- Email: [email protected]
Functional brain networks in Parkinson's disease
Parkinsonism & Related Disorders Parkinsonism and Related Disorders 8 (2001) 91±94
www.elsevier.com/locate/parkreldis
Functional brain networks in Parkinson's disease Masafumi Fukuda, Christine Edwards, David Eidelberg* Department of Neurology and Neurosurgery, New York University School of Medicine, Functional Brain Imaging Laboratory, North Shore-LIJ Research Institute, Manhasset, NY, USA
Abstract With the advent of new methods of network analysis, we have utilized metabolic data acquired through positron emission tomography (PET) to identify disease-related patterns of functional pathology in the movement disorders. In Parkinson's disease (PD), we have used [ 18F]-¯uorodeoxyglucose (FDG)/PET to identify a disease-related regional metabolic covariance pattern characterized by lentiform and thalamic hypermetabolism associated with regional metabolic decrements in the lateral premotor cortex, the supplementary motor area, the dorsolateral prefrontal cortex, and the parieto-occipital association regions. The expression of this network is modulated in a predictable fashion by levodopa therapy and by stereotaxic interventions for PD. We have extended this network analytical approach from studies of glucose metabolism in the resting state to dynamic studies of brain activation during motor performance. These PET studies utilized [ 15O]±water (H2 15O) to measure cerebral blood ¯ow activation responses during the execution of simple and complex motor tasks. In addition to the modulation of abnormal resting metabolic networks, effective PD therapy can enhance brain activation responses during motor execution, with speci®c regional associations with improvements in timing and spatial accuracy. This approach is also useful in identifying speci®c brain networks mediating the learning of sequential information. We have found that the normal relationship between brain networks and learning performance are altered in the earliest stages of PD with a functional shift from striatal to cortical processing. Brain activation PET studies during therapeutic interventions for PD demonstrate how normal brain-behavior relationships can be restored with successful therapy. Thus, functional brain imaging with network analysis can provide insights into the mechanistic basis of basal ganglia disorders and their treatment. q 2001 Elsevier Science Ltd. All rights reserved. Keywords: Parkinson's disease; PET; Brain networks; Glucose metabolism; Cerebral blood ¯ow
1. Introduction Functional brain imaging techniques have revolutionized the conceptual framework for exploring the mechanistic substrates of the movement disorders. We have developed specialized network analytical methods, in conjunction with positron emission tomography (PET), to examine hypotheses concerning alterations in brain-behavior relationships in Parkinson's disease (PD) and related disorders. Although the neurochemical lesion of PD is mainly localized to the nigrostriatal dopamine system, the effects of dopamine loss are widespread and affect several discrete parallel corticostriatopallidal-thalamocortical (CSPTC) circuits as crucial elements of motor control [1]. We have employed advanced PET techniques with network analysis to determine: (1) whether functional brain connectivity in PD patients and normal subjects conform to the anatomy of the postulated loops and (2) how the functional activity of these circuits * Corresponding author. Tel.: 11-516-562-2498; fax: 11-516-562-1008. E-mail address: [email protected] (D. Eidelberg).
relates to motor performance in PD patients studied at baseline and during effective therapeutic interventions. In this review, we will focus on the validation of the PDrelated metabolic networks and their modulation during dopaminergic pharmacotherapy and surgical interventions for parkinsonism. 2. Metabolic network mapping in PD We have developed and applied a statistical modeling approach for the identi®cation and quanti®cation of patterns of regional metabolic covariation in PD patients scanned with [ 18F]-¯uorodeoxyglucose (FDG)/PET [2]. This approach, known as the Scaled Subpro®le Model (SSM; [3,4]), is a form of principal component analysis (PCA) and can be used to identify patterns of regional covariation in functional imaging data. In SSM modeling, PCA is employed to identify regional covariance patterns from regional metabolic datasets obtained from combined samples of patients and normal controls. This form of analysis is blind to subject class designation, and utilizes the variance across the entire population to
1353-8020/01/$ - see front matter q 2001 Elsevier Science Ltd. All rights reserved. PII: S 1353-802 0(01)00022-0
92
M. Fukuda et al. / Parkinsonism and Related Disorders 8 (2001) 91±94
