Absolute

内容需要下载文档才能查看

Original Research Article

ement Geriatr Cogn Disord 2008;26:89–100 D

DOI:10.1159/000144044 ccepted: April 8, 2008 A

Published online: July 11, 2008

bsolute Quantification in Proton Magnetic A

Resonance Spectroscopy Is Superior to Relative Ratio to Discriminate Alzheimer’s Disease from Binswanger’s Disease

Toshiyuki Watanabe a, b Akihiko Shiino b Ichiro Akiguchi a

b Center of Neurological and Cerebrovascular Diseases, Takeda Hospital, Kyoto , and DepartmentofNeurosurgery,Shiga University of Medical Science, Otsu , Japan

a

ey Words K

1H magnetic resonance spectroscopy ? N-acetylaspartate ? Alzheimer’s disease ? Binswanger’s disease ? Hippocampus ? Periventricular and deep white matter

IntroductionAbstractBackground/Aims: Although many proton magnetic reso-

nance spectroscopy studies have assessed the relative ratios of brain metabolites from patients with dementia, absolute quantification is rare. The aim of this study is to compare the diagnostic accuracy of these 2 methods in proton magnetic resonance spectroscopy in discriminating Alzheimer’s dis-ease (AD) and Binswanger’s disease (BD) from healthy con-trols (HC). Methods: The subjects were 30 AD patients, 13 BD patients and 26 HC subjects. Single-voxel proton MR spectra at short echo times were acquired from 8 volumes of inter-est. Results: At 80% specificity, the absolute N-acetylaspar-tate concentration in the hippocampus was the most sensi-tive measure to discriminate AD from HC, and the absolute N-acetylaspartate concentration in the anterior periventric-ular and deep white matter to differentiate BD from HC and AD. No relative ratio using creatine as a reference had a sen-sitivity over 80% at 80% specificity. The cause of disparities between the 2 methods was attributed to fluctuations in the creatine concentration. Conclusion: Our study revealed that absolute quantification is superior to relative ratio to differ-entiate AD and BD from HC. Copyright © 2008 S. Karger AG, Basel

lzheimer’s disease (AD) and Binswanger’s disease A

(BD) are both common forms of dementia. AD is a de-generative disease and can be pathologically character-ized by the formation of neurofibrillary tangles and se-nile plaques [1, 2] . BD, on the other hand, is considered to be a small-vessel disease, and is characterized by a com-bination of multiple lacunar infarcts and diffuse white

[3, 4] .matter lesions

Proton magnetic resonance spectroscopy (1H-MRS), a

noninvasive tool, has been used to measure metabolite concentrations in patients with neurological diseases. Among various substances assessed using 1H-MRS, func-tional significances in 3 kinds of metabolite have been discussed at length in the literature reporting on cognitive disorders: N-acetylaspartate (NAA), myoinositol (MI) and choline compounds (Cho). Many studies have esti-mated the metabolite concentrations in patients with AD [5, 6] . Most of them, including our previous study [7],eval-

uated the relative ratio of NAA/creatine (Cr) or MI/Cr,

[8–10].assuming that the concentration of Cr was stable

However, the Cr concentration has in fact been reported as variable with respect to the patient’s age, the site in the

[11, 12] . It is therefore brain and the clinical stage of AD

desirable to measure the absolute concentrations of these metabolites in the hippocampus, which is an important anatomical structure for memory consolidation and is be-[1, 2] .lieved to be damaged in early-stage AD

ToshiyukiWatanabeHigashi-Shiokouji 841-5, Shimogyo-ku

Kyoto, 600-8558 (Japan)

Tel. +81 75 361 1351, Fax +81 75 361 7602

E-Mailtw1019@ya2.so-net.ne.jp 2008 S. Karger AG, Basel©

1420–8008/08/0261–0089$24.50/0

Fax +41 61 306 12 34E-Mail karger@http://wendang.chazidian.com

Accessibleonlineat:http://wendang.chazidian.com/dem

cF ig. 1. Locations of VOIs for 1H-MRS. a Bilateral hippocampi. b Bilateral anterior and posterior PDWM. cPCG.dOccipitallobe. In the PCG and the occipital lobe, VOIs were located in para-median positions. H

owever, absolute measurements in this region are technically challenging. Due to its proximity to air and bone within the cranium, difficulties in water suppres-sion and obtaining magnetic field homogeneity have led to low resolution and poor signal-to-noise ratio MR spec-tra from the hippocampus [13] . On the other hand, recent functional studies, includ-ing PET and SPECT, suggest that a decrease in the me-tabolism or perfusion of the posterior cingulate gyrus (PCG) is an early sign of AD

