates to the autosomal dominant DYT1 mutation on chromosome 9q34, representing a GAG deletion within the coding area for torsinA (de Carvalho Aguiar and Ozelius, 2002). The link between the presence of dystonia mutations and clinical manifestations of disease is not well understood.

Abnormalities of the sensorimotor system in dystonia have been described at multiple levels (e.g. Berardelli et al., 1998; Quartarone et al., 2006). An overarching concept emerges from studies of focal dystonia relating the disorder to increased cortical excitability, decreased cortical inhibition, maladaptive cortical plasticity and impaired sensorimotor integration (Quartarone et al., 2006). Increased cortical excitability has been shown in hereditary dystonia, regardless of clinical penetrance (Edwards et al., 2003), in unaffected body parts of focal dystonia patients (Sommer et al., 2002), and in relatives of patients with sporadic dystonia (Walsh et al., 2007). Overall, these studies suggest that this feature represents a trait characteristic of dystonia. These observations were obtained in neurophysiological studies conducted in the resting state or during the performance of simple motor paradigms, such as flexion and extension at single joints. In this regard, functional imaging has extended these findings by delineating corresponding metabolic abnormalities at rest or abnormalities of regional brain activation during the execution of simple movements. For instance, trait characteristics have been demonstrated in a variety of imaging investigations of DYT1 mutation carriers. Metabolic abnormalities of the basal ganglia, supplementary motor area and cerebellum are evident in DYT1 mutation carriers scanned in the rest state (Eidelberg et al., 1998; Trošt et al., 2002; Carbon et al., 2004a; see e.g. Carbon and Eidelberg, 2009 for review). Microstructural studies with magnetic resonance diffusion tensor imaging suggest that these functional changes relate to an underlying maldevelopment of cerebellar outflow pathways in mutation carriers (Carbon et al., 2008b; cf. Argyelan et al., 2009). Indeed, we have found increased activation of secondary motor regions and inferior parietal cortex in non-manifesting DYT1 carriers, accompanied by reduced activation of the right inferior cerebellum (Ghilardi et al., 2003).

In contrast to these studies in genetically homogenous populations, functional imaging of motor activation responses in sporadic dystonia has yielded inconsistent findings. Although many of these studies reported activation abnormalities within the sensorimotor system, no consistent regional pattern emerged. Despite neurophysiological evidence of disinhibited sensorimotor cortex activity in dystonia, abnormally increased sensorimotor cortex activation has been demonstrated in some studies (Odergren et al., 1998; Pujol et al., 2000; Preibisch et al., 2001; Blood et al., 2004; Lerner et al., 2004; Hu et al., 2006), while decreased activation in this region was reported in others (Ceballos-Baumann et al., 1995b; Ibanez et al., 1999; Oga et al., 2002; Haslinger et al., 2005; Dresel et al., 2006). Similar discrepancies exist regarding activation changes in secondary motor regions. Moreover, to date, few imaging studies have examined the functional changes occurring in dystonia at a systems level (e.g. Ibanez et al., 1999; Carbon et al., 2004b).

In the current study, we used a new multivariate network-modelling approach for activation (Habeck et al., 2005; Moeller and Habeck, 2006) to quantify the activity of the sensorimotor system as a whole in dystonia mutation carriers and in control subjects. In addition to clinically affected subjects, we studied non-manifesting mutation carriers in order to distinguish intrinsic genotypic abnormalities from symptom-related changes. A combined behavioural and neuroimaging approach was used to: (i) identify abnormal movement characteristics during the performance of a repetitive reaching task in DYT1 carriers; (ii) identify abnormal motor activation responses in DYT1 carriers; and (iii) quantify sensorimotor network activity in DYT1 carriers and controls in 15O-labelled water (Embedded Image) PET scans acquired during movement and in a controlled audio-visual state without movement.

Materials and methods

This study was divided into three parts: (i) a behavioural study of repetitive reaching movements; (ii) a univariate voxel-wise analysis of regional cerebral blood flow (rCBF) measured during repetitive reaching movements and in a non-motor audio-visual control condition; and (iii) a multivariate spatial covariance analysis of the rCBF data to quantify sensorimotor network activity in the same conditions.


The following groups of right-handed subjects were included.

  1. DYT1 mutation carriers: 11 clinically manifesting mutation carriers [42.8 ± 15.5 years (mean ± SD)] and 10 non-manifesting mutation carriers (age 51.5 ± 14.3 years; manifesting versus non-manifesting comparison P > 0.2, Student’s t-test). Two of the manifesting subjects chose to participate in the behavioural testing, but not in the PET studies. Thus, a total of nine affected carriers (age 46.1.8 ± 15.1 years; manifesting versus non-manifesting comparison P > 0.4, Student’s t-test) participated in the imaging experiments in addition to the behavioural studies.

  2. Healthy controls: 12 healthy control subjects (age 44.7 ± 12.7 years; P > 0.3, Student’s t-tests, for age comparison with manifesting and non-manifesting DYT1 carriers) served as controls for both the behavioural and imaging experiments.

