NeuroImage 59 (2012) 2743–2750

Contents lists available at SciVerse ScienceDirect

NeuroImage

j ourna l homepage: www.e lsev ie r .com/ locate /yn img

Cortical hypoperfusion in Parkinson's disease assessed using arterial spin labeledperfusion MRI

María A. Fernández-Seara a,⁎, Elisa Mengual b,c, Marta Vidorreta a, Maite Aznárez-Sanado a,Francis R. Loayza a, Federico Villagra a, Jaione Irigoyen a, María A. Pastor a

a Neuroimaging Laboratory, Division of Neuroscience, Center for Applied Medical Research (CIMA), University of Navarra, Pamplona, Spainb Neuroanatomy of Basal Ganglia Laboratory, Division of Neuroscience, Center for Applied Medical Research (CIMA), University of Navarra, Pamplona, Spainc Department of Anatomy, Medical School, University of Navarra, Pamplona, Spain

⁎ Corresponding author at: Center for Applied MedicalPio XII, 55, 31008 Pamplona, Spain. Fax: +34 948 194715

E-mail address: mfseara@unav.es (M.A. Fernández-S

1053-8119/$ – see front matter © 2011 Elsevier Inc. Alldoi:10.1016/j.neuroimage.2011.10.033

a b s t r a c t

a r t i c l e i n f o

Article history:Received 29 August 2011Revised 6 October 2011Accepted 10 October 2011Available online 18 October 2011

Keywords:PerfusionArterial spin labelingParkinson's diseaseGlobal mean normalization

Alterations in cerebral perfusion and metabolism in Parkinson's disease have been assessed in several studies,using nuclear imaging techniques and more recently magnetic resonance imaging. However, to date there isno consensus in the literature regarding the extent and the magnitude of these alterations. In this work, ar-terial spin labeled perfusion MRI was employed to quantify absolute cerebral blood flow in a group of early-to-moderate Parkinson's disease patients and age-matched healthy controls. Perfusion comparisons betweenthe two groups showed that Parkinson's disease is characterized by wide-spread cortical hypoperfusion. Sub-cortically, hypoperfusion was also found in the caudate nucleus. This pattern of hypoperfusion could be relat-ed to cognitive dysfunctions that have been previously observed even at the disease early stages. The presentresults were obtained by means of whole brain voxel-wise comparisons of absolute perfusion values, usingstatistical parametric mapping, thus avoiding the potentially biased global mean normalization procedure.In addition, this work demonstrates that between-group comparison of relative perfusion values after globalmean normalization, introduced artifactual relative perfusion increases, where absolute perfusion was in factpreserved. This has implications for perfusion studies of other brain disorders.

© 2011 Elsevier Inc. All rights reserved.

Introduction

Cerebral perfusion disturbances have been observed in a variety ofneurodegenerative and psychiatric disorders. Due to the well-knownphenomenon of neurovascular coupling, changes in cerebral bloodflow (CBF) can be directly related to regional metabolism and neuralactivity (Raichle, 1998). Thus, CBF measurements could potentiallybecome biomarkers for early disease diagnosis, and also serve to followup disease progression and treatment response.

Cerebral glucose consumption (CMRglc) and perfusion have beenpreviously assessed in Parkinson's disease (PD) using positron emissiontomography (PET) (Eckert et al., 2007), single photon emission comput-ed tomography (SPECT) (Amorim et al., 2007) and,more recently, perfu-sion magnetic resonance imaging (MRI) (Melzer et al., 2011). Previousstudies with PET and SPECT have found alterations in metabolism andperfusion in PD with a pattern of relative cortical decreases concurrentwith subcortical increases extending into large areas of the basal gangliaand thalamus (Eckert et al., 2007). However, recent publications havequestioned the presence of such subcortical increases, arguing thatthey are an artifact introduced by global mean normalization of the

Research, University of Navarra,.eara).

rights reserved.

data (Borghammer et al., 2010; Borghammer et al., 2009a; Borghammeret al., 2009b). Normalization is commonly used in PET and SPECT studiesand indeed required when absolute quantification is not feasible. Evenwhen absolute quantification can be performed, both perfusion and glu-cose consumption measurements from PET and SPECT have great inter-individual variabilitywith large coefficients of variation (Borghammer etal., 2008), which makes the detection of small between-group differ-ences difficult. However, normalization to the globalmean is only appro-priatewhen there is no difference in globalmean values between groups,an assumption that seems to be violated in comparisons of PD patientsand healthy controls (Borghammer et al., 2010).

Although anatomical studies in experimental models of PD using2-deoxyglucose (2DG) autoradiography have also reported increasedmetabolism in several subcortical and brainstem structures (see forreview (Borghammer et al., 2009a; Obeso et al., 2008)), these struc-tures are generally too small to be detected with the resolutionachievable using conventional PET and SPECT cameras (Cassidy andRadda, 2005).

Arterial spin labeled (ASL) perfusionMRI (Detre et al., 1992;Williamset al., 1992) has been recently introduced as a non-invasive alternativefor perfusion measurements in PD (Fernandez-Seara et al., 2009;Kamagata et al., 2011; Ma et al., 2010; Melzer et al., 2011). The ASLtechnique utilizes electromagnetically labeled arterial blood wateras an endogenous tracer, yielding quantitative CBF measurements

2744 M.A. Fernández-Seara et al. / NeuroImage 59 (2012) 2743–2750

in well-characterized physiological units of mL·100 g−1 min−1. Re-cent technical advances in ASL have increased the sensitivity of themethod and, as a result, absolute CBF measurements can now beobtained with high reliability in both young and elderly subjects(Xu et al., 2010). Using quantitative ASL methods, a recent studyhas reported absolute perfusion decreases in PD in diverse corticalareas, while subcortical perfusion appeared preserved (Melzer etal., 2011). This study, while being the largest study of perfusion inPD to date, involved a heterogeneous group of patients that also in-cluded patients with dementia.

In the current study, an optimized ASL technique (Fernandez-Seara et al., 2008a) has been used to quantify absolute CBF in a homo-geneous group of early-to-moderate PD patients and age-matchedhealthy controls. We hypothesized that perfusion alterations in PDwould consist in cortical decreases and that those alterations couldbe detected using statistical parametric mapping analysis with thistechnique. In addition, comparisons of relative CBF values were alsocarried out to evaluate the effect of global mean normalization.

