See

discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/49801023

Association of Olfactory Dysfunction and Brain.

Metabolism in Parkinson's Disease

ARTICLE in MOVEMENT DISORDERS MARCH 2011

Impact Factor: 5.68 DOI: 10.1002/mds.23602 Source: PubMed

CITATIONS READS

25 45

13 AUTHORS, INCLUDING:

Atsushi Takeda Takafumi Hasegawa

National Hospital Organization Sendai-Nishit Tohoku University
167 PUBLICATIONS 7,141 CITATIONS 68 PUBLICATIONS 1,339 CITATIONS

SEE PROFILE SEE PROFILE

Shoki Takahashi
Tohoku University
329 PUBLICATIONS 3,489 CITATIONS

SEE PROFILE

Available from: Takafumi Hasegawa

Retrieved on: 03 November 2015
 RESEARCH ARTICLE

Association of Olfactory Dysfunction and Brain. Metabolism in

Parkinsons Disease
Toru Baba, MD,1 Atsushi Takeda, MD,1* Akio Kikuchi, MD,1 Yoshiyuki Nishio, MD,2 Yoshiyuki Hosokai, MD,2
Kazumi Hirayama, MD,2,3 Takafumi Hasegawa, MD,1 Naoto Sugeno, MD,1 Kyoko Suzuki, MD,2,4 Etsuro Mori, MD,2
Shoki Takahashi, MD,5 Hiroshi Fukuda, MD,6 and Yasuto Itoyama, MD1

1
Department of Neurology, Tohoku University Graduate School of Medicine, Sendai, Japan
2
Department of Behavioral Neurology and Cognitive Neuroscience, Tohoku University Graduate School of Medicine, Sendai, Japan
3
Department of Occupational Therapy, Yamagata Prefectural University of Health Sciences, Yamagata, Japan
4
Department of Clinical Neuroscience, Yamagata University, Graduate School of Medical Science, Yamagata, Japan
5
Department of Diagnostic Radiology, Tohoku University Graduate School of Medicine, Sendai, Japan
6
Department of Nuclear Medicine and Radiology, Institute of Development, Aging and Cancer, Tohoku University, Sendai, Japan

A B S T R A C T : Hyposmia is one of the cardinal early
symptoms of Parkinson disease (PD). Accumulating clini-
cal and pathological evidence suggests that dysfunction
of the olfactory-related cortices may be responsible for
the impaired olfactory processing observed in PD; how-
ever, there are no clear data showing a direct association
between altered brain metabolism and hyposmia in PD.
In this study, we evaluated brain glucose metabolism and
smell-identication ability in 69 Japanese patients with
nondemented PD. Olfactory function was assessed using
the Odor Stick Identication Test for Japanese. The re-
gional cerebral metabolic rate of glucose consumption at
least-squares method. We found that olfactory dysfunc-
tion was closely related to cognitive dysfunction, includ-
ing memory impairment. Moreover, brainbehavior partial
least-squares analysis revealed that odor-identication
performance was closely associated with broad cortical
dysfunction, including dysfunction of the piriform cortex
and amygdala. Our results suggest that the cognitive
decit in olfactory perception is an important aspect
of hyposmia in PD and that this decit is caused by
altered brain metabolism in the amygdala and piriform
cortex. V
C 2011 Movement Disorder Society

rest was measured using 18F-uorodeoxyglucose posi-

tron emission tomography and was analyzed using SPM- Key Words: hyposmia; spatial covariance analysis;
based group comparisons and the brainbehavior partial cluster analysis; PET, amygdala; piriform cortex

Hyposmia is now recognized as one of the major
nonmotor symptoms of Parkinson disease (PD).1,2
Postmortem studies have demonstrated that the olfac-
tory bulb is one of the earliest affected sites in PD pa-
thology.3,4 In addition, studies have demonstrated that
the amygdala and olfactory cortices are preferentially
affected in PD,57 and reduced activities in these
------------------------------------------------------------
*Correspondence to: Atsushi Takeda, Department of Neurology,
Tohoku University Graduate School of Medicine, 1-1 Seiryo-machi,
Aoba-ku, Sendai 980-8575, Japan; atakeda@em.neurol.med.tohoku.
ac.jp
Relevant conicts of interest/nancial disclosures: Nothing to report.
Full nancial disclosures and author roles may be found in the online
version of this article.
Received: 31 August 2010; Revised: 24 November 2010; Accepted:
29 November 2010
Published online 31 January 2011 in Wiley Online Library
(wileyonlinelibrary.com). DOI: 10.1002/mds.23602
regions during olfactory perception have been demon-
strated in PD patients with hyposmia.810 Therefore, it
is plausible that dysfunction of the olfactory-related
cortices may be responsible for hyposmia in PD. To
date, however, no studies have clearly demonstrated a
direct association between altered resting brain metab-
olism and olfactory performance in PD.
In the present study, we investigated the possible
relationships among olfactory impairment, representa-
tive clinical features, and resting-state brain metabo-
lism in PD.
Patients and Methods
Subjects
We studied 69 patients with PD in Hoehn and Yahr
(HY) stages IIII.11,12 Enrolled patients were 5575
years old, with disease onset after age 40. Exclusion

