Changes in network activity with the progression of
Chaorui Huang,1,2ChengkeTang,1,2Andrew Feigin,1,2Martin Lesser,3Yilong Ma,1,2Michael Pourfar,1,2
Vijay Dhawan1,2and David Eidelberg1,2
1Center for Neurosciences,The Feinstein Institute for Medical Research, North Shore-Long Island Jewish Health System,
Manhasset, NY,2Departments of Neurology and Medicine, North Shore University Hospital and NewY ork University
School of Medicine, NewYork, NYand3Biostatistics Unit,The Feinstein Institute for Medical Research, North Shore-Long
Island Jewish Health System, Manhasset, NY,USA
Correspondence to: David Eidelberg, MD, Center for Neurosciences, Feinstein Institute for Medical Research, North
Shore-Long Island Jewish Health System, 350 Community Drive, Manhasset, NY11030,USA
Parkinson’s disease (PD) is associated with abnormal activity in spatially distributed neural systems mediating
the motor and cognitive manifestations of this disorder. Metabolic PET studies have demonstrated that this
illness is characterized by a set of reproducible functional brain networks that correlate with these clinical
features.The time at which these abnormalities appearis unknown, as is theirrelationship to concurrent clinical
and dopaminergic indices of disease progression.
In this longitudinal study,15 early stage PD patients (age 58.0?10.2 years;Hoehn and Yahr Stage1.2?0.3) were
enrolled within 2 years of diagnosis. The subjects underwent multitracer PET imaging at baseline, 24 and
48 months. At each timepoint they were scanned with [18F]-fluorodeoxyglucose (FDG) to assess longitudinal
changes in regional glucose utilization and in the expression of the PD-related motor (PDRP) and cognitive
metabolic covariance patterns (PDCP). At each timepoint the subjects also underwent PET imaging with
[18F]-fluoropropyl ?CIT (FP-CIT) to quantify longitudinal changes in caudate and putamen dopamine transpor-
ter (DAT) binding.Regional metabolic changes across the three timepoints were localized using statistical para-
metric mapping (SPM). Longitudinal changes in regional metabolism and network activity, caudate/putamen
DAT binding, and Unified Parkinson’s Disease Rating Scale (UPDRS) motor ratings were assessed using
repeated measures analysis of variance (RMANOVA). Relationships between these measures of disease
progression were assessed by computing within-subject correlation coefficients.
We found that disease progression was associated with increasing metabolism in the subthalamic nucleus (STN)
and internal globus pallidus (GPi) (P50.001), as well as in the dorsal pons and primary motor cortex (P50.0001).
Advancing disease was also associated with declining metabolism in the prefrontal and inferior parietal regions
(P50.001).PDRPexpressionwas elevated atbaselinerelative tohealthycontrolsubjects (P50.04), andincreased
progressivelyover time (P50.0001).PDCP activity alsoincreased with time (P50.0001).However, these changes
in network activity were slower than for the PDRP (P50.04), reaching abnormal levels only at the final
timepoint. Changes in PDRP activity, but not PDCP activity, correlated with concurrent declines in striatal
DAT binding (P50.01) and increases in motor ratings (P50.005). Significant within-subject correlations
(P50.01) were also evident between the latter two progression indices.
The early stages of PD are associated with progressive increases and decreases in regional metabolism at key
nodes of the motor and cognitive networks that characterize the illness. Potential disease-modifying therapies
may alter the time course of one or both of these abnormal networks.
Keywords: Parkinson’s disease; PET;18F-fluorodeoxyglucose (FDG); glucose metabolism; network analysis
Abbreviations: SPM¼statistical parametric mapping; FDR¼false discovery rate; STN¼subthalamic nucleus
Received November 3. 2006. Revised March16, 2007 . Accepted March 20, 2007
doi:10.1093/brain/awm086 Brain (2007) Page1of13
? The Author (2007).Publishedby Oxford University Pressonbehalfofthe Guarantorsof Brain. Allrightsreserved.For Permissions, please email: firstname.lastname@example.org
Brain Advance Access published April 30, 2007
by guest on June 1, 2013
Accurate and comprehensive descriptions of the natural
history of Parkinson’s disease (PD) are critical to the
assessment of new therapies for this disorder. Knowledge of
the rate of disease progression, particularly at early phases
of the illness, is essential for the design of clinical trials
aimed at evaluating potential neuroprotective treatment
strategies. However, such determinations can be difficult
when based solely upon clinical assessments, especially if
different manifestations of disease do not evolve in parallel.
For instance, the rate of progression in standardized clinical
ratings is different for PD patients with tremor- and gait-
predominant presentations (Jankovic and Kapadia, 2001).
These subpopulations also appear to differ in their propen-
sity to develop cognitive impairment (Zetusky et al., 1985;
Alves et al., 2006). Additionally, the rate of progression for
a given clino-pathological feature may not be constant.
Indeed, this phenomenon has been observed in PD, with
accelerated nigral dopamine loss at early stages of the
disease (Fearnley and Lees, 1991; cf. Morrish et al., 1998;
Hilker et al., 2003). Furthermore, at a specific clinical stage,
progression can vary across individual subjects depending
upon demographic factors such as age of onset (cf.
Nakamura et al., 2001; Alves et al., 2005). The observed
rate of clinical deterioration can also be affected by
concurrent symptomatic therapy, especially when the effects
of treatment are long lasting (Fahn et al., 2004).
Because of these considerations, attention has turned
toward the use of imaging biomarkers as a potentially more
objective and accurate means of gauging disease progression
(see e.g. Brooks et al., 2003; Au et al., 2005 for reviews).
nigrostriatal dopaminergic functioning have been included
in several recent clinical trials designed to evaluate disease
modifying therapies (Marek et al., 2003; Whone et al., 2003;
Fahn et al., 2004). However, the conflicting nature of the
findings has suggested a complex relationship between the
clinical and imaging-based measures (Ravina et al., 2005).
Furthermore, while suitable for the in vivo assessment of
nigrostriatal function, these radiotracer approaches are not
designed to capture the effects of disease on physiologically
relevant neural systems, including
Other imaging strategies can be used to assess changes
analysis has been used extensively with metabolic imaging
to study the pathophysiology of parkinsonism and its
treatment (e.g. Asanuma et al., 2006; Tros ˇt et al., 2006; for
reviews see Carbon et al., 2003a; Eckert and Eidelberg,
2005). The motor features of PD are associated with the
expression of an abnormal metabolic pattern characterized
by increased pallido-thalamic and pontine activity, asso-
ciated withrelative reductions
regions (Eidelberg et al., 1994; 1997; cf. Carbon et al.,
2003a). This PD-related spatial covariance pattern (PDRP)
has thus far been detected in eight independent patient
populations (Moeller et al., 1999; Feigin et al., 2002; Lozza
et al., 2004; Asanuma et al., 2005; Eckert et al., 2007).