[14, 15] . Some 1H-MRS stud-ies also investigated the relative ratios of metabolites in the PCG [8–10] . From a pathophysiological point of view, it is interesting to measure the metabolite concentrations of the same patients with AD in both the hippocampus and PCG in vivo at one time.

In contrast to AD, which is a degenerative disease, BD is considered a subtype of subcortical ischemic vascular dementia [3] . Ischemic changes in the periventricular and deep white matter (PDWM), as well as in the hippocam-pus and related structures, are thought to involve cogni-tive disturbances observed in BD patients

[3] .B

y means of LCModel [16] , which is fully automatic and free from subjective interactions, we assessed the ab-90

Dement Geriatr Cogn Disord 2008;26:89–100

solute metabolite concentrations for AD and BD patients, and then compared these results with the relative ratios of metabolites using Cr as a reference.

T

his is the first study to measure the absolute concen-trations of metabolites in the bilateral hippocampi, the PCG and the bilateral PDWM all from the same patients with dementia, revealing disparities between methods using absolute quantification and relative ratios. Part of

this study was published as a preliminary report

[17] .M

ethods Subjects All patients were recruited consecutively and prospectively for longitudinal studies of cognitive disturbances at our outpatient clinic. They underwent comprehensive diagnostic evaluations, including medical histories, neurological and psychiatric exami-nations, neuropsychological testing, laboratory tests, as well as brain MRI and SPECT. The exclusionary criteria were medical histories of cortical stroke or other major neurological diseases, thyroid dysfunctions, seizures, alcohol abuse and psychiatric dis-orders. Dementia was diagnosed in accordance with the Diagnos-tic and Statistical Manual of Mental Disorders – Fourth Edition [ 18] . Patients with AD fulfilled the NINCDS-ADRDA criteria for probable AD [19] , and those with BD met the criteria for possible

BD proposed by Bennett et al. [20] . All BD patients also had clin-ical and neuroimaging findings enumerated by Caplan [21].De-mentia severity was assessed with the Clinical Dementia Rating Scale [22] , and only patients with mild to moderate severity (Clin-ical Dementia Rating 1 or 2) were studied. Cognitive function was examined with the Mini-Mental State Examination(MMSE) [23],Japanese Wechsler Adult Intelligence Scale-Revised (WAIS-R)

[24] and Japanese Wechsler Memory Scale-Revised (WMS-R) [

25] .F

ifty-three patients with AD and 18 with BD were recruited between April 2001 and December 2006. On the basis of the

above-mentioned criteria, however, we studied 30 patients with AD, 13 patients with BD and 26 age-matched healthy control (HC) subjects. The HC subjects were volunteers for the study proj-ect of mild cognitive impairment in our facilities. None of the

control subjects had any severe neurological or medical diseases. The protocol was approved by the local ethics committee, and in-formed consent was obtained from all participants and their re-sponsible caregivers in accordance with the Declaration of Hel-sinki.

MRI and 1H-MRS

B

oth MRI and 1H-MRS were performed on a 1.5-tesla appa-ratus (Signa Horizon LX, General Electric Medical Systems, Mil-waukee, Wisc., USA) using a standard head coil suited for MRI and MRS. The protocol for structural MRI included axial FLAIR images with repetition time (TR)/echo time (TE)/inversion time (TI) = 4,000/124/2,000 ms with a slice thickness of 4 mm. The signal hyperintensities in PDWM in axial FLAIR images were es-timated according to the rating scale for white matter lesions by

Wahlund et al.