Movement Disorders, Vol. 26, No. 4, 2011 621

B A B A E T A L .
criteria were any history of other neurological or psy-
chiatric diseases, any focal brain lesions diagnosed by
MRI, any family history of parkinsonism, and probable
dementia as dened by a score <25 on the MiniMental
State Examination (MMSE). This study was part of a 3-
year prospective study of patients with PD at Tohoku
University, the design of which has been previously
described.13,14 After a full explanation of the study,
written informed consent was obtained from all partici-
pants according to the Declaration of Helsinki. The
study was approved by the Ethical Committee of
Tohoku University Graduate School of Medicine.
Olfactory Testing
The odor-identication performance of each subject
was measured using the Odor Stick Identication Test
for Japanese (OSIT-J, Daiichi Yakuhin, Co., Ltd., To-
kyo, Japan), which consists of 12 odorants familiar to
the Japanese.15 This test has been successfully applied for
the assessment of odor identication ability in Japanese
PD patients.16 The procedure for the OSIT-J has been
previously described.15 In the present study, the olfaction
stages were dened as PD with severe hyposmia
(PDSH, OSIT-J score <5), PD with moderate hypo-
smia (PDMH, 5
OSIT-J score
7), and PD with nor-
mal olfactory function (PDNO, OSIT-J score >7).
Neuropsychological Assessment
We assessed memory and visuoperceptual abilities
because patients with PD are known to experience dif-
culty in performing tasks that call on these func-
tions.17 We were unable to employ a thorough screen
of cognitive function because of time constraints in
the outpatient clinic. Short-term memory was eval-
uated using the word-recall task of the Alzheimers
Disease Assessment Scale (ADAS), and the total num-
ber of correct answers was dened as the word-recall
score. Visual perception was assessed using the over-
lapping-gure identication test, which consists of 10
cards, each card bearing 4 images of common objects.
In each trial, patients were instructed to nd all 4
images on a card. The total number of correctly iden-
tied objects from 10 trials was dened as the gure reduced between-subject variation in global metabolic
identication score.
Cluster Analysis
Hierarchical clustering with the correlation distance
and Wards method was performed to classify clinical
symptoms of 69 PD patients based on the degree of
similarity. We used the R version 2.10.1 software
environment (R Development Core Team, 2008). The
variables were onset age, disease duration, HY stage,
UPDRS3, levodopa equivalent dose, MMSE, word
recall score, gure identication score, and OSIT-J
score. Next, multiscale bootstrap resampling was
applied for each cluster using the R package Pvclust
622 Movement Disorders, Vol. 26, No. 4, 2011
software18,19 to assess the certainty associated with
each cluster (the ShimodairaHasegawa test). Pvclust
calculates an approximately unbiased probability value
(AU P value). In this study, multiscale random boot-
strap resampling of 10,000 iterations was performed,
and the distance was calculated using Wards method
on a correlation-based dissimilarity matrix. An AU
P > 95% was considered statistically signicant.
PET Imaging Acquisition
Regional cerebral glucose metabolism (rCMRGlc)
was measured using 18F- uorodeoxyglucose positron
emission tomography (18F-FDG PET). Each patient
fasted for at least 5 hours before PET scanning. Scans
were performed using a Siemens biograph Duo PET/
computed tomography scanner (Siemens Medical Sys-
tem Inc., Hoffman Estates, IL). A 185218 MBq injec-
tion of 18F-FDG was administered intravenously under
resting conditions (ie, with eyes closed and wearing an
eye mask). We used a 10-minute static acquisition
protocol beginning 60 minutes after the injection of
18
F-FDG. The in-plane and axial resolutions of the
scanner were 3.38 mm full width at half maximum
(FWHM). Image reconstruction was performed using
an ordered subset expectation maximization (16 sub-
sets) and a 6-iteration reconstruction algorithm (Gaus-
sian lter; lter FWHM, 2.0 mm). Attenuation
correction was performed with the built-in CT scan.
Voxel-Based Comparison of Metabolism
Each scan was preprocessed before statistical analy-
sis using the SPM5 software (Wellcome Department of
Cognitive Neurology, London, UK) running under
MATLAB R2007b (MathWorks Inc., Sherborn, MA).
All images were normalized to the 18F-FDG template
and smoothed with a 10-mm Gaussian kernel. Re-
gional metabolic abnormalities were located in all par-
ticipants and subgroups by comparing their metabolic
rates at each voxel with comparable values determined
for the 11 age-matched control subjects. These com-
parisons were achieved using the 2-sample t test
option in SPM5, with age and sex as covariates. We
rates by proportionally scaling each image. The statis-
tical threshold selected for this analysis was P < .001
(uncorrected), with an extent threshold of 100 voxels.
BrainBehavior Partial Least-Squares Analysis
We then performed brainbehavior partial least-
squares (PLS) analysis to identify distributed patterns of
brain activity that covaried with standardized clinical
ratings (onset age, HY stage, UPDRS3, levodopa equiva-
lent dose, MMSE score, word recall score, gure identi-
cation score, and OSIT-J score) using PLSgui software
(ftp.rotman-baycrest.on.ca/pub/Randy/pls) in the total
PD group (MMSE > 24) and the nondemented PD group
 H Y P O S M I A A N D B R A I N M E T A B O L I S M I N P D

