In this study we have investigated the pathophysiology of two idiopathic focal dystonias: hand cramp with excessive cocontractions of agonist and antagonist hand or forearm muscles during specific tasks, such as writing, and facial dystonia manifested by involuntary eyelid spasms (blepharospasm) and lower facial and jaw spasms (oromandibular dystonia). We used positron emission tomography (PET) to measure thein vivo binding of the dopaminergic radioligand [18F]spiperone in putamen in 21 patients with these two focal dystonias and compared the findings with those from 13 normals. We measured regional cerebral blood flow and blood volume in each subject as well as the radiolabeled metabolites of [18F]spiperone in arterial blood. A stereotactic method of localization, independent of the appearance of the images, was used to identify the putamen in all of the PET images. We analyzed the PET and arterial blood data with a validated nonsteady-state tracer kinetic model representing the in vivo behavior of the radioligand. An index of binding called the combined forward rate constant was decreased by 29% in dystonics, as compared with normals (p < 0.05). There were no significant differences between dystonics and normals in regional blood flow, blood volume, nonspecific binding, permeability–surface area product of [18F]spiperone or the dissociation rate constant. These findings are consistent with a decrease of dopamine D2-like binding in putamen and are the first demonstration of a receptor abnormality in idiopathic dystonia. These results have important implications for the pathophysiology of dystonia as well as for function of the basal ganglia.
Dystonia is a syndrome of repetitive or sustained involuntary muscle contractions that frequently produce twisting, repetitive movements and abnormal postures (Fahn, 1988). Idiopathic dystonias are distinguished from secondary dystonias (such as those caused by birth injury, stroke, or drug reaction) by the lack of identifiable etiology (Calne and Lang, 1988) and can be classified by affected body part. Generalized idiopathic dystonia often begins in childhood, whereas focal dystonias more frequently start in adult life (Marsden and Harrison, 1974; Fahn, 1988; Greene et al., 1995). There are several types of idiopathic focal dystonia, including blepharospasm and hand cramp. Blepharospasm refers to spasms of involuntary eyelid closure that can be sufficiently severe to render a person functionally blind. Some patients also may have lower facial dystonic spasms; the combination is called cranial dystonia. Dystonic hand cramp is produced by excessive cocontractions of agonist and antagonist hand or forearm muscles during specific tasks, such as writing (writer’s cramp) or typing (typist’s cramp) (Sheehy and Marsden, 1982; Cohen and Hallett, 1988). Clinically, one might suspect that the idiopathic focal dystonias share a common pathophysiology, because there is frequent overlap of symptoms in individual patients (Jankovic et al., 1991), but the exact etiological relationship among the idiopathic focal dystonias remains unclear (Micheli et al., 1994) despite recent advances in genetics (Ozelius et al., 1989, 1992; Waddy et al., 1991; Bressman et al., 1994; Kramer et al., 1994; Greene et al., 1995).
Numerous reports have described structural abnormalities in basal ganglia contralateral to the symptomatic side in hemidystonic patients (Grimes et al., 1982; Demierre and Rondot, 1983; Pettigrew and Jankovic, 1985). Computed tomography (CT) and magnetic resonance imaging (MRI) have revealed putamenal lesions in patients with secondary dystonias (Marsden et al., 1985; Fross et al., 1987; Obeso and Gimenez-Roldan, 1988; Rutledge et al., 1988; Krauss et al., 1992;Bhatia and Marsden, 1994; Lee and Rinne, 1994), and we found an abnormality of blood flow and oxygen metabolism in the putamen contralateral to the side of the body affected by a post-traumatic paroxysmal hemidystonia despite completely normal brain MRI, CT, and angiogram (Perlmutter and Raichle, 1984). Furthermore, high field-strength MRI demonstrated prolonged T 2times in the lentiform nucleus in idiopathic torticollis (Schneider et al., 1994). Overall, these studies suggest that the putamen is a likely site of pathophysiology in dystonia.
