Dopaminergic Dysfunction in Midbrain Dystonia: Anatomoclinical Study Using 3-Dimensional Magnetic Resonance Imaging and Fluorodopa F 18 Positron Emission Tomography

Objective  To determine the role of damage to neuronal systems, especially the dopaminergic system, in patients with symptomatic dystonia and mesencephalic lesions.

Design  Stereotaxic magnetic resonance imaging analysis and positron emission tomography after the administration of fluorodopa F 18.

Patients  Of a group of 48 patients with unilateral dystonia following a stroke, 7 patients with a well-defined midbrain lesion were selected.

Results  All patients had unilateral dystonic posture of an upper extremity and cerebellar dysmetria or hypotonia. Cerebellar tremor was present in 1 patient. Two patients had resting and postural tremor, which showed a marked improvement with treatment with levodopa. In patients with dystonia only, dopaminergic lesions were mostly confined to the ventromesial mesencephalon and red nucleus area, including the substantia nigra and nigrostriatal and cerebellothalamic fibers. Dystonia was severe and did not resolve with time in patients with lesions involving the nigrostriatal pathway, and the degree of dopaminergic denervation revealed by positron emission tomography was correlated with the severity of dystonia. In patients with resting and postural tremor, lesions of the dopaminergic structures were larger and located more laterally and dorsally in the pars compacta, the perirubral and retrorubral areas, and extending to the central tegmental tract.

Conclusions  Dopaminergic dysfunction plays a role in the occurrence and severity of midbrain dystonia, and additional lesions to dopaminergic neurons in the perirubral and retrorubral areas result in tremor that responds to levodopa treatment.

LESIONS AFFECTING the striatopallidal complex and thalamus are responsible for most cases of symptomatic focal dystonia.^1-3 Dystonia may occasionally be caused by midbrain lesions^4,5 affecting the dopaminergic nigrostriatal system, the red nucleus area, and the superior cerebellar peduncle.^6,7 In patients with a midbrain lesion, dystonia is frequently associated with tremor,¹ which usually responds to treatment with levodopa.⁷ That dopaminergic cell loss has a role in the cause of resting and postural tremor has been suggested both experimentally^8,9 and clinically.^7,10 In contrast, it is unclear whether midbrain structures contribute to dystonia, with or without tremor.

We used stereotaxic analysis of 3-dimensional (3-D) T₁-weighted magnetic resonance imaging (MRI) and positron emission tomographic (PET) examination with fluorodopa F 18 ([¹⁸F]) to investigate which structures are responsible for dystonia that occurs alone or with tremor.

Patients were selected for this study if they had both unilateral, localized symptomatic dystonia and the presence of a well-defined lesion due to a midbrain stroke. Of 48 patients with unilateral poststroke dystonia observed in our department between 1989 and 1997, 7 fulfilled the criteria. The clinical features and topography of the dystonia and associated movement disorders and the cause of the stroke are summarized in Table 1. Dystonia was defined as sustained muscle contractions that caused twisting and repetitive movements or abnormal postures.¹¹ Polygraphic electromyographic recordings and accelerometry measurements were obtained in all patients with associated tremor (Table 1). The clinical assessment and polygraphic electromyographic recordings were made blind to the 3-D MRI analysis of the anatomical location of the lesion. Dystonia was evaluated using the Fahn and Marsden scale,¹² and akinesia and rigidity tremor were evaluated with the scores on the motor subscales of the Unified Parkinson's Disease Rating Scale.¹³ The baseline score was obtained 12 hours after a patient had withdrawn from antiparkinsonian treatment. The treated score was measured when the patient had attained maximal improvement of motor signs. The dystonia score is the sum of the scores for each part of the body of the provoking factor (0 indicates no dystonia; 4, dystonia present at rest)×the severity factor (0 indicates no dystonia; 4, dystonia present most of the time). Maximal possible score is 120.

All studies were performed on a 1.5-T MRI unit. After a scout sequence was taken, the following 3 series of scans were made at each examination: sagittal, T₁-weighted; coronal, 3-D Fourier-transform spoiled-gradient acquisition at the steady state, T₁-weighted (1.5-mm slice thickness); and axial or coronal, T₂-weighted. Compared with computed tomographic scans or standard MRI, 3-D T₁-weighted MRI sections present several advantages: thin sections (1.5 mm), reducing partial voluming; multiplanar analysis; and reformatted images.