TABLE 1. Demographic and clinical proles of subjects smia (PDMH), and 17 were classied as having PD
with normal olfactory function (PDNO); see Table 1
PDNO PDMH PDSH
(n 17) (n 19) (n 33) P*
and Fig. 1). In the subgroup of 40 nondemented
patients (MMSE > 28), 14 were classied as having
Age, y 62.5 6 7.7 64.1 6 6.7 66.0 6 6.3 .22 PDSH, 14 were classied as having PDMH, and 12
Age of disease 55.9 6 9.3 58.5 6 7.3 61.4 6 7.2 .06 were classied as having PDNO. The distributions of
onset
OSIT-J scores for all cases (MMSE > 24) and for non-
Duration, y 6.7 6 6.4 5.5 6 6.2 4.6 6 3.3 .41
HY scale 2.2 6 0.8 2.4 6 0.6 2.5 6 0.5 .21 demented cases were nearly identical (Fig. 1). Statisti-
UPDRS 3 16.8 6 8.6 19.1 6 7.2 17.2 6 7.1 .61 cally signicant differences (1-way analysis of variance,
L-Dopa equivalent 294.7 6 217.5 303.9 6 264.2 326.5 6 205.4 .88 P < .05) were obtained for MMSE score, word recall
dose score, and gure identication score between the
MMSE 28.8 6 1.4 28.9 6 1.1 27.7 6 1.8 .01
PDSH and PDNO groups (Table 1).
Word recall score 22.1 6 2.7 19.8 6 3.7 17.7 6 3.6 .00
Figure identication 33.7 6 3.2 31.6 6 4.8 29.9 6 5.5 .00
score Clustering Solutions
OSIT-J score 8.8 6 0.9 5.9 6 0.7 2.3 6 1.4 .00 The results obtained using hierarchical clustering
revealed a group of symptoms that demonstrated simi-
Data are given as mean 6 SD.
Abbreviations: HY, Hoehn and Yahr; UPDRS, Unied Parkinsons Disease larity relationships (Fig. 2). One large cluster compris-
Rating Scale; MMSE, MiniMental State Examination; OSIT-J, Odor Stick ing MMSE, OSIT-J, word recall, and gure
Identication Test for Japanese.
*One-way analysis of variance. identication scores was found to be signicant. In
this cluster, the OSIT-J score and the word-recall score
(MMSE > 28). A full explanation of PLS has been pro- showed the closest relationship. Other clinical parame-
vided elsewhere.20 Briey, based on the covariance ters, such as degree of motor impairment and duration
between global metabolism and clinical ratings, PLS of disease, failed to show signicant correlations with
extracts new sets of variables (latent variables) that best olfactory performance.
reect the brainbehavior relationship. Each extracted
latent variable is associated with a brain activity pattern. Voxel-Based Comparison of Metabolism
Each brain voxel has a weight on each latent variable, The results of the between-group comparisons of
known as a salience, which indicates in what way that metabolic proles are shown in Figure 3. Compared
voxel is related to the latent variable. A salience can be with normal controls, the total PD group exhibited
positive or negative depending on whether the voxel rCMRGlc reduction bilaterally in the medial
shows a positive or negative relationship with the overall
pattern identied by the latent variable. Multiplying the
rCMRGlc value in each brain voxel for each subject by
the salience for that voxel, and summing across all vox-
els, gives a brain score for each clinical feature on a
given latent variable. The correlation between brain
scores and clinical ratings indicates the relationship
between brain activity patterns and clinical features. In
this study, the statistical signicance of each latent vari-
able was assessed using a permutation test.20 Five hun-
dred random permutations of data were performed, and
the statistical threshold chosen for this analysis was P <
.05. We also assessed the reliability of the salience (for
particular voxels and clinical ratings) for each latent vari-
able via bootstrap estimation of the standard errors.21
All saliences were submitted to a bootstrap resampling
of 100 iterations. We considered a salience signicant if
the ratio of salience to standard error was greater than
2.0 (bootstrap ratio).