The DYT1 mutation carriers were recruited and genetically tested through the Neurology Department at Beth Israel Medical Centre in New York. The control cohort consisted of patient spouses and healthy volunteers from the community. Given the very low prevalence of DYT1 mutation in the general population, the likelihood that any one of these healthy subjects was a carrier was negligible (0.1%) (Ghilardi et al., 2003). Data on the genotype-specific aspects of brain activation in a subgroup of the non-manifesting DYT1 cohort have appeared previously (Ghilardi et al., 2003). (Elements of these previously published non-manifesting DYT1 activation data were included in this study to compare with corresponding data from the manifesting DYT1 carriers. Genotype-related effects in non-manifesting DYT1 carriers were reported in detail in the earlier publication.)

The clinical characteristics of the affected subjects are provided in Table 1. Exclusion criteria for all subjects were: (i) past history of additional neurological illness; (ii) prior or current exposure to neuroleptic agents or drug use; (iii) past medical history of hypertension, cardiovascular disease or diabetes mellitus; and (iv) abnormal MRI. For controls and non-manifesting subjects, the following additional exclusion criteria were applied: (i) current use of psychotropic medication; (ii) abnormal neurological examination; and (iii) past history of dystonic symptoms.

View this table:
Table 1

Clinical characteristics of the manifesting DYT1 mutation carriers

SubjectAge (years)Symptom duration (years)Distribution (body part)aBFMDRS motor scoreMedicationsbCCW pace, Hz
12313Generalized (Lg, rA)17THX, baclofen1
22411Focal (rA)1THX1
34031Generalized (A, Lg, T)33THX, baclofen, zonisamide1
4c2414Generalized (C, A, Lg, T)68Ethopropazine, baclofen2
5c3115Focal (rA)4None1
64132Generalized (C, A, Lg, T)54THX1
74336Generalized (C, A, Lg, T)17THX, levodopa, zonisamide1
85750Generalized (C, rA, rLg)23Topiramate1.16
95745Generalized (C, rA, Lg)6.5Levodopa1.3
106454Generalized (C, A, Lg, T)35THX2
116557Multifocal (A, Lg)19None1
  • a r = right; Lg = leg; A = arm; T = trunk; C = craniofacial.

  • b THX = trihexphenidyl.

  • c Motor testing only, no imaging.


All subjects performed a simple motor execution task involving repetitive counterclockwise reaching movements (CCW) and an audio-visual (AV) control task involving AV perception of the target display without movement (Ghilardi et al., 2003). These two tasks were performed in a pseudo-random order over subjects to counter time and order effects. All subjects were trained to perform the motor task outside the scanner 1–7 days prior to the imaging session. For CCW, subjects moved a cursor on a digitizing tablet with their right hand out and back from a central starting point to one of eight radially arrayed targets. Subjects were instructed to make movements with sharp reversal inside each target. Targets appeared on a computer screen in synchrony with a tone at a constant interval of 1–2 s. The default tone interval was 1 s and was determined during the training session. The interval was lengthened only if the subject could not correctly reach >60% of the targets within the time frame. The individual paces of the movements are provided in Table 1. The movement extent was kept constant across all individuals and measurements. During the AV condition (analogous to the S task employed in previous studies, Ghilardi et al., 2000), subjects remained immobile but experienced comparable AV stimuli as during the activation task. Screen targets, cursor images and tones were presented to the subjects asynchronously and irregularly in equal numbers to those used in the motor task. Although in the AV condition, subjects merely observed the screen and no movement was required during the imaging sessions, the rate of target appearance was matched to the rate used for CCW during imaging sessions.

Performance measures

To quantify motor performance during CCW, tangential velocities and accelerations were computed from hand position data. Automatic routines were used to identify the time and position at movement onsets, peak velocities, peak accelerations and reversals. Cursor positions at direction reversals were taken as the aimed end points for each movement; those at zero, calculated backwards from peak velocities, define onsets.

The following spatial and temporal characteristics were quantified for all movements: (i) movement time (the time from movement onset to the end point); (ii) spatial error (the shortest distance of the end point of the movement trajectory from the centre of the target, which the movement trajectory targeted); and (iii) onset time (the time from target and tone presentation to movement onset). Negative values indicate responses that are initiated before the tone, as required in this task.

All movements in which targets were reached within a time window of 250 ms prior to and after each tone were classified as correct. The percentage of correct movements was considered as a global measure of motor performance. Movement time, spatial error and onset time were calculated for all correct movements and used as analytical performance measures. We also computed a variability measure for movement time defined as its standard deviation (SD) over each 90 s trial.

Imaging studies


Nineteen DYT1 carriers (9 manifesting DYT1, age 46.1.8 ± 15.1 years; 10 non-manifesting DYT1, 51.5 ± 14.3 years, manifesting versus non-manifesting comparison P > 0.4, Student’s t-test) performed the CCW and AV tasks while undergoing Embedded Image PET imaging. Scans from 12 healthy volunteers were used as controls for the motor activation imaging experiments (age 44.7 ± 12.7 years; comparisons with either group of DYT1 carriers P > 0.3, Student’s t-tests). To identify a spatial covariance pattern corresponding to the normal motor activation network, we utilized a separate set of scans acquired in 18 different healthy volunteer subjects (age 37.2 ± 13.4 years). Each of these normal subjects performed 2–3 task repetitions during PET scanning, yielding a total of 78 scans (39 CCW scans, 39 AV scans) for pattern derivation (see below).