Several lines of evidence suggest that abnormalities of dopaminergic pathways also play a key role in the underlying pathophysiology of dystonia (Garver et al., 1976; Ashizawa et al., 1980; Kolbe et al., 1981; Rupniak et al., 1986; Poewe et al., 1988; Perlmutter et al., 1993; Playford et al., 1993; Nygaard, 1995), but the specific nature of the abnormality remains unclear. We have designed this study to investigate whether there is an abnormality of D2-like dopaminergic specific binding sites in putamen in patients with idiopathic focal dystonia. We have limited our study to include only those with hand cramp and facial dystonia, because patients with these conditions can lie within the positron emission tomography (PET) scanner for the required time without sedation, substantial movement, or discomfort.
MATERIALS AND METHODS
Subjects. We studied 21 patients with dystonia (16 women, mean age = 59 ± 14 years, range 25–79), including 14 with cranial dystonia and 7 with dystonic hand cramp, as well as 12 normals (6 women, mean age = 53 ± 19, range 21–76) recruited from the Movement Disorders and Neuro-ophthamology clinics at Washington University, the Movement Disorders Clinic at the Baylor College of Medicine, and referrals from the Benign Essential Blepharospasms and the Dystonia Medical Research Foundation. The patients did not have dystonia in other parts of the body. None of the patients or normals had other neurological or psychiatric disease. All had a Mini Mental State Examination score >26 (Folstein et al., 1975) and a Hamilton Rating Scale for Depression score <6 (Hamilton, 1960). No subjects were taking medications known to affect dopamine receptor binding. Some of the patients had been treated with botulinum toxin A injections directly into affected muscles, as described in Table1. Additional details of the subjects are included in Table 1. Patients with dystonia and normals were studied concurrently. These studies were approved by the Human Studies Committee of Washington University and by the Radioactive Drug Research Committee (US Food and Drug Administration). Each subject provided written informed consent before participation.
MRI. Each subject had an MRI of the brain with the Siemens Vision 1.5 T Magnetom scanner. The following pulse sequences were used: MPRAGE (TR = 9.7 msec, TE = 4 msec, flip angle = 12, time = 6:36, pixel size = 1 × 1 × 1.25 mm), turbo spin echo (TR = 4100 msec,TE = 110 msec, flip angle = 140, time = 5:48, pixel size = 1.19 × 1 × 1.25 mm), and proton density (TR = 3675 msec, TE = 20, 90 msec, flip angle = 90, time = 8:54, pixel size = 1.3 × 0.90 × 1.3 mm). A midsagittal scoutT 1-weighted spin echo sequence was used to identify midline structures such as the inner table of the skull and anterior and posterior commissures.
PET. PET studies were done with the Siemens 953b in the two-dimensional mode with 31 simultaneous slices with 3.4 mm center-to-center slice separation (Spinks et al., 1988; Mazoyer et al., 1991). Attenuation factors were measured for each subject by using rotating rod sources of [68Ge]/[68Ga]. Reconstructed transaxial resolution of the emission images was ∼12 mm, and axial resolution was ∼4.2 mm. Each reconstructed voxel is ∼2 × 2 × 3.4 mm.
Protocol. All subjects were videotaped on the day of the PET studies. Subjects were placed in the PET scanner, a 20-gauge catheter was inserted into an arm vein for injection of radiopharmaceuticals, and a similar catheter was inserted into a radial artery, after local anesthesia with lidocaine, for sampling arterial blood. The head was positioned to include imaging from the top of the striatum to the lower portions of the cerebellum. We placed ear plugs with radio-opaque markers to confirm that the head was not rotated about the anterior–posterior or vertical axes and then stabilized the head with a polyform plastic mask molded to the subject’s head. A lateral skull radiograph taken with a reference PET slice marked by a radio-opaque wire provided a permanent record of the patient’s position (Fox et al., 1985). For each emission scan, the eyes were closed and the ears not further occluded. We measured regional cerebral blood volume (rCBV) with a 5 min scan beginning 2 min after inhalation of 50–100 mCi of C15O and regional cerebral blood flow (rCBF) with a 40 sec scan after injection of 50 mCi of H2 15O (Herscovitch et al., 1983; Raichle et al., 1983; Martin et al., 1987;Videen et al., 1987). Radioligand binding was measured with [18F]spiperone ([18F]SP), because this was the only dopaminergic radioligand available to us for human studies at the time these studies began. Five milliliters of arterial blood were taken for measurement of the free fraction (f 1; dimensionless) of [18F]SP in blood (done in duplicate or triplicate) with a centrifree technique (Perlmutter et al., 1986). After adequate time allotted for decay of 15O, as much as 5 mCi of no-carrier-added [18F]SP containing <1 μg of unlabeled ligand (specific activity >2000 Ci/mmol) was injected intravenously, and PET scans were begun immediately. Initial scans were 60 sec and increased up to 10 min to maintain adequate counts for statistical accuracy (Perlmutter et al., 1987). During these scans, ∼35 arterial blood samples were collected to measure total radioactivity, and 11 of these samples were assayed in duplicate for the fraction of [18F] activity that represented unmetabolized [18F]SP (Perlmutter et al., 1986). Subjects were observed continuously throughout the PET procedures. Some of the patients had minimal blepharospasm, but there were no other movements seen.
