Areas with statistically significant enhanced activation in rostral supplementary motor area (rSMA) and premotor cortex (PMC) (ipsilateral to stimulation and contralateral to movement) in the patient group during subthalamic nucleus stimulation, shown as statistical parametric mapping projections with a cutoff of P<.001, while patients performed random joystick movements at 0.33 Hz. Adjusted regional cerebral blood flow (rCBF) of the peak enhanced activation in PMC (coordinates −28, 6, 54) ipsilateral to the stimulating electrode and rSMA (coordinates 8, 20, 56) at rest (R) and during the activation condition (A) with (on) and without (off) STN stimulation is depicted.
Decreases in resting activity with subthalamic nucleus stimulation on compared with off, shown as statistical parametric mapping projections with a cutoff of P<.001. Adjusted regional cerebral blood flow (rCBF) of the peak rCBF decrease in primary motor cortex ipsilateral to the stimulating electrode (coordinates −36, −28, 64) with (on) and without (off) stimulation is depicted.
Ceballos-Baumann AO, Boecker H, Bartenstein P, von Falkenhayn I, Riescher H, Conrad B, Moringlane JR, Alesch F. A Positron Emission Tomographic Study of Subthalamic Nucleus Stimulation in Parkinson DiseaseEnhanced Movement-Related Activity of Motor-Association Cortex and Decreased Motor Cortex Resting Activity. Arch Neurol. 1999;56(8):997-1003. doi:10.1001/archneur.56.8.997
Long-term high-frequency stimulation of the subthalamic nucleus (STN) improves akinesia in Parkinson disease. The neural correlates of STN stimulation are not well understood. Positron emission tomography can be applied to the in vivo study of the mechanisms of deep brain stimulation.
To study changes in regional cerebral blood flow as an index of synaptic activity in patients with Parkinson disease with effective STN stimulation on and off during rest and movement.
Eight patients with Parkinson disease who had electrodes implanted in the STN underwent 12 measurements of regional cerebral blood flow with water O 15 positron emission tomography at rest and during performance of paced freely selected joystick movements, both with and without STN stimulation (3 scans per experimental condition). Motor performance and reaction and movement times were monitored. Statistical parametric mapping was used to compare changes in regional cerebral blood flow between conditions and differences in activation.
All patients showed improvement in reaction and movement times during scans with the stimulator on. As predicted, increases in activation of rostral supplementary motor area and premotor cortex ipsilateral to stimulation were observed when stimulation was on during contralateral movement (P<.001). Unpredicted observations included decreases in regional cerebral blood flow in primary motor cortex at rest induced by STN stimulation.
Stimulation of the STN reduces the movement-related impairment of frontal motor association areas and the inappropriate motor cortex resting activity in Parkinson disease.
STEREOTAXIC procedures in Parkinson disease (PD) are currently undergoing a renaissance. Medial pallidotomy has the ability to relieve levodopa-associated dyskinesias as well as akinesia, tremor, and rigidity.1,2 Long-term high-frequency stimulation of the globus pallidus internus (GPi) has recently been suggested as a nondestructive and reversible alternative to pallidotomy.3,4
Overactivity of excitatory neurons from subthalamic nucleus (STN) to the GPi has been shown in monkeys rendered parkinsonian by 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP).5 Lesions and high-frequency stimulation of the STN in this animal model reduce akinesia, rigidity, and tremor.6 Surgery aiming at the STN was then suggested as a treatment for PD in humans, as hyperactivity of STN projections to the GPi seems to represent a crucial feature of parkinsonism in animal models of PD.7 Eventually, STN stimulation was shown to be safe and effective for the treatment of akinesia and rigidity.8
The combination of deep brain stimulation and positron emission tomography (PET) activation studies offers a powerful technique for assessing the effects of discrete perturbations at different target structures throughout the basal ganglia–thalamocortical circuitries on regional cerebral blood flow (rCBF), an index of local synaptic activity.9 Patients can undergo repetitive scans with stimulation off and on in a single PET study, and the task-specific effects of stimulation on segregated neuronal systems can be studied throughout the brain.