All Embedded Image PET scans were performed using the GE Advance tomograph at North Shore University Hospital as described in detail elsewhere (Carbon et al., 2008a). Motor tasks were performed with the dominant right hand, an intravenous catheter for Embedded Image delivery was placed in the non-dominant left hand. For each task run, relative rCBF was estimated using a slow bolus method in which 10 mCi of Embedded Image (Ferrieri et al., 1994) in 4 ml saline was injected by automatic pump over 16 s (15 ml/min) followed by a manual 3 ml saline flush. Embedded Image was produced remotely and delivered by the automatic pump (Nakamura et al., 2001). Dynamic 3D PET data acquisition began at the time of radioactivity arrival in the brain and continued for 80 s thereafter. Reconstructed PET images were corrected for random coincidences, electronic dead time and tissue attenuation by transmission scans. A single scalar correction was used to compensate for scatter effects.

Each subject was scanned while performing the CCW and AV tasks in randomized order with the dominant right arm. Each task was repeated 2–3 times per subject. Digital recordings of the hand position were acquired for every run during the motor task. For the manifesting DYT1 carriers, anti-dystonic and psychotropic medications were discontinued for at least 12 h before imaging. These subjects were also rated according to the Burke-Fahn-Marsden Dystonia Rating Scale (BFMDRS) at the time of imaging. Ethical permission for these studies was obtained from the Institutional Review Board of North Shore University Hospital. Written consent was obtained from each subject following detailed explanation of the procedures.

Univariate analysis: statistical parametric mapping

Imaging data processing was performed using statistical parametric mapping (SPM)-5 software (Institute of Neurology, London, UK; http://www.fil.ion.ucl.ac.uk/spm/software/spm5/). Standard pre-processing (realignment, spatial normalizing, smoothing with 10 _ 10 _ 10 mm) was applied followed by univariate and multivariate analyses.

Voxel-wise comparisons were performed using the flexible factorial model implemented in SPM5. We controlled for the effect of potential global blood flow differences by a global normalization to 50 ml/min/dl. SPM{t} maps were generated to assess the effects of task (AV, CCW) and group (control, non-manifesting DYT1, manifesting DYT1), and the interaction of these factors. To control for potential confounds resulting from the variability in task pace, the intertone interval was entered as a covariate of no interest. Contrasts were defined according to the specific questions. In particular, searches for abnormal motor activation responses in manifesting DYT1 were performed in the flexible factorial model (i) as contrasts of task-specific activation [manifesting DYT1 (CCW–AV) versus control (CCW–AV)]; and (ii) as within-condition contrasts between groups. In addition to searching for within-condition group effects in the CCW condition, we also explored group effects in the AV condition. To optimize the power of our inference, we restricted our search volume to areas showing a significant main effect of motor activation (CCW > AV). Crucially, this contrast is orthogonal to the simple main effect of group within the AV conditions and cannot bias our inference. For all voxel-based analyses, results were considered significant at P < 0.05 (false discovery rate corrected) (Genovese et al., 2002), with a cluster extent cut-off of [ke] > 80 voxels. Coordinates were reported in the standard anatomical space developed at the Montreal Neurological Institute.

To assess the effects of genotype in the clusters that differed in manifesting DYT1 and controls, we extended the results of the voxel-based searches by incorporating a post hoc volume-of-interest analysis. To address the possibility that non-manifesting DYT1 takes an intermediate position between manifesting and controls, volume-of-interest analysis was applied to all the significant clusters identified in the voxel-based comparisons of manifesting DYT1 and controls. To do this, we averaged data within spherical volumes-of-interest (10 mm diameter) centred on the maxima and submaxima of the statistical parametric mapping clusters detected in the manifesting DYT1 and control contrasts. These measures were compared using ANOVA, followed by Tukey–Kramer honestly significant difference for pairwise group comparisons (controls, non-manifesting DYT1, manifesting DYT1). The resulting activations were also depicted graphically.

Multivariate analysis: ordinal trends

In addition to univariate assessments of between-group differences in regional activation, we employed a multivariate approach to examine group differences at the network level. Specifically, we sought to determine whether the activity of the normal sensorimotor network is elevated in dystonia mutation carriers, and whether this increase is greater in manifesting relative to non-manifesting gene-positive subjects. To test this hypothesis, we first identified and validated a significant activation pattern in healthy subjects. This spatial covariance pattern was derived using Embedded Image PET scans from a group of healthy subjects that was independent of those used as controls for comparison with the gene carriers (see above).

