- Email: [email protected]
FDG PET in the differential diagnosis of parkinsonian disorders
www.elsevier.com/locate/ynimg NeuroImage 26 (2005) 912 – 921
FDG PET in the differential diagnosis of parkinsonian disorders Thomas Eckert,a,b Anna Barnes,a Vijay Dhawan,a,c,d Steve Frucht,e Mark F. Gordon,f Andrew S. Feigin,a,c,d and D. Eidelberga,c,d,* a
Center for Neurosciences, Institute for Medical Research, North Shore Long-Island Jewish Health System, Manhasset, NY 11030, USA Departments of Neurology II and Psychiatry, University of Magdeburg, Germany c Department of Neurology, North Shore University Hospital, Manhasset, NY 11030, USA d New York University School of Medicine, New York, NY 10016, USA e Neurologic Institute, Columbia-Presbyterian Medical Center, New York, NY 10032, USA f Department of Neurology, Long Island Jewish Medical Center, New Hyde Park, NY 11040, USA b
Received 18 November 2004; revised 3 February 2005; accepted 2 March 2005 Available online 26 April 2005 The differential diagnosis of parkinsonian disorders can be challenging, especially early in the disease course. PET imaging with [18F]fluorodeoxyglucose (FDG) has been used to identify characteristic patterns of regional glucose metabolism in patient cohorts with idiopathic Parkinson’s disease (PD), as well as variant forms of parkinsonism such as multiple system atrophy (MSA), progressive supranuclear palsy (PSP), and corticobasal degeneration (CBGD). In this study, we assessed the utility of FDG PET in the differential diagnosis of individual patients with clinical parkinsonism. 135 parkinsonian patients were referred for FDG PET to determine whether their diagnosis could be made accurately based upon their scans. Imaging-based diagnosis was obtained by visual assessment of the individual scans and also by computer-assisted interpretation. The results were compared with 2-year follow-up clinical assessments made by independent movement disorders specialists who were blinded to the original PET findings. We found that blinded computer assessment agreed with clinical diagnosis in 92.4% of all subjects (97.7% early PD, 91.6% late PD, 96% MSA, 85% PSP, 90.1% CBGD, 86.5% healthy control subjects). Concordance of visual inspection with clinical diagnosis was achieved in 85.4% of the patients scanned (88.4% early PD, 97.2% late PD, 76% MSA, 60% PSP, 90.9% CBGD, 90.9% healthy control subjects). This study demonstrates that FDG PET performed at the time of initial referral for parkinsonism accurately predicted the clinical diagnosis of individual patients made at subsequent follow-up. Computer-assisted methodologies may be particularly helpful in situations where experienced readers of FDG PET images are not readily available. D 2005 Elsevier Inc. All rights reserved. Keywords: Parkinson’s disease; Multiple system atrophy; Progressive supranuclear palsy; FDG PET; Differential diagnosis
* Corresponding author. Center for Neurosciences, Institute for Medical Research, North Shore-Long Island Jewish Health System, 350 Community Drive, Manhasset, NY 11030, USA. E-mail address: [email protected] (D. Eidelberg). Available online on ScienceDirect (www.sciencedirect.com). 1053-8119/$ - see front matter D 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.neuroimage.2005.03.012
Introduction Idiopathic Parkinson’s disease (PD) is a frequent diagnosis in the elderly (Moghal et al., 1994). However, only approximately 75% of parkinsonian patients prove to have PD at autopsy (Hughes et al., 1992b). The most common alternative diagnoses include multiple system atrophy (MSA), progressive supranuclear palsy (PSP), and corticobasal degeneration (CBGD) (Litvan, 1999). The precise differential diagnosis of parkinsonian syndromes is important because prognosis (Wenning et al., 1997) and treatment options (Tintner and Jankovic, 2002; Wenning et al., 2003) can differ substantially for PD and its clinical variants. However, differentiating parkinsonian syndromes by clinical means has proved unsatisfactory, especially early on in the course of the disease (Hughes et al., 2002; Litvan et al., 1998; Osaki et al., 2002). A variety of methods have been employed to improve the accuracy of differential diagnosis in patients with parkinsonism (Eckert et al., 2004). A number of studies using PET imaging with [18F]-fluorodeoxyglucose (FDG) have described characteristic patterns of glucose metabolism in patients with PD (Eidelberg et al., 1994; Moeller et al., 1999), MSA (Antonini et al., 1998; Eidelberg et al., 1993; Taniwaki et al., 2002), PSP (Blin et al., 1990; Foster et al., 1988), and CBGD (Blin et al., 1992; Eidelberg et al., 1991; Laureys et al., 1999). However, this technique has not been employed to assess diagnostic accuracy in individual patients with presumptive PD or variant forms of parkinsonism (e.g., Eckert and Eidelberg, 2004). To explore the utility of FDG PET in the ascertainment of single cases, we determined whether blinded scan interpretation was concordant with the ultimate clinical diagnosis. In a recent review of the clinical and neuropathological data from patients with parkinsonism (Hughes et al., 2002), the clinical impression of a movement disorder specialist proved to be more accurate than strict reliance upon specific sets of diagnostic criteria. We therefore selected the clinical diagnosis made by two independent movement disorder specialists as our ‘‘gold standard’’ for validating the imaging-based
T. Eckert et al. / NeuroImage 26 (2005) 912 – 921 Table 1 Clinical characteristics of control subjects and PD, MSA, PSP, and CBGD patients N Age at scan (male/female) in years, mean T SD in years Controls PD Early Late MSA PSP CBGD
Disease duration at scan in years, mean T SD in years
Follow-up time after FDG PET scanning in years, mean T SD in years –
22 (8/14)
56.3 T 11.5
–
43 36 25 20 11
56.7 60.2 59.6 69.0 68.0
T T T T T
2.2 9.8 3.9 2.9 3.3
(26/17) (19/17) (11/14) (10/10) (3/8)
10.9 8.9 8.9 7 9.4
T T T T T
1.2 3.5 1.4 1.2 1.0
2.1 1.9 2.3 1.9 2.3
T T T T T
1.1 0.7 1.6 0.6 1.5
Shown is a summary of clinical and demographical findings in controls and patients. Rows indicate the group (PD = idiopathic Parkinson’s disease, MSA = multiple system atrophy, PSP = progressive supranuclear palsy, CBGD = corticobasal degeneration).
diagnosis. In the present study, we compared the clinical categorization to blinded single case ascertainments made with FDG PET by Fvisual reading_ by a trained expert and by Fcomputer-supported reading_ by a non-expert using statistical parametric mapping (SPM) as a decision aid (Barnes et al., 2000; Signorini et al., 1999).
Methods Patients We studied 135 patients (age 60.6 T 10.4 years) referred for FDG PET imaging to assist in the differential diagnosis of parkinsonism. In these subjects, the clinical diagnosis was uncertain at the time of referral for imaging. This group of patients represented a subset of a larger group of 205 patients referred for PET imaging for the differential diagnosis of parkinsonism. Of the total cohort, 53 patients were excluded for lack of follow-up and 17 were excluded because of structural abnormalities on MRI. A probable clinical diagnosis was determined at follow-up by two independent movement disorder specialists (SF, TE) who were blinded to the imaging diagnosis (see below). The follow-up clinical assessments were conducted an average of 2.1 years after the initial PET imaging. Patients were included in this study only if a diagnosis of PD or a variant syndrome (MSA, PSP, or CBGD) could be achieved by the ascertaining clinicians at follow-up, and if no additional brain pathology (e.g., white matter changes, ischemia) was detected on routine MRI. Of the 135 patients, 43 were considered likely to have early stage PD (<5 years duration), 36 late stage PD (5 year duration), 25 MSA, 20 PSP, and 11 CBGD. In two of the patients thought to have MSA, the diagnosis was subsequently confirmed at postmortem. In addition, 22 healthy control subjects were included in the analysis. Demographic data for the patient groups and healthy control subjects are presented in Table 1. Ethical permission for this study was obtained from the Institutional Review Board of the North Shore-Long Island Jewish Health System. FDG PET imaging Subjects fasted overnight before PET scanning; antiparkinsonian medications were withheld at least 12 h before the PET
913
investigations. 185 – 370 MBq (5 – 10 mCi) FDG in 4 ml saline was injected intravenously. Subjects were scanned using the GE Advance tomograph (General Electric Medical Systems, Milwaukee, WI, USA) at North Shore University Hospital. This eight-ring bismuth germanate scanner provided 35 two-dimensional image planes with an axial field of view of 14.5 cm and transaxial resolution of 4.2 mm (full width half maximum) in all directions. The detailed performance characteristics of this instrument have been described elsewhere (Dhawan et al., 1998). All studies were performed with the subject’s eyes open in a dimly lit room with minimal auditory stimulation. Diagnostic evaluation Visual reading For routine visual reading of the PET images, the scans were reconstructed, corrected for attenuation, and smoothed for each subject. FDG PET images were displayed as a series of 35 transaxial slices scaled to a common maximum in a standard color scale. In addition, a composite slice was generated by summing the central four of the six slices in which the striatum was visualized (thickness approximately 16 mm). Striatal and subcortical structures are compared visually to the activity in the overall cortical strip, cerebellum, and occipital cortex by an expert reader (VD) blinded to the clinical information. Diagnostic criteria are based on the published literature (for a review, see Dhawan and Eidelberg, 2003). These feature decreased metabolism of the basal ganglia and cerebellum for MSA, medial frontal decreases especially at superior levels for PSP, marked asymmetric cortical and basal ganglia metabolism for CBGD, and increased metabolism of the lentiform nucleus in relation to the cortex in clinical PD. Each single patient FDG PET image was diagnosed as either a control subject, PD, MSA, PSP, or CBGD patient. Computer-supported reading Image processing and analysis All image processing and analyses were performed using Statistical Parametric Mapping (SPM99, Wellcome Department of Cognitive Neurology, London, UK) running in MATLAB (R12, Mathworks, Natick, MA). The images were spatially normalized after which an isotropic 10 mm full width half maximum smoothing filter was applied to the images (Poline et al., 1995). Creating the characteristic disease templates We developed template statistical maps to aid in PET scan interpretation. This was accomplished by randomly selecting eight subjects from each of the PD, MSA, and PSP groups and five members from the smaller CBGD group. These patient cohorts were cross-matched for age and disease duration. Scans from each of the cohorts were compared to those from 10 agematched control subjects using the general linear model in the SPM99 (Friston et al., 1995). For the PD and CBGD groups, images of predominately left side affected subjects were flipped, so that all affected sides were on the right. Analysis of covariance (ANCOVA) was used for the purposes of intensity normalization to account for differences in global flow across subjects. A threshold of 80% of the mean global counts was used to identify voxels representative of gray matter and therefore to be included in the statistical analysis. A t test was then performed at all these voxels. Any voxels showing increased or decreased glucose
914
T. Eckert et al. / NeuroImage 26 (2005) 912 – 921
metabolism in the patient groups compared to the control group above the statistical threshold of P < 0.01 (peak and extent uncorrected for multiple statistical comparisons) were overlaid onto the T1 MR template image provided by the SPM99 software (Friston et al., 1996). This threshold was chosen because it yielded the maps of abnormal regional glucose metabolism that conformed closely to published patterns for each of the
syndromes. These canonical maps (see Fig. 1) were then used as templates to assist in the differential diagnosis of single scans. Characterizing single subject scans To characterize single patient scans, each patient or control subject’s PET image was compared statistically to the reference group of 10 healthy control subjects providing a map of statistically
Fig. 1. Detailed templates (whole brain) of abnormal glucose metabolism in patients with (A) PD, (B) MSA, (C) PSP, and (D) CBGD identified by group analysis using SPM ( P < 0.05). At this relaxed threshold the defining and supportive features become evident. Increased glucose metabolism is indicated by Fhot_ colors and decreased metabolism by Fwinter_ colors. For image analysis, the hemispheres opposite the clinically more affected body sides appeared on the right (for larger views, see www.neuroscience-nslij.org).