We hypothesized that increases in activation of motor association cortex coupled with improvement of akinesia would occur in patients with PD during STN stimulation. Like surgery targeting the GPi, STN stimulation aims to reduce the overactivity of the GPi by decreasing overactive excitatory STN projections that otherwise reinforce the inhibitory GPi overactivity directed toward the thalamus. Therefore, reduction of STN overactivity would be expected to reduce inhibitory GPi output to thalamic relay nuclei and, in turn, disinhibit the ventral thalamus and facilitate thalamic excitation of premotor and prefrontal cortical areas.
The motor paradigm used in the present study has been previously used to demonstrate impaired activation of the striatum and mesial frontal cortex in PD10,11 and increased activation of prefrontal and premotor regions in acquired and idiopathic dystonia.12,13 Herein, we report the changes in rCBF occurring in patients with PD associated with STN stimulation during resting and movement-related brain activity.
Nine patients with medically intractable PD, all levodopa responsive (7 men and 2 women; mean age, 53.8±8.2 years; range, 42-72 years), were selected for this study. The patients had suffered from parkinsonian symptoms for 11±6 years (range, 4-20 years). All had initially experienced onset of symptoms on one side, and this side was more affected than the other when stimulators were off. Assessment on the Unified Parkinson's Disease Rating Scale was blinded and performed at the PET center independently from the neurosurgical teams at least 4 months after implantation of the electrodes. Patients were assessed with the stimulator on and off by means of the motor Unified Parkinson's Disease Rating Scale at different days in randomized order in the "practically defined off period" (after a drug withdrawal of more than 12 hours). Three patients (patients 5, 6, and 9) stopped taking dopaminergic medication after implantation of the STN electrodes. Two patients (patients 4 and 7) could be weaned from their apomorphine hydrochloride pumps and be treated with moderate doses of dopaminergics after effective stimulation. The clinical details of the patients, their Unified Parkinson's Disease Rating Scale scores, and medication are detailed in Table 1 and their stimulation variables during the PET study in Table 2.
The study was approved by the ethics committee. All subjects gave written informed consent before the PET study. Permission to administer radioactive substances was obtained from the radiation protection authorities.
Neurosurgery was performed in Homburg, Germany (2 patients), and Vienna, Austria (7 patients). All subjects gave written informed consent before implantation of the stimulating systems. The electrode implantation took place under stereotaxic conditions, with the patient under local anesthesia, with the use of computed tomography and/or magnetic resonance imaging as well as ventriculography. The primary anatomical target was the center of the STN, based on the Talairach diagram at a laterality of 10 to 12 mm. A computer program using both ventriculographic and magnetic resonance imaging data was used to calculate this target.14 The neurophysiological exploration of the target was done in a standard way with specially designed semimicroelectrodes (PLS; Inomed, Teningen, Germany) at intervals of 2 mm. Intraoperative test stimulation was done at a rate of 130 Hz and a pulse rate of 50 microseconds, with a variable current flow (0-5 mA). The effect of stimulation was checked by finger tapping and pronation-supination for bradykinesia, and active and passive flexion-extension of the hand for rigidity. After final identification of the target, the testing electrode was removed and replaced by a permanent quadripolar (Medtronic 338X; Medtronic Inc, Minneapolis, Minn). The electrodes were first externalized via an extension cable. This step was followed by a postoperative screening phase of several days for test stimulation. After successful test stimulation, a pulse generator was implanted. The externalized extension cable was removed and the electrode was connected to the pulse generator, which was located in a surgically prepared pocket in the pectoral region.