Materials and methods

Subjects

Twenty-five early to moderate stage PD patients without demen-tia and thirty-four healthy age-matched controls participated in thestudy, approved by the Ethics Research Committee of the Universityof Navarra, after signing a written informed consent. The healthy vol-unteers had no past or present history of neurological or psychiatricdisorders. The patients had been diagnosed with idiopathic Parkinson'sdisease according to the United Kingdom Parkinson's Disease SocietyBrain Bank criteria (Hughes et al., 1992). The diagnosis was confirmedby a neurologist, specialized in movement disorders. They were stagedaccording to the Hoehn and Yahr scale and their motor symptomswereevaluated during the ON state, under medication, using the motor ex-amination score of the Unified Parkinson's Disease Rating Scale part III(motor UPDRS) (Martinez-Martin et al., 1994). To assess cognitive im-pairment, the Mini Mental State Examination (MMSE) was used in thepatient group (Folstein et al., 1975). Only 2 patients showedmild cogni-tive impairment as determined by their scores in the test. At the time ofthe study, the patientsweremedicated as follows (medication (numberof patients)): dopamine receptor agonist (3); agonist+monoamine ox-idase B (MAO-B) inhibitor (2); levodopa (1); levodopa+agonist (8);levodopa+MAO-B inhibitor (2); levodopa+agonist+MAO-B inhibi-tor (7); levodopa+agonist+amantadine (1); levodopa+agonist+amantadine+MAO-B inhibitor (1). Patients were scanned under med-ication in ON state. See Table 1 for detailed information on the subjects.

Scanning protocol

Studies were performed on a 3 T Trio TIM (Siemens AG, Erlangen,Germany) using a 12-channel head array. Subjects were instructed toremain still with eyes open during the scanning session. T1-weighted

Table 1Demographic and clinical characteristics of the study groups.

Patients Controls

N (female:male) 25 (7:18) 34 (12:22)Age, years: mean (SD); [range] 63.2 (6.6); [53–76] 63.5 (6.6); [53–76]UPDRS ON: mean (SD); [range] 12.5 (5.3); [4–22]Disease duration, years: mean(SD); [range]

5.6 (3.4); [1.5–12]

Hoehn andYahr stage ON: mean(SD); [range]

1.84 (0.37); [1–2]

MMSE score ON: mean (SD); [range] 27.9 (2.3); [23–30]Initially affected body side [Right:Left] [15:10]

anatomical images were acquired with a MPRAGE sequence, with thefollowing imaging parameters: resolution=1 mm isotropic,FOV=192×256 mm2, matrix=192×256, 160 axial slices, TR/TE/TI=1620/3.87/950 ms, flip angle=15°.

Perfusion MRI data were acquired using a pseudo-continuous ASL(PCASL) technique with a background-suppressed 3D GRASE read-outsequence, modified to achieve late inflow delay (Fernandez-Seara etal., 2008b), with imaging parameters: resolution=4×4×7mm3,FOV=250×188×112 mm3, 16 nominal partitions with 13%oversampling, 5/8 partial Fourier, measured partitions=11, matrixsize=64×49, BW=2790 Hz/pixel, GE spacing=0.4 ms (with rampsampling), SE spacing=26 ms, read-out time=270 ms, TE=55.7 msand TR=3.5 s. Two non-selective inversion pulses (15.35 ms durationand 220 mG amplitude) were added for background suppression (BS)with inversion times TI1=1800ms; TI2=500 ms. The PCASL pulseconsisted of 1536 selective radio-frequency (RF) pulses (Hanning win-dow, peak B1=53mG, duration=500 μs and G=0.6 G/cm, labelingduration=1600 ms, post-labeling delay=1530 ms). For the control,the RF phase alternated from 0 to 180°. Bipolar gradients (b=5 s/mm2) were added to suppress intravascular signal. The inversionplanewas offset 8 cm from the center of the FOV in the head–foot direc-tion, so that it was located at the base of the cerebellum to achieve goodlabeling efficiency. The imaging slab covered the cerebrum and thesuperior part of the cerebellum. 50 label/control pairs were acquiredin a scan time of 6 min. A short scan of 5 label/control pairs was per-formed using the same sequence without background suppression toacquire control images needed for calculation of CBF. Raw data weresaved and reconstructed off-line.

Data processing and analysis

Data analysis was performed using SPM (version 8, WellcomeTrust Center for Neuroimaging, University College London, UK) andcustom scripts in Matlab (Mathworks, MA, USA).

Voxel-based morphometryThe T1-weighted anatomical images were segmented, using a uni-

fied segmentation procedure (Ashburner and Friston, 2005) and astudy-specific gray matter template was generated using the Diffeo-morphic Anatomical Registration through Exponentiated Lie algebra(DARTEL) algorithm (Ashburner, 2007). The individual gray matterimages were normalized to this template and modulated, preservingthe total amount of signal from each region in the image. The modu-lated normalized gray matter images were mapped into MNI spaceusing an affine transformation and spatially smoothed with an8 mm full-width at half-maximum Gaussian kernel. Total intra-cranial volume of each subject was computed by adding voxels inthe segmented gray and white matter and cerebrospinal fluid images.

Voxel-wise comparison of gray matter volume between patientand control groups was carried out using a two-sample t-test (un-equal variances), with age, sex and intra-cranial volume as covariatesof no-interest. A cluster-corrected p value of 0.05 was employed, de-termined using a non-stationarity correction toolbox (Hayasaka et al.,2004).

Perfusion dataRaw ASL images acquired with BS and images acquired without BS

were realigned and co-registered to the anatomical dataset. Forty-nine perfusion weighted images were obtained by pair-wise subtrac-tion of the background-suppressed label and control images, after dis-carding the first pair, because the signal had not reached the steadystate. Perfusion weighted images identified as outliers were eliminatedfrom the time series and replaced by the average of the two adjacentimages. A perfusion image was considered an outlier when the globalperfusion signal differed from the mean of the perfusion image seriesby ±2 standard deviations (Wang et al., 2008). A mask was generated

2745M.A. Fernández-Seara et al. / NeuroImage 59 (2012) 2743–2750

from the unsuppressed mean control image, using a brain extractiontool in MRIcron (www.mricro.com) and manual adjustment whenneeded. This mask was applied to the mean perfusion weighted imageto remove out-of-brain voxels. Subsequently, a CBFmapwas computedfrom the masked mean perfusion weighted image, based on a singlecompartment ASL perfusion model, as described in Wang et al.(2005).A whole brain mean CBF value was calculated for each subject by aver-aging CBF voxel values. Differences in whole brain mean CBF betweenpatient and control groups were assessed by means of a linear regres-sion analysis with group, age and sex as independent variables.