T. Eckert et al. / NeuroImage 26 (2005) 912 – 921
915
Fig. 1 (continued).
different regions of glucose metabolism in the manner described above. Similarly all voxels showing increased or decreased glucose metabolism at a threshold of P < 0.05 (peak and extent uncorrected for multiple statistical comparisons) compared to the control group were also overlaid onto the T1 MRI template image provided by SPM99 and saved as a portable document format (PDF) image to be read by a non-expert. Increased glucose metabolism is shown in Fhot_ colors and decreased glucose metabolism in Fwinter_ colors. The PDF images were viewed on a standard PC screen by a nonexpert reader (AB) blinded to all patient’s clinical information.
Prior to the evaluation, the reader was trained in the evaluation method of the single patient statistical maps using a newly developed set of criteria for differentiating parkinsonism syndromes based upon characteristic patterns of regional metabolism (see Table 2). These criteria were formulated on the basis of descriptions of abnormal glucose utilization in PD (Eidelberg et al., 1994; Moeller et al., 1999), MSA (Antonini et al., 1998; Eidelberg et al., 1993; Taniwaki et al., 2002), PSP (Blin et al., 1990; Foster et al., 1988), and CBGD (Blin et al., 1992; Eidelberg et al., 1991; Laureys et al., 1999) patients. The reader compared each single patient PDF image to the
916
T. Eckert et al. / NeuroImage 26 (2005) 912 – 921
Fig. 1 (continued).
defining features of these criteria and each PDF image was then characterized as either PD, MSA, PSP, or CBGD. If no defining features of disease were apparent, the scan was classified as belonging to the control group. The non-expert did not view the original PET scans. The single patient imaging diagnosis of Fvisual reading_ and FSPM-supported reading_ were then compared to each patient’s final clinical diagnosis (i.e., the diagnostic Fgold standard_). The sensitivity and specificity of both image-based methods were then calculated from these results using a 2 2 table method (Lijmer et al., 1999).
Results Characteristic patterns of abnormal regional metabolism PD We found that the hallmark of glucose metabolism in the PD cohort was increased metabolism in the putamen and globus pallidus (Fig. 1A). The finding was present bilaterally without regard to the more affected side. Hypermetabolic areas are also observed in the ventral thalamus, and in the motor cortex and cerebellum. Abnormal reductions in glucose metabolism in the
T. Eckert et al. / NeuroImage 26 (2005) 912 – 921
917
Fig. 1 (continued).
PD group were present bilaterally in parietal and occipital association areas, and in the dorsolateral prefrontal cortex (DLPFC). MSA The MSA cohort exhibited a metabolic pattern characterized by marked bilateral reductions in the lentiform nuclei and in the cerebellum (see Fig. 1B and Table 2). Metabolic reductions in the lentiform nucleus were seen in all presumptive MSA patients. While glucose metabolism of the cerebellum was reduced in all patients with clinical ataxia, it was also reduced in a number of
subjects without signs of cerebellar dysfunction. Reduced regional metabolism was also observed in the brainstem. Relative metabolic increases were noted bilaterally in the thalamus; and in parietal, frontal, and temporal cortical regions. PSP The distinguishing feature of the PSP group was the presence of metabolic decrements in midline frontal regions and in the brainstem (see Fig. 1C and Table 2). Decreased metabolism was also noted in the superior frontal and insular areas, and in the caudate nucleus. Relative metabolic increases were observed in the
918
T. Eckert et al. / NeuroImage 26 (2005) 912 – 921
Table 2 Characteristic glucose metabolism patterns of parkinsonian syndromes, defining features are used for classification of FSPM-supported diagnosis_ of single patients Parkinson’s disease Defining features & hypermetabolism of the dorsolateral putamen & no defining features of other parkinsonian syndromes Supporting features & thalamic hypermetabolism & cerebellar hypermetabolism & bilateral hypermetabolism of cortical motor areas & bilateral hypometabolism of the parietal cortex Multiple system atrophy Defining features & bilateral hypometabolism of the putamen & cerebellar hypometabolism Supporting features & bilateral hypermetabolism of frontal and superior parietal cortical areas & bilateral hypermetabolism of the thalamus & hypometabolism of the brainstem Progressive supranuclear palsy Defining features & hypometabolism of the brainstem & hypometabolism of the midline frontal cortex & no defining features of MSA or CBGD Supporting features & bilateral hypermetabolism of cortical motor areas & bilateral hypermetabolism of the parietal cortex & bilateral hypermetabolism of the thalamus & bilateral hypometabolism of the caudate nucleus Corticobasal degeneration Defining features & asymmetrical cortical activation: relative hypometabolism contralateral to the most affected side & asymmetrical basal ganglia metabolism: relative hypometabolism contralateral to the most affected side Supporting features & one-sided cortical hypometabolism, especially of the insular cortex & hypometabolism of paraventricular areas & hypometabolism of the brain stem and midline frontal areas
thalamus, and in parietal, occipital, and temporal regions, as well as in motor cortex and infero-lateral frontal regions. CBGD The pattern of glucose metabolism in CBGD was characterized by a distinctive asymmetrical distribution of radiotracer uptake. Relative metabolic reductions were present in many cortical areas, including the insula and in the basal ganglia contralateral to the most affected side (see Fig. 1D and Table 2). Conversely, relatively increased glucose metabolism in the CBGD group compared to controls was observed in the cortex and basal ganglia ipsilateral to the most affected side. Diagnostic classification of single patient scans FVisual reading_ Using the Fvisual reading_ method to evaluate PET images, imaging diagnosis accorded with the clinical diagnosis in 85.4% of all subjects. Specifically, this concordance was 88.4% for early PD, 97.2% for late PD, 76% for MSA, 60% for PSP, and 90.9% for CBGD. 90.9% of healthy volunteers were correctly categorized by
this method of scan interpretation (for detailed information, see Tables 3 and 4). A sensitivity of 100% and specificity of 91% was obtained in identifying early PD patients (<5 years duration of symptoms) with respect to normal control subjects. Similarly, a sensitivity of 92% and specificity of 95% was achieved for PD patients versus subjects with presumed atypical parkinsonian syndromes. FSPM-supported reading_ Using the FSPM-supported reading_ method to evaluate PET images based on the classification criteria described in Table 2, correct imaging diagnosis was obtained in 92.4% of all subjects. Concordance with clinical diagnosis was 97.7% for early PD, 91.6% for late PD, 96% for MSA, 85% for PSP, 90.1% for CBGD, and 86.5% for healthy control subjects (for detailed information, see Tables 3 and 4). A sensitivity of 100% and specificity of 86% was obtained in identifying early PD patients (<5 years duration of symptoms) as compared with the normal controls. A sensitivity of 96% and specificity of 91% was achieved for PD patients as compared with patients with presumed atypical parkinsonism.