All patients with PD underwent 12 sequential rCBF scans with water oxygen 15 (H215O) PET. The light was dimmed. The stimulator contralateral to the side of initial akinetic symptoms in the course of the disease was switched on and off. The 4 experimental conditions were arranged as 6 pairs (AB and CD) of scans, randomized to ABCDABCDABCD or CDABCDABCDAB. The conditions were as follows: (A) rest with STN stimulators off: the stimulators were switched off 10 minutes before the scan and left off for scan B; (B) joystick movements in freely selected directions paced by a tone at 3.0- to 3.5-second intervals with STN stimulators off; (C) rest with STN stimulator on: the stimulator was switched on 10 minutes before the scan and left on for scan D; and (D) joystick movements with STN stimulator on, otherwise same as B.
Patients were instructed to make 1 joystick movement for each pacing tone, to choose a different direction (forward, back, left, or right) of movement on each occasion, and to avoid repetitive patterns of movements. Patients practiced the movement task beforehand to ensure correct performance with eyes closed. We computed mean reaction time and movement time during the scans with the stimulator on and off and compared the data by means of a paired 2-tailed t test. During the rest condition, subjects were asked to relax with their arms and hands in the most comfortable position possible with eyes closed. They were told beforehand that they would be hearing tones as during the activation scans. All subjects were closely observed and videotaped during the scans to detect head movement or dyskinesias.
The PET measurements were performed with a PET scanner (Siemens 951R/31; Siemens CTI, Knoxville, Tenn) in 3-dimensional mode with a total axial field of view of 10.5 cm and no interplane dead space. To measure rCBF, 277.5 MBq of H215O was administered intravenously over 30 seconds with a semibolus injection by means of an infusion pump. Single frames were acquired for 60 seconds starting with the appearance of the tracer in the brain. The pacing tones commenced 10 seconds before actual scanning both for the resting and the activated state. The interval between successive H215O administrations was 10 minutes. A 20-minute transmission scan that used rotating rods of germanium 68–gallium 68 was performed for attenuation correction with the septa in place. Two-dimensional blank and transmission scans (septa extended) were used to reconstruct a 3-dimensional attenuation map. Oblique lines of coincidence for which the attenuation correction factor had not been measured were obtained by forward projection through the 3-dimensional map.15 After corrections for randoms, dead time, and scatter, images were reconstructed by filtered back-projection with a Hanning filter (cutoff frequency, 0.4 cycles per projection element), resulting in 31 slices with a 128×128-pixel matrix (pixel size, 2.0 mm) and interplane separation of 3.375 mm.
Image processing was carried out with computers (Sun SPARC 2; Sun Computers Europe Inc, Surrey, England) with the use of PRO MATLAB (MathWorks Inc, Natick, Mass). Realignment, normalization into Talairach stereotaxic space,16 intersubject averaging, and statistical analysis was performed with established software17- 20 (SPM96b; Wellcome Department of Cognitive Neurology, London, England). Each image was smoothed with an isotropic gaussian kernel of 12 mm to increase signal-to-noise ratio. Global blood flow was normalized by scaling across the entire data set to a grand mean of 50 mL/100 mL per minute.
We tested for relative increases and decreases in rCBF by categorical comparisons of on vs off STN stimulation at rest. Relative differences in activation were assessed by comparing rCBF changes (ie, movement vs rest) during STN stimulation on with the movement-associated rCBF increases during STN stimulation off.
All comparisons were specified by appropriately weighted categorical contrasts and performed on a voxel-by-voxel basis by means of analysis of variance. This generated statistical parametric mapping (t) maps for the rCBF changes associated with each comparison. For the comparison of the activation effects, the statistical parametric mapping (t) maps were subsequently transformed into statistical parametric mapping (z) maps, and the level of significance of areas of activation was assessed by the peak height of their foci by means of estimations based on the theory of random gaussian fields.20 Significance was accepted if voxels survived an uncorrected threshold of P<.001.