Moreover, we have recently demonstrated that PDRP
expression is highly reproducible in individual subjects,
with stable network activity over hours to weeks (Ma et al.,
2007). In addition to providing accurate discrimination
between PD patients and controls (Asanuma et al., 2005;
Eckert et al., 2007; Ma et al., 2007), PDRP expression has
been found consistently to correlate with standardized
motor ratings (Lozza et al., 2004; Asanuma et al., 2006;
cf. Eidelberg et al., 1994, 1995) as well as with symptom
duration (Moeller and Eidelberg, 1997). Thus, this measure
of pathological network activity can provide a useful
quantitative descriptor of advancing motor dysfunction
Network analysis has also provided unique insights into
the mechanisms underlying abnormal cognitive functioning
in PD. In a study of regional glucose metabolism in non-
demented patients, we used a covariance mapping approach
to identify a significant pattern associated with executive
dysfunction (Mentis et al., 2002). These FDG PET data
were the basis for a recent comprehensive spatial covariance
analysis in which neuropsychological performance in PD
patients was found to be correlated with the activity of
a distinct metabolic network that was topographically
unrelated to the PDRP (Huang et al., 2007; cf. Lozza
et al., 2004). This PD-related cognitive pattern (PDCP) was
characterized by reduced metabolic activity in prefrontal
and parietal cortex, associated with relative increases in
the cerebellum and dentate nuclei. PDCP expression
in individual patients correlated consistently with per-
formance on tests of memory and executive function
and has recently been found to be elevated in PD patients
with neuropsychologically defined criteria for minimal
cognitive impairment (Huang et al., in press). These
network values were also found to be highly reproducible
over an 8-week period and were not altered by routine
antiparkinsonian therapies such as levodopa or subthalamic
nucleus (STN) deep brain stimulation (Huang et al.,
2007; cf. Asanuma et al., 2006). These findings suggest
that this network may have utility as an objective biomarker
of cognitive functioning at early clinical stages of the
Although substantial information exists linking these
metabolic networks to the motor and cognitive manifesta-
tions of PD, little is known about the actual time course of
their evolution during the early phases of the illness. In this
regard, it would be useful to measure the rate at which their
expression changes with time, as well as to examine the
relationship between these changes and concurrent clinical
and dopaminergic imaging measures of disease progression.
To address these issues, we performed a longitudinal multi-
tracer PET imaging study of early stage PD patients
followed over a 4-year period.
Page 2 of13Brain (2007)C. Huang et al.
by guest on June 1, 2013
Materials and methods
Fifteen right-handed PD patients [11 males and 4 females,
age: 58.0?10.2 years (mean?SD)] participated in this
longitudinal study. A diagnosis of PD was made according to
the UK Brain Bank criteria (Hughes et al., 1992); patients
were enrolled within 2 years of diagnosis. At baseline, the
patients had predominantly unilateral motor signs (Hoehn
and Yahr Stage: 1.2?0.3, mean?SD). Nine patients had
predominant right-sided limb involvement; the remaining
six had predominant involvement of the left limbs. Clinical
and imaging evaluations were conducted at the first visit and
were repeated approximately 2 years later (mean interval
2.1?0.6 years). Ten of the participants returned for further
assessment at a third timepoint approximately 4 years from
baseline (mean interval 3.9?0.7 years). Of the five subjects
who did not return for the final timepoint, three relocated
and two refused to participate in further testing.
Of the 15 original subjects, eight were drug-naı ¨ve at
baseline; the remaining seven patients were chronically
treated with anti-parkinsonian medication at the time of
enrollment. The medicated patients were on levodopa singly
(n¼1), levodopa in combination with selegiline and/or
dopamine agonists (n¼3), or on monotherapy with either
selegiline (n¼1) or dopamine agonists (n¼2). Seven of the
initially unmedicated patients required treatment by the
second timepoint, receiving either levodopa (n¼3) or
dopamine agonists (n¼4). By the third timepoint, all
participants required levodopa singly or in combination
with a dopamine agonist.
Positron emission tomography
At each timepoint, subjects underwent PET imaging with
18F-fluorodeoxyglucose (FDG) to quantify regional glucose
metabolism and18F-fluoropropyl ?CIT (FPCIT) to measure
caudate and putamen dopamine transporter (DAT) binding
(Kazumata et al., 1998; Ma et al., 2002). At each timepoint,
the two scans were separated by an average interval of
approximately 3 weeks (mean interval: 27, 16 and 13 days
for the three timepoints, respectively). They fasted over-
night before each imaging session; all antiparkinsonian
medications were held for at least 12h before the PET
procedures. The subjects were evaluated according to the
Unified Parkinson’s Disease Rating Scale (UPDRS, Fahn
et al., 1987) immediately prior to imaging.
PET imaging was performed in 3D mode using a GE
Advance tomograph (General Electric; Milwaukee, WI,
USA). The 18-ring bismuth germanate scanner provide
35 image planes with an axial field of view of 14.5cm and
an intrinsic resolution of 4.2mm (FWHM) in all directions.
PET scans were conducted in a dimly lit room with
minimal auditory stimulation with eyes open. Patients were
positioned in the scanner using a stereoadapter with 3D
laser alignment with reference to the orbitomeatal line;
identical stereoadapter and laser settings were used in each
imaging session. Ethical permission for these procedures
was obtained from the Institute Review Board of North
Shore University Hospital. Written consent was obtained
from each subject with detailed explanation of the
Striatal DAT binding
The FPCIT PET scans from each subject were realigned to
the image frame acquired at 40min post injection to
correct for possible motion artefact (Ma et al., 2002).
Regions-of-interest (ROIs) were placed on the caudate and
putamen bilaterally and on the occipital cortex (Carbon
et al., 2004). The ROIs were defined on a single slice
summed over the striatal sections and were individually
adjusted for each subject. All three scans were realigned to
the baseline scan so that the same ROI template was used
across the three timepoints. At each timepoint, caudate
and putamen DAT binding was estimated for each hemi-
sphere by the striatal-to-occipital ratio (SOR), defined as
(striatum?occipital)/occipital counts in a single 10min
frame beginning at 90min after the tracer injection
(Kazumata et al., 1998; Ma et al., 2002). These values
were averaged across hemispheres and compared with
analogous values from 10 healthy volunteers (4 males and
6 females; age 60.0?9.9 years).
Regional and global rates of glucose metabolism were
computed on a voxel basis for each FDG PET scan utilizing
a single arterial sampling method (Takikawa et al., 1993).
The metabolic images were then processed using SPM99
(Wellcome Department of Cognitive Neurology, Institute of
Neurology, London) running on Matlab 6.5 (Mathworks,
Inc., Natick, MA). The scans from each subject were realigned
and spatially normalized to a Talairach-based FDG PET
template. The normalized data were then smoothed using
a Gaussian kernel at FWHM¼10mm. Global normalization
was performed using proportional scaling.
Longitudinal changes in regional metabolism for the entire
cohort (n¼15) were assessed across the three timepoints
using the ‘multisubjects, conditions and covariates’ model in
SPM. Contrasts defined as ‘?1 0.5 0.5’ and ‘1 ?0.5 ?0.5’
were used to assess increasing and decreasing brain
metabolism over time, respectively. Changes were considered
significant at a threshold of P50.05 corrected at the cluster
level, as well as at a false discovery rate (FDR)-corrected
threshold of P50.05at the voxel level. Coordinates were
reported in the standard anatomical space developed at the
Montreal Neurological Institute. The localization of each
reported cluster was confirmed using the Talairach space
utility (available at http://www.ihb.spb.ru/?pet_lab/TSU/
TSUMain.html). We used the atlas of Schmahmann (2000)
to localize clusters within the cerebellum and related
Metabolic progression in early PD Brain (2007)Page 3 of13
by guest on June 1, 2013
For each significant cluster, we performed a post hoc
analysis in which metabolic activity at each timepoint was
measured within a sphere (radius¼4mm) centred on
the peak voxel. Each regional value was ratio normalized
by the global metabolic rate measured in that scan. These
measures were then adjusted to the group mean by
multiplication with the average global value across subjects.