Results
Demographic Data and Subgroup Evaluations
Of the 69 PD patients assessed in this study, 33 were
classied as having PD with severe hyposmia (PDSH),
19 were classied as having PD with moderate hypo-
FIG. 1. Scatter and box plots of the OSIT-J scores in patients with
PD. A scatter plot of OSIT-J scores is shown on the left, and the box
plots of OSIT-J scores in the total PD group (MMSE > 24; open and
closed circles) and the nondemented group (MMSE > 28; open circle)
are presented on the right. The olfaction stages were defined as PD
with severe hyposmia (PD1SH, OSIT-J score <5), PD with moderate
hyposmia (PD1MH, 5
OSIT-J score
7), and PD with normal olfac-
tory function (PD1NO, OSIT-J score >7).

Movement Disorders, Vol. 26, No. 4, 2011 623

B A B A E T A L .
FIG. 2. Hierarchical clustering and ShimodairaHasegawa test of clini-
cal features in the total PD group (n 5 69). Distance was calculated
using Wards method on a correlation-based dissimilarity matrix, and
multiscale random bootstrap resampling of 10,000 iterations was per-
formed. Approximately unbiased probability values (AU P values) are
presented at branch connections as percentages, and cluster labels
are presented below the branches. Clinical features in the same clus-
ter were considered to have high similarities, and clusters with AU P >
95% were considered significant (indicated by the dashed rectangle).
We observed only 1 large cluster, which was composed of olfactory
and cognitive functions.
prefrontal cortex (mPFC), dorsolateral prefrontal cor-
tex (DLPFC), medial occipital cortex, and lateral pari-
eto-temporo-occipital area (Fig. 3a). In subanalyses of
total PD patients, the PDSH group showed broader
occipital hypometabolism (Fig. 3b), whereas the
PDNO group showed rCMRGlc reduction only in
the bilateral DLPFC (Fig. 3c). In further subanalyses
of the nondemented PD patients, neither the total
cases (Fig. 3d) nor the normosmic subgroup (data not
shown) showed apparent metabolic changes, but the
nondemented PDSH subgroup showed rCMRGlc
reduction in the medial occipital cortex (Fig. 3e).
BrainBehavior Partial Least-Squares Analysis
Brainbehavior PLS analysis allowed detection of spe-
cic topographic patterns of metabolic activity, which
primarily reected underlying pathological processes
and enabled quantication of the correlation of each
topographic pattern with clinical features. In 69 PD
patients, 2 signicant topographic patterns were
obtained using this method. Topographic pattern 1a
(permutation P < .001) could explain 58.3% of the cor-
relation matrix for brain metabolism and behavior, and
18.7% could be explained by topographic pattern 2a
(permutation P < .001). In the subanalysis of nonde-
mented cases, similar topographic patterns were
detected; 49.8% of the correlation matrix for brain me-
tabolism and behavior could be explained by topo-
graphic pattern 1b (permutation P < 0.001), and
624 Movement Disorders, Vol. 26, No. 4, 2011
18.1% could be explained by topographic pattern 2b
(permutation P .022). The brain regions constituting
these topographic patterns are shown in Figure 4a,b.
Patterns 1a and 1b were both characterized by posi-
tive salience in the bilateral mPFC, DLPFC, medial
occipital cortex, piriform cortex, cingulate cortex, lat-
eral parieto-occipito-temporal area, and caudate. They
were also characterized by negative salience in the puta-
men, globus pallidus, pons, cerebellum, and paracentral
regions. In addition to these areas, pattern 1b demon-
strated positive salience in the bilateral amygdala. Pat-
terns 2a and 2b, which were essentially identical, were
characterized by positive salience in the bilateral mPFC
and orbitofrontal cortex (OFC) and the antero-medial
temporal and the anterior cingulate regions. The rela-
tionships between these topographic patterns and clini-
cal measures are shown in Figure 4c,d. The OSIT-J and
word recall scores were highly correlated with topo-
graphic patterns 1a and 1b, and onset age was corre-
lated with patterns 2a and 2b (Fig. 4c,d).
Discussion
Although no participants were demented in this
study, the PDSH group showed a tendency toward
slightly more severe impairments in general cognitive,
memory, and visuoperceptual functions compared with
the PDNO group (Table 1). A close correlation
between hyposmia and cognitive impairments in PD
was further supported by cluster analysis and the Shi-
modairaHasegawa test (Fig. 2). Although cluster anal-
ysis has been used in some studies investigating
PD,22,23 the signicance of the clustering results was
difcult to assess. The ShimodairaHasegawa test was
developed as a remedy for this problem and enabled us
to assess the uncertainty of the clustering results by cal-
culating approximately unbiased probability values.18
The present study demonstrated that olfactory, mem-
ory, general intellectual, and visuoperceptual impair-
ments in PD converged into 1 large cluster that was
statistically independent of motor symptoms. This nd-
ing is consistent with past studies that have demon-
strated that olfactory decit is not correlated with
motor impairment.24 Furthermore, in the present study,
olfactory dysfunction was most closely correlated with
memory impairment, which is consistent with a recent
study demonstrating close correlation between olfactory
and memory impairments.25 Based on these observa-
tions, it is plausible that hyposmia in PD is not an inde-
pendent symptom. Indeed, hyposmia may be associated
with other cognitive dysfunctions and may share some
underlying mechanism with memory dysfunction.
To investigate the pathophysiology of hyposmia in
PD, we analyzed the FDG-PET data obtained for our
participants. As glucose metabolism at rest is thought to
be mainly affected by regional synaptic metabolism,26
the pattern of glucose metabolism obtained by PET
 H Y P O S M I A A N D B R A I N M E T A B O L I S M I N P D