identify speci®c patterns associated with the disease state. SSM/PCA network analysis produces subject scores that quantify the degree to which each subject expresses each pattern. Subject scores for the identi®ed disease-related patterns can be independently validated with measures of task performance or behavior. Principal components (PCs) obtained from the analysis represent metabolic covariance patterns re¯ecting aspects of functional connectivity between brain regions [5,6]. Using this method, we have found that PD is associated with the expression of a speci®c metabolic covariance topography characterized by lentiform and thalamic hypermetabolism associated with regional metabolic decrements in the lateral premotor cortex (PMC), the supplementary motor area (SMA), the dorsolateral prefrontal cortex (DLPFC), and the parieto-occipital association regions [5,7]. The spatial topography of this PD-related pattern (PDRP) is consistent with the notion of excessive pallido-thalamic inhibition as the main functional substrate of parkinsonian bradykinesia [1]. We also developed a modi®cation of SSM/PCA for computing the expression of this network in individual subjects. This algorithm, known as Topographic Pro®le Rating (TPR), is used to compute subject scores for the PDRP on a prospective individual patient basis [8]. We have employed this approach to validate the PDRP as an imaging marker for the differential diagnosis and for the assessment of disease severity in PD [8,9]. In addition, to assess the reproducibility of this PDRP, we compared its topography with that of disease-related covariance patterns identi®ed in other patient populations. The results showed that the PDRP identi®ed in the original cohort was highly correlated with each of the corresponding topographies identi®ed in the other populations. Prospectively computed subject scores for the PDRP accurately discriminated PD patients from controls in all four populations [10]. These ®ndings indicate that the PDRP is highly reproducible across patient populations and tomographs and represents a robust metabolic marker of metabolic pathology in parkinsonism. 3. Modulation of metabolic networks during antiParkinsonian treatment Posterolateral ventral pallidotomy (PVP) has been shown to signi®cantly improve akinetic symptoms in PD as well as to relieve dyskinesia associated with levodopa administration [11±14]. Recently, deep brain high frequency stimulation (DBS) has been applied to the internal globus pallidus (GPi) because of the advantage of avoiding permanent side effects due to an ablative lesion, therefore allowing for a reversible amelioration of parkinsonian symptoms [15±17]. Despite the growing use of these surgical interventions, their mechanism of achieving therapeutic bene®t is not fully understood. The ameliorative effects of PVP and GPi DBS have been attributed to the reduction of excessive inhibitory out¯ow from the GPi [18].
We originally reported eight PD patients undergoing pallidotomy who were scanned with FDG/PET preoperatively and 6 months postoperatively [19]. To quantify potential modulations in the expression of motor networks by pallidotomy, we applied SSM/PCA to operative differences in regional glucose metabolism. We found that the topography identi®ed in this analysis closely resembled the PDRP, being characterized by a postoperative decline in the lentiform and thalamic metabolism ipsilateral to the surgical side associated with bilateral increases in supplementary motor area (SMA) metabolism. Subject scores for this pattern correlated signi®cantly with improvements in both contralateral and ipsilateral limb performance. These ®ndings indicate that unilateral PVP has broad ranging effects of brain function. Pallidal ablation can modulate the activity of spatially distributed functionally interconnected brain regions remote from the lesion site. Importantly, the topography of operative change following pallidal ablation conforms well to that predicted based upon the previously described PDRP network. Because local rates of glucose are a re¯ection of net afferent synaptic activity, it is reasonable to expect a decline in thalamic metabolism subsequent to surgical interference with pallidofugal inhibitory projections to the thalamus. Similarly, the metabolic increases in motor cortical areas occurring with pallidotomy may be related to enhanced cortical afferent synaptic activity from the ventral thalamus following surgical reduction in pallidothalamic inhibitory output [1]. Indeed, we have shown that spontaneous GPi single unit activity recorded intraoperatively during pallidotomy correlated signi®cantly with preoperative measures of thalamic glucose utilization obtained in the same patients under comparable behavioral conditions [6]. Additionally, we found that GPi ®ring rates correlated