The CBF maps were normalized to the study-specific gray mattertemplate previously generated, and mapped into MNI space usingan affine transformation, re-sliced to an isotropic voxel size of 2 mmand spatially smoothed using an 6 mm full-width at half-maximumGaussian kernel.

Whole brain voxel-wise statistical analysis of the CBF data was per-formed using a two-sample t-test (unequal variances), with age and sexas covariates of no-interest. A gray matter mask generated from thestudy-specific graymatter template, including voxels with a probabilityof being gray matter larger than 5%, was applied. Coverage of the cere-bellumwas variable from subject to subject, so this regionwas excludedfrom the analysis using an additional mask. Areas of perfusion changebetween patient and control groups were identified using a cluster-corrected p valueb0.05, determined using a non-stationarity correctiontoolbox (Hayasaka et al., 2004). An exploratory analysiswas also carriedout with a significance value of pb0.005, uncorrected for multiple com-parisons and a minimum cluster size of 30. Anatomical labeling wasperformed using the SPM Anatomy toolbox (Eickhoff et al., 2005) andthe Talairach Daemon (Lancaster et al., 2000).

In order to determine the magnitude of the perfusion differencesbetween patients and controls in anatomical structures, sphericalregions of interest (ROIs) (radius=3 mm) were defined from thethresholded SPM (t) map, centered in every local maximum. CBFvalues in these ROIs were extracted from the normalized CBF imagesfor each subject. Individual CBF values were averaged to obtain groupmeans. Percentage CBF difference was computed as the difference ingroup means divided by the mean CBF in the control group.

In addition to the whole brain voxel-wise comparison, a region ofinterest (ROI) statistical analysiswas also carried out to further examineperfusion changes in subcortical structures. CBF values were measuredin the basal ganglia (caudate nucleus, putamen and external and

Table 2Areas of perfusion deficit in PD patients (non-stationarity corrected cluster p valueb0.05). Lorected voxel p value. % CBF decrease: CBF decrease in patients relative to controls, computed

Cluster size Region (BA) MNI coordi

9328 Left lingual gyrus (18) −2Left lingual gyrus (18) −2Right middle occipital gyrus 36Left precuneus, superior parietal lobule (7A) −8Right middle occipital gyrus 36Right middle occipital gyrus 36Right postcentral gyrus 12Right superior occipital gyrus 18Left precuneus, superior parietal lobule (7P) −8Right superior occipital gyrus 28Left paracentral lobule − 14

3500 Right superior frontal gyrus (6) 12Right superior frontal gyrus (6) 6Right superior frontal gyrus (8) 34Right superior frontal gyrus (6) 2Left superior frontal gyrus (6) −6Left superior frontal gyrus (6) −24Left precentral gyrus (44) −44Left middle frontal gyrus (6) −36Left superior medial gyrus (8) −8Right middle frontal gyrus (6) 46Right middle frontal gyrus (8) 54

internal globus pallidus) and thalamus, in unilateral ROIs that com-prised the full extent of each structure as defined by the masks in theWFU PickAtlas toolbox (Maldjian et al., 2003). Unilateral CBF valuesextracted from patients with predominant left body side affectationwere transposed to the other hemisphere, to simulate that all pa-tients had right-sided symptoms, thus avoiding potential confoundsof asymmetric brain affectation. Significant differences in CBFbetween patient and control groups were assessed in a linear regres-sion analysis with group, age and sex as independent variables, witha Bonferroni corrected p value of 0.05. The ROI statistical tests werecarried out using SPSS 15.

Finally, in order to evaluate the effects of normalizing the data tothe global mean, a second whole brain voxel-wise statistical analysisof the CBF data was performed using global mean normalization,implemented as proportional scaling in SPM.

Results

Perfusion data

CBF quantificationThe calculated whole brain mean CBF in the control group was

38.9±6.2 mL·100 g−1 min−1 (mean±standard deviation). In thepatient group, the mean CBF was 36.9±6.5, a value reduced by 5.1%with respect to the control group. However, the linear regressionanalysis showed that group was not a significant predictor of wholebrain mean CBF (p=0.196).

Parkinson's disease related perfusion deficitsThe voxel-wise analysis of absolute CBF maps revealed areas of

hypoperfusion in the cortex of PD patients. These areas are listed inTable 2 and depicted in Fig. 1. Bilateral perfusion decreases werefound in frontal, parietal and occipital areas. In the frontal lobe, perfu-sion was reduced in the superior and middle frontal gyri as well as inthe precentral gyrus, affecting Brodmann areas (BA) 6, 8 and 9. Area 6was affected both on the lateral surface of the brain, which function-ally corresponds to the premotor cortex (Matelli et al., 2004), as wellas in the mesial wall of the hemisphere, where the region of hypoper-fusion was located rostral to the level of the anterior commisure, cor-responding to the pre-supplementary motor area (pre-SMA) (Picardand Strick, 2001; Zilles et al., 1996). In contrast, perfusion was

cal maxima are shown in MNI coordinates (in mm). BA: Brodmann area. p unc.: uncor-in a spherical ROI (radius=3 mm) centered in the coordinates of the local maximum.

nates x, y, z T p unc. % CBF decrease

−94 −12 3.63 0.0005 36.35−92 −18 3.51 0.0005 41.37−84 22 3.50 0.0005 19.18−60 70 3.49 0.0005 33.99−80 30 3.47 0.001 21.54−82 26 3.44 0.001 19.68−34 74 3.38 0.001 27.48−84 20 3.33 0.001 22.69−78 42 3.30 0.001 20.80−80 24 3.20 0.001 22.69−32 74 3.18 0.001 25.544 68 3.30 0.001 35.932 68 3.29 0.001 31.3628 56 3.29 0.001 33.7614 66 3.29 0.001 37.9416 64 3.28 0.001 35.136 68 3.27 0.001 35.828 48 3.21 0.001 19.034 56 3.17 0.001 20.8046 52 3.09 0.002 29.9412 50 3.04 0.002 22.6324 40 2.97 0.002 23.33

Fig. 1. Surface projection of clusters with significantly decreased absolute CBF in PD patients compared to healthy controls.