FIG. 3. Brain. regions showing reduced glucose metabolism in PD subgroups compared with the normal control group. a: Total PD group showed
reduced metabolism bilaterally in the medial prefrontal cortex, dorsolateral prefrontal cortex (DLPFC), medial occipital cortex, and lateral parieto-
temporo-occipital area. b: Hyposmic PD group demonstrated additional, broader occipital hypometabolism. c: Normosmic PD group showed meta-
bolic reductions only in the bilateral DLPFC. d: Nondemented PD group showed no apparent metabolic changes. e: Nondemented PD1SH group
showed metabolic reductions in the medial occipital cortex (P < .001 uncorrected with an extent threshold of 100 voxels).

seems to reect the distribution of the pathological pro-
cess that alters neuropils and causes functional decits
of the synaptic activities. Therefore, it is plausible that
functional decits and metabolic alterations are closely
connected to each other. Compared with the normal
control group, the total PD group showed mild meta-
bolic reduction in the mPFC, DLPFC, medial occipital
cortices, and lateral parieto-temporo-occipital area
(Fig. 3a). These ndings are consistent with previous
studies14,27,28 and indicate that our participants were

Movement Disorders, Vol. 26, No. 4, 2011 625

B A B A E T A L .

FIG. 4. Correlation between clinical features and topographic patterns. Two orthogonal topographic patterns derived from glucose metabolic scans
at rest in (a) the total PD group (MMSE > 24) and (b) the nondemented PD group (MMSE > 28). Relationships between topographic patterns and
clinical ratings in (c) the total PD group (MMSE > 24) and (d) the nondemented PD group (MMSE > 28). Coordinates along the x, y, and z axes refer
to Montreal Neurological Institute standard stereotactic space. The bootstrap ratio is the ratio of the voxel salience value to its standard error (esti-
mated from 100 bootstrap samples). It is a measure of the stability of high (red scale) and low (blue scale) metabolic values, which constitute the
topographic pattern. The correlation between a topographic pattern and a clinical rating was not considered significant (n.s.) if the 95% CI crossed
zero, as determined by estimation from 100 bootstrap samples.