Data analysis. All volumes of interest (VOIs) were identified by an observer blinded to subject diagnosis. For each subject, we started with the same coordinates for the center of putamen identified on a stereotactic atlas of the brain (Talairach and Tournoux, 1988), then transferred these coordinates to the appropriate single PET slice by a stereotactic method of localization (Fox et al., 1985), and finally expanded the VOI to include putamenal activity on the slices immediately above and below. The same-sized region was outlined on all slices (9 × 5 voxels), and then the regional values were averaged across the three slices and for the right and left putamen. A single hemispheric cerebellar value was averaged from left- and right-sided regions (5 × 5 voxels each) identified on three PET slices. The VOIs were held in a constant position for all frames of the dynamic collection made after injection of [18F]SP and for the CBV and CBF images. The [18F] tissue activity curves were decay-corrected to the time of injection of [18F]SP. We calculated radioligand binding with a tracer kinetic model previously described (Perlmutter et al., 1986), validated (Perlmutter et al., 1989, 1991) and applied to human studies (Perlmutter et al., 1987). Briefly, PET and arterial blood data were analyzed with a nonsteady-state two-compartment model to estimate the free fraction of radioligand in the cerebellum (f 2; dimensionless) as a measure of the nonspecific binding. This value was assumed to be the same in the putamen, and then a three-compartment, three-parameter nonsteady model was used to estimate the local permeability–surface area product (PS) for [18F]SP at the blood–brain barrier, the combined forward rate constant (CFRC) of [18F]SP (this equals the apparent maximum number of specific binding sites times the association rate constant of [18F]SP for the specific sites) as well as the dissociation rate constant of [18F]SP–receptor complex. The assumptions and limitations of this approach have been described in detail, including the test–retest variability of calculations of the relevant binding variables (Mintun et al., 1984;Perlmutter et al., 1986, 1987, 1989, 1991).
Statistical analyses. Results were compared between dystonics and normals with unpaired t tests. Because there was only a comparison of mean values from a single VOI value, there was no correction for multiple comparisons.
No subject had a gross abnormality on MRI scan of the brain. Cerebellar blood flow and blood volume, putamenal blood flow and blood volume, and the measured free fraction of [18F]SP in blood are listed in Table 2. There were no statistical differences between patients and normals. The typical time course of total radioactivity measured in the arterial blood after injection of [18F]SP is shown in Figure 1. Most of the area under the curve is within the first few minutes, and it is necessary to sample this part of the curve adequately to permit accurate parameter estimation (Perlmutter et al., 1986). Nearly one-half of the radioactivity in arterial blood by 30 min after [18F]SP injection represents radiolabeled metabolites of [18F]SP rather than [18F]SP itself, as shown in the inset of Figure 1. An example of the time-dependent measurements of regional radioactivity within the putamen (averaged left and right side) is demonstrated in Figure 2. The parameter estimation method finds the optimal estimates of the unknown variables to make the tracer kinetic model equations fit the observed tissue activity points. The closeness of fit of the model to the data also is shown in Figure 2.