Discussion The present study has shown that FDG PET has a very good ability to differentiate PD, MSA, PSP, and CBGD patients as well as healthy control subjects on a single case basis. There have been a number of previous studies that have identified characteristic glucose metabolism patterns in patients with PD (Eidelberg et al., 1994; Moeller et al., 1999), MSA (Antonini et al., 1998; De Volder et al., 1989; Eidelberg et al., 1993; Otsuka et al., 1996; Taniwaki et al., 2002), PSP (Blin et al., 1990; Foster et al., 1988; Piccini et al., 2001), and CBGD (Blin et al., 1992; Eidelberg et al., 1991; Laureys et al., 1999) with respect to normal control subjects. These earlier FDG PET studies focused on the description of group comparisons of glucose metabolism between patient and control groups, without application to the differential diagnosis of individual subjects. The current study used SPM to create disease-related templates in which regional features were in close agreement with those previously reported. Our single case approach showed that these patterns can potentially be used diagnostically in that correct classification can be achieved relative to the final clinical assessment in over 90% of subjects. Correct diagnosis of patients with parkinsonism is of practical importance, especially at early stages of disease. If possible, these patients should be informed about their prognosis, which may differ considerably depending upon differential diagnosis (Diamond et al., 1987; Golbe et al., 1988; Wenning et al., 1997). Additionally, there are major differences in treatment options. While PD patients tend to demonstrate ongoing responsiveness to dopaminergic medication (Martin and Wieler, 2003; Olanow, 2002; Tintner and Jankovic, 2002), atypical parkinsonian patients are quite variable in this regard (Hughes et al., 1992a; Parati et al., 1993; Wenning et al., 2003). PD and its variants may also differ with regard to surgical response. Deep brain stimulation (DBS) and ablation of the subthalamic nucleus (STN) or the internal globus pallidus (GP) are well-established approaches for the treatment of advanced PD (Krack et al., 2000). However, patients with atypical parkinsonian syndromes appear not to profit from these surgical interventions (Chou et al., 2004; Lezcano et al., 2004; Tarsy et al., 2003; VisserVandewalle et al., 2003). Correct diagnosis can be used to optimize
T. Eckert et al. / NeuroImage 26 (2005) 912 – 921
919
Table 3 Diagnostic classification matrix for visual reading and SPM-supported reading Clinical classification (n)
Visual reading Controls
PD
Controls (22) PD early (43) PD late (36) MSA (25) PSP (20) CBGD (11)
20 0 0 1 0 1
0 38 35 0 1 0
Clinical classification (n)
SPM-supported reading
Controls (22) PD early (43) PD late (36) MSA (25) PSP (20) CBGD (11)
(90.9%) (0%) (0%) (4%) (0%) (9.1%)
Controls
PD
19 0 1 1 0 0
1 42 33 0 3 1
(86.5%) (0%) (2.8%) (4%) (0%) (0%)
(0%) (88.4%) (97.2%) (0%) (5%) (0%)
(4.5%) (97.7%) (91.6%) (0%) (15%) (9.1%)
MSA
PSP
CBGD
0 0 0 19 3 0
2 2 0 1 12 0
0 3 1 4 4 10
(0%) (0%) (0%) (76%) (15%) (0%)
(9.1%) (4.6%) (0%) (4%) (60%) (0%)
(0%) (7%) (2.8%) (16%) (20%) (90.9%)
MSA
PSP
CBGD
1 0 0 24 0 0
1 1 0 0 17 0
0 0 2 0 0 10
(4.5%) (0%) (0%) (96%) (0%) (0%)
(4.5%) (2.3%) (0%) (0%) (85%) (0%)
(0%) (0%) (5.6%) (0%) (0%) (90.1%)
Overall correct diagnosis: visual reading 85.4%, SPM suppported reading 92.4%. Summary of the image guided diagnosis using visual reading and SPMsupported reading. Shown is the classification of subjects due to the FDG PET imaging guided diagnosis with respect to the clinical diagnosis. Rows represent the clinical diagnosis and columns the diagnosis predicted by FDG PET imaging. Bold style indicates correct diagnosis (PD = idiopathic Parkinson’s disease, MSA = multiple system atrophy, PSP = progressive supranuclear palsy, CBGD = corticobasal degeneration).
treatment and to reduce the risks attendant to potentially ineffective treatment strategies. Our case-by-case study shows that considerable accuracy can be achieved with FDG PET to discriminate classical from potentially less responsive forms of parkinsonism. Accurate differential diagnosis is also of particular importance in the conduct of treatment trials in parkinsonism. Inadvertent inclusion of atypical patients into pharmacological trials for PD is likely to reduce statistical power by increasing the heterogeneity of the treatment cohorts. Trials of potential neuroprotective agents for PD (Parkinson-Study-Group, 2002; Whone et al., 2003; Fahn et al., 2004) have focused on early stage disease, in which progression may be more rapid (Bonnet et al., 1987; Fearnley and Lees, 1991; Lee et al., 1994). However, differential diagnosis of parkinsonism based on clinical features is especially challenging at early disease stages. A recent neuropathological study revealed that while correct clinical diagnosis at follow-up was 85%, the initial clinical impression was revised in 60% of subjects (Hughes et al., 2002). Another postmortem study showed that the sensitivity for correct clinical diagnosis of MSA at the initial clinical visit was only 22% (Osaki et al., 2002). Our data suggest that accurate differential diagnosis in patients with parkinsonism supported by the use of imaging techniques as FDG PET may improve the power of future clinical trials by promoting group homogeneity. Our newly defined diagnostic FDG imaging criteria (Table 2) provides a useful diagnostic imaging classification for individual patients. Indeed, the results of Fcomputer-supported reading_ with an overall correct diagnosis of 92.4% by a non-expert investigator were superior to the results of Fvisual reading_ with an overall correct diagnosis of 85.4% by an experienced interpreter of neurological PET scans. Applying this SPM-based approach, it is possible to normalize the PET images, to statistically compare them to a group of images from healthy controls, and to overlay the resulting maps of abnormal glucose utilization so as to make them easily visible for single case interpretation. The procedures used in this study were all automated, blinded, and investigator independent. Performing a voxel-based comparison has the distinct
advantage of not relying upon prior assumptions regarding the anatomical distribution of the metabolic abnormalities associated with each syndrome. By contrast, visual evaluation of scans depends on the experience of the reader in qualitatively assessing neurological PET scans. Our results suggest that Fcomputersupported reading_ may actually be more accurate and requires less examiner knowledge than visual inspection by a trained expert. This attribute may enhance the utility of FDG PET in non-expert imaging centers. In this respect, fully automated, investigatorindependent methods of image evaluation are desirable, although these ‘‘black box’’ techniques generally require a suitable FDG control group. SPM-supported image analysis for single patients can potentially be used with other neuroimaging techniques, such as [99mTc]-ethylene cysteinate dimer (ECD) SPECT (Feigin et al.,
Table 4 Diagnostic values of FDG PET using visual reading and SPM-supported reading Visual reading
PPV NPV Sensitivity Specificity
PD
MSA
PSP
CBGD
Controls
0.99 0.93 0.92 0.99
0.86 0.96 0.76 0.98
0.70 0.94 0.60 0.96
0.45 0.99 0.91 0.92
0.91 0.99 0.91 0.99
SPM-supported reading
PPV NPV Sensitivity Specificity
PD
MSA
PSP
CBGD
Controls
0.94 0.95 0.95 0.94
0.96 0.99 0.96 0.99
0.89 0.98 0.85 0.99
0.83 0.99 0.91 0.99
0.90 0.98 0.86 0.99
Diagnostic values of correct imaging diagnosis using visual reading and SPM-supported reading referred to the clinical diagnosis (PD = idiopathic Parkinson’s disease, MSA = multiple system atrophy, PSP = progressive supranuclear palsy, CBGD = corticobasal degeneration, PPV = positive predictive value, NPV = negative predictive value).
920
T. Eckert et al. / NeuroImage 26 (2005) 912 – 921
2002; Van Laere et al., 2004), perfusion MRI or magnetization transfer imaging (Eckert et al., 2004). In summary, this study demonstrates the utility of FDG PET imaging in the differential diagnosis of patients with parkinsonism using an evidence-based medicine approach. This new method may contribute to early differential diagnosis in clinically ambiguous cases of parkinsonism, or to confirm a diagnosis of PD in therapeutic trials. The development of fully automated diagnostic techniques is likely to be valuable in future studies of PD and related disorders. However, we would like to emphasize that this new approach cannot substitute for clinical evaluation in the assessment of parkinsonian disorders.
