Patients and methods

SUBJECTS AND EXPERIMENTAL DESIGN CONDITIONS

After giving informed consent four patients (one woman, aged 68 years; three men, aged 41, 56, and 68 years) with definite writing tremor participated in the study. The data from the 68 year old man had to be excluded as the fMRI was overshadowed by tremor artifacts. The female patient had had writing tremor for 6 years and reported an additional slight tremor of the right hand during car driving and eating. The 41 year old male patient had recognised the first symptoms of writing tremor 11 years ago and a slight tremor when grasping small objects such as a spoon. Both showed no dystonic postures during writing. In the 56 year old male patient with writing tremor for 9 years, writing tremor was associated with a slight hyperextension of the wrist and abduction in the shoulder. All patients were right handed with no neurological abnormalities except for writing tremor and had no dystonic movements in the right arm while at rest. Brain MRI of the female patient displayed small vascular lesions of the right temporal lobe; MRI of all other subjects was normal. There was no previous neuroleptic medication and no family history for movement disorders. None of the patients received any medication at the time of fMRI that could interfere with central activities.

The control group consisted of 10 healthy subjects (five women, five men, mean age: 34.1 years, range: 28–43 years).

Patients and controls were asked to either rest or write with their right hand according to the instructions given. Each patient wrote the same sentence (“keine Rose ohne Dornen” (no rose without thorns)) to make sure that writing was performed as a complex but automatic and overlearned motor task requiring no decision making. They were trained before the investigation to write in a supine position with eyes closed and as little movement of the arm and head as possible.

The paradigm consisted of alternating a 30 second period of rest and a 30 second period of writing. The rest/writing cycle was repeated three times. Every 3 seconds the subjects were instructed to either write or rest.

MRI DATA ACQUISITION

All experiments were performed on a 1.5 Tesla Magnetom VISION (Siemens, Erlangen, Germany) whole body MRI system equipped with a circular polarised volume head coil. Initially a set of localised images was acquired to position the imaging slices.

For functional imaging 60 EPI multislice data sets were acquired (TE 45 ms, TR 3 seconds, flip angle 90°, acquisition time for the whole paradigm 3 minutes). Each multislice data set contained 16 transverse slices (slice thickness 5 mm, interslice gap 2.5 mm, matrix 64×64, FOV 25 cm). At the beginning of the session for each subject a set of 16 T1 weighted anatomical images was acquired in the same slice positions.

DATA ANALYSIS

All images were analysed using SPM96 Software (Wellcome Department of Cognitive Neurology, London, UK; available athttp://www.fil.ion.ucl.uk).25-27 The first three data sets of each time series were discarded to eliminate the influence of magnetic saturation effects. All remaining EPI volumes were then realigned to the first volume28 to correct for movement between scans. This spatial transformation is essentially a six parameter rigid body transformation (three translations, three rotations). To ensure that the structural and functional images were in the same anatomical space, the T1 weighted anatomical images were coregistered to a mean EPI data set which was previously calculated from the realigned time series.28 Finally, the anatomical image was spatially normalised28 to a standard template,29 30 using a 12 parameter affine transformation. The same transformation was then applied to the EPI time series data. The images were smoothed with a Gaussian kernel of 10 mm full width at half maximum (FWHM) for single subject and group analysis. The images were proportionally scaled to remove the effect of global signal differences between subjects.

STATISTICAL ANALYSIS

Statistical analysis was performed using the general linear model as employed by SPM96. The different conditions (rest and writing) were modelled as boxcar functions convolved with the haemodynamic response function.26 Aliased low frequency noise caused by cardiac and respiratory effects were modelled using cosine waves of 120 to 240 seconds. The remaining autocorrelation of the residual errors was removed by temporal smoothing with a 2.8 second FWHM gaussian kernel. Statistical maps (Z statistics) were created by applying appropriate contrasts to the parameter estimates for each condition. The contrast used was +1 for writing and -1 for rest. This contrast shows voxels that have a significantly larger signal during writing relative to rest. The resulting statistical parametric maps were interpreted by referring to the gaussian random field theory because of the need to correct for multiple independent comparisons. Data were analyzed for the group of subjects with writing tremor and for the group of healthy volunteers separately. As only data of three patients with writing tremor were analyzed, no group comparison was performed for patients with writing tremor.

Results

Except for small vascular lesions of the right temporal lobe in one patient, MRI of all subjects investigated was normal. The field of view comprised the whole cerebrum and the cerebellum.

Data of activated areas (statistical and anatomical data) of patients with writing tremor are given in table 1 and for the control group in table 2. The tables summarise the coordinates of first order maxima of activated clusters as well as the second order maxima within clusters, which are especially important in clusters covering large areas. The statistics (height threshold p=0.00001, extent threshold p=0.001) are displayed as maximum intensity projections on a “glass brain” in normalised Talairach space: figure 1 A and B show the activated brain areas of patients with writing tremor and controls. Figure 2 A and B display the respective activation maps projected on the surface of the brain.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Activated brain areas displayed as maximum intensity projections during writing in patients with writing tremor (A) and healthy controls (B). The results have been displayed as statistical maps in the sagittal, coronal, and transverse projection according to the atlas of Talairach and Tournoux.30

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Activation maps projected on the surface of the brains from the bottom and the top as well as of both hemispheres in patients with writing tremor (A) and healthy controls (B).

PATIENTS WITH WRITING TREMOR AND CONTROLS

Patients with writing tremor and controls in common showed a marked activation of the left sensorimotor cortex (SMC) with a remarkably consistent pattern, of area 44, and of the right cerebellar hemisphere, which was larger in patients with writing tremor.