metabolically typical, as observed in nonadvanced PD.
Furthermore, the hyposmic PD group showed more
prominent occipital hypometabolism, even in nonde-
mented cases (Fig. 3e). These results suggested that ol-
factory dysfunction is a sensitive marker of brain
hypometabolism in PD.
Next, we performed brainbehavior PLS analysis
and identied 2 signicant topographic patterns of
metabolic alteration at rest that correlated with repre-
sentative clinical features in PD, including olfactory
dysfunction (Fig. 4). A topographic pattern of glucose
metabolism at rest is thought to reect a group of
structural and functional types of damage to the neu-
ropil caused by pathological changes. A recent study
used PLS analysis to demonstrate correlations among
topographic patterns and cognitive and mood func-
tions in moderate to advanced PD.17 Other studies
that used similar spatial covariance analysis conrmed
the existence of a disease-specic metabolic pattern at
rest in PD.29,30 In the present study, we demonstrated
that 2 signicant topographic patterns could be
detected even in nonadvanced PD (Fig. 4a,b). Further-
more, we found that the expression of topographic
patterns 1a and 1b was positively correlated with ol-
factory and memory performance. Topographic pat-
terns 1a and 1b were both characterized by altered
metabolism in multiple brain regions, including the
mPFC, DLPFC, piriform cortex, cingulate cortex, lat-
eral parieto-occipito-temporal area, and caudate.
These data suggest that hyposmia and memory impair-
ment are 2 reliable predictors of cortical hypometabo-
lism in PD.