References Antonini, A., Kazumata, K., Feigin, A., et al., 1998. Differential diagnosis of parkinsonism with [18F]fluorodeoxyglucose and PET. Mov. Disord. 13, 268 – 274. Barnes, A., Lusman, D., Patterson, J., Brown, D., Wyper, D., 2000. The use of statistical parametric mapping (SPM96) as a decision aid in the differential diagnosis of dementia using 99mTc-HMPAO SPECT. Behav. Neurol. 12, 77 – 86. Blin, J., Baron, J.C., Dubois, B., et al., 1990. Positron emission tomography study in progressive supranuclear palsy. Brain hypometabolic pattern and clinicometabolic correlations. Arch. Neurol. 47, 747 – 752. Blin, J., Vidailhet, M.J., Pillon, B., Dubois, B., Feve, J.R., Agid, Y., 1992. Corticobasal degeneration: decreased and asymmetrical glucose consumption as studied with PET. Mov. Disord. 7, 348 – 354. Bonnet, A.M., Loria, Y., Saint-Hilaire, M.H., Lhermitte, F., Agid, Y., 1987. Does long-term aggravation of Parkinson’s disease result from nondopaminergic lesions? Neurology 37, 1539 – 1542. Chou, K.L., Forman, M.S., Trojanowski, J.Q., Hurtig, H.I., Baltuch, G.H., 2004. Subthalamic nucleus deep brain stimulation in a patient with levodopa-responsive multiple system atrophy, Case report. J. Neurosurg. 100, 553 – 556. De Volder, A.G., Francart, J., Laterre, C., et al., 1989. Decreased glucose utilization in the striatum and frontal lobe in probable striatonigral degeneration. Ann. Neurol. 26, 239 – 247. Dhawan, V., Eidelberg, D., 2003. PET imaging in Parkinson’s disease. In: Sinha, K.K., Chandra, P., Jha, D.K. (Eds.), Advances in Clinical Neurosciences. The Catholic Press, Jharkhand, India, pp. 251 – 276. Dhawan, V., Kazumata, K., Robeson, W., et al., 1998. Quantitative Brain PET. Comparison of 2D and 3D Acquisitions on the GE Advance Scanner. Clin. Positron Imaging 1, 135 – 144. Diamond, S.G., Markham, C.H., Hoehn, M.M., McDowell, F.H., Muenter, M.D., 1987. Multi-center study of Parkinson mortality with early versus later dopa treatment. Ann. Neurol. 22, 8 – 12. Eckert, T., Eidelberg, D., 2004. The role of functional neuroimaging in the differential diagnosis of idiopathic Parkinson’s disease and multiple system atrophy. Clin. Auton. Res. 14, 84 – 91. Eckert, T., Sailer, M., Kaufmann, J., et al., 2004. Differentiation of idiopathic Parkinson’s disease, multiple system atrophy, progressive supranuclear palsy, and healthy controls using magnetization transfer imaging. NeuroImage 21, 229 – 235. Eidelberg, D., Dhawan, V., Moeller, J.R., et al., 1991. The metabolic landscape of cortico-basal ganglionic degeneration: regional asymmetries studied with positron emission tomography. J. Neurol., Neurosurg. Psychiatry 54, 856 – 862. Eidelberg, D., Takikawa, S., Moeller, J.R., et al., 1993. Striatal hypometabolism distinguishes striatonigral degeneration from Parkinson’s disease. Ann. Neurol. 33, 518 – 527. Eidelberg, D., Moeller, J.R., Dhawan, V., et al., 1994. The metabolic topography of parkinsonism. J. Cereb. Blood Flow. Metab. 14, 783 – 801.
Fahn, S., Oakes, D., Shoulson, I., Kieburtz, K., Rudolph, A., Lang, A., et al., 2004. Levodopa and the progression of Parkinson’s disease. N. Engl. J. Med. 351, 2498 – 2508. Fearnley, J.M., Lees, A.J., 1991. Ageing and Parkinson’s disease: substantia nigra regional selectivity. Brain 114, 2283 – 2301. Feigin, A., Antonini, A., Fukuda, M., et al., 2002. Tc-99m ethylene cysteinate dimer SPECT in the differential diagnosis of parkinsonism. Mov. Disord. 17, 1265 – 1270. Foster, N.L., Gilman, S., Berent, S., Morin, E.M., Brown, M.B., Koeppe, R.A., 1988. Cerebral hypometabolism in progressive supranuclear palsy studied with positron emission tomography. Ann. Neurol. 24, 399 – 406. Friston, K.J., Holmes, A.P., Worsley, K.J., Poline, J.P., Frith, C.D., Frackowiak, R.S.J., 1995. Statistical parametric maps in functional imaging: a general linear approach. Hum. Brain Mapp. 2, 189 – 210. Friston, K.J., Holmes, A., Poline, J.B., Price, C.J., Frith, C.D., 1996. Detecting activations in PET and fMRI: levels of inference and power. NeuroImage 4, 223 – 235. Golbe, L.I., Davis, P.H., Schoenberg, B.S., Duvoisin, R.C., 1988. Prevalence and natural history of progressive supranuclear palsy. Neurology 38, 1031 – 1034. Hughes, A.J., Colosimo, C., Kleedorfer, B., Daniel, S.E., Lees, A.J., 1992a. The dopaminergic response in multiple system atrophy. J. Neurol., Neurosurg. Psychiatry 55, 1009 – 1013. Hughes, A.J., Daniel, S.E., Kilford, L., Lees, A.J., 1992b. Accuracy of clinical diagnosis of idiopathic Parkinson’s disease: a clinico-pathological study of 100 cases. J. Neurol., Neurosurg. Psychiatry 55, 181 – 184. Hughes, A.J., Daniel, S.E., Ben-Shlomo, Y., Lees, A.J., 2002. The accuracy of diagnosis of parkinsonian syndromes in a specialist movement disorder service. Brain 125, 861 – 870. Krack, P., Poepping, M., Weinert, D., Schrader, B., Deuschl, G., 2000. Thalamic, pallidal, or subthalamic surgery for Parkinson’s disease? J. Neurol. 247, 122 – 134. Laureys, S., Salmon, E., Garraux, G., et al., 1999. Fluorodopa uptake and glucose metabolism in early stages of corticobasal degeneration. J. Neurol. 246, 1151 – 1158. Lee, C.S., Schulzer, M., Mak, E.K., et al., 1994. Clinical observations on the rate of progression of idiopathic parkinsonism. Brain 117, 501 – 507. Lezcano, E., Gomez-Esteban, J.C., Zarranz, J.J., et al., 2004. Parkinson’s disease-like presentation of multiple system atrophy with poor response to STN stimulation: a clinicopathological case report. Mov. Disord. 19, 973 – 977. Lijmer, J.G., Mol, B.W., Heisterkamp, S., et al., 1999. Empirical evidence of design-related bias in studies of diagnostic tests. JAMA 282, 1061 – 1066. Litvan, I., 1999. Recent advances in atypical parkinsonian disorders. Curr. Opin. Neurol. 12, 441 – 446. Litvan, I., Booth, V., Wenning, G.K., et al., 1998. Retrospective application of a set of clinical diagnostic criteria for the diagnosis of multiple system atrophy. J. Neural Transm. 105, 217 – 227. Martin, W.R., Wieler, M., 2003. Treatment of Parkinson’s disease. Can. J. Neurol. Sci. 30, S27 – S33. Moeller, J.R., Nakamura, T., Mentis, M.J., et al., 1999. Reproducibility of regional metabolic covariance patterns: comparison of four populations. J. Nucl. Med. 40, 1264 – 1269. Moghal, S., Rajput, A.H., D’Arcy, C., Rajput, R., 1994. Prevalence of movement disorders in elderly community residents. Neuroepidemiology 13, 175 – 178. Olanow, C.W., 2002. Surgical therapy for Parkinson’s disease. Eur. J. Neurol. 9, 31 – 39. Osaki, Y., Wenning, G.K., Daniel, S.E., et al., 2002. Do published criteria improve clinical diagnostic accuracy in multiple system atrophy? Neurology 59, 1486 – 1491. Otsuka, M., Ichiya, Y., Kuwabara, Y., et al., 1996. Glucose metabolism in the cortical and subcortical brain structures in multiple system atrophy and Parkinson’s disease: a positron emission tomographic study. J. Neurol. Sci. 144, 77 – 83.
T. Eckert et al. / NeuroImage 26 (2005) 912 – 921 Parati, E.A., Fetoni, V., Geminiani, G.C., et al., 1993. Response to l-DOPA in multiple system atrophy. Clin. Neuropharmacol. 16, 139 – 144. Parkinson-Study-Group, 2003. Does levodopa slow or hasten the rate of progression of Parkinson disease? The result of the elldopa trial. Am. Acad. Neurol. 60, A80 – A81. Piccini, P., de Yebenez, J., Lees, A.J., et al., 2001. Familial progressive supranuclear palsy: detection of subclinical cases using 18F-dopa and 18 Fluorodeoxyglucose positron emission tomography. Arch. Neurol. 58, 1846 – 1851. Poline, J.B., Worsley, K.J., Holmes, A.P., Frackowiak, R.S., Friston, K.J., 1995. Estimating smoothness in statistical parametric maps: variability of P values. J. Comput. Assist. Tomogr. 19, 788 – 796. Signorini, M., Paulesu, E., Friston, K., et al., 1999. Rapid assessment of regional cerebral metabolic abnormalities in single subjects with quantitative and nonquantitative [18F]FDG PET: a clinical validation of statistical parametric mapping. NeuroImage 9, 63 – 80. Taniwaki, T., Nakagawa, M., Yamada, T., et al., 2002. Cerebral metabolic changes in early multiple system atrophy: a PET study. J. Neurol. Sci. 200, 79 – 84. Tarsy, D., Apetauerova, D., Ryan, P., Norregaard, T., 2003. Adverse effects
921
of subthalamic nucleus DBS in a patient with multiple system atrophy. Neurology 61, 247 – 249. Tintner, R., Jankovic, J., 2002. Treatment options for Parkinson’s disease. Curr. Opin. Neurol. 15, 467 – 476. Van Laere, K., Santens, P., Bosman, T., De Reuck, J., Mortelmans, L., Dierckx, R., 2004. Statistical parametric mapping of (99m)Tc-ECD SPECT in idiopathic Parkinson’s disease and multiple system atrophy with predominant parkinsonian features: correlation with clinical parameters. J. Nucl. Med. 45, 933 – 942. Visser-Vandewalle, V., Temel, Y., Colle, H., van der Linden, C., 2003. Bilateral high-frequency stimulation of the subthalamic nucleus in patients with multiple system atrophy-parkinsonism. Report of four cases. J. Neurosurg. 98, 882 – 887. Wenning, G.K., Tison, F., Ben-Shlomo, Y., Daniel, S.E., Quinn, N., 1997. Multiple system atrophy: a review of 203 pathologically proven cases. Mov. Disord. 12, 133 – 147. Wenning, G.K., Geser, F., Stampfer-Kountchev, M., Tison, F., 2003. Multiple system atrophy: an update. Mov. Disord. 18, S34 – S42. Whone, A.L., Watts, R.L., Stoessl, A.J., et al., 2003. Slower progression of Parkinson’s disease with ropinirole versus levodopa: the REAL-PET study. Ann. Neurol. 54, 93 – 101.