with a signi®cant metabolic network comprising the pallidum and its major out¯ow projections to the ventral tier and intralaminar thalamic nuclei, and to the brainstem. Modulation of functional activity in these regions is likely to be the substrate of clinical improvement with PVP. We have recently noted reversible PDRP network suppression with deep brain stimulation (DBS) of the internal globus pallidus (GPi; [20]) and with intravenous levodopa infusion (iv LD; [21]). The degree of change in this metabolic pattern with DBS and ivLD correlated with improvements in motor performance subsequent to these interventions. These ®ndings indicate that surgical or pharmacological intervention can reduce the abnormal expression of pathological brain networks in PD patients. 4. Effects of therapy on brain networks during motor execution We have extended the network analytical approach from studies of glucose metabolism in the resting state to studies of brain activation during motor performance. These PET studies were conducted utilizing [ 15O] H2O to measure
M. Fukuda et al. / Parkinsonism and Related Disorders 8 (2001) 91±94
regional cerebral blood ¯ow (rCBF) during the performance of simple and complex motor tasks. DBS offers the opportunity of inducing a reversible alteration in functional brain circuitry. It is therefore a useful experimental method to assess the modulation of structure-functional relationships during the successful treatment of PD. Recently, we studied six patients with advanced PD who performed a kinematically controlled motor execution task [22] on and off GPi stimulation [23]. GPi DBS resulted in signi®cant rCBF increases in the sensorimotor cortex (SMC) contralateral to the moving hand and bilaterally in SMA. Motor performance during PET imaging was improved by GPi DBS, especially with regard to signi®cant reductions in both timing and spatial errors. Improvements in timing error correlated with rCBF changes in the ipsilateral SMC and ventral thalamus, as well as in the contralateral cerebellum. By contrast, improvements in spatial accuracy correlated with rCBF changes in both cerebellar hemispheres as well as in the SMC ipsilateral to stimulation. Thus, GPi DBS can enhance activation responses within the motor CSPTC loop. Improvements in motor performance correlate with the functional activity of speci®c nodes within this circuit, and as well as in cerebello-cortical pathways. 5. Effect of therapy on brain networks during motor sequence learning Impaired learning of motor sequences is abnormal in PD, even at its earliest clinical stages [24,25]. We focused upon motor sequence learning during PET imaging as a means of quantifying abnormal brain-behavior relationships in PD patients at baseline and during therapy. We utilized kinematically controlled motor learning tasks [22,26] in which performance was quanti®ed psychophysically during the scanning epoch. We found that in untreated early stage PD patients, learning performance was decreased compared to age-matched normal controls. For PD patients to achieve a normal level of learning performance, greater brain activation was needed in the left DLPFC, preSMA, and bilaterally in the superior parietal cortex [26,27]. These ®ndings suggest that in PD increased activation of cortico-cortical pathways can compensate for striatal-cortical dysfunction during the learning of sequential information. Additionally, we assessed the relationship between different forms of antiparkinsonian treatment in terms of their effects on complex motor behavior and regional brain activation. We found that both ivLD and GPi DBS improved motor performance to comparable degrees. Nonetheless, GPi DBS improved sequence-learning performance while a trend toward worsening performance was noted during levodopa administration. In keeping with these behavioral ®ndings, PET imaging during the motor learning task revealed bilateral increases in the dorsolateral prefrontal cortex (Brodmann area 9), premotor cortex (Brodmann area 6), and precuneus (Brodmann area 7). By contrast,
93
ivLD did not signi®cantly alter brain activation during sequence learning, despite achieving equivalent motoric bene®t. These ®ndings suggest that direct modulation of higher order associative and prefrontal CSPTC loops may be obtained with localized stereotaxic procedures but not with typical pharmacotherapeutic approaches. Distinct pathophysiologic mechanisms may underlie the therapeutic ef®cacy of different treatment strategies for parkinsonism. 6. Conclusion Functional brain imaging with network analysis has allowed investigators to gain novel insights into the mechanistic basis of basal ganglia disorders. Quantitative functional brain imaging may also serve as a useful tool in objectively assessing the ef®cacy of therapeutic intervention for PD.