The data were plotted and displayed with respect to
reference values from 15 healthy volunteer subjects (8 males
and 7 females; age 56.7?12.3 years).
Using a network quantification approach (Ma et al., 2007;
Spetsieris et al., 2006), we assessed longitudinal changes in
the expression of the two disease-related spatial covariance
patterns that we had previously identified and validated in
the metabolic data of PD patients. The first pattern was
identified during a voxel-driven principal components
analysis (PCA) of FDG PET scans from a combined
group of PD patients and age-matched healthy volunteers
(Ma et al., 2007). This PD-related motor pattern (PDRP;
Fig. 1A) was characterized by increased pallidothalamic and
pontine metabolic activity associated with relative metabolic
reductions in the premotor and supplementary motor areas
and parieto-occipital association regions. We have recently
described a separate spatial covariance pattern that is
associated with memory and executive functioning in non-
demented PD patients (Huang et al., 2007). This PD-related
Fig.1 Parkinson’s Disease-Related Spatial Covariance Patterns. (A) Parkinson’s Disease-Related Pattern (PDRP).This motor-related
metabolic pattern was identified by network analysis of [18F]-fluorodeoxyglucose (FDG) positron emission tomography (PET) scans from
33 PD patients and 33 age-matched healthy volunteers (Ma et al., 2007).This pattern was characterized by relative increases in
pallidothalamic, pontocerebellar and motor cortical/supplementary motor area (SMA) metabolic activity (top), associated with reductions
in the lateral premotor and posterior parietal areas (bottom). PDRP expression was significantly increased in the PD cohort (P50.001)
compared to controls. (B) Parkinson’s Disease-Related Cognitive Pattern (PDCP).This cognition-related metabolic pattern was identified
in the network analysis of FDG PETscans from15 non-demented PD patients with mild-moderate motor symptoms (Huang et al., 2007).
This pattern was characterized by relative hypometabolism of dorsolateral prefrontal cortex, rostral supplementary motor area (preSMA)
and superior parietal regions, associated with relative cerebellar/dentate nucleus metabolic increases. PDCP expression correlated
significantly (P50.01) with psychometric indices of memory and executive functioning. [The displays represent voxels that contribute
significantly (P50.001) to each of the two networks.Voxels with positive region weights (metabolic increases) are colour-coded from red
to yellow; those with negative region weights (metabolic decreases) are colour-coded from blue to purple.]
Page 4 of13 Brain (2007) C. Huang et al.
by guest on June 1, 2013
cognitive pattern (PDCP; Fig. 1B) was characterized by
reductions in prefrontal and parietal metabolism associated
with relative increases in the cerebellum and dentate nuclei.
In all subjects, the expression of the PDRP and PDCP
networks was separately quantified at each timepoint using
a fully automated voxel-based algorithm (software avai-
lable at http://neuroscience-nslij.org/Methods/software.html,
Spetsieris et al., 2006; Ma et al., 2007). All network
computations were performed blind to subject, timepoint
(0, 24 or 48 months), baseline treatment status (drug naı ¨ve
or initially treated) and disease severity (UPDRS motor
ratings). Reference values for PDRP and PDCP expression
were computed in the same group of 15 age-matched
healthy volunteer subjects that was used for the post hoc
analysis of the SPM results (see earlier). Subject scores for
the entire cohort (PD patients and healthy controls) were
z-transformed and offset so that the control mean was zero.
PDRP and PDCP scores computed in the baseline scans
(n¼15) were separately compared with control values using
two-tailed Student’s t-tests. One-way repeated measure
analysis of variance (RMANOVA) was performed on the
data of all 15 subjects acquired at the three timepoints to
evaluate longitudinal changes in UPDRS motor ratings,
mean left–right caudate/putamen DAT binding, regional
metabolic measures and PDRP/PDCP network activity.
These analyses were performed using an iterative method
to fit a mixed model for each RMANOVA (Searle et al.,
1992). This approach allowed for the inclusion of subjects
with missing data to provide a more precise and reliable
estimation of the data at each timepoint, thereby increasing
the power of the whole model (cf. Ellis et al., 2000). Post hoc
comparisons were performed between timepoints (e.g. 2–1,
3–2, 3–1) using Tukey’s HSD tests. To examine the effects of
initial medication status on baseline motor ratings, striatal
DAT binding and network expression, we compared these
measures in the drug-naı ¨ve and initially medicated sub-
groups. We also determined whether progression rates for
each of these measures differed for these two subgroups. This
was accomplished by using a two-way RMANOVA that
included initial medication status as the between-subject
variable and timepoints as the repeated within-subject
The degree of linearity in the changes in the clinical,
dopaminergic imaging and network measures with disease
regression model to the data from the 10 participants
who were scanned at all three timepoints. R2values were
computed by regression for each subject; the mean value
was used to represent the overall linearity of change in each
measure over time. If linearity was evident (i.e. mean
R240.9), the annual rate of change for each individual was
computed by dividing the difference in the measure
between timepoints 1 and 3 by the corresponding time
fittinga simple linear
interval. The mean rate of change across subjects was used
to represent the rate of change per year for the PD group.
Additionally, we used the Wilcoxon signed rank test to
compare the R2values for the PDRP and PDCP scores as a
means of assessing the relative linearity of the trajectories of
the two measures.
coefficients (Bland and Altman, 1995) to determine the
relationship between changes in the clinical and imaging
progression indices over the three longitudinal timepoints.
For all analyses, the significance level was set at P50.05.
(SAS Institute Inc.).
Off-state UPDRS motor ratings at the three timepoints are
presented in Fig. 2. These values increased significantly over
time [F(2,17)¼23.2, P50.0001; RMANOVA]. Post hoc
testing with respect to baseline revealed significant increases
in motor ratings at both the second (P50.001) and third
timepoints (P50.0001), with a trend toward an increase
between these timepoints (P¼0.08). Overall, the increase in
UPDRS motor ratingswas linear
R2¼0.91), with an average rate of 2.1 points per year.
over time (mean
Striatal DAT binding
Caudate and putamen DAT binding values at the three
timepoints are presented in Table 1 and Fig. 3. At baseline,
DAT binding was 97.3% (P¼0.79) and 54.7% (P50.001)
of the normal mean for the caudate and putamen,
Fig. 2 Mean off-state Unified Parkinson’s Disease Rating Scale
(UPDRS) motor ratings at baseline, 24 and 48 months.These
scores increased linearly over time (P50.0001; RMANOVA), at
a rate of 2.1units per year (see text). [Bars represent the standard
error at each timepoint.]
Metabolic progression in early PDBrain (2007) Page 5 of13
by guest on June 1, 2013
respectively. DAT binding for the two measures fell to
88.5% (P¼0.21) and 48.1% (P50.001) of normal at 24
months, and 77.4% (P50.02) and 40.0% (P50.001) at 48
months. These measures declined significantly over time
[caudate: F(2,17)¼10.2; putamen: F(2,17)¼8.4, P50.003;
RMANOVA]. Post hoc testing revealed significant reduc-
tions in DAT binding at the third timepoint relative to
baseline (caudate: P50.001; putamen: P50.005), with a
trend toward decline at the second timepoint (caudate:
P¼0.06; putamen: P¼0.07). Declines in striatal DAT
binding between the second and third timepoints were not
significant (caudate: P¼0.12; putamen: P¼0.23).