FDG PET in the differential diagnosis of parkinsonian disorders Thomas Eckert,a,b Anna Barnes,a Vijay Dhawan,a,c,d Steve Frucht,e Mark F. Gordon,f Andrew S. Feigin,a,c,d and D. Eidelberga,c,d,* a
Center for Neurosciences, Institute for Medical Research, North Shore Long-Island Jewish Health System, Manhasset, NY 11030, USA Departments of Neurology II and Psychiatry, University of Magdeburg, Germany c Department of Neurology, North Shore University Hospital, Manhasset, NY 11030, USA d New York University School of Medicine, New York, NY 10016, USA e Neurologic Institute, Columbia-Presbyterian Medical Center, New York, NY 10032, USA f Department of Neurology, Long Island Jewish Medical Center, New Hyde Park, NY 11040, USA b
Received 18 November 2004; revised 3 February 2005; accepted 2 March 2005 Available online 26 April 2005 The differential diagnosis of parkinsonian disorders can be challenging, especially early in the disease course. PET imaging with [18F]fluorodeoxyglucose (FDG) has been used to identify characteristic patterns of regional glucose metabolism in patient cohorts with idiopathic Parkinson’s disease (PD), as well as variant forms of parkinsonism such as multiple system atrophy (MSA), progressive supranuclear palsy (PSP), and corticobasal degeneration (CBGD). In this study, we assessed the utility of FDG PET in the differential diagnosis of individual patients with clinical parkinsonism. 135 parkinsonian patients were referred for FDG PET to determine whether their diagnosis could be made accurately based upon their scans. Imaging-based diagnosis was obtained by visual assessment of the individual scans and also by computer-assisted interpretation. The results were compared with 2-year follow-up clinical assessments made by independent movement disorders specialists who were blinded to the original PET findings. We found that blinded computer assessment agreed with clinical diagnosis in 92.4% of all subjects (97.7% early PD, 91.6% late PD, 96% MSA, 85% PSP, 90.1% CBGD, 86.5% healthy control subjects). Concordance of visual inspection with clinical diagnosis was achieved in 85.4% of the patients scanned (88.4% early PD, 97.2% late PD, 76% MSA, 60% PSP, 90.9% CBGD, 90.9% healthy control subjects). This study demonstrates that FDG PET performed at the time of initial referral for parkinsonism accurately predicted the clinical diagnosis of individual patients made at subsequent follow-up. Computer-assisted methodologies may be particularly helpful in situations where experienced readers of FDG PET images are not readily available. D 2005 Elsevier Inc. All rights reserved. Keywords: Parkinson’s disease; Multiple system atrophy; Progressive supranuclear palsy; FDG PET; Differential diagnosis
* Corresponding author. Center for Neurosciences, Institute for Medical Research, North Shore-Long Island Jewish Health System, 350 Community Drive, Manhasset, NY 11030, USA. E-mail address: [email protected] (D. Eidelberg). Available online on ScienceDirect (www.sciencedirect.com). 1053-8119/$ - see front matter D 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.neuroimage.2005.03.012
Introduction Idiopathic Parkinson’s disease (PD) is a frequent diagnosis in the elderly (Moghal et al., 1994). However, only approximately 75% of parkinsonian patients prove to have PD at autopsy (Hughes et al., 1992b). The most common alternative diagnoses include multiple system atrophy (MSA), progressive supranuclear palsy (PSP), and corticobasal degeneration (CBGD) (Litvan, 1999). The precise differential diagnosis of parkinsonian syndromes is important because prognosis (Wenning et al., 1997) and treatment options (Tintner and Jankovic, 2002; Wenning et al., 2003) can differ substantially for PD and its clinical variants. However, differentiating parkinsonian syndromes by clinical means has proved unsatisfactory, especially early on in the course of the disease (Hughes et al., 2002; Litvan et al., 1998; Osaki et al., 2002). A variety of methods have been employed to improve the accuracy of differential diagnosis in patients with parkinsonism (Eckert et al., 2004). A number of studies using PET imaging with [18F]-fluorodeoxyglucose (FDG) have described characteristic patterns of glucose metabolism in patients with PD (Eidelberg et al., 1994; Moeller et al., 1999), MSA (Antonini et al., 1998; Eidelberg et al., 1993; Taniwaki et al., 2002), PSP (Blin et al., 1990; Foster et al., 1988), and CBGD (Blin et al., 1992; Eidelberg et al., 1991; Laureys et al., 1999). However, this technique has not been employed to assess diagnostic accuracy in individual patients with presumptive PD or variant forms of parkinsonism (e.g., Eckert and Eidelberg, 2004). To explore the utility of FDG PET in the ascertainment of single cases, we determined whether blinded scan interpretation was concordant with the ultimate clinical diagnosis. In a recent review of the clinical and neuropathological data from patients with parkinsonism (Hughes et al., 2002), the clinical impression of a movement disorder specialist proved to be more accurate than strict reliance upon specific sets of diagnostic criteria. We therefore selected the clinical diagnosis made by two independent movement disorder specialists as our ‘‘gold standard’’ for validating the imaging-based
T. Eckert et al. / NeuroImage 26 (2005) 912 – 921 Table 1 Clinical characteristics of control subjects and PD, MSA, PSP, and CBGD patients N Age at scan (male/female) in years, mean T SD in years Controls PD Early Late MSA PSP CBGD
Disease duration at scan in years, mean T SD in years
Follow-up time after FDG PET scanning in years, mean T SD in years –
22 (8/14)
56.3 T 11.5
–
43 36 25 20 11
56.7 60.2 59.6 69.0 68.0
T T T T T
2.2 9.8 3.9 2.9 3.3
(26/17) (19/17) (11/14) (10/10) (3/8)
10.9 8.9 8.9 7 9.4
T T T T T
1.2 3.5 1.4 1.2 1.0
2.1 1.9 2.3 1.9 2.3
T T T T T
1.1 0.7 1.6 0.6 1.5
Shown is a summary of clinical and demographical findings in controls and patients. Rows indicate the group (PD = idiopathic Parkinson’s disease, MSA = multiple system atrophy, PSP = progressive supranuclear palsy, CBGD = corticobasal degeneration).
diagnosis. In the present study, we compared the clinical categorization to blinded single case ascertainments made with FDG PET by Fvisual reading_ by a trained expert and by Fcomputer-supported reading_ by a non-expert using statistical parametric mapping (SPM) as a decision aid (Barnes et al., 2000; Signorini et al., 1999).
Methods Patients We studied 135 patients (age 60.6 T 10.4 years) referred for FDG PET imaging to assist in the differential diagnosis of parkinsonism. In these subjects, the clinical diagnosis was uncertain at the time of referral for imaging. This group of patients represented a subset of a larger group of 205 patients referred for PET imaging for the differential diagnosis of parkinsonism. Of the total cohort, 53 patients were excluded for lack of follow-up and 17 were excluded because of structural abnormalities on MRI. A probable clinical diagnosis was determined at follow-up by two independent movement disorder specialists (SF, TE) who were blinded to the imaging diagnosis (see below). The follow-up clinical assessments were conducted an average of 2.1 years after the initial PET imaging. Patients were included in this study only if a diagnosis of PD or a variant syndrome (MSA, PSP, or CBGD) could be achieved by the ascertaining clinicians at follow-up, and if no additional brain pathology (e.g., white matter changes, ischemia) was detected on routine MRI. Of the 135 patients, 43 were considered likely to have early stage PD (<5 years duration), 36 late stage PD (5 year duration), 25 MSA, 20 PSP, and 11 CBGD. In two of the patients thought to have MSA, the diagnosis was subsequently confirmed at postmortem. In addition, 22 healthy control subjects were included in the analysis. Demographic data for the patient groups and healthy control subjects are presented in Table 1. Ethical permission for this study was obtained from the Institutional Review Board of the North Shore-Long Island Jewish Health System. FDG PET imaging Subjects fasted overnight before PET scanning; antiparkinsonian medications were withheld at least 12 h before the PET
913
investigations. 185 – 370 MBq (5 – 10 mCi) FDG in 4 ml saline was injected intravenously. Subjects were scanned using the GE Advance tomograph (General Electric Medical Systems, Milwaukee, WI, USA) at North Shore University Hospital. This eight-ring bismuth germanate scanner provided 35 two-dimensional image planes with an axial field of view of 14.5 cm and transaxial resolution of 4.2 mm (full width half maximum) in all directions. The detailed performance characteristics of this instrument have been described elsewhere (Dhawan et al., 1998). All studies were performed with the subject’s eyes open in a dimly lit room with minimal auditory stimulation. Diagnostic evaluation Visual reading For routine visual reading of the PET images, the scans were reconstructed, corrected for attenuation, and smoothed for each subject. FDG PET images were displayed as a series of 35 transaxial slices scaled to a common maximum in a standard color scale. In addition, a composite slice was generated by summing the central four of the six slices in which the striatum was visualized (thickness approximately 16 mm). Striatal and subcortical structures are compared visually to the activity in the overall cortical strip, cerebellum, and occipital cortex by an expert reader (VD) blinded to the clinical information. Diagnostic criteria are based on the published literature (for a review, see Dhawan and Eidelberg, 2003). These feature decreased metabolism of the basal ganglia and cerebellum for MSA, medial frontal decreases especially at superior levels for PSP, marked asymmetric cortical and basal ganglia metabolism for CBGD, and increased metabolism of the lentiform nucleus in relation to the cortex in clinical PD. Each single patient FDG PET image was diagnosed as either a control subject, PD, MSA, PSP, or CBGD patient. Computer-supported reading Image processing and analysis All image processing and analyses were performed using Statistical Parametric Mapping (SPM99, Wellcome Department of Cognitive Neurology, London, UK) running in MATLAB (R12, Mathworks, Natick, MA). The images were spatially normalized after which an isotropic 10 mm full width half maximum smoothing filter was applied to the images (Poline et al., 1995). Creating the characteristic disease templates We developed template statistical maps to aid in PET scan interpretation. This was accomplished by randomly selecting eight subjects from each of the PD, MSA, and PSP groups and five members from the smaller CBGD group. These patient cohorts were cross-matched for age and disease duration. Scans from each of the cohorts were compared to those from 10 agematched control subjects using the general linear model in the SPM99 (Friston et al., 1995). For the PD and CBGD groups, images of predominately left side affected subjects were flipped, so that all affected sides were on the right. Analysis of covariance (ANCOVA) was used for the purposes of intensity normalization to account for differences in global flow across subjects. A threshold of 80% of the mean global counts was used to identify voxels representative of gray matter and therefore to be included in the statistical analysis. A t test was then performed at all these voxels. Any voxels showing increased or decreased glucose
914
T. Eckert et al. / NeuroImage 26 (2005) 912 – 921
metabolism in the patient groups compared to the control group above the statistical threshold of P < 0.01 (peak and extent uncorrected for multiple statistical comparisons) were overlaid onto the T1 MR template image provided by the SPM99 software (Friston et al., 1996). This threshold was chosen because it yielded the maps of abnormal regional glucose metabolism that conformed closely to published patterns for each of the
syndromes. These canonical maps (see Fig. 1) were then used as templates to assist in the differential diagnosis of single scans. Characterizing single subject scans To characterize single patient scans, each patient or control subject’s PET image was compared statistically to the reference group of 10 healthy control subjects providing a map of statistically
Fig. 1. Detailed templates (whole brain) of abnormal glucose metabolism in patients with (A) PD, (B) MSA, (C) PSP, and (D) CBGD identified by group analysis using SPM ( P < 0.05). At this relaxed threshold the defining and supportive features become evident. Increased glucose metabolism is indicated by Fhot_ colors and decreased metabolism by Fwinter_ colors. For image analysis, the hemispheres opposite the clinically more affected body sides appeared on the right (for larger views, see www.neuroscience-nslij.org).