The procedure used to identify this pattern was based on ordinal trends canonical variate analysis (OrT/CVA) (Habeck et al., 2005). OrT/CVA is a supervised principal component analysis designed to identify spatial covariance patterns that change in expression within subject across experimental conditions. The property of consistent change in pattern expression across tasks/conditions is called an ‘ordinal trend’. This attribute differs from routine assessments of task-related changes at the group mean level, employing mass univariate procedures such as statistical parametric mapping (Moeller and Habeck, 2006). In addition to detecting one or more significant activation patterns, OrT/CVA quantifies the expression of the resulting pattern(s) in each subject and condition. The significance of OrT/CVA patterns is assessed by permutation testing of the relevant subject scores to exclude the possibility that the observed changes across conditions had occurred by chance. Likewise, the reliability of the regional contributions to the pattern (i.e. the voxel weights) is assessed by bootstrap estimation (Efron and Tibshirani, 1994). The test–retest reliability of the subject scores measured in each condition is assessed across repeat runs conducted within the imaging session. Once validated, pattern expression can be quantified on a prospective case basis in scans obtained in new subject cohorts.

In this study, OrT/CVA was applied to 39 pairs of CCW and AV scans (a total of 78 Embedded Image PET images from the independent normal cohort of 18 subjects described above). This analysis was used to identify significant and stable normal motor-related activation pattern topography. The expression of this pattern was then quantified prospectively in the scans of the DYT1 carriers and in the second group of healthy controls that was age-matched to the gene-positive subjects. Thus, these computations were performed on an individual case basis in each of the scans (CCW and AV) separately acquired in the manifesting DYT1, non-manifesting DYT1 and control subjects. For each experimental condition, between-group differences were assessed using ANOVA followed by post hoc tests (Tukey’s honestly significant difference for data with equal variance and Dunnett’s T3 test for data with unequal variance). For the manifesting DYT1 group, we correlated pattern expression in each condition with BFMDRS motor ratings acquired at the time of PET. Correlations were assessed using Spearman rank correlation coefficients. JMP (SAS Institute Inc., Cary, NC, USA) or the Statistical Package for the Social Sciences (SPSS, Inc., Chicago, IL, USA) were used for statistical analyses. All results were considered significant for P < 0.05.


Behavioural findings

General motor task performance, expressed as the percentage of correctly hit targets, was reduced in manifesting DYT1 (49.4 ± 11.4%, mean ± SD) compared to controls as well as to their non-manifesting counterparts (controls: 85.1 ± 7.2%; non-manifesting DYT1: 77.4 ± 4.8; P = 0.004, Supplementary Table S1). For the correct movements, there was a significant group difference in movement time variability (SD). This measure was higher in manifesting DYT1 relative to age-matched controls and non-manifesting DYT1 carriers (movement time_SD: manifesting: 103.6 ± 9.0 ms; controls: 60.5 ± 12.0 ms; non-manifesting DYT1: 69.7 ± 11.3; P = 0.003). This indicates greater irregularity of the repetitive reaching movements in the affected carriers. The remaining temporal and spatial movement descriptors did not differ across groups (P ≥ 0.15 for all comparisons). In line with our earlier results (Ghilardi et al., 2003), all measures of motor task performance were normal in non-manifesting mutation carriers (Supplementary Table S1).

Imaging findings

Statistical parametric mapping

Within-group comparisons

In manifesting DYT1, significant motor activation responses (CCW > AV) were found in the left sensorimotor cortex (BA 1–4) extending into adjacent premotor (BA 6, dorsal premotor cortex and supplementary motor area), preparietal (BA 5), parietal (BA 40) and cingulate cortices (BA 24) (Supplementary Table S2). Significant activation responses were also detected in the left thalamus (ventral postero-lateral nucleus, ventral postero-medial nucleus), in the putamen bilaterally, and in the right superior cerebellar cortex (lobulus IV–VI), extending into the dentate nucleus, vermis and the left superior cerebellar cortex. Similar activation responses were found in both the non-manifesting DYT1 and healthy volunteer groups (Supplementary Table S2).

Between-group comparisons

Relative to controls, manifesting DYT1 exhibited increased motor activation (Fig. 1A–C) in the left sensorimotor cortex extending into adjacent secondary motor, preparietal and cingulate areas and into the homologous region of the right hemisphere (Panel A in Table 2). No significant decreases in motor activation responses were found in manifesting DYT1 relative to controls. Except for those regions with abnormal increases in motor activation in manifesting DYT1, there were no other significant regional differences on comparison of manifesting DYT1 to non-manifesting DYT1. Comparisons of motor task-specific activation between groups (Panel B in Table 2) showed an increase in the inferior parietal cortex in manifesting DYT1. In this region (Fig. 1D), significant deactivation (CCW < AV) was present during task performance in healthy subjects. The magnitude of this deactivation response was attenuated in non-manifesting DYT1 carriers and failed to occur in their affected counterparts.