Motor performance improved with STN stimulation (Table 1). Individual task performance measures with the contralateral STN stimulation switched on during PET scanning are shown in Table 3. The error rate in joystick movements was not significantly different between STN stimulation on and off conditions. The mean reaction and movement times were, on average, 24.6% and 28.9% longer when the contralateral STN stimulators were off (reaction time, P<.005, and movement time, P<.001, paired t test).
Relative increases of movement-associated activation ipsilateral to the STN electrode and contralateral to movement were observed in mesial frontal cortex (2 adjacent clusters totaling 147 voxels) and premotor cortex (106 voxels) (Table 4; Figure 1). The peak of enhanced activation in the mesial frontal cortex was in Brodmann area 6 bordering to Brodmann area 8 in the Talairach atlas (rostral supplementary motor area [SMA]). When the threshold was lowered to P<.01 (not included in Table 4 and Figure 1), the 2 adjacent clusters merged and extended posteriorly into the area just anterior to the vertical line perpendicular to the anterior commissure line. At this lower threshold of of P<.01, relative increases in contralateral dorsal prefrontal cortex activation were observed with the stimulator on (peak at coordinates 50, 16, and 40; cluster size, 201; z score, 3.3).
We also found decreased activation in caudal SMA and in a minor focus corresponding to lower primary motor cortex.
Stimulation of the STN led to ipsilateral increases at rest in ventral thalamus and globus pallidus internus according to the coordinates in the Talairach space (Table 5). The region formed a contiguous cluster containing 133 voxels with 2 foci of activation. The foci corresponded to ventrolateral thalamus and to globus pallidus. We also found small foci of rCBF increases in ipsilateral parietal and parieto-occipital cortex as well as in contralateral dorsolateral prefrontal cortex.
At rest we found an area including 620 voxels of STN stimulator–associated rCBF decrease in ipsilateral primary motor cortex (Figure 2). There was also a small focus of decrease in anterior cingulate cortex in 23 voxels.
There were 2 main findings associated with STN stimulation in PD: first, as predicted, rostral premotor areas (lateral area 6 and rostral SMA) and, at a lower level of significance, dorsal prefrontal cortex, showed enhanced movement-associated activation coupled with improvement in akinesia during STN stimulation. Second, not predicted beforehand, we observed that STN stimulation induced ipsilateral resting rCBF decreases in primary motor cortex.
The finding of STN-induced enhanced activation of motor association areas (rostral SMA, premotor cortex) during volitional movements, along with the improvement in akinesia, is consistent with our hypothesis that STN stimulation improves activity of motor association cortex, possibly by reducing inappropriate excitation of the STN on inhibitory pallidothalamic projections. Positron emission tomographic studies have shown that SMA, in particular its rostral part (pre-SMA), subserves the central control of internally generated sequential movements.21 The SMA is functionally underactive in PD, but its function can be restored after administering dopaminergic agents,10,22,23 transplantation of embryonic tissue,24 and pallidotomy.25- 27 The present study shows that restoration of SMA activity (mainly rostral) is also a feature of STN stimulation and adds further support to the pivotal role of the STN nucleus in the motor deficit of PD.
Our results are in reasonable agreement with the pattern of movement-associated rCBF changes caused by STN stimulation described by Limousin et al.28 In both studies, movement-associated increased activation was centered around the SMA, although the peak changes in mesial frontal cortex were more rostral (pre-SMA) in our study. We also found, in contrast to Limousin et al, enhanced activation of the lateral premotor cortex ipsilateral to stimulation. It is likely that differences in patient selection, stimulation variables, and slight variation in targets within the STN may account for the differences in the frontal pattern of movement-associated rCBF changes.