Acknowledgements Supported by NIH RO1 NS 35069, RO1 NS 37564, and by K24 NS 02101, as well as grants from the National Parkinson Foundation, the American Parkinson Disease Association, and the Dystonia Medical Research Foundation. Dr Fukuda is a Veola T. Kerr Fellow of the Parkinson Disease Foundation. References [1] Wichmann T, DeLong MR. Functional and pathophysiological models of the basal ganglia. Curr Opin Neurobiol 1996;6:751±8. [2] Eidelberg D. Functional brain networks in movement disorders. Curr Opin Neurol 1998;11:319±26. [3] Moeller JR, Strother SC. A regional covariance approach to the analysis of functional patterns in positron emission tomographic data. J Cereb Blood Flow Metab 1991;11:A121±35. [4] Alexander G, Moeller J. Application of the scaled subpro®le model to functional imaging in neuropsychiatric disorders: A principal component approach to modeling brain function in disease. Hum Brain Mapp 1994;2:1±16. [5] Eidelberg D, Moeller JR, Dhawan V, et al. The metabolic topography of parkinsonism. J Cereb Blood Flow Metab 1994;14:783±801. [6] Eidelberg D, Moeller JR, Kazumata K, et al. Metabolic correlates of pallidal neuronal activity in Parkinson's disease. Brain 1997;120:1315±24. [7] Eidelberg D, Moeller JR, Dhawan V, et al. The metabolic anatomy of Parkinson's disease: complementary [ 18F]¯uorodeoxyglucose and [ 18F]¯uorodopa positron emission tomographic studies. Mov Disord 1990;5:203±13. [8] Eidelberg D, Moeller JR, Ishikawa T, et al. Assessment of disease severity in parkinsonism with ¯uorine-18-¯uorodeoxyglucose and PET. J Nucl Med 1995;36:378±83. [9] Eidelberg D, Moeller JR, Ishikawa T, et al. Early differential diagnosis of Parkinson's disease with 18F-¯uorodeoxyglucose and positron emission tomography. Neurology 1995;45:1995± 2004. [10] Moeller JR, Nakamura T, Mentis MJ, et al. Reproducibility of regional metabolic covariance patterns: comparison of four populations. J Nucl Med 1999;40:1264±9. [11] Laitinen LV, Bergenheim AT, Hariz MI. Leksell's posteroventral
94
[12] [13] [14] [15] [16] [17] [18] [19]
M. Fukuda et al. / Parkinsonism and Related Disorders 8 (2001) 91±94 pallidotomy in the treatment of Parkinson's disease. J Neurosurg 1992;76:53±61. Lozano AM, Lang AE, Galvez-Jimenez N, et al. Effect of GPi pallidotomy on motor function in Parkinson's disease. Lancet 1995;346:1383±7. Baron MS, Vitek JL, Bakay RA, et al. Treatment of advanced Parkinson's disease by posterior GPi pallidotomy: 1-year results of a pilot study. Ann Neurol 1996;40:355±66. Dogali M, Fazzini E, Kolodny E, et al. Stereotactic ventral pallidotomy for Parkinson's disease. Neurology 1995;45:753±61. Gross C, Rougier A, Guehl D, Boraud T, Julien J, Bioulac B. Highfrequency stimulation of the globus pallidus internalis in Parkinson's disease: a study of seven cases. J Neurosurg 1997;87:491±8. Pahwa R, Wilkinson S, Smith D, Lyons K, Miyawaki E, Koller WC. High-frequency stimulation of the globus pallidus for the treatment of Parkinson's disease. Neurology 1997;49:249±53. Tronnier VM, Fogel W, Kronenbuerger M, Steinvorth S. Pallidal stimulation: an alternative to pallidotomy? J Neurosurg 1997;87:700±5. Marsden CD, Obeso JA. The functions of the basal ganglia and the paradox of stereotaxic surgery in Parkinson's disease. Brain 1994;117:877±97. Eidelberg D, Moeller JR, Ishikawa T, et al. Regional metabolic correlates of surgical outcome following unilateral pallidotomy for Parkinson's disease. Ann Neurol 1996;39:450±9.
[20] Fukuda M, Mentis M, Dhawan V, Lyons K, Koller W, Eidelberg D. Metabolic correlates of clinical outcome with pallidal deep brain stimulation for Parkinson's disease. Neurology 2000;54:A283. [21] Feigin A, Mentis M, Dhawan V, Margouleff C, Eidelberg D. Levodopa infusion reduced the expression of a Parkinson's disease metabolic brain network: A FDG PET study. Neurology 2000;54:A113±4. [22] Ghilardi M, Ghez C, Moeller J, Dhawan V, Eidelberg D. Patterns of regional brain activation associated with different aspects of motor learning. Brain Research 2000;871:127±45. [23] Fukuda M, Mentis M, Ghilardi M, et al. Functional correlates of pallidal stimulation for Parkinson's disease. Annals of Neurology 2001;49(2):155±65. [24] Brown RG, Marsden CD. Cognitive function in Parkinson's disease: from description to theory. Trends Neurosci 1990;13:21± 29. [25] Dubois B, Pillon B. Cognitive de®cits in Parkinson's disease. J Neurol 1997;244:2±8. [26] Nakamura T, Ghilardi MF, Mentis MJ, et al. Functional networks in motor sequence learning: abnormal topographies in Parkinson's disease. Human Brain Mapping 2001 in press. [27] Nakamura T, Mentis M, Dhawan V, et al. Abnormal motor sequence learning in early stage Parkinson's disease: a PET study. J Nucl Med 1999;40:266.