Figure 1

Brain regions with abnormal activation in manifesting DYT1 carriers (Table 2, see text). SPM{t} maps (left) were superimposed on a single-subject MRI T1 template (x, y, z coordinates in MNI space indicate the position of the slice). The colour scale represents t-values thresholded at 3.5 corresponding to P < 0.05 false discovery rate corrected. Bar diagrams (right) of adjusted rCBF measured during performance of the motor task (CCW) or in the non-motor audio-visual control condition (AV) (see text). For each volume-of-interest, rCBF values (mean ± SE) were displayed for manifesting (MAN) DYT1 (dark grey), non-manifesting (NM) DYT1 (light grey), and control subjects (white). Increased motor activation was present in affected carriers in the left sensorimotor cortex (SMC, A), supplementary motor area (SMA, B), and in the dorsal premotor cortex (dPMC, C). Loss of task-related deactivation was evident in the inferior parietal lobule (PARi, D). The activation abnormalities in the sensorimotor cortex and supplementary motor area were specific for clinically affected DYT1 carriers. In dorsal premotor cortex and PARi, abnormal activation was also present in non-manifesting gene carriers. In the non-motor AV condition, rCBF was elevated in the right cerebellum (E), with significant rCBF elevations in both manifesting and non-manifesting DYT1 carriers. In contrast, in the left sensorimotor cortex (F), rCBF was elevated only in the affected mutation carriers. (Significant P-values of post hoc tests are denoted by asterisks).

View this table:
Table 2

Brain regions with significant differences in regional cerebral blood flow between manifesting DYT1 carriers and controls

Voxel-based comparisonsa: manifesting DYT1 versus controlsVolume of interest-based comparisonsd: manifesting DYT1, non-manifesting DYT1 and controls
Brain regionZmaxCoordinates (MNI)Cluster extentP < 0.05
A. Manifesting DYT1 (CCW) > controls (CCW)b
    L SMC (BA1–4), pre-parietal (BA5), premotor (BA6), parietal (BA7, BA40) and cingulate cortices (BA24, 31); bilateral SMA (BA6)c5.4−54−28524981MAN > NM =C
    L cingulate (submax., BA 24)c5.2−8−846MAN > NM = C
    L SMA (submax., BA 6)4.6−12−3068MAN > NM = C
    R SMA (submax., BA 6)4.512−2072MAN > NM > C
B. Manifesting DYT1 (CCW–AV) > controls (CCW–AV)
    L inferior parietal lobule (BA 40)c5.1−36−4854128MAN > NM = C
    R inferior parietal lobule (BA 40)c5.040−504698MAN > NM > C
C. Manifesting DYT1 (AV) > controls (AV)b
    R anterior cerebellumc4.916−44−14478MAN > NM > C
    L anterior cerebellum4.6−12−50−2467MAN > NM > C
    L SMC (BA 3, 4)3.9−30−2866164MAN > NM = C
D. Manifesting DYT1 (AV) < controls (AV)b
    R DLPFC (BA 9)4.140104288MAN = NM < C
  • a Group differences in rCBF recorded during reaching movements (CCW) are presented in A and B. rCBF differences recorded in the audio-visual reference condition (AV) are presented in C and D. The left most columns reflect local maxima identified in statistical parametric mapping contrasts of only manifesting DYT1 versus controls. Right-hand columns show the results of the comparisons including non-manifesting DYT1 in addition to manifesting DYT1 and controls (see text). Volume-of-interest measurements at the maximum identified in the respective statistical parametric mapping comparisons of manifesting DYT1 and controls were used for these ANOVAs.

  • b Comparisons in [CCW–AV] mask of the study population.

  • c Family-wise error-corrected P < 0.05.

  • d Tukey–Kramer honestly significant difference.

  • BA = Brodmann area; C = controls; DLPFC =dorsolateral prefrontal cortex; L = left; MAN = manifesting DYT1; MNI = Montreal Neurological Institute space; NM = non-manifesting DYT1; R = right; SMA = supplementary motor area; SMC = sensorimotor cortex.

To address the effect of genotype on the abnormal manifesting DYT1 activations, volume-of-interest comparisons were performed on the data from the manifesting DYT1, non-manifesting DYT1 and control groups (see the ‘Material and methods’ section). Abnormal activation of the left sensorimotor cortex, supplementary motor area, cingulate and superior parietal regions (Panel A in Table 2) was specific for the affected carriers, and was not present in the non-manifesting carriers (Fig. 1A and B). In contrast, motor activation of the left dorsal premotor cortex was abnormally increased in DYT1 mutation carriers irrespective of penetrance. In this region (Fig. 1C), non-manifesting DYT1 carriers assumed an intermediate position between affected carriers and controls. Similar changes were present in the right supplementary motor area.

Analyses of the volume-of-interest data for each region confirmed that the significant group differences during motor task performance were not attributable to underlying group differences in rCBF recorded in the AV reference condition.

Significant differences were also present in group comparisons of rCBF recorded in the non-motor AV condition. In this condition, manifesting DYT1 carriers exhibited abnormally increased rCBF in the anterior cerebellum bilaterally and in the left sensorimotor cortex (Panel C in Table 2). In the right cerebellum (Fig. 1E), volume-of-interest analysis revealed stepwise rCBF increases in non-manifesting and manifesting DYT1 carriers relative to controls. In contrast, in this condition, abnormally increased left sensorimotor cortex activation (Fig. 1F) was evident only in the affected carriers. Significant reductions in rCBF recorded in the AV condition were also present in the right dorso-lateral prefrontal cortex of manifesting and non-manifesting DYT1 carriers relative to controls (Panel D in Table 2).