626 Movement Disorders, Vol. 26, No. 4, 2011

 H Y P O S M I A
The nding that olfactory and memory dysfunctions
in PD were correlated with the same metabolic pattern
suggested that these 2 symptoms have a common
pathological etiology. Convergent evidence from path-
ological57 and imaging studies810,31 has suggested
that dysfunction of the amygdala and piriform cortex
is responsible for olfactory impairment in PD. Further-
more, a recent study demonstrated that central cholin-
ergic dysfunction in the amygdala and hippocampus
correlated with olfactory and memory impairment in
nondemented PD.25 Our data showed that the per-
formance of olfaction and memory was highly corre-
lated with topographic patterns 1a and 1b. With
respect to olfaction-related brain regions, patterns 1a
and 1b both included metabolic changes in the piri-
form cortex, and pattern 1b encompassed metabolic
changes in the amygdala (Fig. 4a,b). Previous studies
have suggested that the link between olfaction and
memory is mediated by the amygdala32 and the piri-
form cortex.33 Taken together, these ndings indicate
that dysfunction of these areas, which may be due in
part to cholinergic dysfunction, may be responsible for
the olfactory and memory impairment in nondemented
PD. It was noteworthy that although PLS analysis
demonstrated relative hypometabolism of the piriform
cortex and amygdala in hyposmic PD patients, the
parametric t test implemented in SPM5 failed to detect
signicant metabolic reductions in these areas (Fig.
3b,e). These results are explained by the nding that
multivariate methods such as PLS are more suitable
for assessing metabolic network abnormalities than is
SPMs t test,34 but further study is needed to clarify
these points. Interestingly, the correlation between
metabolic changes in the amygdala and olfactory per-
formance was observed only in topographic pattern
1b, which suggested that the hypometabolism in the
amygdala was one of the earliest metabolic alterations
in the PD brain and reached a nadir at later stages.
Recent clinical studies have demonstrated that
higher onset age is the best predictor of PD pro-
gression rate.35,36 In the present study, onset age
showed comparatively close relationships with all
topographic patterns (Fig. 4c,d), which suggested
that a higher onset age might be associated with
broad cortical metabolic reduction. Furthermore,
recent pathological studies indicated that later-onset
PD patients often exhibit early limbic and neocorti-
cal involvement and develop dementia.37,38 Our
data revealed that topographic patterns 2a and 2b
represented metabolic abnormalities in the limbic
and prefrontal cortices, consistent with those patho-
logical studies and also suggested that metabolic
changes in these areas were present even in nonde-
mented PD patients. Taken together, the present
data suggest that hyposmia, memory impairment,
and later-onset age are associated with brain hypo-
metabolism affecting broad regions and could
A N D B R A I N M E T A B O L I S M I N P D
potentially predict later development of dementia in
PD. Because atrophy correction was not performed
in this study, there is a possibility that our data
may reect the combined effect of atrophy and met-
abolic reduction. However, we veried that all par-
ticipants did not exhibit apparent focal brain
atrophies, and previous studies comparing resting
glucose metabolism with and without voxel-based
atrophy correction revealed that regional changes of
glucose metabolism measured by PET are not sim-
ply a reection of focal brain atrophy.39 Thus, it is
expected that the effect of circumscribed atrophy
on metabolic reduction would be relatively small.
However, further study is needed to conrm this
point.
In conclusion, we investigated the possible relation-
ships among olfactory impairment, representative clin-
ical features, and resting-state brain metabolism in
PD. The results suggested that odor-identication
impairment in PD is closely associated with cognitive
dysfunctions, especially memory impairment. Further
analysis revealed that odor-identication decit and
memory impairment are closely associated with dis-
ease-specic metabolic changes including changes in
the amygdala and piriform cortex. Moreover, we dem-
onstrated that onset age is a highly accurate predictor
of cortical hypometabolism even in nondemented PD
patients. The results of this study shed light on the
complex involvement of the brain in PD and may
facilitate the development of better management pro-
tocols for patients with PD.
References
1. Langston JW. The Parkinsons complex: parkinsonism is just the
tip of the iceberg. Ann Neurol. 2006;59:591596.
2. Haehner A, Boesveldt S, Berendse HW, et al. Prevalence of smell
loss in Parkinsons diseasea multicenter study. Parkinsonism
Relat Disord. 2009;15:490494.
3. Braak H, Del Tredici K, Rub U, de Vos RA, Jansen Steur EN,
Braak E. Staging of brain pathology related to sporadic Parkinsons
disease. Neurobiol Aging. 2003;24:197211.
4. Del Tredici K, Rub U, De Vos RA, Bohl JR, Braak H. Where does
parkinson disease pathology begin in the brain?. J Neuropathol
Exp Neurol. 2002;61:413426.
5. Harding AJ, Stimson E, Henderson JM, Halliday GM. Clinical cor-
relates of selective pathology in the amygdala of patients with Par-
kinsons disease. Brain. 2002;125:24312445.
6. Hubbard PS, Esiri MM, Reading M, McShane R, Nagy Z. Alpha-
synuclein pathology in the olfactory pathways of dementia
patients. J Anat. 2007;211:117124.
7. Silveira-Moriyama L, Holton JL, Kingsbury A, et al. Regional dif-