Variables estimated from the tracer kinetic model, including the calculated free fraction of [18F]SP in brain tissue, the permeability–surface area product (PS) for [18F]SP in cerebellum, PS for [18F]SP in striatum, and dissociation rate constant of [18F]SP, are in Table 3. There were no statistically significant differences between dystonics and normals except for the combined forward rate constant, which was ∼29% lower in dystonics compared with normals (p < 0.05), as shown in Table 3. We found no significant difference between the combined forward rate constant for dystonic hand cramp and cranial dystonia (p > 0.2). It is important to note that a change in the combined forward rate constant is consistent with a change of the association rate constant, the maximum number of specific binding sites (B max), or both. The parameter estimation of the dissociation rate constant has substantially more noise than for the combined forward rate constant. This is indicated in Table 3 with the greater variance of the estimates of the dissociation rate constant compared with the variance of the estimates of the combined forward rate constant (i.e., higher mean coefficient of variation for the variable estimates). In part, this is attributable to the greater number of association rate events compared with dissociation events that occur during a 3 hr PET study. For this reason, we have chosen to report the combined forward rate constant as the index of binding rather than the binding potential (this equals combined forward rate constant/dissociation rate constant), which would incorporate the additional uncertainty of the dissociation rate constant estimate (Mintun et al., 1984; Perlmutter et al., 1986). The uncertainty or greater variance of the estimates for the dissociation rate constant limits the detection of statistical differences between groups.
Limitations imposed by potential radiation exposure of subjects and reduced brain penetration of [18F]SP, as compared with other more promising radioligands (Moerlein et al., 1995), reduce the signal-to-noise ratio of PET-based measurements, which tends to increase the variance of the estimated unknown binding variables. To reduce the variance of the estimates, we combined data from the left and right putamen, thus doubling the regional counts to improve signal-to-noise by ∼40%. This compromise obfuscates any potential side-to-side differences in a seemingly unilateral condition, such as hand cramp. However, numerous other physiological measurements in hand cramp patients have shown bilateral abnormalities (Panizza et al., 1989, 1990; Tempel and Perlmutter, 1993; Chen et al., 1995; Van Der Kamp et al., 1995), and many patients progress to bilateral hand cramp. Of course, there is no compelling reason to suspect that the patients with bilateral facial dystonia have a unilateral brain abnormality. Thus, we believe that this is a reasonable compromise, given the nature of the data.
Our findings are not likely to be affected by previous treatment of dystonia. Most hand cramp patients had not been treated before the PET, although most blepharospasm patients had previous local injections of botulinum toxin A but were not exposed to oral drugs (Table 1). The direct effects of the toxin probably are limited to the periphery, with blockade of presynaptic release of acetylcholine at the neuromuscular junction (Hamian and Walker, 1994) and minimal penetration of the blood–brain barrier (Black and Dolly, 1987). Although the numbers are small, we did not find a statistical difference in the combined forward rate constant between those patients treated with botulinum versus those not previously treated (p > 0.2). The effects of oral medications are not likely to have influenced these findings either. We compared the combined forward rate constant in dystonic patients treated only with botulinum or aspirin (n = 9) with normals meeting the same criteria (n = 8), and the CFRC was still lower for the dystonics (0.209 ± 0.058), as compared with the normals (0.270 ± 0.142).
We found decreased [18F]SP binding in putamen in patients with facial or hand dystonia, the first demonstration of a receptor abnormality in idiopathic dystonia. This has important implications for the pathophysiology of dystonia as well as for the function of the basal ganglia.
Our findings must be interpreted cautiously, because [18F]SP-specific binding is relatively nonselective. [18F]SP-specific binding in the nonhuman primate putamen comprises ∼74% to D2-like and 26% to serotonergic S2 sites (Perlmutter et al., 1991). Furthermore, because [18F]SP binds to D2-like receptors, [18F]SP binding could reflect a change in D2- or D3-specific sites. D4 binding sites are less likely to be relevant, because they are much less numerous in primate putamen (Seeman et al., 1993).