www.elsevier.com/locate/parkreldis
Functional brain networks in Parkinson's disease Masafumi Fukuda, Christine Edwards, David Eidelberg* Department of Neurology and Neurosurgery, New York University School of Medicine, Functional Brain Imaging Laboratory, North Shore-LIJ Research Institute, Manhasset, NY, USA
Abstract With the advent of new methods of network analysis, we have utilized metabolic data acquired through positron emission tomography (PET) to identify disease-related patterns of functional pathology in the movement disorders. In Parkinson's disease (PD), we have used [ 18F]-¯uorodeoxyglucose (FDG)/PET to identify a disease-related regional metabolic covariance pattern characterized by lentiform and thalamic hypermetabolism associated with regional metabolic decrements in the lateral premotor cortex, the supplementary motor area, the dorsolateral prefrontal cortex, and the parieto-occipital association regions. The expression of this network is modulated in a predictable fashion by levodopa therapy and by stereotaxic interventions for PD. We have extended this network analytical approach from studies of glucose metabolism in the resting state to dynamic studies of brain activation during motor performance. These PET studies utilized [ 15O]±water (H2 15O) to measure cerebral blood ¯ow activation responses during the execution of simple and complex motor tasks. In addition to the modulation of abnormal resting metabolic networks, effective PD therapy can enhance brain activation responses during motor execution, with speci®c regional associations with improvements in timing and spatial accuracy. This approach is also useful in identifying speci®c brain networks mediating the learning of sequential information. We have found that the normal relationship between brain networks and learning performance are altered in the earliest stages of PD with a functional shift from striatal to cortical processing. Brain activation PET studies during therapeutic interventions for PD demonstrate how normal brain-behavior relationships can be restored with successful therapy. Thus, functional brain imaging with network analysis can provide insights into the mechanistic basis of basal ganglia disorders and their treatment. q 2001 Elsevier Science Ltd. All rights reserved. Keywords: Parkinson's disease; PET; Brain networks; Glucose metabolism; Cerebral blood ¯ow
1. Introduction Functional brain imaging techniques have revolutionized the conceptual framework for exploring the mechanistic substrates of the movement disorders. We have developed specialized network analytical methods, in conjunction with positron emission tomography (PET), to examine hypotheses concerning alterations in brain-behavior relationships in Parkinson's disease (PD) and related disorders. Although the neurochemical lesion of PD is mainly localized to the nigrostriatal dopamine system, the effects of dopamine loss are widespread and affect several discrete parallel corticostriatopallidal-thalamocortical (CSPTC) circuits as crucial elements of motor control [1]. We have employed advanced PET techniques with network analysis to determine: (1) whether functional brain connectivity in PD patients and normal subjects conform to the anatomy of the postulated loops and (2) how the functional activity of these circuits * Corresponding author. Tel.: 11-516-562-2498; fax: 11-516-562-1008. E-mail address: [email protected] (D. Eidelberg).
relates to motor performance in PD patients studied at baseline and during effective therapeutic interventions. In this review, we will focus on the validation of the PDrelated metabolic networks and their modulation during dopaminergic pharmacotherapy and surgical interventions for parkinsonism. 2. Metabolic network mapping in PD We have developed and applied a statistical modeling approach for the identi®cation and quanti®cation of patterns of regional metabolic covariation in PD patients scanned with [ 18F]-¯uorodeoxyglucose (FDG)/PET [2]. This approach, known as the Scaled Subpro®le Model (SSM; [3,4]), is a form of principal component analysis (PCA) and can be used to identify patterns of regional covariation in functional imaging data. In SSM modeling, PCA is employed to identify regional covariance patterns from regional metabolic datasets obtained from combined samples of patients and normal controls. This form of analysis is blind to subject class designation, and utilizes the variance across the entire population to
1353-8020/01/$ - see front matter q 2001 Elsevier Science Ltd. All rights reserved. PII: S 1353-802 0(01)00022-0
92
M. Fukuda et al. / Parkinsonism and Related Disorders 8 (2001) 91±94