A fludeoxyglucose F 18 PET study of covariance patterns of resting glucose metabolism in patients with PD demonstrated increased glucose metabolism after pallidotomy in ipsilateral dorsal prefrontal cortex, primary motor cortex, and lateral premotor cortex along with decreases in thalamic and lentiform metabolism.29 No relative increases in resting SMA metabolism were detected in that study. However, in an rCBF PET study on the effects of GPi stimulation, Davis et al4 showed increased rCBF in rostral SMA compared with baseline without stimulation. These authors required their patients with GPi stimulators to count silently during all scans, and therefore a direct comparison of these premotor rCBF increases with our rest condition may not be appropriate. However, our study design allowed us to examine task-specific changes in the activated motor system, and during STN stimulation we found movement-related enhanced activation in premotor cortex with no rCBF increases at baseline (rest). This could mean that STN, in contrast to GPi, stimulation is quite specific in movement-related disinhibition of thalamocortical circuits.
We also found decreased movement-associated activity in caudal SMA and primary motor cortex along with the increased activation in premotor and prefrontal cortex. Interestingly, this pattern of overactivity of frontal association areas and underactivity in motor executive cortex (mainly caudal SMA) has already been described in idiopathic dystonia.12 Parallels between idiopathic dystonia and STN stimulation in PD also occur clinically: STN stimulation may induce dystonia,30 and bradykinesia is common in idiopathic dystonia.
We found rCBF decreases in primary motor cortex associated with STN stimulation during the resting condition, while akinesia improved. This pattern of resting rCBF changes caused by STN stimulation was similar to that described by Limousin et al.28 In contrast to their study, we observed statistically significant enhanced resting activity during stimulation in an area corresponding to the globus pallidus–ventral thalamus in the stereotaxic atlas. The predominance of unipolar stimulation in our study may represent the simple explanation of these subcortical increases in resting flow because of spread of current in adjacent pathways to the STN. However, other theories could be advanced here. Orthodromic activation, namely via the fasciculus thalamicus or the ansa lenticularis, or through a direct projection from STN to thalamus,31 could enhance synaptic activity in the thalamus. The increases in activity in the pallidal area may be caused by increased inhibitory activity on the internal pallidum through anterodromic stimulation or backfiring toward the external pallidal segment from STN, a mechanism of action of STN stimulation that was already suggested by Limousin et al.8
It has been argued that the strong direct excitatory projection from the cerebral cortex to STN could contribute to the increased neuronal activity in the STN after dopamine loss.32,33 This implies that rCBF decreases in primary motor cortex at rest could represent the cortical consequence of the functional inactivation of the STN on the system subserving the motor cortex efferents to STN. Alternatively, STN stimulation leads to less afferent noise from premotor areas because of improvement of signaling in thalamofrontal circuitry.
It could also be argued that the decrease in motor cortex activity at rest simply represents a decrease in proprioceptive input caused by stimulation-induced decrease in rigidity and off-dystonia. However, off-dystonia was minor in our patients and did not develop acutely after the stimulator was switched off. The same applies to rigidity. Moreover, the variability across individuals would argue against a relation between changes in these signs and the robust rCBF changes observed in the motor cortex ipsilateral to the stimulation.
Our study showed that improvement of akinesia with STN stimulation is coupled with movement-associated increased activation in lateral premotor cortex and rostral SMA. The decrease in motor cortex activity with STN stimulation at rest may represent relative normalization of defective cortical excitability in PD.
Accepted for publication September 16, 1998.
This study was supported by Sonderforschungsbereich SFB 462 "Sensomotorik" from the Deutsche Forschungsgemeinschaft, Bonn, and the Deutsche Parkinson Vereinigung e. V., Neuss, Germany.
We thank our radiochemistry group and cyclotron staff for their reliable supply of radiopharmaceuticals; Frank Muntz and Evi Dickmann for methodological assistance; and Sylvia Fürst, Claudia Kolligs, and Colletta Kruschke for their technical assistance.
Reprints: Andrés O. Ceballos-Baumann, MD, Department of Neurology, Technische Universität München, Möhlstr 28, D-81675 Munich, Germany (e-mail: firstname.lastname@example.org).