T. Eckert et al. / NeuroImage 26 (2005) 912 – 921
915
Fig. 1 (continued).
different regions of glucose metabolism in the manner described above. Similarly all voxels showing increased or decreased glucose metabolism at a threshold of P < 0.05 (peak and extent uncorrected for multiple statistical comparisons) compared to the control group were also overlaid onto the T1 MRI template image provided by SPM99 and saved as a portable document format (PDF) image to be read by a non-expert. Increased glucose metabolism is shown in Fhot_ colors and decreased glucose metabolism in Fwinter_ colors. The PDF images were viewed on a standard PC screen by a nonexpert reader (AB) blinded to all patient’s clinical information.
Prior to the evaluation, the reader was trained in the evaluation method of the single patient statistical maps using a newly developed set of criteria for differentiating parkinsonism syndromes based upon characteristic patterns of regional metabolism (see Table 2). These criteria were formulated on the basis of descriptions of abnormal glucose utilization in PD (Eidelberg et al., 1994; Moeller et al., 1999), MSA (Antonini et al., 1998; Eidelberg et al., 1993; Taniwaki et al., 2002), PSP (Blin et al., 1990; Foster et al., 1988), and CBGD (Blin et al., 1992; Eidelberg et al., 1991; Laureys et al., 1999) patients. The reader compared each single patient PDF image to the
916
T. Eckert et al. / NeuroImage 26 (2005) 912 – 921
Fig. 1 (continued).
defining features of these criteria and each PDF image was then characterized as either PD, MSA, PSP, or CBGD. If no defining features of disease were apparent, the scan was classified as belonging to the control group. The non-expert did not view the original PET scans. The single patient imaging diagnosis of Fvisual reading_ and FSPM-supported reading_ were then compared to each patient’s final clinical diagnosis (i.e., the diagnostic Fgold standard_). The sensitivity and specificity of both image-based methods were then calculated from these results using a 2 2 table method (Lijmer et al., 1999).
Results Characteristic patterns of abnormal regional metabolism PD We found that the hallmark of glucose metabolism in the PD cohort was increased metabolism in the putamen and globus pallidus (Fig. 1A). The finding was present bilaterally without regard to the more affected side. Hypermetabolic areas are also observed in the ventral thalamus, and in the motor cortex and cerebellum. Abnormal reductions in glucose metabolism in the
T. Eckert et al. / NeuroImage 26 (2005) 912 – 921
917
Fig. 1 (continued).
PD group were present bilaterally in parietal and occipital association areas, and in the dorsolateral prefrontal cortex (DLPFC). MSA The MSA cohort exhibited a metabolic pattern characterized by marked bilateral reductions in the lentiform nuclei and in the cerebellum (see Fig. 1B and Table 2). Metabolic reductions in the lentiform nucleus were seen in all presumptive MSA patients. While glucose metabolism of the cerebellum was reduced in all patients with clinical ataxia, it was also reduced in a number of
subjects without signs of cerebellar dysfunction. Reduced regional metabolism was also observed in the brainstem. Relative metabolic increases were noted bilaterally in the thalamus; and in parietal, frontal, and temporal cortical regions. PSP The distinguishing feature of the PSP group was the presence of metabolic decrements in midline frontal regions and in the brainstem (see Fig. 1C and Table 2). Decreased metabolism was also noted in the superior frontal and insular areas, and in the caudate nucleus. Relative metabolic increases were observed in the
918
T. Eckert et al. / NeuroImage 26 (2005) 912 – 921
Table 2 Characteristic glucose metabolism patterns of parkinsonian syndromes, defining features are used for classification of FSPM-supported diagnosis_ of single patients Parkinson’s disease Defining features & hypermetabolism of the dorsolateral putamen & no defining features of other parkinsonian syndromes Supporting features & thalamic hypermetabolism & cerebellar hypermetabolism & bilateral hypermetabolism of cortical motor areas & bilateral hypometabolism of the parietal cortex Multiple system atrophy Defining features & bilateral hypometabolism of the putamen & cerebellar hypometabolism Supporting features & bilateral hypermetabolism of frontal and superior parietal cortical areas & bilateral hypermetabolism of the thalamus & hypometabolism of the brainstem Progressive supranuclear palsy Defining features & hypometabolism of the brainstem & hypometabolism of the midline frontal cortex & no defining features of MSA or CBGD Supporting features & bilateral hypermetabolism of cortical motor areas & bilateral hypermetabolism of the parietal cortex & bilateral hypermetabolism of the thalamus & bilateral hypometabolism of the caudate nucleus Corticobasal degeneration Defining features & asymmetrical cortical activation: relative hypometabolism contralateral to the most affected side & asymmetrical basal ganglia metabolism: relative hypometabolism contralateral to the most affected side Supporting features & one-sided cortical hypometabolism, especially of the insular cortex & hypometabolism of paraventricular areas & hypometabolism of the brain stem and midline frontal areas
thalamus, and in parietal, occipital, and temporal regions, as well as in motor cortex and infero-lateral frontal regions. CBGD The pattern of glucose metabolism in CBGD was characterized by a distinctive asymmetrical distribution of radiotracer uptake. Relative metabolic reductions were present in many cortical areas, including the insula and in the basal ganglia contralateral to the most affected side (see Fig. 1D and Table 2). Conversely, relatively increased glucose metabolism in the CBGD group compared to controls was observed in the cortex and basal ganglia ipsilateral to the most affected side. Diagnostic classification of single patient scans FVisual reading_ Using the Fvisual reading_ method to evaluate PET images, imaging diagnosis accorded with the clinical diagnosis in 85.4% of all subjects. Specifically, this concordance was 88.4% for early PD, 97.2% for late PD, 76% for MSA, 60% for PSP, and 90.9% for CBGD. 90.9% of healthy volunteers were correctly categorized by
this method of scan interpretation (for detailed information, see Tables 3 and 4). A sensitivity of 100% and specificity of 91% was obtained in identifying early PD patients (<5 years duration of symptoms) with respect to normal control subjects. Similarly, a sensitivity of 92% and specificity of 95% was achieved for PD patients versus subjects with presumed atypical parkinsonian syndromes. FSPM-supported reading_ Using the FSPM-supported reading_ method to evaluate PET images based on the classification criteria described in Table 2, correct imaging diagnosis was obtained in 92.4% of all subjects. Concordance with clinical diagnosis was 97.7% for early PD, 91.6% for late PD, 96% for MSA, 85% for PSP, 90.1% for CBGD, and 86.5% for healthy control subjects (for detailed information, see Tables 3 and 4). A sensitivity of 100% and specificity of 86% was obtained in identifying early PD patients (<5 years duration of symptoms) as compared with the normal controls. A sensitivity of 96% and specificity of 91% was achieved for PD patients as compared with patients with presumed atypical parkinsonism.