The stereotaxic procedures were conducted with the 3-D Fourier-transform spoiled-gradient acquisition sequence, using a Voxtool software workstation system (Advantage Windows Workstation; General Electric Co, Milwaukee, Wis). Images were reformatted sections, independent of the subject's head position, and were strictly symmetrical in the axial, sagittal, or coronal planes. Axial and sagittal sections were reconstructed parallel to the axis of the anterior and posterior commissures, and coronal reconstructions perpendicular to the axis of the anterior and posterior commissures were performed (contiguous 1-mm-thick slices). The line of the anterior and posterior commissures was superimposed on each slice (determined as the shortest distance between opposing surfaces of the commissures). These reconstructions allowed the calculation of the coordinates of length, height, and laterality of the lesions compared with the bicommissural line. Measurements were corrected for the width of the third ventricle, the height of the thalamus, and the length of the line of the anterior and posterior commissures. These coordinates were analyzed according to the atlas of Hassler. The 3-D MRI analysis of the lesions was made blind to the clinical assessment.

All patients were examined using a high-resolution brain PET scanner (ECAT 953B/31; CTI, Knoxville, Tenn) that collects 31 simultaneous 5-mm-thick planes, 3.38 mm apart, with an in-plane resolution of 6 mm.¹⁴ None of the patients was taking dopaminergic medication at the time of the PET study. Benserazide, a peripheral blocker of the dopa decarboxylase, was given in a dose of 50 mg 1 hour before tracer was administered. The patients were positioned and maintained using an individually molded head holder. The correction for tissue attenuation of 511-keV gamma radiation was measured. Scanning was initiated immediately after the intravenous administration of [¹⁸F]fluorodopa, 130 to 333 MBq. Nine time frames were acquired in the ensuing 90 minutes.^15,16

The time frames collected from 30 to 90 minutes were summed to create an integrated image such as the one presented in Figure 1. Circular regions of interest (ROIs; diameter=10 mm) were placed by visual inspection above the head of the caudate nucleus (1 ROI) and the putamen (3 ROIs) in each plane where these structures were visible.^15,16

A circular ROI was placed on the occipital region at the same level. Regional time-activity curves were obtained, and [¹⁸F]fluorodopa uptake constants (K_i per minute) were determined in the caudate nucleus and the putamen using a multiple-time graphic analysis and the occipital activity as a nonspecific input function.^15,17,18 The results obtained in the patients were compared with the K_i values measured using the same methods in the caudate nucleus and the putamen of 11 control subjects (age, 36 ± 6 [mean±SD] years) and 13 patients (age, 59±8 years) having severe Parkinson disease (stages IV-V on the Hoehn and Yahr scale).¹⁵

A 56-year-old man with a medical history of diabetes mellitus and hypercholesterolemia had a stroke at age 55 years. For 2 to 3 days, he had dizziness, gait ataxia, and diplopia. At that time, the patient had a complete left third nerve palsy, mild cerebellar ataxia on the tandem-walk test, cerebellar dysmetria, and adiadochokinesis of the right arm. No sensory loss was detected. Two days after the stroke, dystonia was observed. At the time of the study, he had mild dystonic posture with hyperextension and abduction of the fingers of the right hand on action. When he walked, internal rotation of the arm and flexion of the wrist and the fingers associated with dystonic hyperextension of the big toe were present. The dystonia score was 6. An MRI showed a left paramedian midbrain infarct (Figure 2, B). The results of echocardiography and cerebral angiography were normal. The patient was treated with aspirin (250 mg/d). One year after the study, the patient's condition improved, and he considered himself normal. Dystonia was observed only when he walked.

A 63-year-old man with a medical history of cardiac infarct and heavy smoking had a first transient ischemic stroke at age 53 years. Clinical signs, including clumsiness, left arm paresthesia, diplopia, and dysarthria, persisted for several hours. At age 54 years, he had right hemiplegia plus coma. A computed tomographic scan showed an old infarct of the right middle cerebral artery (perisylvian cortical region) associated with a recent infarct of the vertebrobasilar system involving the left temporo-occipital region of the posterior cerebral artery, the bilateral paramedian thalamic, and the left midbrain. Echocardiography showed apical cardiac hypokinesia. Cerebral angiography did not show significant arterial stenosis. The patient was treated with warfarin. The patient's hemiparesis resolved completely. Six months after the stroke, the patient complained of tremor in his right arm. Intermittent resting tremor was observed in the right hand (flexor digitorum, with little involvement of the extensor digitorum) and arm (largely in the biceps). On polygraphic recordings, bursts of tremor (145-200 milliseconds) were separated by silent periods, and alternating agonist-antagonist tremor was observed. Tremor was clinically similar to parkinsonian tremor, although the mean frequency was 4.0 to 4.5 Hz on accelerometry. When the arms were extended, the frequency increased to 5.5 Hz, and tremor became less regular. During voluntary movements, tremor increased greatly in amplitude.