[26] . As shown in figure 1 , volumes of interests Watanabe

/Shiino

/Akiguchi

where CC is the corrected concentration of a certain metabolite, Tabe 1. Clinical characteristics and neuropsychological test

scores of AD, BD and HC subjects

(VOIs) for 1H-MRS were assigned to 8 areas in the brain; the bi-lateral anterior PDWM, bilateral posterior PDWM, bilateral hip-pocampi, PCG and occipital lobe. Gray matter volumes in the

PCG and the occipital lobe were located in paramedian positions spanning about 20 mm across the interhemispheric fissure. White matter volumes were situated between the inferior frontal sulcus and the inferior precentral sulcus for the anterior PDWM, and between the central sulcus and the superior temporal sulcus for the posterior PDWM. To establish the VOIs for the hippocampi, sagittal images were obtained, followed by tilted T 2-weighted(TR/TE = 4,000/107 ms) coronal images perpendicular to the long axis of the hippocampus with a slice thickness of 3 mm. A rect-angular VOI was placed along the long axis of the hippocampus and contained the hippocampal head and body about 20 mm from just posterior to the amygdala. The VOIs were about 3 ml for the right and left hippocampi, 5 ml for the PCG, 6 ml for the oc-cipital lobe and 4 ml for the PDWM. With TR/TE = 2,000/30 ms and 192 signal averages, 1H-MRS spectra were acquired with point-resolved spectroscopy volume selection. Special care was taken to avoid susceptibility effects and to minimize partial vol-ume effects with other tissues. Spectral analyses were performed using a spectroscopy software package for the Sun Workstation. Absolute concentrations were calculated by LCModel [16].Cor-rections for gray and white matter with respect to T 1 and T 2re-laxation times were performed according to data by Kreis et al. [11] and for hippocampi by Choi et al. [27] . As to the hippocam-pus, the voxel contained variable amounts of CSF due to tissue atrophy. We compensated for CSF by calculating the T 2compo-nents of water in each coronal slice using the following formula:C C = UC [ ? (Ti + Fi)/ ? Ti], i = 1, 2, 3, 4, 5, 6

1 H-MRS of AD and BD UC the uncorrected concentration, Ti the number of pixels from the hippocampal tissue in each slice contained in the VOI and Fi the number of pixels from CSF in the same slice.

StatisticalAnalysesA

ll comparisons of the means from the demographic data were tested by analysis of variance (ANOVA) followed by Scheffe’s post hoc analysis. Concerning WAIS-R and WMS-R, some pa-tients had scores below the lower limits in some tests, prohibiting calculation of the mean value. We therefore used the median as a measure of central tendency and the Mann-Whitney U test was employed. As for the metabolite concentrations obtained with 1H-MRS, group differences were analyzed by 3-factorial multi-variate analysis of variance (MANOVA) using voxels ( !8) and metabolites ( ! 4) as within factors and as between factors for the 3 groups, followed by post hoc univariate ANOVA with Scheffe’s test to determine differences between the groups. The statistical analyses were performed with SPSS version 13.0 (SPSS Inc., Chi-cago, Ill., USA). A p value ! 0.05 was considered to be statistically significant.

T

o compare the diagnostic accuracy of the 2 methods (abso-lute quantification and relative ratio) in 1H-MRS in discriminat-ing AD and BD from HC, the sensitivity of the MRS measures was calculated at a fixed specificity of 80%.

R

esults T

he demographic and clinical characteristics of the subjects are summarized in table 1 . One-way ANOVA showed significant differences in the mean score of the MMSE among these 3 groups, and the post hoc analysis (Scheffe’s test) disclosed significant differences between the HC and AD groups (p ! 0.001), and between the HC and BD groups (p ! 0.001), but no differences between the AD and BD groups (p = 0.62). Table 1 also shows the re-sults from the Japanese versions of the WMS-R and WAIS-R. Whereas the WAIS-R scores were relatively pre-served in both dementia groups, those of the WMS-R were remarkably decreased. The Mann-Whitney U test disclosed that the BD group was significantly superior to the AD group in verbal memory, general memory and delayed recall. There was no difference between the 2 groups in visual memory, attention or the subtests in the WAIS-R.

Axial MR (FLAIR) images ( fig. 2 )showed different characteristics between the AD and BD groups. The BD patients had severe signal hyperintensities in the PDWM and multiple lacunes in the thalamus, basal ganglia and bilateral PDWM ( fig. 2 a), whereas the AD patients only had punctate hyperintensities in the basal ganglia and PDWM ( fig. 2 b). According to the rating scale for white

matter lesions by Wahlund et al.