2746 M.A. Fernández-Seara et al. / NeuroImage 59 (2012) 2743–2750

preserved in the caudal portion of themesial area 6, where SMA properhas been located (Picard and Strick, 2001), and in most of area 4. Perfu-sion deficits extended to areas 8 and 9 in the prefrontal cortex (Petridesand Pandya, 2004). In the parietal lobe, decreases were located in thelateral superior parietal lobule and medial precuneus (BA 7), and in asmall region of the postcentral gyrus. In the occipital lobe, decreaseswere found in the lingual, superior and middle occipital gyri, thus

Fig. 2. Unthresholded T statistic maps corresponding to the contrast patients minus contronegative T values are displayed, representing increased and decreased CBF in the patient g

affecting both primary and mainly secondary visual cortices. Themagnitude of the perfusion deficits ranged from 20 to 40% relative tothe control group CBF. These values are in the range of perfusiondecreases measured in a recent ASL study of PD (Melzer et al., 2011).The exploratory analysis using an uncorrected voxel-wise p valuerevealed two additional clusters of decreased perfusion in the rightcaudate nucleus (k=51, [x, y, z]=[16, 6, 12], T=3.69, uncorrected

ls, obtained by comparison of absolute CBF (a) and relative CBF (b). Both positive androup, respectively. The value of the T statistic is shown on the color scale.

Table 3Areas of atrophy in PD patients (non-stationarity corrected cluster p valueb0.05). Local maxima are shown in MNI coordinates (in mm). BA: Brodmann area. p unc.: uncorrectedvoxel p value.

Cluster size Region (BA) MNI coordinates x, y, z T p unc.

12,037 Right precentral gyrus (6) 18 −26 74 4.02 0.0005Right precentral gyrus (6) 26 −24 66 3.65 0.0005Right precuneus, superior parietal lobule (7A) 12 −57 45 3.64 0.0005Left superior frontal gyrus (6) 0 0 68 3.54 0.0005Left superior parietal lobule (7A) −26 −65 59 3.51 0.0005Left precentral gyrus (6) −29 −5 63 3.44 0.001Left superior frontal gyrus (6) −26 −3 60 3.43 0.001Left paracentral lobule (6) −11 −21 72 3.43 0.001Left paracentral lobule (6) −14 −26 75 3.41 0.001Left superior parietal lobule (7A) −15 −65 50 3.36 0.001Left precentral gyrus (6) −30 −20 60 3.04 0.002

2747M.A. Fernández-Seara et al. / NeuroImage 59 (2012) 2743–2750

voxel p valueb0.0005, % CBF decrease=14.28; k=32, [x, y, z]=[12, 20,−4], T=3.08, uncorrected voxel p valueb0.002, % CBF de-crease=14.57). The results of voxel-wise comparisons did not showany area of increased perfusion in the patients compared to the controlsubjects. The statistical analysis in the subcortical ROIs showed no sig-nificant differences in CBF between patient and control groups in thebasal ganglia or thalamus, although there were trends of decreasedCBF in both left and right caudate nuclei (for details, see SupplementaryTable 1).

Fig. 3. (a) Surface projection of clusters with significantly decreased gray matter volume in P(in blue) and clusters of significant atrophy (in red), superimposed on anatomical T1-weigh

Effects of global mean normalizationFig. 2 shows the T statistic maps without thresholding for the con-

trast patients minus controls, obtained in the comparison of absoluteCBF values (Fig. 2a) and relative CBF values (Fig. 2b). Note that bothpositive and negative T values are displayed, representing increasedand decreased CBF in the patient group, respectively.

Visual inspection of the unthresholded T statisticmaps revealed thatglobal mean normalization led to the appearance of large areas of arti-factual hyperperfusion in the patient group (Fig. 2b), in regions where

D patients compared to healthy controls. (b) Clusters of significant hypoperfusion in PDted images. Regions of hypoperfusion and atrophy overlap are shown in pink.

2748 M.A. Fernández-Seara et al. / NeuroImage 59 (2012) 2743–2750

absolute perfusion was in fact conserved (Fig. 2a). Increased absoluteperfusion was only observed in two bilateral subcortical clusters cover-ing a small fraction of the globus pallidus and a small fraction of thethalamus, with the local maxima located in the ventral lateral nucleusof the thalamus in the left hemisphere and in the lateral globus pallidusin the right hemisphere. Additional clusters of cortical hyperperfusionwere found in the insulae and hippocampi (see Fig. 2a). Note thatthese perfusion increases were not significant.

Voxel-based morphometry

Gray matter atrophy in the patient group was found in frontal andparietal areas (see Table 3 and Fig. 3a). In the frontal lobe, atrophicareaswere located in the superior frontal and precentral gyri, affectingthe premotor cortex, pre-supplementary and supplementary motorareas and a small region of the primary motor cortex. In the parietallobe, atrophy was detected in the superior parietal lobule. Atrophicareas overlapped with areas of low cortical perfusion in the premotorcortex, the pre-supplementarymotor area and themesial superior pa-rietal lobule (shown in pink in Fig. 3b).

Discussion

Assessment of absolute cerebral perfusion in PD using the ASLtechnique yielded a pattern of cortical perfusion deficit, affectingfrontal, parietal and occipital areas. Our results largely confirm priorstudies of perfusion and metabolism measuring absolute CBF orCMRglc (Berding et al., 2001; Bohnen et al., 1999; Hu et al., 2000;Mito et al., 2005). In particular, they are mostly in agreement withthe recent study of Melzer et al.(2011) that evaluated perfusionusing ASL in a large and heterogeneous group of PD patients, includ-ing patients with dementia. In their study however, Melzer et al. usedprincipal component analysis, which required the a priori subtractionof the data mean. This technique yielded a Parkinson's disease relatedperfusion pattern, reproducing results obtained in previous studies ofPD employing network analysis (Poston and Eidelberg, 2009). Althoughthey were able to quantify absolute perfusion in regions of interest de-fined by this pattern, the pattern obtained can be potentially biased bythe de-meaning procedure, given the systematic group differencesthat exist in the global mean (Borghammer et al., 2010).