Discussion The present study has shown that FDG PET has a very good ability to differentiate PD, MSA, PSP, and CBGD patients as well as healthy control subjects on a single case basis. There have been a number of previous studies that have identified characteristic glucose metabolism patterns in patients with PD (Eidelberg et al., 1994; Moeller et al., 1999), MSA (Antonini et al., 1998; De Volder et al., 1989; Eidelberg et al., 1993; Otsuka et al., 1996; Taniwaki et al., 2002), PSP (Blin et al., 1990; Foster et al., 1988; Piccini et al., 2001), and CBGD (Blin et al., 1992; Eidelberg et al., 1991; Laureys et al., 1999) with respect to normal control subjects. These earlier FDG PET studies focused on the description of group comparisons of glucose metabolism between patient and control groups, without application to the differential diagnosis of individual subjects. The current study used SPM to create disease-related templates in which regional features were in close agreement with those previously reported. Our single case approach showed that these patterns can potentially be used diagnostically in that correct classification can be achieved relative to the final clinical assessment in over 90% of subjects. Correct diagnosis of patients with parkinsonism is of practical importance, especially at early stages of disease. If possible, these patients should be informed about their prognosis, which may differ considerably depending upon differential diagnosis (Diamond et al., 1987; Golbe et al., 1988; Wenning et al., 1997). Additionally, there are major differences in treatment options. While PD patients tend to demonstrate ongoing responsiveness to dopaminergic medication (Martin and Wieler, 2003; Olanow, 2002; Tintner and Jankovic, 2002), atypical parkinsonian patients are quite variable in this regard (Hughes et al., 1992a; Parati et al., 1993; Wenning et al., 2003). PD and its variants may also differ with regard to surgical response. Deep brain stimulation (DBS) and ablation of the subthalamic nucleus (STN) or the internal globus pallidus (GP) are well-established approaches for the treatment of advanced PD (Krack et al., 2000). However, patients with atypical parkinsonian syndromes appear not to profit from these surgical interventions (Chou et al., 2004; Lezcano et al., 2004; Tarsy et al., 2003; VisserVandewalle et al., 2003). Correct diagnosis can be used to optimize
T. Eckert et al. / NeuroImage 26 (2005) 912 – 921
919
Table 3 Diagnostic classification matrix for visual reading and SPM-supported reading Clinical classification (n)
Visual reading Controls
PD
Controls (22) PD early (43) PD late (36) MSA (25) PSP (20) CBGD (11)
20 0 0 1 0 1
0 38 35 0 1 0
Clinical classification (n)
SPM-supported reading
Controls (22) PD early (43) PD late (36) MSA (25) PSP (20) CBGD (11)
(90.9%) (0%) (0%) (4%) (0%) (9.1%)
Controls
PD
19 0 1 1 0 0
1 42 33 0 3 1
(86.5%) (0%) (2.8%) (4%) (0%) (0%)
(0%) (88.4%) (97.2%) (0%) (5%) (0%)
(4.5%) (97.7%) (91.6%) (0%) (15%) (9.1%)
MSA
PSP
CBGD
0 0 0 19 3 0
2 2 0 1 12 0
0 3 1 4 4 10
(0%) (0%) (0%) (76%) (15%) (0%)
(9.1%) (4.6%) (0%) (4%) (60%) (0%)
(0%) (7%) (2.8%) (16%) (20%) (90.9%)
MSA
PSP
CBGD
1 0 0 24 0 0
1 1 0 0 17 0
0 0 2 0 0 10
(4.5%) (0%) (0%) (96%) (0%) (0%)
(4.5%) (2.3%) (0%) (0%) (85%) (0%)
(0%) (0%) (5.6%) (0%) (0%) (90.1%)
Overall correct diagnosis: visual reading 85.4%, SPM suppported reading 92.4%. Summary of the image guided diagnosis using visual reading and SPMsupported reading. Shown is the classification of subjects due to the FDG PET imaging guided diagnosis with respect to the clinical diagnosis. Rows represent the clinical diagnosis and columns the diagnosis predicted by FDG PET imaging. Bold style indicates correct diagnosis (PD = idiopathic Parkinson’s disease, MSA = multiple system atrophy, PSP = progressive supranuclear palsy, CBGD = corticobasal degeneration).
treatment and to reduce the risks attendant to potentially ineffective treatment strategies. Our case-by-case study shows that considerable accuracy can be achieved with FDG PET to discriminate classical from potentially less responsive forms of parkinsonism. Accurate differential diagnosis is also of particular importance in the conduct of treatment trials in parkinsonism. Inadvertent inclusion of atypical patients into pharmacological trials for PD is likely to reduce statistical power by increasing the heterogeneity of the treatment cohorts. Trials of potential neuroprotective agents for PD (Parkinson-Study-Group, 2002; Whone et al., 2003; Fahn et al., 2004) have focused on early stage disease, in which progression may be more rapid (Bonnet et al., 1987; Fearnley and Lees, 1991; Lee et al., 1994). However, differential diagnosis of parkinsonism based on clinical features is especially challenging at early disease stages. A recent neuropathological study revealed that while correct clinical diagnosis at follow-up was 85%, the initial clinical impression was revised in 60% of subjects (Hughes et al., 2002). Another postmortem study showed that the sensitivity for correct clinical diagnosis of MSA at the initial clinical visit was only 22% (Osaki et al., 2002). Our data suggest that accurate differential diagnosis in patients with parkinsonism supported by the use of imaging techniques as FDG PET may improve the power of future clinical trials by promoting group homogeneity. Our newly defined diagnostic FDG imaging criteria (Table 2) provides a useful diagnostic imaging classification for individual patients. Indeed, the results of Fcomputer-supported reading_ with an overall correct diagnosis of 92.4% by a non-expert investigator were superior to the results of Fvisual reading_ with an overall correct diagnosis of 85.4% by an experienced interpreter of neurological PET scans. Applying this SPM-based approach, it is possible to normalize the PET images, to statistically compare them to a group of images from healthy controls, and to overlay the resulting maps of abnormal glucose utilization so as to make them easily visible for single case interpretation. The procedures used in this study were all automated, blinded, and investigator independent. Performing a voxel-based comparison has the distinct
advantage of not relying upon prior assumptions regarding the anatomical distribution of the metabolic abnormalities associated with each syndrome. By contrast, visual evaluation of scans depends on the experience of the reader in qualitatively assessing neurological PET scans. Our results suggest that Fcomputersupported reading_ may actually be more accurate and requires less examiner knowledge than visual inspection by a trained expert. This attribute may enhance the utility of FDG PET in non-expert imaging centers. In this respect, fully automated, investigatorindependent methods of image evaluation are desirable, although these ‘‘black box’’ techniques generally require a suitable FDG control group. SPM-supported image analysis for single patients can potentially be used with other neuroimaging techniques, such as [99mTc]-ethylene cysteinate dimer (ECD) SPECT (Feigin et al.,
Table 4 Diagnostic values of FDG PET using visual reading and SPM-supported reading Visual reading
PPV NPV Sensitivity Specificity
PD
MSA
PSP
CBGD
Controls
0.99 0.93 0.92 0.99
0.86 0.96 0.76 0.98
0.70 0.94 0.60 0.96
0.45 0.99 0.91 0.92
0.91 0.99 0.91 0.99
SPM-supported reading
PPV NPV Sensitivity Specificity
PD
MSA
PSP
CBGD
Controls
0.94 0.95 0.95 0.94
0.96 0.99 0.96 0.99
0.89 0.98 0.85 0.99
0.83 0.99 0.91 0.99
0.90 0.98 0.86 0.99
Diagnostic values of correct imaging diagnosis using visual reading and SPM-supported reading referred to the clinical diagnosis (PD = idiopathic Parkinson’s disease, MSA = multiple system atrophy, PSP = progressive supranuclear palsy, CBGD = corticobasal degeneration, PPV = positive predictive value, NPV = negative predictive value).
920
T. Eckert et al. / NeuroImage 26 (2005) 912 – 921
2002; Van Laere et al., 2004), perfusion MRI or magnetization transfer imaging (Eckert et al., 2004). In summary, this study demonstrates the utility of FDG PET imaging in the differential diagnosis of patients with parkinsonism using an evidence-based medicine approach. This new method may contribute to early differential diagnosis in clinically ambiguous cases of parkinsonism, or to confirm a diagnosis of PD in therapeutic trials. The development of fully automated diagnostic techniques is likely to be valuable in future studies of PD and related disorders. However, we would like to emphasize that this new approach cannot substitute for clinical evaluation in the assessment of parkinsonian disorders.