The decline in caudate and putamen DAT binding was
linear (mean R2¼0.94 for both regions), corresponding to
an average rate of 5.2 and 3.9% of normal per year for the
two regions, respectively. Declines for the caudate and
putamen were highly intercorrelated (r¼0.91, P50.0001),
and both correlated with concurrent increases in UPDRS
motor ratings (caudate: r¼?0.60, P50.01; putamen:
Regional glucose metabolism
Global rates of glucose metabolism in the PD patients
did not differ from normal (P40.3) at any of the three
timepoints (PD: 7.31?1.55, 6.91?1.32, 6.40?1.90mg/
min/100g at baseline, 24 months and 48 months; normal:
7.03?1.38mg/min/100g). These measures did not change
significant metabolic changes over time are presented in
Table 2 and Fig. 4. Progressive metabolic increases
(P50.001; Fig. 4A, top) were detected in the left STN,
and in the right internal globus pallidus (GPi) and the
adjacent subthalamic region.
(P50.0001) were also present in the dorsal pons, in the
vicinity of the pedunculopontine nucleus (PPN) and the
adjacent cerebellar hemisphere. Additionally, regional meta-
bolism in the left primary motor cortex and the adjacent
supplementary motor area (SMA) increased between the
first and second timepoints, remaining stable throughout
the rest of the follow-up period. At baseline, significant
elevations relative to controls were present only in the
dorsal pontine region (P50.05). Regional metabolism in
the GPi and motor cortex reached abnormal levels (P50.05
and P50.01, respectively) at the second timepoint. All the
increasing regions, including the STN, attained supernormal
levels of metabolic activity (P50.01) by the final timepoint.
Declines in regional metabolism with advancing disease
(P50.001; Fig. 4A, bottom) were present bilaterally in the
prefrontal region (BA 9/10) and in the inferior parietal
lobule (BA 40). Despite the highly significant decline in the
former region, local metabolic activity did not fall to
subnormal levels during the course of the study. In the
latter region, metabolic reductions reached significance
(P50.05) only at the final timepoint. Although all the
reported regions exhibiting longitudinal change (Table 2)
were localized to a single hemisphere, in each case
significant metabolic progression (P50.005) was noted in
homologous areas of the opposite hemisphere.
Mean values for PDRP and PDCP expression at the three
timepoints are displayed in Fig. 5. At baseline, PDRP
expression was significantly higher in the PD patients
relative to controls (t¼2.24; P50.04), whereas PDCP
expression did not differ across the two groups (t¼0.75;
P¼0.46). Both networks exhibited significant changes
RMANOVA]. Post hoc testing revealed an increase in
PDRP expression from baseline to the second (P50.05)
and third timepoints (P50.0001), and between the second
and third timepoints (P50.0001). In contrast, PDCP
expression did not change between baseline and the
second timepoint (P¼0.31), but was elevated at the third
timepoint with respect to baseline and second timepoint
T able1 Striatal dopamine transporter binding at baseline,
24 and 48 months
Baseline 24 Months48 Months Controls
aMean striato-occipital ratio (SOR) values?SD (see text).
?P50.05,??P50.001, compared to age-matched healthy control
values (Student’s t-tests).
Fig. 3 Mean striatal DAT binding at baseline, 24 and 48 months.
Values for the caudate (squares) and putamen (triangles) are
represented as percent of the normal mean value for each region
(see text). In both regions, DAT binding declined linearly over time
(P50.003; RMANOVA). [Bars represent the standard error ateach
Page 6 of13Brain (2007)C. Huang et al.
by guest on June 1, 2013
values P50.005), and with respect to healthy control values
Although a significant correlation was noted between
changes in PDRP and PDCP expression over time (r¼0.81,
P50.0001), the two network trajectories proved to be
different (mean R2¼0.77 and 0.56 for the PDRP and
PDCP, respectively; P50.04, Wilcoxon signed rank test).
This is consistent with the relatively faster increase in PDRP
activity that was observed over time. The progressive
increases in PDRP expression significantly correlated with
concurrent increases in UPDRS motor ratings (r¼0.62,
P50.005) and declines in striatal DAT binding (caudate:
Longitudinal changes in PDCP expression did not correlate
with advancing motor disability or with concurrent changes
in striatal DAT binding.
Effect of initial treatment
At baseline, there were no significant differences between
treated and drug-naı ¨ve patients in PDRP/PDCP expression
or caudate/putamen DAT binding (P40.19). However, the
UPDRS motor ratings than their drug-naı ¨ve counterparts
(P50.02). Two-way RMANOVA revealed no significant
main effect of initial treatment status (P40.27) and no
interaction effect of initial medication status and time
(P40.14) for the UPDRS motor ratings, caudate/putamen
Thus, rates of progression in these measures are similar
for drug-naı ¨ve and initially treated patients.
In this longitudinal PET study, we found that early stage
PD is associated with progressive changes in regional
metabolism at key nodes of the two functional brain
networks that characterize this disorder. Increasing regional
metabolism was observed within elements of the motor-
related PDRP network, which evolved in parallel with
deterioration in striatal DAT binding and UPDRS motor
ratings. In contrast, the declines in cortical metabolism that
were seen with advancing disease were associated with
changes in the activity of the cognition-related PDCP
network. Significant elevations in PDCP activity occurred
late in the follow-up period, and did not correlate with
motor ratings or with dopaminergic imaging measures of
disease progression. These findings indicate that the
functioning of motor- and cognition-related neural systems
is dissociated at the earliest symptomatic phases of the
Changes in glucose metabolism with disease
A whole-brain search of the longitudinal FDG PET data
revealed progressive increases in the metabolic activity of
the STN, GPi, PPN and motor cortex, regions that have
been demonstrated experimentally to mediate the motor
manifestations of parkinsonism (Parent and Hazrati, 1995;
Middleton and Strick, 2000; Wichman and DeLong, 2003).
In contrast, declining metabolic activity was observed in
prefrontal and parietal association cortex, areas associated
with cognitive dysfunction in this disorder (Huang et al.,
2007; see Carbon and Marie ´, 2003 for review).
The role of the STN in the pathophysiology of PD and its
treatment has been reviewed elsewhere (Hamani et al.,
2004; cf. Asanuma et al., 2006; Tros ˇt et al., 2006). Increased
neural activity in the STN is a consistent feature of
both clinical and experimental parkinsonism. Nonetheless,
regional glucose metabolism has been found to be reduced
in this region when measured in animal models (Palombo
et al., 1990; Carlson et al., 1999). Regional glucose
utilization predominantly reflects local synaptic activity
T able 2 Regions with significant longitudinal changes in glucose metabolism.
Mean adjusted regional glucose metabolism (mg/min/100g)
xyz Baseline (n¼15)24 Months
Paracentral Gyrus (BA 4/6)
Globus Pallidus (internal)
Medial Frontal Lobe (BA 9/10)
Inferior Parietal Lobule (BA 39/40) ?56 ?56
30 ?60 ?46
aMontreal Neurological Institute (MNI) standard space; BA¼Brodmann Area.
bAccording to the atlas of Schmahmann (Schmahmann et al., 2000).
cSPM, P-values corrected at the cluster level. All maxima were significant at P50.01, FDR-corrected at the voxel level.
?P50.05,??P50.01,???P50.001, compared to age-matched healthy control values (Student’s t-tests).