Ordinal trends analysis

Normal movement-related activation pattern

OrT/CVA was applied to the set of 39 cerebral blood flow scan pairs from an independent cohort of healthy subjects (age 37.2 ± 13.4 years) who performed the CCW and AV tasks (see above). An activation pattern was identified that exhibited an ordinal trend in that its expression in CCW was greater than in AV on every run (P < 0.0001, permutation test; Fig. 2A). This normal movement-related pattern (Fig. 2B) was characterized by covarying activation of the left sensorimotor cortex (BA 1–4), premotor cortices (BA 6; dorsal premotor cortex, supplementary motor area), cingulate cortex (BA 24, 31), superior parietal (BA 5, 7), and inferior parietal (BA 40) cortices, and the occipital association region (BA 18) (Table 3). Additional regions with positive network contributions (i.e. network-related regional increases in CCW relative to AV) were identified in the left anterior cerebellar cortex, right inferior cerebellar cortex, vermis and deep cerebellar nuclei (Fig. 2A). Negative network contributions (i.e. network-related regional decreases in CCW relative to AV) were present in the right temporal (BA 21) and inferior parietal (BA 40) regions, and bilaterally in the dorsolateral prefrontal cortex (BA 46, 9) (Table 3). Voxel weights on these network regions were demonstrated to be reliable by bootstrap estimation (integrity check value extrema −5.11 to 7.85, P < 0.0001). Subject scores on this pattern exhibited high within-session reproducibility in CCW scans (intraclass correlation coefficient 0.85, 95% CI 0.71–0.93; P < 0.0001), and in the AV scans (intraclass correlation coefficient 0.98, 95% CI 0.95–0.99; P < 0.0001).

Figure 2

The normal motor-related activation pattern. (A) Normal motor-related activation pattern (NMRP) identified by ordinal trends canonical variates analysis (OrT/CVA) of 78 Embedded Image PET scans (39 CCW- and 39 AV) acquired in 18 healthy volunteer subjects (see text). This spatial covariance pattern was characterized by activation of the left sensorimotor cortex (SMC), premotor cortex (dPMC) and inferior parietal cortex (Table 3). Additional regional contributions to network activity were found bilaterally in the cerebellar vermis and hemispheres. [The colour scale represents positive voxel weights thresholded at Z = 3.09, corresponding to regions that contributed significantly (P < 0.001) to network activity and which were also reliable (P < 0.001) on bootstrap estimation.] (B) Normal motor-related activation pattern expression in the subjects comprising the original derivation sample. For all subjects and runs, the expression of this pattern increased during the performance of the motor task (P < 0.0001; see text).

View this table:
Table 3

Regions with significant contributions to the normal motor-related activation pattern

Brain regionCoordinates (MNI)
Positive loadingsa
    L sensorimotor cortex (BA 1–4)7.0−37−4360
    L SMC5.9−36−1850
    L dorsal premotor cortex (BA6)7.6−22−1671
    Supplementary motor area (BA 6)4.9−2−1750
    R cingulate cortex (BA 24)4.36−1450
    L cingulate cortex (BA 31)3.5−8−2641
    L superior parietal (BA 5, BA 7)4.2−22−4864
    L inferior parietal (BA 40)4.3−54−3234
    Occipital association cortex (BA 18, 31)3.8−1−7714
    L anterior cerebellar cortex4.1−20−60−26
    R deep cerebellar nuclei3.222−41−33
    R inferior cerebellar cortex3.422−60−44
Negative loadingsa
    R temporal cortex (BA 21)3.760−44−4
    R inferior parietal (BA 40)4.260−5424
    R DLPFC (BA 46)3.4483016
    R DLPFC (BA 9)3.738746
    L DLPFC (BA 9)3.5−46746
  • a Regional contributions to normal motor-related activation pattern activity were considered significant for Z > 3.09, P < 0.001 (see text).

  • BA = Brodmann area; DLPFC = dorsolateral prefrontal cortex; L = left; MNI = Montreal Neurological Institute space; R = right; SMC = sensorimotor cortex.

Abnormal normal motor-related activation pattern expression in DYT1 carriers

Normal motor-related activation pattern expression was computed prospectively in scans from the manifesting and non-manifesting DYT1 carriers and the healthy control subjects acquired in the CCW and AV conditions. An ordinal trend across conditions (CCW > AV) was evident for each of these groups (P < 0.001, paired t-tests), in that pattern expression was elevated in CCW relative to AV for each scan pair, without a single violation in the prospective control cohort and in non-manifesting DYT1 (Supplementary Fig. S1). In manifesting DYT1, there were three violations in 22 scan pairs. For both conditions, significant differences in normal motor-related activation pattern expression were evident across the three groups (P < 0.001, ANOVA; Fig. 3A). Post hoc testing revealed increased normal motor-related activation pattern expression in manifesting DYT1 relative to non-manifesting DYT1 and controls for both the AV and CCW conditions (P < 0.05, Dunett’s T3). Notably, normal motor-related activation pattern expression for non-manifesting DYT1 was within the range of control values for CCW scans (Fig. 3A, left). In contrast, for AV scans (Fig. 3A, right), pattern expression was significantly elevated in non-manifesting DYT1 carriers relative to controls (P < 0.05, Dunett’s T3).