References Antonini, A., Kazumata, K., Feigin, A., et al., 1998. Differential diagnosis of parkinsonism with [18F]fluorodeoxyglucose and PET. Mov. Disord. 13, 268 – 274. Barnes, A., Lusman, D., Patterson, J., Brown, D., Wyper, D., 2000. The use of statistical parametric mapping (SPM96) as a decision aid in the differential diagnosis of dementia using 99mTc-HMPAO SPECT. Behav. Neurol. 12, 77 – 86. Blin, J., Baron, J.C., Dubois, B., et al., 1990. Positron emission tomography study in progressive supranuclear palsy. Brain hypometabolic pattern and clinicometabolic correlations. Arch. Neurol. 47, 747 – 752. Blin, J., Vidailhet, M.J., Pillon, B., Dubois, B., Feve, J.R., Agid, Y., 1992. Corticobasal degeneration: decreased and asymmetrical glucose consumption as studied with PET. Mov. Disord. 7, 348 – 354. Bonnet, A.M., Loria, Y., Saint-Hilaire, M.H., Lhermitte, F., Agid, Y., 1987. Does long-term aggravation of Parkinson’s disease result from nondopaminergic lesions? Neurology 37, 1539 – 1542. Chou, K.L., Forman, M.S., Trojanowski, J.Q., Hurtig, H.I., Baltuch, G.H., 2004. Subthalamic nucleus deep brain stimulation in a patient with levodopa-responsive multiple system atrophy, Case report. J. Neurosurg. 100, 553 – 556. De Volder, A.G., Francart, J., Laterre, C., et al., 1989. Decreased glucose utilization in the striatum and frontal lobe in probable striatonigral degeneration. Ann. Neurol. 26, 239 – 247. Dhawan, V., Eidelberg, D., 2003. PET imaging in Parkinson’s disease. In: Sinha, K.K., Chandra, P., Jha, D.K. (Eds.), Advances in Clinical Neurosciences. The Catholic Press, Jharkhand, India, pp. 251 – 276. Dhawan, V., Kazumata, K., Robeson, W., et al., 1998. Quantitative Brain PET. Comparison of 2D and 3D Acquisitions on the GE Advance Scanner. Clin. Positron Imaging 1, 135 – 144. Diamond, S.G., Markham, C.H., Hoehn, M.M., McDowell, F.H., Muenter, M.D., 1987. Multi-center study of Parkinson mortality with early versus later dopa treatment. Ann. Neurol. 22, 8 – 12. Eckert, T., Eidelberg, D., 2004. The role of functional neuroimaging in the differential diagnosis of idiopathic Parkinson’s disease and multiple system atrophy. Clin. Auton. Res. 14, 84 – 91. Eckert, T., Sailer, M., Kaufmann, J., et al., 2004. Differentiation of idiopathic Parkinson’s disease, multiple system atrophy, progressive supranuclear palsy, and healthy controls using magnetization transfer imaging. NeuroImage 21, 229 – 235. Eidelberg, D., Dhawan, V., Moeller, J.R., et al., 1991. The metabolic landscape of cortico-basal ganglionic degeneration: regional asymmetries studied with positron emission tomography. J. Neurol., Neurosurg. Psychiatry 54, 856 – 862. Eidelberg, D., Takikawa, S., Moeller, J.R., et al., 1993. Striatal hypometabolism distinguishes striatonigral degeneration from Parkinson’s disease. Ann. Neurol. 33, 518 – 527. Eidelberg, D., Moeller, J.R., Dhawan, V., et al., 1994. The metabolic topography of parkinsonism. J. Cereb. Blood Flow. Metab. 14, 783 – 801.
Fahn, S., Oakes, D., Shoulson, I., Kieburtz, K., Rudolph, A., Lang, A., et al., 2004. Levodopa and the progression of Parkinson’s disease. N. Engl. J. Med. 351, 2498 – 2508. Fearnley, J.M., Lees, A.J., 1991. Ageing and Parkinson’s disease: substantia nigra regional selectivity. Brain 114, 2283 – 2301. Feigin, A., Antonini, A., Fukuda, M., et al., 2002. Tc-99m ethylene cysteinate dimer SPECT in the differential diagnosis of parkinsonism. Mov. Disord. 17, 1265 – 1270. Foster, N.L., Gilman, S., Berent, S., Morin, E.M., Brown, M.B., Koeppe, R.A., 1988. Cerebral hypometabolism in progressive supranuclear palsy studied with positron emission tomography. Ann. Neurol. 24, 399 – 406. Friston, K.J., Holmes, A.P., Worsley, K.J., Poline, J.P., Frith, C.D., Frackowiak, R.S.J., 1995. Statistical parametric maps in functional imaging: a general linear approach. Hum. Brain Mapp. 2, 189 – 210. Friston, K.J., Holmes, A., Poline, J.B., Price, C.J., Frith, C.D., 1996. Detecting activations in PET and fMRI: levels of inference and power. NeuroImage 4, 223 – 235. Golbe, L.I., Davis, P.H., Schoenberg, B.S., Duvoisin, R.C., 1988. Prevalence and natural history of progressive supranuclear palsy. Neurology 38, 1031 – 1034. Hughes, A.J., Colosimo, C., Kleedorfer, B., Daniel, S.E., Lees, A.J., 1992a. The dopaminergic response in multiple system atrophy. J. Neurol., Neurosurg. Psychiatry 55, 1009 – 1013. Hughes, A.J., Daniel, S.E., Kilford, L., Lees, A.J., 1992b. Accuracy of clinical diagnosis of idiopathic Parkinson’s disease: a clinico-pathological study of 100 cases. J. Neurol., Neurosurg. Psychiatry 55, 181 – 184. Hughes, A.J., Daniel, S.E., Ben-Shlomo, Y., Lees, A.J., 2002. The accuracy of diagnosis of parkinsonian syndromes in a specialist movement disorder service. Brain 125, 861 – 870. Krack, P., Poepping, M., Weinert, D., Schrader, B., Deuschl, G., 2000. Thalamic, pallidal, or subthalamic surgery for Parkinson’s disease? J. Neurol. 247, 122 – 134. Laureys, S., Salmon, E., Garraux, G., et al., 1999. Fluorodopa uptake and glucose metabolism in early stages of corticobasal degeneration. J. Neurol. 246, 1151 – 1158. Lee, C.S., Schulzer, M., Mak, E.K., et al., 1994. Clinical observations on the rate of progression of idiopathic parkinsonism. Brain 117, 501 – 507. Lezcano, E., Gomez-Esteban, J.C., Zarranz, J.J., et al., 2004. Parkinson’s disease-like presentation of multiple system atrophy with poor response to STN stimulation: a clinicopathological case report. Mov. Disord. 19, 973 – 977. Lijmer, J.G., Mol, B.W., Heisterkamp, S., et al., 1999. Empirical evidence of design-related bias in studies of diagnostic tests. JAMA 282, 1061 – 1066. Litvan, I., 1999. Recent advances in atypical parkinsonian disorders. Curr. Opin. Neurol. 12, 441 – 446. Litvan, I., Booth, V., Wenning, G.K., et al., 1998. Retrospective application of a set of clinical diagnostic criteria for the diagnosis of multiple system atrophy. J. Neural Transm. 105, 217 – 227. Martin, W.R., Wieler, M., 2003. Treatment of Parkinson’s disease. Can. J. Neurol. Sci. 30, S27 – S33. Moeller, J.R., Nakamura, T., Mentis, M.J., et al., 1999. Reproducibility of regional metabolic covariance patterns: comparison of four populations. J. Nucl. Med. 40, 1264 – 1269. Moghal, S., Rajput, A.H., D’Arcy, C., Rajput, R., 1994. Prevalence of movement disorders in elderly community residents. Neuroepidemiology 13, 175 – 178. Olanow, C.W., 2002. Surgical therapy for Parkinson’s disease. Eur. J. Neurol. 9, 31 – 39. Osaki, Y., Wenning, G.K., Daniel, S.E., et al., 2002. Do published criteria improve clinical diagnostic accuracy in multiple system atrophy? Neurology 59, 1486 – 1491. Otsuka, M., Ichiya, Y., Kuwabara, Y., et al., 1996. Glucose metabolism in the cortical and subcortical brain structures in multiple system atrophy and Parkinson’s disease: a positron emission tomographic study. J. Neurol. Sci. 144, 77 – 83.
T. Eckert et al. / NeuroImage 26 (2005) 912 – 921 Parati, E.A., Fetoni, V., Geminiani, G.C., et al., 1993. Response to l-DOPA in multiple system atrophy. Clin. Neuropharmacol. 16, 139 – 144. Parkinson-Study-Group, 2003. Does levodopa slow or hasten the rate of progression of Parkinson disease? The result of the elldopa trial. Am. Acad. Neurol. 60, A80 – A81. Piccini, P., de Yebenez, J., Lees, A.J., et al., 2001. Familial progressive supranuclear palsy: detection of subclinical cases using 18F-dopa and 18 Fluorodeoxyglucose positron emission tomography. Arch. Neurol. 58, 1846 – 1851. Poline, J.B., Worsley, K.J., Holmes, A.P., Frackowiak, R.S., Friston, K.J., 1995. Estimating smoothness in statistical parametric maps: variability of P values. J. Comput. Assist. Tomogr. 19, 788 – 796. Signorini, M., Paulesu, E., Friston, K., et al., 1999. Rapid assessment of regional cerebral metabolic abnormalities in single subjects with quantitative and nonquantitative [18F]FDG PET: a clinical validation of statistical parametric mapping. NeuroImage 9, 63 – 80. Taniwaki, T., Nakagawa, M., Yamada, T., et al., 2002. Cerebral metabolic changes in early multiple system atrophy: a PET study. J. Neurol. Sci. 200, 79 – 84. Tarsy, D., Apetauerova, D., Ryan, P., Norregaard, T., 2003. Adverse effects
921
of subthalamic nucleus DBS in a patient with multiple system atrophy. Neurology 61, 247 – 249. Tintner, R., Jankovic, J., 2002. Treatment options for Parkinson’s disease. Curr. Opin. Neurol. 15, 467 – 476. Van Laere, K., Santens, P., Bosman, T., De Reuck, J., Mortelmans, L., Dierckx, R., 2004. Statistical parametric mapping of (99m)Tc-ECD SPECT in idiopathic Parkinson’s disease and multiple system atrophy with predominant parkinsonian features: correlation with clinical parameters. J. Nucl. Med. 45, 933 – 942. Visser-Vandewalle, V., Temel, Y., Colle, H., van der Linden, C., 2003. Bilateral high-frequency stimulation of the subthalamic nucleus in patients with multiple system atrophy-parkinsonism. Report of four cases. J. Neurosurg. 98, 882 – 887. Wenning, G.K., Tison, F., Ben-Shlomo, Y., Daniel, S.E., Quinn, N., 1997. Multiple system atrophy: a review of 203 pathologically proven cases. Mov. Disord. 12, 133 – 147. Wenning, G.K., Geser, F., Stampfer-Kountchev, M., Tison, F., 2003. Multiple system atrophy: an update. Mov. Disord. 18, S34 – S42. Whone, A.L., Watts, R.L., Stoessl, A.J., et al., 2003. Slower progression of Parkinson’s disease with ropinirole versus levodopa: the REAL-PET study. Ann. Neurol. 54, 93 – 101.