Metabolic progression in early PDBrain (2007) Page 7 of13
by guest on June 1, 2013
Fig. 4 (A) Voxel-based analysis of longitudinal changes in regional metabolic activity. Metabolic increases with disease progression (top) are displayed using a red^yellow scale.
Progressive metabolic declines (bottom) are displayed using a blue^purple scale. Both displays were superimposed on a single-subject MRI brain template and thresholded at t¼3.48,
P¼0.001 (peak voxel, uncorrected). (B) Displays of the metabolic data for these individual regions at each timepoint (Table 2).The significance level (P-value) of the repeated measures
analysis of variance (RMANOVA) is presented for each region (see text). [The coordinates refer to the Montreal Neurological Institute (MNI) standard space.GPi: internal globus
pallidus, STN: subthalamic nucleus, BA: Brodmann area.]
by guest on June 1, 2013http://brain.oxfordjournals.org/ Downloaded from
and the biochemical maintenance processes of the rest state
(Jueptner and Weiller, 1995; Eidelberg et al., 1997).
Therefore, the decline in STN metabolism in experimental
parkinsonism has been viewed as a consequence of
a reduction in inhibitory afferents from the external
pallidum. However, other lines of evidence have pointed
to an increase in STN metabolic activity in parkinsonism
(Hirsch et al., 2000), attributed to overactivity of excitatory
projections to this region from the PPN, intralaminar
thalamus and the motor cortex (cf. Parent and Hazrati,
1995). Our observation of steady increases in STN
metabolism with advancing disease supports the latter
notion. In this vein, the progressive increases in GPi
metabolism that were also observed in the current study are
likely the consequence of increased STN output to this area.
Although these pallidal increases were discerned in the rest
state, they appear to correlate with changes in the neural
activity of this region during movement. Indeed, we have
recently reported progressive increases in pallidal activation
during the performance of a motor task that was
kinematically controlled across the first two timepoints
(Carbon et al., 2007).
It is noteworthy that the primary motor cortex and the
PPN, two prominent sources of excitatory input to STN,
also exhibited increases in glucose utilization with advanc-
ing disease. Hypersynchrony of motor cortical neuronal
activity has been observed as a prominent feature of
nigral dopamine depletionin parkinsonianprimates
(Goldberg et al., 2002). This phenomenon may be the
basis for the localized increases in metabolism in this region
that we have observed in untreated PD patients (Asanuma
et al., 2006). The time course data indicated that metabolic
increases in the motor cortex reached significance with
respect to healthy control values at the second timepoint,
while in the STN regional metabolism attained an abnormal
level only at the end of the study. This is consistent with
the notion that in parkinsonism, the STN functions as
a modulator of this cortically generated synchronous
high-frequency oscillation (Levy et al., 2002).
Unlikethe other regions
consumption in the course of disease progression, regional
metabolism in the primary motor cortex and SMA did not
rise continuously over time (see Fig. 4B, middle right panel).
Rather, metabolic activity in this region rose initially and
then remained stable at an elevated value relative to the
healthy controls. Interestingly, these changes were more
closely associated with concurrent UPDRS tremor ratings
(r¼0.73, P50.001; Bland–Altman correlation coefficient)
than those occurring in the other regions with longitudinal
increases in metabolic activity (r¼0.49, 0.48 and 0.32 for
the PPN, STN and GPi, respectively). We note that tremor
ratings in our subjects also reached an early plateau
(mean?SE: 0.4?0.2; 1.9?0.8; 2.2?0.6 for the three
timepoints). It is therefore likely that the metabolic changes
in the primary motor cortex/SMA reflected progression in
this specific disease manifestation.
The origin of the progressive metabolic increases in the
PPN is less clear. Similar changes have been observed in
both primate and rodent models of parkinsonism (Palombo
et al., 1990; Carlson et al., 1999), and have been attributed
to overactive input to this structure from the globus
pallidus/substantia nigra pars reticularis (SNr), or from the
STN. It is interesting however, that the PPN was the only
progressively increasing brain region in which glucose
metabolism was abnormal at baseline, while in GPi and
STN, metabolism reached supernormal levels only at
subsequent timepoints. Indeed, it is possible that this
early change in PPN metabolic activity is a compensatory
response to nigral disease by which dopamine release in the
striatum is enhanced (Blaha and Winn, 1993). In contrast,
such an early metabolic effect is not present in the STN to
support a compensatory role for this structure in the
presymptomatic period (cf. Bezard et al., 2003). Although
ventral thalamic metabolic activity has been found to
correlate consistently with spontaneous GPi firing rates in
PD patients (Eidelberg et al., 1997), firm physiological
inferences cannot always be made based upon in vivo
metabolic measurements. Conclusions regarding the locali-
zation and time course of potential compensatory responses
in PD patients will require the assessment of early and late
changes in metabolic activity in the hemispheres ipsilateral
and contralateral to the initially affected body side.
Our voxel-based approach also revealed several areas
in which metabolic activity declined significantly with
with increasing glucose
Fig. 5 Mean network activity at baseline, 24 and 48 months.
Values for the PD-related motor and cognitive spatial covariance
patterns (PDRP and PDCP; see Fig.1) were computed at each
timepoint and displayed relative to the mean for15 age-matched
healthy subjects (see text). Network activity increased significantly
over time for both patterns (P50.0001; RMANOVA), with the
PDRP progressing faster than the PDCP (P50.04). Relative to
controls, PDRP activity in the patient group was elevated at all
three timepoints, while PDCP activity reached abnormal levels
only at the final timepoint. [Bars represent the standard error at
Metabolic progression in early PDBrain (2007)Page 9 of13
by guest on June 1, 2013
advancing disease. These regions, the prefrontal cortex and
inferior parietal lobule, are involved mainly in cognitive
functioning and metabolic reductions, and both areas have
been found to correlate with impaired memory and
executive function in non-demented PD patients (Huang
et al., 2007; cf. Carbon and Marie ´, 2003). Also relevant is
that despite highly significant progressive declines over
time, metabolic activity in these regions declined to only
marginally abnormal levels by the end of the study. This is
consistent with the early, mild disease that typified our
patients. Indeed, although serial psychometric testing was
not performed in this study, there was no gross indication
of cognitive impairment in these subjects. Nonetheless, the
continuous decline in neural function that was observed in
these regions is compatible with an evolving cortically based
neurodegenerative process, the histopathological basis for
which is currently unknown (see Emre, 2004; Braak et al.,
2005; cf. Huang et al., 2007). Alternatively, these focal
metabolic changes may reflect alterations in the activity of
cognition-related neural networks involving these brain
areas (cf. Carbon et al., 2003b; 2004).
Development of abnormal network activity
In addition to identifying areas with progressively increas-
ing or decreasing metabolism in early stage PD, we assessed
longitudinal changes in the activity of functional networks
associated with this disorder. Although motor signs were
comparatively mild in our patient cohort, PDRP activity
was abnormally elevated throughout the period of observa-
tion, even at the initial timepoint. Thus, despite a paucity
of regional abnormalities at baseline (see earlier), functional
connectivity within motor pathways is likely to have been
altered by this time.
In previous studies, we have consistently found that
PDRP scores correlate with clinical severity ratings (e.g.
Eidelberg et al., 1995; Feigin et al., 2002; Lozza et al., 2004;
Asanuma et al., 2006). Thus, it is not surprising that
network activity in individual patients increased over time
progression obtained by clinical
dopaminergic imaging. Specifically, longitudinal changes
in PDRP expression were associated with increases in motor
UPDRS ratings and declines in striatal DAT binding.