identify speci®c patterns associated with the disease state. SSM/PCA network analysis produces subject scores that quantify the degree to which each subject expresses each pattern. Subject scores for the identi®ed disease-related patterns can be independently validated with measures of task performance or behavior. Principal components (PCs) obtained from the analysis represent metabolic covariance patterns re¯ecting aspects of functional connectivity between brain regions [5,6]. Using this method, we have found that PD is associated with the expression of a speci®c metabolic covariance topography characterized by lentiform and thalamic hypermetabolism associated with regional metabolic decrements in the lateral premotor cortex (PMC), the supplementary motor area (SMA), the dorsolateral prefrontal cortex (DLPFC), and the parieto-occipital association regions [5,7]. The spatial topography of this PD-related pattern (PDRP) is consistent with the notion of excessive pallido-thalamic inhibition as the main functional substrate of parkinsonian bradykinesia [1]. We also developed a modi®cation of SSM/PCA for computing the expression of this network in individual subjects. This algorithm, known as Topographic Pro®le Rating (TPR), is used to compute subject scores for the PDRP on a prospective individual patient basis [8]. We have employed this approach to validate the PDRP as an imaging marker for the differential diagnosis and for the assessment of disease severity in PD [8,9]. In addition, to assess the reproducibility of this PDRP, we compared its topography with that of disease-related covariance patterns identi®ed in other patient populations. The results showed that the PDRP identi®ed in the original cohort was highly correlated with each of the corresponding topographies identi®ed in the other populations. Prospectively computed subject scores for the PDRP accurately discriminated PD patients from controls in all four populations [10]. These ®ndings indicate that the PDRP is highly reproducible across patient populations and tomographs and represents a robust metabolic marker of metabolic pathology in parkinsonism. 3. Modulation of metabolic networks during antiParkinsonian treatment Posterolateral ventral pallidotomy (PVP) has been shown to signi®cantly improve akinetic symptoms in PD as well as to relieve dyskinesia associated with levodopa administration [11±14]. Recently, deep brain high frequency stimulation (DBS) has been applied to the internal globus pallidus (GPi) because of the advantage of avoiding permanent side effects due to an ablative lesion, therefore allowing for a reversible amelioration of parkinsonian symptoms [15±17]. Despite the growing use of these surgical interventions, their mechanism of achieving therapeutic bene®t is not fully understood. The ameliorative effects of PVP and GPi DBS have been attributed to the reduction of excessive inhibitory out¯ow from the GPi [18].
We originally reported eight PD patients undergoing pallidotomy who were scanned with FDG/PET preoperatively and 6 months postoperatively [19]. To quantify potential modulations in the expression of motor networks by pallidotomy, we applied SSM/PCA to operative differences in regional glucose metabolism. We found that the topography identi®ed in this analysis closely resembled the PDRP, being characterized by a postoperative decline in the lentiform and thalamic metabolism ipsilateral to the surgical side associated with bilateral increases in supplementary motor area (SMA) metabolism. Subject scores for this pattern correlated signi®cantly with improvements in both contralateral and ipsilateral limb performance. These ®ndings indicate that unilateral PVP has broad ranging effects of brain function. Pallidal ablation can modulate the activity of spatially distributed functionally interconnected brain regions remote from the lesion site. Importantly, the topography of operative change following pallidal ablation conforms well to that predicted based upon the previously described PDRP network. Because local rates of glucose are a re¯ection of net afferent synaptic activity, it is reasonable to expect a decline in thalamic metabolism subsequent to surgical interference with pallidofugal inhibitory projections to the thalamus. Similarly, the metabolic increases in motor cortical areas occurring with pallidotomy may be related to enhanced cortical afferent synaptic activity from the ventral thalamus following surgical reduction in pallidothalamic inhibitory output [1]. Indeed, we have shown that spontaneous GPi single unit activity recorded intraoperatively during pallidotomy correlated signi®cantly with preoperative measures of thalamic glucose utilization obtained in the same patients under comparable behavioral conditions [6]. Additionally, we found that GPi ®ring rates correlated with a signi®cant metabolic network comprising the