ferences in the severity of Lewy body pathology across the olfac-
tory cortex. Neurosci Lett. 2009;453:7780.
8. Masaoka Y, Yoshimura N, Inoue M, Kawamura M, Homma I.
Impairment of odor recognition in Parkinsons disease caused by
weak activations of the orbitofrontal cortex. Neurosci Lett. 2007;
412:4550.
9. Takeda A, Saito N, Baba T, et al. Functional imaging studies of
hyposmia in Parkinsons disease. J Neurol Sci. 2010;289:3639.
10. Westermann B, Wattendorf E, Schwerdtfeger U, et al. Functional
imaging of the cerebral olfactory system in patients with Parkin-
sons disease. J Neurol Neurosurg Psychiatry. 2008;79:1924.
Movement Disorders, Vol. 26, No. 4, 2011 627
B A B A E T A L .
11. Hoehn MM, Yahr MD. Parkinsonism: onset, progression and mor-
tality. Neurology. 1967;17:427442.
12. Hughes AJ, Daniel SE, Kilford L, Lees AJ. Accuracy of clinical di-
agnosis of idiopathic Parkinsons disease: a clinico-pathological
study of 100 cases. J Neurol Neurosurg Psychiatry. 1992;55:
181184.
13. Abe N, Fujii T, Hirayama K, et al. Do parkinsonian patients have
trouble telling lies? The neurobiological basis of deceptive behav-
iour. Brain. 2009;132:13861395.
14. Hosokai Y, Nishio Y, Hirayama K, et al. Distinct patterns of re-
gional cerebral glucose metabolism in Parkinsons disease with
and without mild cognitive impairment. Mov Disord. 2009;24:
854862.
15. Saito S, Ayabe-Kanamura S, Takashima Y, et al. Development of a
smell identication test using a novel stick-type odor presentation
kit. Chem Senses. 2006;31:379391.
16. Iijima M, Kobayakawa T, Saito S, et al. Smell identication in Jap-
anese Parkinsons disease patients: using the odor stick identica-
tion test for Japanese subjects. Intern Med. 2008;47:18871892.
17. Mentis MJ, McIntosh AR, Perrine K, et al. Relationships among
the metabolic patterns that correlate with mnemonic, visuospatial,
and mood symptoms in Parkinsons disease. Am J Psychiatry.
2002;159:746754.
18. Shimodaira H. An approximately unbiased test of phylogenetic
tree selection. Syst Biol. 2002;51:492508.
19. Suzuki R, Shimodaira H. Pvclust: an R package for assessing the
uncertainty in hierarchical clustering. Bioinformatics. 2006;22:
15401542.
20. McIntosh AR, Bookstein FL, Haxby JV, Grady CL. Spatial pattern
analysis of functional brain images using partial least squares. Neu-
roimage. 1996;3:143157.
21. McIntosh AR, Rajah MN, Lobaugh NJ. Interactions of prefrontal
cortex in relation to awareness in sensory learning. Science. 1999;
284:15311533.
22. Graham JM, Sagar HJ. A data-driven approach to the study of het-
erogeneity in idiopathic Parkinsons disease: identication of three
distinct subtypes. Mov Disord. 1999;14:1020.
23. Lewis SJ, Foltynie T, Blackwell AD, Robbins TW, Owen AM,
Barker RA. Heterogeneity of Parkinsons disease in the early clini-
cal stages using a data driven approach. J Neurol Neurosurg. Psy-
chiatry 2005;76:343348.
24. Doty RL, Deems DA, Stellar S. Olfactory dysfunction in parkin-
sonism: a general decit unrelated to neurologic signs, disease
stage, or disease duration. Neurology. 1988;38:12371244.
628 Movement Disorders, Vol. 26, No. 4, 2011
25. Bohnen NI, Muller ML, Kotagal V, Koeppe RA, Kilbourn MA,
Albin RL et al. Olfactory dysfunction, central cholinergic integrity
and cognitive impairment in Parkinsons disease. Brain. 2010;133:
17471754.
26. Auker CR, Meszler RM, Carpenter DO. Apparent discrepancy
between single-unit activity and [14C]deoxyglucose labeling in optic
tectum of the rattlesnake. J Neurophysiol. 1983;49:15041516.
27. Eberling JL, Richardson BC, Reed BR, Wolfe N, Jagust WJ. Corti-
cal glucose metabolism in Parkinsons disease without dementia.
Neurobiol Aging. 1994;15:329335.
28. Firbank MJ, Colloby SJ, Burn DJ, McKeith IG, OBrien JT. Re-
gional cerebral blood ow in Parkinsons disease with and without
dementia. Neuroimage. 2003;20:13091319.
29. Eidelberg D, Moeller JR, Dhawan V, et al. The metabolic topogra-
phy of parkinsonism. J Cereb Blood Flow Metab. 1994;14:783801.
30. Tang CC, Poston KL, Eckert T, et al. Differential diagnosis of par-
kinsonism: a metabolic imaging study using pattern analysis. Lan-
cet Neurol. 2010;9:149158.
31. Wattendorf E, Welge-Lussen A, Fiedler K, et al. Olfactory impair-
ment predicts brain atrophy in Parkinsons disease. J Neurosci.
2009;29:1541015413.
32. Herz RS, Eliassen J, Beland S, Souza T. Neuroimaging evidence for
the emotional potency of odor-evoked memory. Neuropsychologia.
2004;42:371378.
33. Gottfried JA, Smith AP, Rugg MD, Dolan RJ. Remembrance of
odors past: human olfactory cortex in cross-modal recognition
memory. Neuron. 2004;42:687695.
34. Eidelberg D. Metabolic brain networks in neurodegenerative disor-
ders: a functional imaging approach. Trends Neurosci. 2009;32:
548557.
35. Post B, Merkus MP, de Haan RJ, Speelman JD. Prognostic factors
for the progression of Parkinsons disease: a systematic review.
Mov Disord. 2007;22:18391851.
36. Obeso JA, Rodriguez-Oroz MC, Goetz CG, et al. Missing pieces in
the Parkinsons disease puzzle. Nat Med. 2010;16:653661.
37. Halliday G, Hely M, Reid W, Morris J. The progression of pathol-
ogy in longitudinally followed patients with Parkinsons disease.
Acta Neuropathol (Berl). 2008;115:409415.
38. Halliday GM, McCann H. The progression of pathology in Parkin-
sons disease. Ann N Y Acad Sci. 2010;1184:188195.
39. Ibanez V, Pietrini P, Alexander GE, Furey ML, Teichberg D, Raja-
pakse JC et al. Regional glucose metabolic abnormalities are not
the result of atrophy in Alzheimers disease. Neurology. 1998;50:
15851593.