Although these indices of disease progression were inter-
correlated, no more than 40% of their variability was
shared between any two of them. Thus, these progression
measures are not interchangeable; each one is likely to
provide unique information concerning the disease process.
We also note that although PDRP activity is reduced by
effective antiparkinsonian therapy (e.g. Asanuma et al.,
2006; cf. Eckert and Eidelberg, 2005), the 12-h medication
washout that was used in this study was adequate for the
assessment of progression with this measure. In other
words, our data suggest that the influence of disease
progression on PDRP expression outweighs the residual
effects on network activity that may be present following
incomplete washout. Rigorous washout studies of chroni-
cally treated patients will be needed to determine the time
course of these residual effects on network activity.
PDCP scores also increased with advancing disease,
correlating significantly with concurrent changes in PDRP
expression. Nonetheless, the trajectories of the two net-
works over time were different, with significant increases in
the former becoming evident only between the second and
third timepoints. Indeed, PDCP activity reached abnormal
levels only at the final timepoint, whereas PDRP activity
was abnormal at baseline. It is conceivable that the
relatively late increase in PDCP expression observed in
this study is a reflection of incipient cognitive impairment
(cf. Huang et al., in press). Further follow-up with serial
psychometrics and network quantification will be needed to
explore this possibility. In any event, the distinctive
trajectories of the two PD-related networks suggest that
discrete pathophysiological mechanisms underlie the motor
and cognitive features of the disease. This notion is further
supported by the dissociation of these patterns in response
to antiparkinsonian therapy (Asanuma et al., 2006; Huang
et al., 2007). These findings suggest a role for FDG PET
imaging in monitoring both motor and cognitive features
of disease progression.
Relationship to clinical and dopaminergic
In this study, we also measured longitudinal changes in
UPDRS motor ratings and striatal DAT binding. Our
findings accord with the results of other studies in which
rates of disease progression were assessed using the UPDRS,
striatal DAT binding indices or both. In our cohort,
off-state motor ratings deteriorated at a rate of 2.1 points
per year, which was comparable to the findings of
other prospective imaging studies (Hilker et al., 2002;
cf. Parkinson-Study-Group, 1993). While our regression-
based estimate was consistent with a linear process, the rise
in the motor ratings was comparatively steeper during the
first 2 years of the study. We note that this rate,
approximating 3 points per year, is slower than that
reported for placebo-treated patients in a recent clinical
trial of the effects of levodopa on disease progression
(Fahn et al., 2004). We attribute this difference to the
younger age of onset and longer duration of symptoms in
our study (Jankovic and Kapadia, 2001; Alves et al., 2005).
This suggests that slowing in the rate of clinical deteriora-
tion may occur with advancing disease.
Likewise, the decline in striatal DAT binding in our
cohort was in line with other longitudinal PET studies in
PD employing this measure of presynaptic nigrostriatal
function (see Au et al., 2005 for review). Our estimates of
loss in striatal binding resembled those measured with
CFT), another radioflourinated PET ligand for quantifying
Page10 of13Brain (2007) C. Huang et al.
by guest on June 1, 2013
DAT binding (Nurmi et al., 2003). We note that in the
prior study, all patients were drug-naı ¨ve at baseline and
that levodopa therapy was initiated in nearly all during the
follow-up period. Similarly, in our study, all the initially
untreated participants required levodopa by the time of the
final imaging session. It has been suggested based upon the
results of the recent ELLDOPA trial (Fahn et al., 2004) that
chronic levodopa therapy can reduce estimates of striatal
DAT binding obtained with radiotracer-based imaging
(cf. Ravina et al., 2005). Therefore, it is conceivable that
the decline in this measure observed in longitudinal studies
of initially untreated subjects may be a reflection of this
pharmacological effect, rather than of true disease progres-
sion. Nonetheless, the presence of a significant negative
correlation between changes in caudate/putamen DAT
binding and motor UPDRS ratings argues against this
possibility. Indeed, our findings are consistent with the
results of previous cross-sectional studies in which correla-
tions of similar magnitude (R2?35%) were reported
between these measures of disease progression (Seibyl
et al., 1995; Asenbaum et al., 1997; Kazumata et al.,
1998). Thus, as with PDRP, imaging descriptors of dopa-
ratings, but not to a degree that would make one or the
other measure redundant.
As with the clinical ratings, the serial assessments of
striatal DAT binding that were obtained were consistent
with a linear process. Interestingly, the decline in putamen
DAT binding between the first and final visits correlated
significantly with baseline values (R2¼0.87, P50.0005).
We note that the presence of such a correlation is typically
indicative of a non-linear process (e.g. Hilker et al., 2005),
which would explain the relatively greater rate of DAT
binding loss seen in the caudate and the ipsilateral
putamen. However, at this stage of disease, the between-
subject variability in putamen DAT binding was relatively
small, with progressive declines in the coefficient of
variation (COV) over time (see Table 1 and error bars in
Fig. 3). Thus, individual differences in baseline binding
values were perhaps less likely to influence the rate of
change than at earlier disease stages. In other words, our
study was conducted at a time in the disease process when
between-subject variability was declining, while the within-
subject rate of change was relatively constant. This scenario
is compatible with an incipient floor effect. In contrast,
during the same time period, there was only minimal
decline in COV for caudate DAT binding. The absence of
an observed floor effect on the caudate measure may
explain why the correlation with PDRP change was stronger
for DAT binding decrements in this region relative to the
putamen. Longer term assessment of these patients with
FP-CIT PET will help determine whether such an effect is
truly present in our data.
Although the number of patients followed in our study
was comparatively small, the progression data were overall
quite robust. Nonetheless, certain aspects of the findings
with motor disability
will require further validation. Specifically, we found that
baseline UPDRS motor ratings were significantly lower in
the seven participants who were on chronic antiparkinso-
nian therapy at the time of enrollment. Six of these subjects
were treated with selegiline and/or dopamine agonists, of
whom three were also on levodopa. Thus, residual central
effects of any of these medications could account for the
lower degree of disability that was observed at baseline,
even following a 12-h washout.
comparable effects were not observed in the imaging
measures. Thus, in this limited, heterogenous treatment
cohort, there was no effect of initial treatment status on
baseline measures of striatal DAT binding or PDRP
expression. Nonetheless, as discussed above, subtle effects
of chronic therapy on either measure cannot be excluded
(cf. Fahn et al., 2004). Our data also suggest that initial
treatment status did not influence subsequent disease
progression, whether assessed clinically or by imaging.
Although based on a relatively small number of subjects,
this observation is consistent with a recent study of the
long-term UPDRS data collected in neuroprotection trials
(Guimaraes et al., 2005). Additionally, despite the major
changes in medication regimen that occurred in the course
of the study, the presence of consistent correlations between
the progression indices suggests that chronic treatment did
not selectively influence any one of these measures. Indeed,
it appears that these biomarkers are more sensitive to
disease progression than to residual treatment effects that
might persist following a 12-h washout. Further studies will
be needed to determine the optimal period of washout
needed to minimize this potential confound.