Artifacts introduced by global mean normalization

The results of our statistical analysis showed that CBF comparisonsbetween patients and controls, using absolute and relative values,yielded vastly different T statistic maps. Examination of these mapsrevealed that global mean normalization introduced artifactual in-creases in perfusion in the patient group, in areaswhere absolute perfu-sion was in fact conserved, and also reduced the statistical significanceof perfusion decreases. These effects are likely due to a systematic re-duction in global perfusion in the patient group, where ratio normaliza-tion by smaller global values yields higher relative values, thus in areaswhere there is no true perfusion change between groups, relative perfu-sion appears to be increased in the patient group and on the other hand,in areaswhere there is a true disease related perfusion decrease, the dif-ference in relative values is reduced and therefore the statistical powerto detect this difference is lower. In our study, we measured a 5% de-crease in global CBF in the PD patients compared to the control group,although this difference did not reach a significant level, likely due toa lack of statistical power. However, Borghammer et al.(2010) havedemonstrated in meta-analyses of perfusion and metabolism studies,that global levels of CBF and CMRglc are generally reduced in PD. Our re-sults help to clarify the controversy existing in the literature regardingthe interpretation of changes in relative perfusion and metabolism inParkinson's disease (Borghammer et al., 2009a; Borghammer et al.,2009b; Ma et al., 2009). They also support the argument in favor of

measuring physiological parameters in absolute values (Grunder,2009). Arterial spin labeled perfusionMRI allows this absolute quantifi-cation, without requiring experimentally demanding procedures.

Cortical hypoperfusion in PD

Our results show perfusion deficits in prefrontal areas (BA 8 and9), dorsal premotor cortex (PMd) and pre-SMA, posterior parietalcortex and parieto-occipital areas. Findings of perfusion changes inthe pre-supplementary motor area in PD patients have been variable,with some authors reporting hypoperfusion (Kikuchi et al., 2001),others reporting hyperperfusion (Tang et al., 2010) and most report-ing no change (Melzer et al., 2011). However, our finding is highlysignificant, with several maxima of hypoperfusion located in thisarea, which is also affected by gray matter volume loss, according tothe morphometry study. Remarkably, a selective loss of pyramidalneurons has been reported in pre-SMA in postmortem brain tissueof PD patients, while all other nonprimary motor cortices were spared(MacDonald and Halliday, 2002). This finding highlights the relevanceof this cortical area in the course of the disease. Anatomical and func-tional data suggest that pre-SMA is more like a prefrontal area than amotor area. In fact, only pre-SMA is interconnected to the prefrontalcortex, contrarily to SMA proper (Bates and Goldman-Rakic, 1993;Lu et al., 1994; Luppino et al., 1993). Functionally, activation of pre-SMA has been associated with cognitive aspects of a variety of tasks(see for review (Picard and Strick, 2001)). These features are alsoshared by the rostral portion of the PMd (Picard and Strick, 2001)which falls within the areas of hypoperfusion detected in the presentwork. Through its connections with the prefrontal cortex, pre-SMAmay have access to the executive-control network that links dorsolat-eral frontal and parietal cortices (Seeley et al., 2007), areas that werealso affected by perfusion deficits. Subcortically, this executive-control network is functionally coupled to the dorsal caudate nucleus,which also showed hypoperfusion. Disruption in this functional net-work may underlie the cognitive deficits that are already present inthe early stages of the disease, affecting specific domains such asworking memory, attention and visuo-spatial processing (Caballol etal., 2007; Possin, 2011). It will be of great interest in future studiesto assess quantitatively specific neuropsychological functions of ourPD patients, in order to determine possible correlations betweenthese cognitive measures and specific perfusion disturbances. Whythese areas show higher or earlier vulnerability as compared toothermotor areas like SMA proper, or primarymotor and sensory cor-tices, is not known. However, pathological studies on disease progres-sion reveal that these primary cortical areas are the last ones to beaffected (Braak et al., 2003).

The cortical perfusion deficit is probably related to gray matter at-rophy in some areas, as shown by the overlap of atrophy and perfu-sion decreases found in our patient group, in the premotor and pre-supplementary motor areas in the frontal lobe, and in the left superiorparietal lobule. However, the perfusion deficit is more extensive thanthe loss of gray matter volume. On the other hand, a few atrophicareas do not appear to be affected by reduced perfusion.

Morphometry studies in early to moderate stage PD patients with-out dementia have reported slight or no gray matter atrophy, whilestudies in late stage PD patients with dementia have found quite ex-tensive gray matter loss in temporal, frontal, occipital and parietalareas (Burton et al., 2004). In our study, atrophy was not present inthe occipital lobe and affected only a small area of the parietal lobe.On the other hand, cortical perfusion appeared to be decreased inmost of these cortical areas, except for the temporal lobes, even thoughpatients were at an early tomoderate stage of the disease. These resultssuggest that functional changes precede structural changes. Thus perfu-sion could be a more sensitive biomarker than gray matter loss in earlyPD.

2749M.A. Fernández-Seara et al. / NeuroImage 59 (2012) 2743–2750

The etiology of the observed cortical hypometabolism remains tobe elucidated. Although degeneration of the nigrostriatal pathway,with the consequent reduction of dopamine in the striatum is the bet-ter known pathological process in Parkinson's disease, there is evi-dence that other nondopaminergic transmitter systems are affectedearly during the disease course, including serotonergic and noradren-ergic systems (Obeso et al., 2010). Cholinergic denervation in pri-mates, following lesions in the basal forebrain, has been reported toinduce wide-spread hypometabolism in the cortex (Kiyosawa et al.,1989). Shimada et al. (2009) have recently shown cholinergic deficitsin the cerebral cortex of PD patients without dementia, which weremost significant in the medial occipital cortex. According to ourdata, the largest perfusion deficits are also localized in the occipitalcortex. While this could be a coincidence, it could alternatively reflectan association between cholinergic dysfunction and cortical hypoper-fusion. Other hypotheses have been proposed that involved deficitsin other neurotransmitter systems (Borghammer et al., 2010), corti-cal Lewy body pathology or Alzheimer pathology (Liepelt et al.,2009), although these are more likely to be relevant at the late dis-ease stages, and in a PD population with signs of dementia (Braaket al., 2003).