Others have found reduced dopaminergic activity in dystonia consistent with the interpretation that reduced [18F]SP binding reflects a change in D2-like binding. For example, Playford et al. (1993) found a 15% mean reduction of [18F]dopa uptake in putamen (and not in caudate) in familial idiopathic dystonia. Others found decreased CSF homovanillic acid (HVA), a major metabolite of CNS dopamine, in one of two patients with cranial dystonia, suggestive of decreased dopamine turnover (Ashizawa et al., 1980), reduced dopamine in GPe (Jankovic et al., 1987) in one patient with cranial dystonia, and reduced striatal dopamine in one of two patients with generalized dystonia (Hornykiewicz et al., 1986). In dopa-responsive dystonia (DRD) there is a remarkable symptomatic response to levodopa. The more common autosomal dominant DRD is associated with a deficiency of an enzyme required for biosynthesis of a cofactor for tyrosine hydroxylase, the rate-limiting enzyme for dopamine production (Nygaard, 1995). The less common autosomal recessive form of DRD is caused by a defect in tyrosine hydroxylase (Knappskog et al., 1995). Furthermore, acute blockade of D2-like receptors with neuroleptics may produce acute dystonic reactions (Garver et al., 1976; Kolbe et al., 1981; Rupniak et al., 1986). Some parkinsonian patients develop dystonia as an early symptom, suggesting that dystonia may result from deficient dopaminergic transmission, because striatal dopamine deficiency produces parkinsonism (Wooten and Trugman, 1989). Similarly, some baboons develop a transient dystonic phase after intracarotidN-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) injection has reduced striatal dopamine levels by >90% (Perlmutter et al., 1993). Because most people with Parkinson’s disease (PD) do not have dystonia, there must be a difference between the nature of the striatal dopamine deficiency in parkinsonism and dystonia. Parkinsonism is associated with severe striatal dopamine deficiency and presumed secondary dysfunction at both D1-like and D2-like dopamine receptors, whereas we suggest a selective dysfunction of D2-like mediated function in dystonia.
If decreased dopaminergic transmission in putamen produces dystonia, how does this fit with current models of basal ganglia function? One model describes multiple cortical–striato–pallido–thalamic–cortical loops with the cortical striate projection fibers of the motor loop predominantly targeting putamen (Alexander et al., 1986; Alexander and Crutcher, 1990; Gerfen et al., 1990; Gerfen, 1992). From there, two major pathways lead to the internal segment of the pallidum (GPi): (1) the direct pathway via inhibitory GABAergic fibers connecting striatum and GPi and (2) the indirect pathway, including inhibitory GABAergic neurons from striatum to the external segment of pallidum (GPe), inhibitory neurons projecting from GPe to subthalamic nucleus (STN), and excitatory neurons projecting from STN to GPi. Both the direct and indirect pathways converge on GPi, which then sends inhibitory GABAergic neurons to ventral anterior thalamus that projects via excitatory neurons to cortical areas, including premotor and motor regions. D2-like receptors predominantly localize to and inhibit the striatopallidal neurons of the indirect pathway projecting to GPe, whereas D1 receptors localize to and facilitate the neurons of the direct pathway that project from striatum to GPi (Gerfen et al., 1990; Gerfen, 1992; Keefe and Gerfen, 1995). We suggest that dystonia occurs after preferential decrease in D2-mediated inhibition of the indirect pathway. This is consistent with decreased [18F]SP binding in putamen and with our previous findings of a 25% reduction in the vibration-induced blood flow responses in primary sensorimotor (PSA) and supplementary motor areas (Tempel and Perlmutter, 1990, 1993). This decreased blood flow response could reflect a primary alteration in cortical activity or alternatively could indicate a change in the cortical–striato–pallidal–thalamic–cortical circuit. A decrease in D2-like inhibitory function would increase the activity of the inhibitory putamen to GPe neurons. This, in turn, would decrease activity of the inhibitory neurons projecting from GPe to STN, leading to increased activity of excitatory neurons projecting from STN to GPi, increased activity of inhibitory neurons projecting from GPi to thalamus, and decreased activity of excitatory neurons projecting from thalamus to cortex. One then might expect that cortical areas would have less activity when interacting with this motor loop through the basal ganglia. Interestingly, we found that a reduced vibration-induced blood flow response in PSA in one patient with a dopa-responsive dystonia (Tempel and Perlmutter, 1990) normalized after levodopa (Perlmutter and Raichle, 1988). We speculate that levodopa modified the PSA response by acting in the putamen, but other sites of action cannot be excluded.