This work was supported by NIH NINDS R01 35069, P50
NS 38370 and by the General Clinical Research Center of
the North Shore-Long Island Jewish Health System (M01
RR 018535). P50 NS 38370. The authors wish to thank Dr
Thomas Chaly for radiochemistry support and Mr Claude
Margouleff for technical assistance in performing the PET
studies. Special thanks to Ms Shivani Rachakonda for data
management and Ms Toni Flanagan for valuable editorial
Alves G, Wentzel-Larsen T, Aarsland D, Larsen JP. Progression of motor
impairment and disability in Parkinson disease: a population-based
study. Neurology 2005; 65: 1436–41.
Alves G, Larsen JP, Emre M, Wentzel-Larsen T, Aarsland D. Changes in
motor subtype and risk for incident dementia in Parkinson’s disease.
Mov Disord 2006; 21: 1123–30.
Asanuma K, Ma Y, Huang C, Carbon-Correll M, Edwards C, Raymond D,
et al. The metabolic pathology of dopa-responsive dystonia. Ann Neurol
2005; 57: 596–600.
Asanuma K, Tang C, Ma Y, Dhawan V, Mattis P, Edwards C, et al.
Network modulation in the treatment of Parkinson’s disease. Brain
2006; 129 (Pt 10): 2667–78.
Metabolic progression in early PDBrain (2007)Page11of13
by guest on June 1, 2013
Asenbaum S, Brucke T, Pirker W, Podreka I, Angelberger P, Wenger S,
et al. Imaging of dopamine transporters with iodine-123-beta-CIT and
SPECT in Parkinson’s disease. J Nucl Med 1997; 38: 1–6.
Au WL, Adams JR, Troiano AR, Stoessl AJ. Parkinson’s disease: in vivo
assessment of disease progression using positron emission tomography.
Brain Res Mol Brain Res 2005; 134: 24–33.
Bezard E, Gross CE, Brotchie JM. Presymptomatic compensation in
Parkinson’s disease is not dopamine-mediated. Trends Neurosci 2003;
Blaha CD, Winn P. Modulation of dopamine efflux in the striatum
following cholinergic stimulation of the substantia nigra in intact and
pedunculopontine tegmental nucleus-lesioned rats. J Neurosci 1993; 13:
Bland JM, Altman DG. Calculating correlation coefficients with repeated
observations: Part 1–Correlation within subjects. Br Med J 1995; 310: 446.
Braak H, Rub U, Jansen Steur EN, Del Tredici K, de Vos RA. Cognitive
status correlates with neuropathologic stage in Parkinson disease.
Neurology 2005; 64: 1404–10.
Brooks DJ, Frey KA, Marek KL, Oakes D, Paty D, Prentice R, et al.
Assessment of neuroimaging techniques as biomarkers of the progres-
sion of Parkinson’s disease. Exp Neurol 2003; 184 (Suppl 1): S68–79.
Carbon M, Marie ´ RM. Functional imaging of cognition in Parkinson’s
disease. Curr Opin Neurol 2003; 16: 475–80.
Carbon M, Edwards C, Eidelberg D. Functional brain imaging in
Parkinson’s disease. Adv Neurol 2003a; 91: 175–81.
Carbon M, Ghilardi MF, Dhawan V, Eidelberg D. Correlates of movement
initiation and velocity in Parkinson’s disease: a longitudinal PET study.
Neuroimage 2007; 34: 361–70.
Carbon M, Ma Y, Barnes A, Dhawan V, Chaly T, Ghilardi MF, et al.
Caudate nucleus: influence of dopaminergic input on sequence
learning and brain activation in parkinsonism. Neuroimage 2004; 21:
Carbon M, Ghilardi MF, Feigin A, Fukuda M, Nakamura T, Dhawan V,
et al. Learning networks in health and Parkinson’s disease: reproduci-
bility and treatment effects. Hum Brain Mapp 2003b; 19: 197–211.
Carlson JD, Pearlstein RD, Buchholz J, Iacono RP, Maeda G. Regional
metabolic changes in the pedunculopontine nucleus of unilateral
6-hydroxydopamine Parkinson’s model rats. Brain Res 1999; 828: 12–9.
Eckert T, Eidelberg D. Neuroimaging and therapeutics in movement
disorders. Neurorx 2005; 2: 361–71.
Eckert T, Van Laere K, Tang C, Lewis DE, Edwards C, Santens P, et al.
Quantification of PD-related network expression with ECD SPECT. Eur
J Nucl Med Mol Imaging 2007, 34:495–501.
Eidelberg D, Moeller JR, Ishikawa T, Dhawan V, Spetsieris P, Chaly T,
et al. Assessment of disease severity in parkinsonism with fluorine-18-
fluorodeoxyglucose and PET. J Nucl Med 1995; 36: 378–83.
Eidelberg D, Moeller JR, Dhawan V, Spetsieris P, Takikawa S, Ishikawa T,
et al. The metabolic topography of parkinsonism. J Cereb Blood Flow
Metab 1994; 14: 783–801.
Eidelberg D, Moeller JR, Kazumata K, Antonini A, Sterio D, Dhawan V,
et al. Metabolic correlates of pallidal neuronal activity in Parkinson’s
disease. Brain 1997; 120: 1315–24.
Ellis SP, Underwood MD, Arango V, Mann JJ. Mixed models and multiple
comparisons in analysis of human neurochemical maps. Psychiatry Res
2000; 99: 111–9.
Emre M. Dementia in Parkinson’s disease: cause and treatment. Curr Opin
Neurol 2004; 17: 399–404.
Fahn S, Elton R. Commitee atmotUPsDRSD. Unified Parkinson’s Disease
Rating Scale. In: Fahn S, Marsden CD, Calne D, editors. Recent
developments in Parkinson’s disease. New York: MacMillan; 1987.
Fahn S, Oakes D, Shoulson I, Kieburtz K, Rudolph A, Lang A, et al.
Levodopa and the progression of Parkinson’s disease. N Engl J Med
2004; 351: 2498–508.
Fearnley JM, Lees AJ. Ageing and Parkinson’s disease: substantia nigra
regional selectivity. Brain 1991; 114 (Pt 5): 2283–301.
Feigin A, Antonini A, Fukuda M, De Notaris R, Benti R, Pezzoli G, et al.
Tc-99m ethylene cysteinate dimer SPECT in the differential diagnosis of
parkinsonism. Mov Disord 2002; 17: 1265–70.
Goldberg JA, Boraud T, Maraton S, Haber SN, Vaadia E, Bergman H.
Enhanced synchronyamong primary
the 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine primate model of
Parkinson’s disease. J Neurosci 2002; 22: 4639–53.
Guimaraes P, Kieburtz K, Goetz CG, Elm JJ, Palesch YY, Huang P, et al.
Non-linearity of Parkinson’s disease progression: implications for
sample size calculations in clinical trials. Clin Trials 2005; 2: 509–18.
Hamani C, Saint-Cyr JA, Fraser J, Kaplitt M, Lozano AM. The
subthalamic nucleus in the context of movement disorders. Brain
2004; 127: 4–20.
Hilker R, Voges J, Thiel A, Ghaemi M, Herholz K, Sturm V, et al. Deep
brain stimulation of the subthalamic nucleus versus levodopa challenge
in Parkinson’s disease: measuring the on- and off-conditions with
FDG-PET. J Neural Transm 2002; 109: 1257–64.
Hilker R, Schweitzer K, Coburger S, Ghaemi M, Weisenbach S, Jacobs AH,
et al. Nonlinear progression of Parkinson disease as determined by serial
positron emission tomographic imaging of striatal fluorodopa F 18
activity. Arch Neurol 2005; 62: 378–82.