Subcortical perfusion abnormalities in PD

The whole brain voxel-wise comparison of absolute CBF valuesrevealed two clusters of decreased perfusion in the caudate nucleusof PD patients, however these clusters did not survive a cluster-cor-rected threshold. The ROI analysis results also supported the findingsof decreased perfusion in the caudate nucleus, in agreement withBerding et al.(2001) and more recently Borghammer et al.(2010),who have also reported metabolic decreases in the caudate nucleus.

Both the whole brain voxel-wise and the ROI analyses failed to re-veal significant absolute CBF changes in the other basal ganglia nucleiand thalamus. The voxel-wise comparison showed bilateral subcorti-cal areas of increased perfusion in the PD patients, that covered asmall fraction of the lateral globus pallidus and thalamus; however,these perfusion increases were not significant. 2DG autoradiographicstudies in animal models of PD have consistently reported increasedmetabolism in the lateral globus pallidus. Pallidal hypermetabolismhas also been observed in a recent high-resolution PET study of PDpatients, in which the authors used reference cluster normalizationmethods (Borghammer et al., 2011). However, no study in PD pa-tients has detected absolute hypermetabolism in the basal ganglia.Detection of absolute perfusion alterations in the small basal ganglianuclei will require the acquisition of CBF maps of higher resolutionand signal-to-noise ratio. Recent technical developments and theavailability of high field MRI scanners will allow increased resolutionand could facilitate the evaluation of perfusion in these small subcor-tical structures (Feinberg et al., 2009).

Influence of medication state

In our study, patients were assessed under medication in ON state.Thus an effect of dopaminergic therapy in the results cannot be dis-carded. Studies comparing perfusion or metabolism in PD in ON andOFF states are scarce. Berding et al.(2001) reported hypometabolismin cortex and caudate nucleus in patients ON and OFF medicationcompared to control subjects, with significant differences betweenON and OFF located only in orbitofrontal cortex. In addition, Liepeltet al. (2009) found no correlation between daily dose of levodopaand cortical hypometabolism. This indicates that our findings areprobably robust to medication state. However, the effect of dopami-nergic therapy on cerebral blood flow requires further investigationand it will be the aim of future work.

Conclusions

The results of the current study contribute to identify the true patternof perfusion alterations in Parkinson's disease. This could have an impacton our understanding of the pathological features of the disease progres-sion. Our results support the hypothesis that cortical perfusion and me-tabolism are reduced in PD, even at the early disease stages, earlierthan most other previous studies have suggested. In addition, this workdemonstrates that the ASL technique has enough sensitivity to detectthis perfusion decreases bymeans of absolute CBF comparisons, withoutthe need for normalization techniques. Normalization to a referenceregion requires the a priori assumption that the reference region isnot affected by the disease and therefore is potentially biased. Similarproblems could be affecting metabolism and perfusion assessments inother brain disorders.

Supplementary materials related to this article can be found on-line at doi:10.1016/j.neuroimage.2011.10.033.

Disclosure/conflict of interest

The authors declare no conflict of interest.

Acknowledgments

Thisworkwas supported by the SpanishMinistry of Science and Inno-vation (grants SAF2008-00678 andRYC-2010-07161) and by theNavarraGovernment Health Department (grant 17/2008 GN Salud). M. Aznárez-Sanado was supported by a grant from the Navarra Government Educa-tion Department.

References

Amorim, B.J., Camargo, E.C., Etchebehere, E.C., 2007. Regional CBF changes in Parkinson'sdisease: the importance of functional neuroimaging analyses. Eur. J. Nucl. Med. Mol.Imaging 34, 1455–1457.

Ashburner, J., 2007. A fast diffeomorphic image registration algorithm. NeuroImage 38,95–113.

Ashburner, J., Friston, K.J., 2005. Unified segmentation. NeuroImage 26, 839–851.Bates, J.F., Goldman-Rakic, P.S., 1993. Prefrontal connections of medial motor areas in

the rhesus monkey. J. Comp. Neurol. 336, 211–228.Berding, G., Odin, P., Brooks, D.J., Nikkhah, G., Matthies, C., Peschel, T., Shing, M., Kolbe,

H., van Den Hoff, J., Fricke, H., Dengler, R., Samii, M., Knapp, W.H., 2001. Resting re-gional cerebral glucose metabolism in advanced Parkinson's disease studied in theoff and on conditions with [(18)F]FDG-PET. Mov. Disord. 16, 1014–1022.

Bohnen, N.I., Minoshima, S., Giordani, B., Frey, K.A., Kuhl, D.E., 1999. Motor correlates ofoccipital glucose hypometabolism in Parkinson's disease without dementia. Neu-rology 52, 541–546.

Borghammer, P., Jonsdottir, K.Y., Cumming, P., Ostergaard, K., Vang, K., Ashkanian, M.,Vafaee, M., Iversen, P., Gjedde, A., 2008. Normalization in PET group comparisonstudies—the importance of a valid reference region. NeuroImage 40, 529–540.

Borghammer, P., Cumming, P., Aanerud, J., Forster, S., Gjedde, A., 2009a. Subcortical eleva-tion ofmetabolism in Parkinson's disease—a critical reappraisal in the context of glob-al mean normalization. NeuroImage 47, 1514–1521.

Borghammer, P., Cumming, P., Aanerud, J., Gjedde, A., 2009b. Artefactual subcorticalhyperperfusion in PET studies normalized to global mean: lessons from Parkin-son's disease. NeuroImage 45, 249–257.

Borghammer, P., Chakravarty, M., Jonsdottir, K.Y., Sato, N., Matsuda, H., Ito, K., Arahata,Y., Kato, T., Gjedde, A., 2010. Cortical hypometabolism and hypoperfusion in Par-kinson's disease is extensive: probably even at early disease stages. Brain Struct.Funct. 214, 303–317.

Borghammer, P., Hansen, S.B., Eggers, C., Chakravarty, M., Vang, K., Aanerud, J., Hilker,R., Heiss, W.D., Rodell, A., Munk, O.L., Keator, D., Gjedde, A., 2011. Glucose metab-olism in small subcortical structures in Parkinson's disease Jun 21 Acta Neurol.Scand.. doi:10.1111/j.1600-0404.2011.01556.x [Epub ahead of print].

Braak, H., Del Tredici, K., Rub, U., de Vos, R.A., Jansen Steur, E.N., Braak, E., 2003. Stagingof brain pathology related to sporadic Parkinson's disease. Neurobiol. Aging 24,197–211.