Nine years later, at the time of the study, the results of polygraphic recordings were unchanged. Magnetic resonance imaging showed a left thalamomesencephalic infarct (Figure 2, C). Slow, regular resting tremor was observed in the right arm. Postural and action tremor of great amplitude was present. A dystonic posture with extension of the fifth digit of the right hand was observed in all positions. The dystonia score was 6. Resting, postural, and action tremors lessened with treatment with levodopa (150 mg, 5 times a day), whereas dystonia was unchanged. During a levodopa test, he had a 70% improvement of the resting (baseline score=1 and treated score=0) and postural (baseline score=3 and treated score=1) components of tremor. In the arm, rigidity (baseline score=1 and treated score=0) and akinesia (baseline score=6 and treated score=2) were also lessened. Mild cerebellar hypotonia was present. Mild hypesthesia was observed in the hand and forearm.

All patients had dystonia affecting the upper extremity or face. With the exception of patients 1 and 6, all patients had mild or moderate dystonia. Patients 3 and 4 had mild dystonia of the hand within 2 to 3 days of the initial stroke. Dystonia lessened in patient 4 and disappeared in patient 3 within 18 months. Patients 2 and 7 had mild dystonia of the fingers, which was noticed only on neurologic examination and disappeared within 3 months. Patient 1 had severe dystonia of the arm and face that developed while recovering from hemiparesis.

Parkinsonism consisting of "plastic" hypertonia associated with bradykinesia in alternative rapid movements was observed in patients 1, 2, 5, and 6. Patients 5 and 6 had 4-Hz distal resting tremor associated with irregular 5.5-Hz postural tremor in the upper extremity. The delay in the onset of tremor after the midbrain lesion varied from 1½ months to 1 year. Dystonia was not the initial symptom in either of these patients and was less disabling than tremor.

The effect of levodopa therapy was evaluated in 3 patients. Patients 3, 4, and 7 considered themselves to be normal and were not treated. Patient 2 refused the treatment. Patients 5 and 6 showed a 70% resolution of resting and postural tremor and of rigidity and akinesia in the arm. In patient 1, a 30% lessening of rigidity of the arm was observed. The reversibility of the effect of treatment was assessed: scores returned to the baseline values after the withdrawal of levodopa. In patients 1, 5, and 6, dystonic postures were unchanged after the long-term administration of levodopa, even when the doses were increased to 750 mg/d.

Cerebellar hypotonia and dysmetria were observed in all patients but patient 1. Coarse cerebellar action tremor of proximal distribution in the upper extremity was observed in patient 7. A neurologic examination showed mild sensory impairment in patients 1, 5, and 7.

Figure 1 shows the extent of the lesions, and Table 2 summarizes the neuroanatomical findings. Lesions affecting the upper midbrain in patients 1 through 6 were contralateral to the dystonic movements. In patient 7, the lesion was ipsilateral to dystonia, being situated in the lower midbrain and pons. The superior cerebellar peduncle was damaged before the decussation, and the ipsilateral superior cerebellum was also involved. In patients 1 through 6, the lesion involved the superior cerebellar peduncle after the decussation (patients 1 and 4-6). In patient 1, the lesion was more ventral, involving only a small part of the superior cerebellar peduncle. In patient 3, these cerebellar fibers were involved at the time of the first MRI examination (Figure 2, E), but the lesion was no longer visible in that area on follow-up scans (Figure 2, F). The red nucleus (mainly the caudal and ventromedial parts) was involved in patients 1 through 6. The medial and parafascicular nuclei of the thalamus (paramedian territory) were involved in patients 1, 2, 5, and 6. In patient 1, the lesion also extended to the most medial part of the ventral oral internal nucleus and intralaminar nuclei. In patient 5, the paramedian thalamic infarction was bilateral.