Others have found functional changes in basal ganglia in either animal models or patients with dystonia. Mitchell et al. (1990) measured 2-deoxyglucose uptake in a monkey made hemiparkinsonian after intracarotid MPTP that had developed peak dose dystonia after chronic treatment with the nonselective dopamine agonist apomorphine. They found increased uptake in caudate, putamen, GPe, GPi, and STN but decreased uptake in regions that receive basal ganglia output, such as ventroanterior/ventrolateral thalamus and lateral habenula. These data are consistent with altered activity of dopaminergic pathways in the basal ganglia, but interpretation of the findings is confounded by the activity of the animal during the uptake phase after 2-deoxyglucose injection. Thus, it is unclear whether the findings reflect the underlying pathophysiology of dystonia, the brain responses to altered motor behavior, or both. Stoessl et al. (1986) found in torticollis patients a loss of significant correlations between thalamic metabolism and metabolism in caudate and lentiform nucleus, suggesting a disruption of the pallidothalamic projections. Karbe et al. (1992), also using PET, found an abnormal pattern of regional metabolism, suggesting altered relationships among basal ganglia, thalamus, and frontal association areas in a heterogeneous group of 15 patients with idiopathic dystonia. Although these studies support abnormalities in basal ganglia circuits, they do not identify the specific nature of such dysfunction. Leenders et al. (1993) did not find a significant abnormality in uptake of [11C]N-methyl-spiperone in six patients mostly with torticollis, although the small number severely hampered identification of a significant result.
One also may view the function of the indirect pathway as broadly inhibiting unwanted movement during an intentional movement and the direct pathway as focally permitting the selected movement (Mink and Thach, 1993). Decreased activity in the indirect pathway could be consistent with normal voluntary initiation of movement but loss of ability to inhibit unwanted involuntary movements in other parts of the body. Clinically, this is typical of dystonia, particularly when it first begins. At that time, involuntary postures and muscle spasms may occur only during a specific motor activity and not at rest. The involuntary spasms spread as the movement persists with the loss of “surround inhibition” (Mink and Thach, 1993). This commonly is seen in writer’s cramp, because the muscle spasms tend to spread from hand to wrist to arm as writing persists (Sheehy and Marsden, 1982). Under such circumstances, we propose that the “braking action” of the basal ganglia, important for inhibition of excessive movements, has gone awry.
Our findings do not explain all types of dystonia. People with PD may develop dystonia not only as an early manifestation of the disease but also in at least three different patterns associated with dopa replacement therapy. Dystonia may occur commonly in the lower extremities when the plasma dopa levels are low but also may occur when plasma levels peak. Finally, dystonia may occur as the effect of an individual dose begins or as it diminishes (Poewe et al., 1988). Proposing pathophysiological mechanisms for these three patterns of dystonia is difficult, given the uncertainty of the relative influence of the different dopamine receptor subtypes on subsequent activities of the direct and indirect pathways.
In summary, we have found a decrease in [18F]SP binding in the putamen of patients with idiopathic adult-onset focal dystonias affecting the face or hand. There was no significant difference between patients with hand and facial dystonia, suggesting that there may be a common mechanism producing both conditions, consistent with the clinical impression that they share a similar pathophysiology (Jankovic et al., 1991). However, because we used large VOIs, our data do not exclude the possibility that different areas of putamen may have different degrees of decreased binding in the two types of dystonia. We propose that the pathophysiology of these dystonias reflects decreased activity in the D2-like mediated function of the indirect pathway of the motor circuit in the basal ganglia. Additional studies with more specific radioligands should help to clarify the nature of this abnormality further. It also would be interesting to determine the relative activity of the putamenal neurons projecting directly to GPi versus those projecting to GPe in animal models of dystonia.
This work was supported by National Institutes of Health Grants NS31001, NS32318, and AA-07466 as well as by the Benign Essential Blepharospasm Foundation, the Greater St. Louis Chapter of the American Parkinson’s Disease Association, the Clinical Hypotheses Research Section of the Charles A. Dana Foundation, the McDonnell Center for the Study of Higher Brain Function, the generosity of Mr. and Mrs. Jefferson Miller, and the Barbara and Sam Murphy Fund. We appreciate useful discussions with Dr. Jonathan Mink and the expert technical assistance of the members of the Division of Radiological Sciences. We also thank Dr. William Hart for referral of some patients who participated in this study.
Correspondence should be addressed to Dr. Joel S. Perlmutter, Division of Radiological Sciences, Washington University School of Medicine, Campus Box 8225, 4525 Scott Avenue, St. Louis, MO 63110.