Burton, E.J., McKeith, I.G., Burn, D.J., Williams, E.D., O'Brien, J.T., 2004. Cerebral atrophyin Parkinson's disease with and without dementia: a comparison with Alzheimer'sdisease, dementia with Lewy bodies and controls. Brain 127, 791–800.

Caballol, N., Marti, M.J., Tolosa, E., 2007. Cognitive dysfunction and dementia in Parkinsondisease. Mov. Disord. 22 (Suppl 17), S358–S366.

Cassidy, P.J., Radda, G.K., 2005. Molecular imaging perspectives. J. R. Soc. Interface 2,133–144.

Detre, J.A., Leigh, J.S., Williams, D.S., Koretsky, A.P., 1992. Perfusion imaging. Magn.Reson. Med. 23, 37–45.

2750 M.A. Fernández-Seara et al. / NeuroImage 59 (2012) 2743–2750

Eckert, T., Tang, C., Eidelberg, D., 2007. Assessment of the progression of Parkinson'sdisease: a metabolic network approach. Lancet Neurol. 6, 926–932.

Eickhoff, S.B., Stephan, K.E., Mohlberg, H., Grefkes, C., Fink, G.R., Amunts, K., Zilles, K.,2005. A new SPM toolbox for combining probabilistic cytoarchitectonic maps andfunctional imaging data. NeuroImage 25, 1325–1335.

Feinberg, D.A., Ramanna, S., Guenther, M., 2009. Evaluation of New ASL 3D GRASE Se-quences Using Parallel Imaging, Segmented and Interleaved k-space at 3T with12- and 32-channel Coils. International Society of Magnetic Resonance in Medicine17th Scientific Meeting, Hawaii, p. 622.

Fernandez-Seara, M.A., Edlow, B.L., Hoang, A., Wang, J., Feinberg, D.A., Detre, J.A., 2008a.Minimizing acquisition time of arterial spin labeling at 3T. Magn. Reson. Med. 59,1467–1471.

Fernandez-Seara, M.A., Wang, J., Feinberg, D.A., Detre, J.A., 2008b. Achieving Late InflowDelay in Pseudo-CASL 3D GRASE Using a Hybridized Labeling and Background Sup-pression Scheme. International Society of Magnetic Resonance in Medicine 16thScientific Meeting, Toronto (Canada, p. 3343.

Fernandez-Seara, M., Aznarez-Sanado, M., Loayza, F., Villagra, F., Pastor, M., 2009.Hypoperfusion in Parkinson's disease assessed by arterial spin labeling perfusionMRI. J. Cereb. Blood Flow Metab. 29, S570–S572.

Folstein, M.F., Folstein, S.E., McHugh, P.R., 1975. “Mini-mental state”. A practical methodfor grading the cognitive state of patients for the clinician. J. Psychiatr. Res. 12,189–198.

Grunder, G., 2009. “Absolute” or “relative”: choosing the right outcome measure inneuroimaging. NeuroImage 45, 258–259.

Hayasaka, S., Phan, K.L., Liberzon, I., Worsley, K.J., Nichols, T.E., 2004. Nonstationary cluster-size inference with random field and permutation methods. NeuroImage 22, 676–687.

Hu, M.T., Taylor-Robinson, S.D., Chaudhuri, K.R., Bell, J.D., Labbe, C., Cunningham, V.J.,Koepp, M.J., Hammers, A., Morris, R.G., Turjanski, N., Brooks, D.J., 2000. Cortical dys-function in non-demented Parkinson's disease patients: a combined (31)P-MRSand (18)FDG-PET study. Brain 123 (Pt 2), 340–352.

Hughes, A.J., Daniel, S.E., Kilford, L., Lees, A.J., 1992. Accuracy of clinical diagnosis of idio-pathic Parkinson's disease: a clinico-pathological study of 100 cases. J. Neurol. Neuro-surg. Psychiatry 55, 181–184.

Kamagata, K., Motoi, Y., Hori, M., Suzuki, M., Nakanishi, A., Shimoji, K., Kyougoku, S.,Kuwatsuru, R., Sasai, K., Abe, O., Mizuno, Y., Aoki, S., Hattori, N., 2011. Posteriorhypoperfusion in Parkinson's disease with and without dementia measured witharterial spin labeling MRI. J. Magn. Reson. Imaging 33, 803–807.

Kikuchi, A., Takeda, A., Kimpara, T., Nakagawa, M., Kawashima, R., Sugiura, M., Kinomura,S., Fukuda, H., Chida, K., Okita, N., Takase, S., Itoyama, Y., 2001. Hypoperfusion in thesupplementary motor area, dorsolateral prefrontal cortex and insular cortex in Par-kinson's disease. J. Neurol. Sci. 193, 29–36.

Kiyosawa, M., Baron, J.C., Hamel, E., Pappata, S., Duverger, D., Riche, D., Mazoyer, B.,Naquet, R., MacKenzie, E.T., 1989. Time course of effects of unilateral lesions ofthe nucleus basalis of Meynert on glucose utilization by the cerebral cortex. Posi-tron tomography in baboons. Brain 112 (Pt 2), 435–455.

Lancaster, J.L., Woldorff, M.G., Parsons, L.M., Liotti, M., Freitas, C.S., Rainey, L., Kochunov,P.V., Nickerson, D., Mikiten, S.A., Fox, P.T., 2000. Automated Talairach atlas labelsfor functional brain mapping. Hum. Brain. Mapp. 10, 120–131.

Liepelt, I., Reimold,M., Maetzler,W., Godau, J., Reischl, G., Gaenslen, A., Herbst, H., Berg, D.,2009. Cortical hypometabolism assessed by a metabolic ratio in Parkinson's diseaseprimarily reflects cognitive deterioration-[18F]FDG-PET. Mov. Disord. 24, 1504–1511.

Lu, M.T., Preston, J.B., Strick, P.L., 1994. Interconnections between the prefrontal cortexand the premotor areas in the frontal lobe. J. Comp. Neurol. 341, 375–392.

Luppino, G., Matelli, M., Camarda, R., Rizzolatti, G., 1993. Corticocortical connections ofarea F3 (SMA-proper) and area F6 (pre-SMA) in the macaque monkey. J. Comp.Neurol. 338, 114–140.