The dopaminergic nigrostriatal system was damaged in patients 1 through 6. In patient 2 (who had mild dystonia at examination), the lesion was situated more superiorly than in the other patients. In patient 3 (who recovered from dystonia), the lesion involved these fibers at the time of the first MRI examination (7 days after the onset). Follow-up MRI scans (18 months later), however, showed shrinkage of the lesion, which was no longer visible in the area of the nigrostriatal pathway. No involvement of this pathway was observed in patient 7. Dopaminergic cell bodies were affected in patients 1, 2, 4, 5, and 6. In patients 1, 2, and 4, the lesion damaged mainly the ventral tegmental area. The perirubral and retrorubral areas and the pars compacta of the substantia nigra showed lesions in only patients 5 and 6 with resting tremor. These 2 patients and patient 7 also had involvement of the central tegmental tract and the medial lemniscus. Last, in patient 1, the lesion was close to the pallidothalamic fibers in the Forel field. Additional lesions were also observed, including right lateral occipitotemporal and supramarginal gyri infarctions (patient 5), left parietal traumatic scar (patient 6), and mild leukoaraiosis (patients 2 and 7).

The [¹⁸F]fluorodopa K_i values obtained in the 7 patients are shown in Table 3, in which the radioactive K_i values obtained in control subjects and patients with severe Parkinson disease are also indicated. Four patients with dystonia (patients 1 and 4-6) had lower K_i values than did controls in the putamen ipsilateral to the mesencephalic lesion and contralateral to the dystonia. In the caudate nucleus ipsilateral to the lesion, the caudate K_i values were also decreased in 3 of these subjects, patient 4 having a caudate K_i in the normal range. In addition, the decrease of K_i values in patient 2 was bilateral in the putamen and contralateral to the lesion in the caudate nucleus, but in this patient, the lesion crossed the midline. In these 5 patients, the mean±SD radioactive K_i values for the ipsilateral caudate nucleus lesion (0.006±0.003 K_i/min) and putamen (0.004±0.002 K_i/min) were in the range of the K_i values observed in patients with severe parkinsonism (Table 3). Conversely, in patients 3 and 7, the striatal K_i values were in the normal range. Dystonic signs tended to be more severe in patients with striatal dopaminergic denervation (dystonia score=6.8±4.1, n=5) than in patients with normal K_i values (dystonia score=1 for patients 3 and 7). This is further suggested by the negative correlation between whole striatal K_i values ipsilateral to the midbrain lesion and the score of dystonia in the 7 patients (df=5, r=−0.93; P<.005) (Figure 3). For the last analysis, the mean striatal K_i was used in view of the correlation between caudate nucleus and putamen K_i values.

Dystonia resulting from a midbrain stroke was found in 7 of 48 patients with unilateral poststroke focal dystonia. In 3 of the 7 patients, tremor was observed: 1 patient had isolated cerebellar tremor, and the other 2 patients had resting and postural tremor that responded to levodopa treatment. Mild cerebellar signs (n=6) and parkinsonism (n=4) were also observed. The lesions involved the cerebellar pathways in all patients, the red nucleus area in 6, and the nigrostriatal dopaminergic system in 5.

Dystonia affected the distal part of the upper extremity without involvement of the lower extremities. A constant feature was the abnormal posture of the hand, with a permanent hyperextension of 1 or more fingers. The severity of dystonia depended on its intensity at the start of the disease: mild dystonia disappeared within a few weeks (patients 2, 3, and 7); when the dystonia was more severe, the evolution varied from mild improvement (patient 4) to an extension of dystonia to include the face (patient 1). These characteristics are difficult to compare with those of previously reported cases. In a revised study of the Benedikt syndrome,⁴ Souques et al¹⁹ described the dystonic postures of the arm and hand with flexed elbow and flexed fingers. Dystonia of mesencephalic origin has rarely been noticed in either computed tomographic or MRI scans^5-7,20,21 or mentioned in pathological studies.^19,22 Dystonia associated with isolated infarct of the pons^23,24 or central pontine myelinolysis²⁵ is rare.

In our study, the sign most frequently associated with dystonia was cerebellar dysmetria, observed in all patients except patient 1. Mild sensory impairment was observed in patients 1, 5, and 7. Parkinsonism—consisting of akinesia and plastic rigidity ipsilateral to dystonia—was present in patients 1, 5, and 6. In patient 2, whose dystonia was transient and noticed only on the initial neurologic examination, parkinsonism was mild. Such parkinsonian features, ipsilateral to dystonia^6,7 and located contralaterally to the predominant mesencephalic lesions, have already been reported. In some patients, however, bilateral parkinsonian manifestations were observed, although dystonia remained unilateral.^10,26

When observed in our group, tremor was of the resting and postural type (patients 5 and 6) and associated with both ipsilateral minimal plastic rigidity and akinesia and cerebellar dysmetria (patient 7). Resting and postural tremor was markedly lessened by levodopa therapy, in contrast to dystonia, where no abatement occurred. Such complex tremor has already been reported following mesencephalic infarction,^7,24,27 hemorrhage,^10,28 or head injury,^29-32 consisting of a combination of resting and postural components, with a proximal and distal distribution^6,7,10,33 and a slow frequency of 3 to 4 Hz.³⁴

Mesencephalic lesions, as seen in our patients, resulted in the varied association of dystonia, cerebellar manifestations, akinesia, plastic rigidity, and resting and postural tremor.