Figure 3

Normal motor-related activation pattern expression in DYT1 mutation carriers. (A) Box-whisker plots of normal motor-related activation pattern expression recorded in the AV reference condition (AV, left) and during movement (right). In the AV condition, pattern expression was elevated in both the manifesting and non-manifesting DYT1 groups, with the latter positioned between the affected carriers and the controls. In the motor condition, significant increases in normal motor-related activation pattern expression were evident only for the affected carriers. [Significant post hoc comparisons with controls (P < 0.05) are denoted by asterisks; see text.] (B) Normal motor-related activation pattern expression in manifesting DYT1 carriers measured in the AV condition correlated with clinical severity ratings according to the BFMDRS. (Black diamonds: subjects without contractions at rest; grey diamonds: subjects with constant or with occasional contractions at rest. The normal mean and range are indicated by the shaded area).

Normal motor-related activation pattern expression measured in the AV scans correlated (r = 0.70, P = 0.04) with concurrent BFMDRS motor ratings (Fig. 3B). These clinical ratings did not correlate significantly with normal motor-related activation pattern expression measured in the CCW scans (r = 0.49, P = 0.2). We note that there was no difference between normal motor-related activation pattern values from the group that was used for pattern derivation (n = 18) and those computed on a prospective case basis in the subsequent control group (controls, n = 12), whether quantified in AV or CCW scans (P > 0.2).


The results of this study demonstrate that overactivity of the sensorimotor network, as manifested by abnormally increased normal motor-related activation pattern expression in DYT1 carriers, is an important and quantifiable trait feature of DYT1 dystonia (Edwards et al., 2003; Rothwell and Huang, 2003; Quartarone et al., 2006). While these abnormalities were evident at the network level, significant regional changes were also discerned, with increased activation of the sensorimotor cortex as the most salient finding in affected carriers. That said, whether clinically penetrant or not, the gene carriers exhibited elevated activation of auxiliary motor and parietal association regions. These findings indicate that in DYT1 dystonia, increased cortical excitability is not restricted to the sensorimotor cortex and that higher order motor and sensory integration areas may be involved as a trait feature of the disorder. Similar findings have been reported in functional imaging studies of sporadic dystonia (Odergren et al., 1998; Preibisch et al., 2001; Dresel et al., 2006; Tecchio et al., 2008), suggesting mechanistic similarity with hereditary forms of the disorder.

The regional activation observed on univariate as well as multivariate image analysis accord with what is known of the neural mechanisms that mediate self-initiated reaching movements (Winstein et al., 1997; Scott, 2003) with goal-directed orientation (Buneo et al., 2002). In this regard, normal motor-related activation pattern expression in individual subjects can be interpreted as a quantitative descriptor of sensorimotor network activity measured in either the motor (CCW) or non-motor (AV) conditions. Of note, pattern expression in the latter condition was found to correlate with independent clinical severity ratings, suggesting the potential of this network measure as an objective treatment biomarker for use in trials of new therapeutic interventions for this disorder. For instance, the efficacy of pallidal deep brain stimulation is well established in primary generalized dystonia (e.g. Vidailhet et al., 2005; Kupsch et al., 2006). However, there is high inter-individual variability in the time course of clinical response and in overall improvement (Alterman et al., 2007; Isaias et al., 2008). Moreover, functional imaging with Embedded Image PET can be advantageous in studies of surgical interventions for dystonia because of the safety concerns involving MRI in patients with implanted deep brain stimulation electrodes. Similar challenges may exist in the assessment of novel interventions for dystonia (e.g. Gonzalez-Alegre et al., 2005; Jinnah and Hess, 2008). Indeed, the development and validation of objective and reliable biomarkers of treatment efficacy is likely to facilitate the evaluation of new therapies (cf. Feigin et al., 2007).

We recognize that the observed correlation between normal motor-related activation pattern activity and clinical severity ratings may be a reflection of the presence of dystonic contractions during the scan. This may be particularly relevant in patients with generalized dystonia and high normal motor-related activation pattern expression in ‘non-motor’ conditions. Nonetheless, abnormal increases in normal motor-related activation pattern activity were also observed in non-manifesting carriers scanned in the AV experimental condition. In this circumstance, the scans were free of the potential confound of involuntary movement. Moreover, motor-related activation pattern values from the two subjects with task-specific dystonia (who were demonstrably free of dystonic contractions during scanning in the AV condition) also fell outside the control range (see Fig. 3). The current observations are entirely preliminary; future studies, particularly in subjects with focal dystonia, will be needed to confirm the correlation between normal motor-related activation pattern scores and concurrent clinical dystonia ratings.