By contrast, patients with writing tremor in addition had a marked activation of the left cerebellar hemisphere. In patients with writing tremor, the activated SMC area extended further caudal and anterior from the primary motor cortex towards the premotor cortex (area 6). This relative overactivity of the motor cortex in writing tremor was also reflected by the peak activation of the sensorimotor cluster in writing tremor, which is located anterior to that of healthy volunteers. Only patients with writing tremor showed an activation of the ipsilateral prefrontal cortex (inferior frontal gyrus; areas 44, 47, and 10) and bilateral parietal lobule (area 40) with a more pronounced activation on the contralateral side.

Discussion

This fMRI study shows that activation patterns of the cerebellum, parietal lobe, and frontal cortex are different in patients with primary writing tremor compared with controls during the task of writing.

CEREBELLAR, PARIETAL, AND FRONTAL LOBE ACTIVATION IN PRIMARY WRITING TREMOR

The bilateral cerebellar activation in writing tremor underscores the pivotal role of the cerebellum in tremor generation already known from patients with essential tremor.15-19 A bilateral cerebellar activation in unilateral postural essential tremor has been demonstrated in PET studies.17 31 Interestingly, bilaterally increased cerebellar rCBF was already visible in patients with essential tremor at rest,16-19 which led to a further bilateral increase in the presence of unilateral involuntary postural tremor, but not during passive wrist oscillation or during voluntary arm extension by control subjects.15-19Moreover, ethanol in alcohol responsive patients with essential tremor19 and stimulation of the contralateral ventral intermediate nucleus (Vim) of the thalamus in patients with unilateral tremor dominant Parkinson's disease32 have led to a significant decrease of previously bilaterally increased cerebellar blood flow. In support of the results of our study a bilateral cerebellar activation has also been described in a previous PET study in patients with writing tremor holding a pen to a paper.24 Therefore a bilateral hyperactivity of the cerebellar networks seems to play an important part also in the pathophysiology of writing tremor. Although patients with writer's cramp and healthy subjects may show a bilateral cerebellar activation during the task of writing, this activation is usually less pronounced compared with patients with essential tremor or writing tremor.22

The cerebellum has been shown to be involved in preparatory and executive activity33 34 and different aspects of cognitive behaviour and affective states.35 It receives input from multiple areas of the cortex including motor, premotor, and parietal areas, conveyed by way of several parallel pathways.35 The output targets project to various cortical areas such as the premotor, prefrontal, and posterior parietal cortex.35-40 These widespread connections suggest that a bilateral cerebellar activity should also be reflected by increased neuronal activity at a cortical level which indeed could be seen as an increased bilateral activation of the inferior parietal lobules. However, biparietal overactivity might not only be the result of an activated bilateral cerebellar loop. It may also reflect increased intrinsic activity of the parietal lobes resulting from an increased task difficulty and effort to perform the task as it is particularly the case in writing tremor. The parietal cortex is reciprocally connected with various isocortical areas of the frontal, temporal, and occipital lobe41 and is involved in the sequential processing of somatotopically organised information received from the adjacent primary somatosensory cortices.35 The inferior parietal lobule corresponds to the secondary somatosensory area and has bilateral but contralateral dominant receptive fields.42 A bilateral activation of secondary somatosensory areas during active movements has recently been associated with the sensorimotor integration using proprioreceptive feedback and auditory cues43-45 and with increased attention.43 46 The increased activation of the ipsilateral SMA could be an additional result of this bilaterally active loop, as it receives major input from the posteroparietal cortex.41

The prefrontal cortex is important for kinaesthetic, motivational, and spatially related functions. In particular, the right prefrontal cortex is activated in subjects choosing either direction or timing of movements.47-49 The activation of the right prefrontal area only seen in patients with writing tremor may therefore reflect the higher neuronal activity required for directing timing of movements under severely disabled circumstances.

Artificial activation patterns as the consequence of tremor in our patients have to be discussed but are very unlikely as motion artifacts were detected by the software, which led to the exclusion of one patient. Furthermore, artifacts would be expected to be scattered all over the brain with a predominance at the vertex. The activated areas shown in the present study clearly correspond to brain areas which are physiologically activated during writing and which can be easily integrated in the pathophysiological concept of tremor disorders.

ACTIVATION PATTERN IN WRITING TREMOR COMPARED WITH WRITER'S CRAMP AND ESSENTIAL TREMOR

There are several similarities in the activation pattern of patients with writing tremor with patients with writer's cramp and essential tremor. The findings of an increased activation of premotor cortical areas and the bilateral but predominantly contralateral cerebellar activation in our patients with writing tremor conform to previous PET studies on patients with writer's cramp.21 22 This points towards a common aetiology of writer's cramp and writing tremor, which is also supported by clinical and electrophysiological findings: both writer's cramp and writing tremor are focal and task specific with a coexistence of focal tremor and dystonia in some affected persons50 51 and even within the same family.52 53 In addition, cocontractions of antagonist muscles in the forearm which produce dystonic movements have been seen clinically and were also evident in EMG recordings in both conditions.11 13

A bilateral parietal and increased bilateral cerebellar activation has been reported in previous PET studies in patients with essential tremor.17 18 24 Particularly, the marked biparietal activation is in line with the hypothesis of writing tremor being a focal variant of essential tremor. This hypothesis is further underscored by various clinical findings such as similar frequencies in both conditions, and the response to propanolol, alcohol, and Vim stimulation.1 3 7 10 54

In conclusion, writing tremor shares several clinical aspects and the pattern of the cortical activation with both writer's cramp and essential tremor, as measured by fMRI and PET. Therefore, writing tremor cannot easily be classified as a variant of either essential tremor or writer's cramp. A distinct category should be proposed integrating hallmarks of both entities.