pallidum and its major out¯ow projections to the ventral tier and intralaminar thalamic nuclei, and to the brainstem. Modulation of functional activity in these regions is likely to be the substrate of clinical improvement with PVP. We have recently noted reversible PDRP network suppression with deep brain stimulation (DBS) of the internal globus pallidus (GPi; [20]) and with intravenous levodopa infusion (iv LD; [21]). The degree of change in this metabolic pattern with DBS and ivLD correlated with improvements in motor performance subsequent to these interventions. These ®ndings indicate that surgical or pharmacological intervention can reduce the abnormal expression of pathological brain networks in PD patients. 4. Effects of therapy on brain networks during motor execution We have extended the network analytical approach from studies of glucose metabolism in the resting state to studies of brain activation during motor performance. These PET studies were conducted utilizing [ 15O] H2O to measure
M. Fukuda et al. / Parkinsonism and Related Disorders 8 (2001) 91±94
regional cerebral blood ¯ow (rCBF) during the performance of simple and complex motor tasks. DBS offers the opportunity of inducing a reversible alteration in functional brain circuitry. It is therefore a useful experimental method to assess the modulation of structure-functional relationships during the successful treatment of PD. Recently, we studied six patients with advanced PD who performed a kinematically controlled motor execution task [22] on and off GPi stimulation [23]. GPi DBS resulted in signi®cant rCBF increases in the sensorimotor cortex (SMC) contralateral to the moving hand and bilaterally in SMA. Motor performance during PET imaging was improved by GPi DBS, especially with regard to signi®cant reductions in both timing and spatial errors. Improvements in timing error correlated with rCBF changes in the ipsilateral SMC and ventral thalamus, as well as in the contralateral cerebellum. By contrast, improvements in spatial accuracy correlated with rCBF changes in both cerebellar hemispheres as well as in the SMC ipsilateral to stimulation. Thus, GPi DBS can enhance activation responses within the motor CSPTC loop. Improvements in motor performance correlate with the functional activity of speci®c nodes within this circuit, and as well as in cerebello-cortical pathways. 5. Effect of therapy on brain networks during motor sequence learning Impaired learning of motor sequences is abnormal in PD, even at its earliest clinical stages [24,25]. We focused upon motor sequence learning during PET imaging as a means of quantifying abnormal brain-behavior relationships in PD patients at baseline and during therapy. We utilized kinematically controlled motor learning tasks [22,26] in which performance was quanti®ed psychophysically during the scanning epoch. We found that in untreated early stage PD patients, learning performance was decreased compared to age-matched normal controls. For PD patients to achieve a normal level of learning performance, greater brain activation was needed in the left DLPFC, preSMA, and bilaterally in the superior parietal cortex [26,27]. These ®ndings suggest that in PD increased activation of cortico-cortical pathways can compensate for striatal-cortical dysfunction during the learning of sequential information. Additionally, we assessed the relationship between different forms of antiparkinsonian treatment in terms of their effects on complex motor behavior and regional brain activation. We found that both ivLD and GPi DBS improved motor performance to comparable degrees. Nonetheless, GPi DBS improved sequence-learning performance while a trend toward worsening performance was noted during levodopa administration. In keeping with these behavioral ®ndings, PET imaging during the motor learning task revealed bilateral increases in the dorsolateral prefrontal cortex (Brodmann area 9), premotor cortex (Brodmann area 6), and precuneus (Brodmann area 7). By contrast,
93
ivLD did not signi®cantly alter brain activation during sequence learning, despite achieving equivalent motoric bene®t. These ®ndings suggest that direct modulation of higher order associative and prefrontal CSPTC loops may be obtained with localized stereotaxic procedures but not with typical pharmacotherapeutic approaches. Distinct pathophysiologic mechanisms may underlie the therapeutic ef®cacy of different treatment strategies for parkinsonism. 6. Conclusion Functional brain imaging with network analysis has allowed investigators to gain novel insights into the mechanistic basis of basal ganglia disorders. Quantitative functional brain imaging may also serve as a useful tool in objectively assessing the ef®cacy of therapeutic intervention for PD.