The current findings are also consistent with impaired sensory function in DYT1 dystonia. We note that the increase in normal motor-related activation pattern expression was proportionately larger in the AV condition relative to movement (CCW). Indeed, patients with DYT1 dystonia exhibited an almost 4-fold increase in pattern expression relative to controls when measured in the AV condition. In contrast, the normal motor-related activation pattern increase was <2-fold when measured during CCW performance. It has long been debated whether dystonia represents a primarily sensory as opposed to motor disorder (Hallett, 1995; cf. Defazio et al., 2007; Tinazzi et al., 2009). In contrast to sporadic focal dystonia, behavioural studies of somatosensory processing in DYT1 carriers have yielded inconsistent findings. Spatial discrimination thresholds have been found to be intact (Molloy et al., 2003), while abnormalities in tactile and visuo-tactile temporal discrimination thresholds have been reported in both manifesting and non-manifesting DYT1 carriers (Fiorio et al., 2007).

Although our results are consistent with previous neurophysiological studies in sporadic as well as DYT1 dystonia, discrepancies with earlier functional imaging studies in sporadic dystonia remain unresolved. Abnormally increased sensorimotor cortex activation has been reported in investigations of focal action-specific dystonia (Odergren et al., 1998; Pujol et al., 2000; Preibisch et al., 2001; Blood et al., 2004; Lerner et al., 2004; Hu et al., 2006) and of secondary dystonia (Ceballos-Baumann et al., 1995a; Lehericy et al., 2004). Nonetheless, abnormally decreased sensorimotor cortex activation has also been reported in focal dystonia (Ceballos-Baumann et al., 1995b; Ibanez et al., 1999; Oga et al., 2002; Haslinger et al., 2005; Dresel et al., 2006), as well as in idiopathic generalized dystonia (Ceballos-Baumann et al., 1995a; Playford et al., 1998). Abnormally increased activation of secondary motor cortices appears to be a consistent finding across the dystonia spectrum (Ceballos-Baumann et al., 1995a, b; 1997; Odergren et al., 1998; Lehericy et al., 2004; Lerner et al., 2004; Hu et al., 2006), although reports also exist of abnormally reduced activation in these areas (Ibanez et al., 1999; Pujol et al., 2000; Haslinger et al., 2005; Dresel et al., 2006). It is possible that these inconsistencies are attributable to differences in experimental paradigms and analytical strategies. In particular, functional MRI studies inherently rely on the difference in blood oxygen level-dependent signal between states as a measure of activation in brain tissue. As a consequence, these studies may potentially ascribe ‘underactivity’ to an area with an abnormal increase in baseline neural activity, or ‘overactivity’ to one with reduced activity at baseline (cf. Argyelan et al., 2009).

It is not known whether increased motor cortical excitability and disorganized sensorimotor integration reflect an intrinsic susceptibility to dystonia, or whether these findings represent a consequence of maladaptive reorganization (Rothwell and Huang, 2003). Supporting the former notion, Edwards and colleagues (2003) demonstrated reductions in intra-cortical inhibition in DYT1 carriers irrespective of clinical penetrance. Increased sensorimotor activity in non-manifesting DYT1 carriers, as evident by elevated normal motor-related activation pattern expression in the AV condition, is likewise consistent with an intrinsic abnormality. Notably, this experimental condition did not require tactile input or motor output. Thus, increases in network activity in this state may be a manifestation of broad changes in higher order multimodal sensory integration. That said, these findings may be specific for the DYT1 genotype. Further studies will be needed to determine their relevance to other forms of primary dystonia.

Lastly, we note that the observed increases of motor cortex activity in DYT1 carriers correlate closely with microstructural changes in the cerebello-thalamic-cortical pathways of the same subjects (Carbon et al., 2008a; Argyelan et al., 2009). Indeed, we have found a tight inverse relationship between cerebellar outflow pathway connectivity and concurrently recorded neural activity in the sensorimotor cortex and premotor regions (Argyelan et al., 2009). In contrast, the changes in sensorimotor cortex activity recorded in DYT1 mutation carriers do not correlate with reductions in striatal D2-receptor binding measured in the same subjects (cf. Carbon et al., 2009). Additional multi-modal imaging experiments will be needed to characterize more precisely the relationships between reduced integrity of cerebellar pathways, elevated sensorimotor cortex neural activity and abnormal dopamine neurotransmission in dystonia patients and in non-manifesting carriers of dystonia mutations.


The National Institutes of Health (R01 NS 047668 to D.E.), the General Clinical Research Centre of The Feinstein Institute for Medical Research [National Centre for Research Resources, a component of the National Institutes of Health (M01 RR 018535)], and NIH/ NIBIB Grant 5R01EB006204 to C.H.

Supplementary material

Supplementary material is available at Brain online.


We thank Dr Thomas Chaly for valuable radiochemistry support and Mr. Claude Margouleff for technical support. We are also indebted to Mrs Deborah Raymond for participant recruitment.


  • Abbreviations:
    audio-visual reference task
    Brodmann area
    Burke-Fahn-Marsden Dystonia Rating Scale
    counterclockwise reaching motor task
    ordinal trends canonical variate analysis
    regional cerebral blood flow
    statistical parametric mapping


View Abstract