In agreement with previous reports,^{5-7,19,22,26,32,33,35} the structures most frequently damaged were the ventromedial mesencephalon, including the ventral tegmental area and the medial part of the substantia nigra, the nigrostriatal pathway, the superior cerebellar peduncle, and the red nucleus area.

The present data suggest that dopaminergic denervation plays a major role in the occurrence of symptomatic dystonia due to a midbrain stroke. First, the 5 patients with the most severe dystonia were those with a marked decrease in [¹⁸F]fluorodopa uptake in the ipsilateral striatum (Table 3 and Figure 1). The K_i values were within the range of those found in patients with severe parkinsonism.¹⁶ Second, there was a correlation between striatal K_i values ipsilateral to the lesion and the dystonia score (Figure 3). Third, the 2 remaining patients, who had normal striatal K_i values, had only mild and regressive dystonia. Moreover, in patient 3, the evolution of dystonia paralleled the regression of the size of the lesion in the ventromedial mesencephalon (Figure 2, E). These data suggest that dystonia is severe and persistent when dopaminergic nuclei and the nigrostriatal pathway are involved. The lack of abatement of dystonia by prolonged high-dose levodopa treatment may be explained by the fact that a lesion of nondopaminergic structures—such as the cerebellothalamic fibers, the serotonergic system,³⁶ and the cholinergic efferent projections of the pedunculopontine nucleus^37,38—also plays a role in midbrain dystonia or that sensorimotor reorganization beyond the nigrostriatal pathways is delayed.³⁹

Lesions of other structures were less consistent. Damage to the thalamus was observed in 4 patients and affected the paramedian territory (patients 1 and 2), as already described.^1-3,5,39-41 Paramedian thalamic infarcts, however, are frequently associated with subthalamomesencephalic lesions.⁵ Thus, it remains unclear in these patients whether dystonia results from damage to the paramedian territory of the thalamus or to the subthalamomesencephalic area. The contribution of rubral and perirubral lesions to dystonia is difficult to assess. Although these structures are almost consistently damaged, as shown in previous reports^6,7,19 and in this study, lesions of this area never occurred in isolation in our patients.

When dystonia was severe and did not improve with time, the lesions mainly involved the nigrostriatal pathway and the structures in the close vicinity. This suggests that direct dopaminergic dysfunction is mandatory for the persistence of dystonia.

Tremor in patient 7 was characteristic of action cerebellar tremor and was associated with hypotonia. This was consistent with the presence of a lesion of the cerebellum and cerebellothalamic fibers in the pons. The movement disorder was thus ipsilateral to the lesion. Patients 5 and 6 had resting and postural tremor, which lessened with levodopa therapy. This suggests that lesions of the nigrostriatal dopaminergic system play a role in the occurrence of such midbrain tremor, as already shown clinically⁷ and experimentally.^8,9 In contrast to patients with isolated dystonia (patients 1-4), patients 5 and 6 had lesions located more laterally and dorsally in the substantia nigra that also affected the perirubral and retrorubral areas (Figure 3 and Figure 4). This is in agreement with the observation that dopaminergic cell loss in the perirubral and retrorubral areas in Parkinson disease preferentially occurs with severe tremor.⁴² It suggests that involvement of this area is essential for the occurrence of the resting and postural component of the tremor. Involvement of the "rubro-olivocerebellorubral loop" and of the medial lemniscus (observed only in patients 5 and 6) may also be necessary for the occurrence of tremor, however. Experimental data in monkeys have shown that only combined lesions of the dopaminergic pathway and the rubro-olivocerebellorubral loop result in sustained tremor.⁹

Patients with dystonia alone and dystonia with tremor had lesions in different parts of the mesencephalon. Tremor occurs when lesions involve both the perirubral and retrorubral dopaminergic system and the rubro-olivocerebellorubral loop.

Accepted for publication October 28, 1998.

Corresponding author: Marie Vidailhet, MD, Department of Neurology, Hôpital Saint Antoine, 184 rue du faubourg Saint Antoine, 75012 Paris, France (e-mail: marie.vidailhet@sat.ap-hop-paris.fr).