Acknowledgements Supported by NIH RO1 NS 35069, RO1 NS 37564, and by K24 NS 02101, as well as grants from the National Parkinson Foundation, the American Parkinson Disease Association, and the Dystonia Medical Research Foundation. Dr Fukuda is a Veola T. Kerr Fellow of the Parkinson Disease Foundation. References [1] Wichmann T, DeLong MR. Functional and pathophysiological models of the basal ganglia. Curr Opin Neurobiol 1996;6:751±8. [2] Eidelberg D. Functional brain networks in movement disorders. Curr Opin Neurol 1998;11:319±26. [3] Moeller JR, Strother SC. A regional covariance approach to the analysis of functional patterns in positron emission tomographic data. J Cereb Blood Flow Metab 1991;11:A121±35. [4] Alexander G, Moeller J. Application of the scaled subpro®le model to functional imaging in neuropsychiatric disorders: A principal component approach to modeling brain function in disease. Hum Brain Mapp 1994;2:1±16. [5] Eidelberg D, Moeller JR, Dhawan V, et al. The metabolic topography of parkinsonism. J Cereb Blood Flow Metab 1994;14:783±801. [6] Eidelberg D, Moeller JR, Kazumata K, et al. Metabolic correlates of pallidal neuronal activity in Parkinson's disease. Brain 1997;120:1315±24. [7] Eidelberg D, Moeller JR, Dhawan V, et al. The metabolic anatomy of Parkinson's disease: complementary [ 18F]¯uorodeoxyglucose and [ 18F]¯uorodopa positron emission tomographic studies. Mov Disord 1990;5:203±13. [8] Eidelberg D, Moeller JR, Ishikawa T, et al. Assessment of disease severity in parkinsonism with ¯uorine-18-¯uorodeoxyglucose and PET. J Nucl Med 1995;36:378±83. [9] Eidelberg D, Moeller JR, Ishikawa T, et al. Early differential diagnosis of Parkinson's disease with 18F-¯uorodeoxyglucose and positron emission tomography. Neurology 1995;45:1995± 2004. [10] Moeller JR, Nakamura T, Mentis MJ, et al. Reproducibility of regional metabolic covariance patterns: comparison of four populations. J Nucl Med 1999;40:1264±9. [11] Laitinen LV, Bergenheim AT, Hariz MI. Leksell's posteroventral
94
[12] [13] [14] [15] [16] [17] [18] [19]
M. Fukuda et al. / Parkinsonism and Related Disorders 8 (2001) 91±94 pallidotomy in the treatment of Parkinson's disease. J Neurosurg 1992;76:53±61. Lozano AM, Lang AE, Galvez-Jimenez N, et al. Effect of GPi pallidotomy on motor function in Parkinson's disease. Lancet 1995;346:1383±7. Baron MS, Vitek JL, Bakay RA, et al. Treatment of advanced Parkinson's disease by posterior GPi pallidotomy: 1-year results of a pilot study. Ann Neurol 1996;40:355±66. Dogali M, Fazzini E, Kolodny E, et al. Stereotactic ventral pallidotomy for Parkinson's disease. Neurology 1995;45:753±61. Gross C, Rougier A, Guehl D, Boraud T, Julien J, Bioulac B. Highfrequency stimulation of the globus pallidus internalis in Parkinson's disease: a study of seven cases. J Neurosurg 1997;87:491±8. Pahwa R, Wilkinson S, Smith D, Lyons K, Miyawaki E, Koller WC. High-frequency stimulation of the globus pallidus for the treatment of Parkinson's disease. Neurology 1997;49:249±53. Tronnier VM, Fogel W, Kronenbuerger M, Steinvorth S. Pallidal stimulation: an alternative to pallidotomy? J Neurosurg 1997;87:700±5. Marsden CD, Obeso JA. The functions of the basal ganglia and the paradox of stereotaxic surgery in Parkinson's disease. Brain 1994;117:877±97. Eidelberg D, Moeller JR, Ishikawa T, et al. Regional metabolic correlates of surgical outcome following unilateral pallidotomy for Parkinson's disease. Ann Neurol 1996;39:450±9.
[20] Fukuda M, Mentis M, Dhawan V, Lyons K, Koller W, Eidelberg D. Metabolic correlates of clinical outcome with pallidal deep brain stimulation for Parkinson's disease. Neurology 2000;54:A283. [21] Feigin A, Mentis M, Dhawan V, Margouleff C, Eidelberg D. Levodopa infusion reduced the expression of a Parkinson's disease metabolic brain network: A FDG PET study. Neurology 2000;54:A113±4. [22] Ghilardi M, Ghez C, Moeller J, Dhawan V, Eidelberg D. Patterns of regional brain activation associated with different aspects of motor learning. Brain Research 2000;871:127±45. [23] Fukuda M, Mentis M, Ghilardi M, et al. Functional correlates of pallidal stimulation for Parkinson's disease. Annals of Neurology 2001;49(2):155±65. [24] Brown RG, Marsden CD. Cognitive function in Parkinson's disease: from description to theory. Trends Neurosci 1990;13:21± 29. [25] Dubois B, Pillon B. Cognitive de®cits in Parkinson's disease. J Neurol 1997;244:2±8. [26] Nakamura T, Ghilardi MF, Mentis MJ, et al. Functional networks in motor sequence learning: abnormal topographies in Parkinson's disease. Human Brain Mapping 2001 in press. [27] Nakamura T, Mentis M, Dhawan V, et al. Abnormal motor sequence learning in early stage Parkinson's disease: a PET study. J Nucl Med 1999;40:266.