Ma, Y., Tang, C., Moeller, J.R., Eidelberg, D., 2009. Abnormal regional brain function inParkinson's disease: truth or fiction? NeuroImage 45, 260–266.

Ma, Y., Huang, C., Dyke, J.P., Pan, H., Alsop, D., Feigin, A., Eidelberg, D., 2010. Parkinson'sdisease spatial covariance pattern: noninvasive quantification with perfusion MRI.J. Cereb. Blood Flow Metab. 30, 505–509.

MacDonald, V., Halliday, G.M., 2002. Selective loss of pyramidal neurons in the pre-supplementary motor cortex in Parkinson's disease. Mov. Disord. 17, 1166–1173.

Maldjian, J.A., Laurienti, P.J., Kraft, R.A., Burdette, J.H., 2003. An automated method forneuroanatomic and cytoarchitectonic atlas-based interrogation of fMRI data sets.NeuroImage 19, 1233–1239.

Martinez-Martin, P., Gil-Nagel, A., Gracia, L.M., Gomez, J.B., Martinez-Sarries, J., Bermejo, F.,1994. Unified Parkinson's Disease Rating Scale characteristics and structure. The Co-operative Multicentric Group. Mov. Disord. 9, 76–83.

Matelli, M., Luppino, G., Geyer, S., Zilles, K., 2004. Motor Cortex. In: Paxinos, G., Mai, J.(Eds.), The Human Nervous System. Elsevier Academic Press, pp. 973–996.

Melzer, T.R., Watts, R., MacAskill, M.R., Pearson, J.F., Rueger, S., Pitcher, T.L., Livingston,L., Graham, C., Keenan, R., Shankaranarayanan, A., Alsop, D.C., Dalrymple-Alford,J.C., Anderson, T.J., 2011. Arterial spin labelling reveals an abnormal cerebral perfusionpattern in Parkinson's disease. Brain 134, 845–855.

Mito, Y., Yoshida, K., Yabe, I., Makino, K., Hirotani, M., Tashiro, K., Kikuchi, S., Sasaki, H.,2005. Brain 3D-SSP SPECT analysis in dementia with Lewy bodies, Parkinson's dis-ease with and without dementia, and Alzheimer's disease. Clin. Neurol. Neurosurg.107, 396–403.

Obeso, J.A., Marin, C., Rodriguez-Oroz, C., Blesa, J., Benitez-Temino, B., Mena-Segovia, J.,Rodriguez, M., Olanow, C.W., 2008. The basal ganglia in Parkinson's disease: cur-rent concepts and unexplained observations. Ann. Neurol. 64 (Suppl 2), S30–S46.

Obeso, J.A., Rodriguez-Oroz, M.C., Goetz, C.G., Marin, C., Kordower, J.H., Rodriguez, M.,Hirsch, E.C., Farrer, M., Schapira, A.H., Halliday, G., 2010. Missing pieces in the Par-kinson's disease puzzle. Nat. Med. 16, 653–661.

Petrides, M., Pandya, D., 2004. The Frontal Cortex. In: Paxinos, G., Mai, J. (Eds.), TheHuman Nervous System. Elsevier Academic Press, pp. 950–972.

Picard, N., Strick, P.L., 2001. Imaging the premotor areas. Curr. Opin. Neurobiol. 11,663–672.

Possin, K.L., 2011. Visual spatial cognition in neurodegenerative disease. Neurocase 16,466–487.

Poston, K.L., Eidelberg, D., 2009. Network biomarkers for the diagnosis and treatmentof movement disorders. Neurobiol. Dis. 35, 141–147.

Raichle, M.E., 1998. Behind the scenes of functional brain imaging: a historical andphysiological perspective. Proc. Natl. Acad. Sci. U. S. A. 95, 765–772.

Seeley, W.W., Menon, V., Schatzberg, A.F., Keller, J., Glover, G.H., Kenna, H., Reiss, A.L.,Greicius, M.D., 2007. Dissociable intrinsic connectivity networks for salience pro-cessing and executive control. J. Neurosci. 27, 2349–2356.

Shimada, H., Hirano, S., Shinotoh, H., Aotsuka, A., Sato, K., Tanaka, N., Ota, T., Asahina,M., Fukushi, K., Kuwabara, S., Hattori, T., Suhara, T., Irie, T., 2009. Mapping ofbrain acetylcholinesterase alterations in Lewy body disease by PET. Neurology73, 273–278.

Tang, C.C., Poston, K.L., Dhawan, V., Eidelberg, D., 2010. Abnormalities in metabolic net-work activity precede the onset of motor symptoms in Parkinson's disease. J. Neu-rosci. 30, 1049–1056.

Wang, J., Zhang, Y., Wolf, R.L., Roc, A.C., Alsop, D.C., Detre, J.A., 2005. Amplitude modu-lated continuous arterial spin labeling perfusion MR with single coil at 3.0 Tesla.Radiology 235, 218–228.

Wang, Z., Aguirre, G.K., Rao, H., Wang, J., Fernandez-Seara, M.A., Childress, A.R., Detre,J.A., 2008. Empirical optimization of ASL data analysis using an ASL data processingtoolbox: ASLtbx. Magn. Reson. Imaging 26, 261–269.

Williams, D.S., Detre, J.A., Leigh, J.S., Koretsky, A.P., 1992. Magnetic resonance imagingof perfusion using spin inversion of arterial water.[erratum appears in Proc. Natl.Acad. Sci. U S A 1992 May 1;89(9):4220]. Proc. Natl. Acad. Sci. U S A 89, 212–216.

Xu, G., Rowley, H.A., Wu, G., Alsop, D.C., Shankaranarayanan, A., Dowling, M., Christian,B.T., Oakes, T.R., Johnson, S.C., 2010. Reliability and precision of pseudo-continuousarterial spin labeling perfusion MRI on 3.0 T and comparison with 15O-water PETin elderly subjects at risk for Alzheimer's disease. NMR Biomed. 23, 286–293.

Zilles, K., Schlaug, G., Geyer, S., Luppino, G., Matelli, M., Qu, M., Schleicher, A., Schormann,T., 1996. Anatomy and transmitter receptors of the supplementarymotor areas in thehuman and nonhuman primate brain. In: Lüders, H.O. (Ed.), Advances in Neurology. :Supplementary Sensorimotor Area, 70. Lippincott-Raven, Philadelphia, pp. 29–43.