[26] , all BD patients ap-Dement Geriatr Cogn Disord 2008;26:89–100

91

a

Fig. 2. Examples of FLAIR images from a patient

with BD ( a ) and a patient with AD ( b ). The BD pa-tient had severe signal hyperintensities in PDWM, and multiple lacunes in the thalamus, basal gan-glia and bilateral PDWM, whereas the AD patient only had punctate hyperintensities in the basal

bganglia and PDWM.

peared to have grade 3 (diffuse involvement of the entire

lesion), whereas no AD patient was graded 3. Both de-mentia groups had mild to moderate hippocampal atro-[28] .phies estimated with voxel-based morphometry

Using 1H-MRS, the absolute concentrations of N-ace-tyl compounds, MI, Cr and Cho were obtained. The N-acetyl compounds contained not only NAA but also N-acetylaspartylglutamate and were expressed as NAA. Representative 1H-MR spectra from each group are shown in figure 3 .

We first describe the results from absolute quantifica-

tion and then show the disparities between absolute con-centrations and relative ratios of metabolites using Cr as a reference. The MI/NAA ratio was also calculated be-cause it has been used as a marker for AD by several au-[5, 6] .thors

Absolute Concentrations T he absolute metabolite concentrations are presented as means 8 SD (millimol/liter) in table 2 and figure 4 . Overall, MANOVA revealed a significant difference among the 3 groups (p ! 0.001). Concerning NAA con-centrations, post hoc univariate ANOVA was performed individually on each region and showed a group differ-ence in all 8 regions. With regard to the bilateral hippo-campi, post hoc analyses disclosed that the AD patients had significantly lower concentrations of NAA as com-pared with the HC subjects (p ! 0.001). The bilateral hip-92

Dement Geriatr Cogn Disord 2008;26:89–100

pocampal NAA concentrations of the BD patients showed intermediate values, which were lower than those of the HC subjects but higher than those of the AD patients. The NAA level of the PCG was not significantly different be-tween the AD and HC (p = 0.141) or between the BD and HC groups (p = 0.52). The occipital NAA concentrations of the AD and BD groups were lower than those of the controls. In the PDWM, the patients with BD had re-markably lower levels of NAA than the HC and AD sub-jects. The patients with AD had intermediate concentra-tions of NAA in the PDWM, which were lower than those of the HC subjects and higher than those of the BD pa-tients. Concerning the patients with BD, the NAA con-centrations from the anterior PDWM were significantly lower than those from the posterior PDWM (p !0.001).As to MI concentrations, post hoc analyses showed no

difference between the HC and AD groups in the bilat-eral hippocampi. In the PCG, the MI concentrations of the AD patients were significantly higher than those of the HC subjects (p = 0.002). In the PDWM, the patients with BD had remarkably lower levels of MI than the HC and AD groups.

With regard to Cho concentrations, post hoc analyses

disclosed that the AD patients had significantly lower levels in the bilateral hippocampi than the HC subjects. In other regions including the PCG and the PDWM, the Cho levels did not differ between the AD and HC groups.

The Cho concentrations of the BD patients in all areas of

Watanabe /Shiino /Akiguchi

HC

MI

Cho

Cr

NAA

MI

Cho

AD

NAA

Cr

MICho

BD

NAA

Cr

LH

LA

PCG

3.02.0Chemical shift (ppm)3.02.0Chemical shift (ppm)3.02.0Chemical shift (ppm)

Fig. 3. Examples of proton spectra obtained from the left hippocampus (LH), left anterior PDWM (LA) and

PCG in a HC subject, a patient with AD and a patient with BD. The concentration of NAA from the LH is low-er in the patients with AD and BD than in the HC subject. In the LA, the BD patient shows a remarkably di-minished NAA concentration compared to the AD and HC subjects. There is no apparent difference in the NAA concentration in the PCG among these 3 subjects. It is, however, difficult from these figures to assess the dif-ferences among these subjects in concentrations of MI, Cho and Cr. All spectra are scaled to the corrected NAA concentration in the PCG of the HC subject.

Table 2. Absolute concentrations of metabolites of AD, BD and HC subjects

1H-MRS of AD and BD

Dement Geriatr Cogn Disord 2008;26:89–100

93