Hilker R, Voges J, Ghaemi M, Lehrke R, Rudolf J, Koulousakis A, et al.
Deep brain stimulation of the subthalamic nucleus does not increase the
striatal dopamine concentration in parkinsonian humans. Mov Disord
2003; 18: 41–8.
Hirsch EC, Perier C, Orieux G, Francois C, Feger J, Yelnik J, et al.
Metabolic effects of nigrostriatal denervation in basal ganglia. Trends
Neurosci 2000; 23: S78–85.
Huang C, Mattis P, Julin P. Identifying functional imaging markers of
mild cognitive impairment in early Alzheimer’s and Parkinson’s disease
using multivariate analysis. Clin Neurosci Res (in press).
Huang C, Mattis P, Tang C, Perrine K, Carbon M, Eidelberg D. Metabolic
brain networks associated with cognitive function in Parkinson’s disease.
Neuroimage 2007; 34: 714–23.
Hughes AJ, Daniel SE, Kilford L, Lees AJ. Accuracy of clinical diagnosis of
idiopathic Parkinson’s disease: a clinico-pathological study of 100 cases.
J Neurol Neurosurg Psychiatry 1992; 55: 181–4.
Jankovic J, Kapadia AS. Functional decline in Parkinson disease. Arch
Neurol 2001; 58: 1611–5.
Jueptner M, Weiller C. Review: does measurement of regional cerebral
blood flow reflect synaptic activity? Implications for PET and fMRI.
Neuroimage 1995; 2: 148–56.
Kazumata K, Dhawan V, Chaly T, Antonini A, Margouleff C, Belakhlef A,
et al. Dopamine transporter imaging with fluorine-18-FPCIT and PET.
J Nucl Med 1998; 39: 1521–30.
Levy R, Ashby P, Hutchison WD, Lang AE, Lozano AM, Dostrovsky JO.
Dependence of subthalamic nucleus oscillations on movement and
dopamine in Parkinson’s disease. Brain 2002; 125: 1196–209.
Lozza C, Baron JC, Eidelberg D, Mentis MJ, Carbon M, Marie RM.
Executive processes in Parkinson’s disease: FDG-PET and network
analysis. Hum Brain Mapp 2004; 22: 236–45.
Ma Y, Tang C, Spetsieres P, Dhawan V, Eidelberg D. Abnormal metabolic
network activity in Parkinson’s disease: Test-retest reproducibility.
J Cereb Blood Flow Metab 2007; 27:501-509.
Ma Y, Dhawan V, Mentis M, Chaly T, Spetsieris PG, Eidelberg D.
Parametric mapping of [18F]FPCIT binding in early stage Parkinson’s
disease: a PET study. Synapse 2002; 45: 125–33.
Marek K, Jennings D, Seibyl J. Dopamine agonists and Parkinson’s disease
progression: what can we learn from neuroimaging studies. Ann Neurol
2003; 53: S160–6.
Mentis MJ, McIntosh AR, Perrine K, Dhawan V, Berlin B, Feigin A, et al.
Relationships among the metabolic patterns that correlate with
mnemonic, visuospatial, and mood symptoms in Parkinson’s disease.
Am J Psychiatry 2002; 159: 746–54.
Middleton FA, Strick PL. Basal ganglia and cerebellar loops: motor and
cognitive circuits. Brain Res Brain Res Rev 2000; 31: 236–50.
Page12 of13Brain (2007)C. Huang et al.
by guest on June 1, 2013
Moeller JR, Eidelberg D. Divergent expression of regional metabolic
topographies in Parkinson’s disease and normal ageing. Brain 1997; 120:
Moeller JR, Nakamura T, Mentis MJ, Dhawan V, Spetsieres P, Antonini A,
et al. Reproducibility of regional metabolic covariance patterns:
comparison of four populations. J Nucl Med 1999; 40: 1264–9.
Morrish PK, Rakshi JS, Bailey DL, Sawle GV, Brooks DJ. Measuring the rate
of progression and estimating the preclinical period of Parkinson’s disease
with [18F]dopa PET. J Neurol Neurosurg Psychiatry 1998; 64: 314–9.
Nakamura T, Dhawan V, Chaly T, Fukuda M, Ma Y, Breeze R, et al.
Blindedpositron emission tomography
implantation for Parkinson’s disease. Ann Neurol 2001; 50: 181–7.
Nurmi E, Bergman J, Eskola O, Solin O, Vahlberg T, Sonninen P, et al.
Progression of dopaminergic hypofunction in striatal subregions in
Parkinson’s disease using [18F]CFT PET. Synapse 2003; 48: 109–15.
Palombo E, Porrino LJ, Bankiewicz KS, Crane AM, Sokoloff L, Kopin IJ.
Local cerebral glucose utilization in monkeys with hemiparkinsonism
induced by intracarotid infusion of the neurotoxin MPTP. J Neurosci
1990; 10: 860–9.
Parent A, Hazrati LN. Functional anatomy of the basal ganglia. I. The
cortico-basal ganglia-thalamo-cortical loop. Brain Res Brain Res Rev
1995; 20: 91–127.
Parkinson-Study-Group. Effects of tocopherol and deprenyl on the
progression of disability in early Parkinson’s disease. The Parkinson
Study Group. N Engl J Med 1993; 328: 176–83.
Ravina B, Eidelberg D, Ahlskog JE, Albin RL, Brooks DJ, Carbon M, et al.
The role of radiotracer imaging in Parkinson disease. Neurology 2005;
Schmahmann JD, Doyon J, Toga AW, Petrides M, Evans AC. MRI atlas of
the human cerebellum. San Diego: Academic Press; 2000.
Searle SS, Casella G, McCulloch CE. Variance components. New York:
John Wiley & Sons; 1992.
Seibyl JP, Marek KL, Quinlan D, Sheff K, Zoghbi S, Zea-Ponce Y, et al.
Decreased single-photon emission computed tomographic [123I]beta-
CIT striatal uptake correlates with symptom severity in Parkinson’s
disease. Ann Neurol 1995; 38: 589–98.
Spetsieris P, Ma Y, Dhawan V, Moeller JR, Eidelberg D. Highly automated
computer-aided diagnosis of neurological disorders using functional
brain imaging. Proc SPIE: Med Imaging 2006; 6144: 61445M1–12.
Takikawa S, Dhawan V, Spetsieris P, Robeson W, Chaly T, Dahl R, et al.
Noninvasive quantitative fluorodeoxyglucose PET studies with an
estimated input function derived from a population-based arterial
blood curve. Radiology 1993; 188: 131–6.
Tros ˇt M, Su S, Su PC, Yen R-F, Tseng H-M, Barnes A, et al. Network
modulation by the subthalamic nucleus in the treatment of Parkinson’s
disease. Neuroimage 2006; 31: 301–07.
Whone AL, Watts RL, Stoessl AJ, Davis M, Reske S, Nahmias C, et al.
Slower progression of Parkinson’s disease with ropinirole versus
levodopa: The REAL-PET study. Ann Neurol 2003; 54: 93–101.
Wichman T, DeLong MR. Pathophysiology of Parkinson’s disease: the
MPTP primate model of the human disorder. Ann NY Acad Sci 2003;
Zetusky WJ, Jankovic J, Pirozzolo FJ. The heterogeneity of Parkinson’s
disease: clinical and prognostic implications. Neurology 1985; 35:
Metabolic progression in early PD Brain (2007) Page13 of13
by guest on June 1, 2013