Dement Neuropsychol 2015 December;9(4):385-393 Original Article

DOI: 10.1590/1980-57642015DN94000385

Analysis of the posterior cingulate

cortex with [18F]FDG-PET and Naa/mI
in mild cognitive impairment
and Alzheimer’s disease
Correlations and differences between the two methods
Artur M.N. Coutinho1, Fábio H.G. Porto3, Poliana F. Zampieri1,
Maria C. Otaduy2, Tíbor R. Perroco3, Maira O. Oliveira3, Rafael F. Nunes1,
Toulouse Leusin Pinheiro2, Cassio M.C. Bottino3, Claudia C. Leite2, Carlos A. Buchpiguel1

ABSTRACT. Reduction of regional brain glucose metabolism (rBGM) measured by [ 18F]FDG-PET in the posterior cingulate
cortex (PCC) has been associated with a higher conversion rate from mild cognitive impairment (MCI) to Alzheimer’s
disease (AD). Magnetic Resonance Spectroscopy (MRS) is a potential biomarker that has disclosed Naa/mI reductions
within the PCC in both MCI and AD. Studies investigating the relationships between the two modalities are scarce.
Objective: To evaluate differences and possible correlations between the findings of rBGM and NAA/mI in the PCC of
individuals with AD, MCI and of cognitively normal volunteers. Methods: Patients diagnosed with AD (N=32) or MCI
(N=27) and cognitively normal older adults (CG, N=28), were submitted to [18F]FDG-PET and MRS to analyze the PCC.
The two methods were compared and possible correlations between the modalities were investigated. Results: The AD
group exhibited rBGM reduction in the PCC when compared to the CG but not in the MCI group. MRS revealed lower
NAA/mI values in the AD group compared to the CG but not in the MCI group. A positive correlation between rBGM and
NAA/mI in the PCC was found. NAA/mI reduction in the PCC differentiated AD patients from control subjects with an area
under the ROC curve of 0.70, while [18F]FDG-PET yielded a value of 0.93. Conclusion: rBGM and Naa/mI in the PCC
were positively correlated in patients with MCI and AD. [18F]FDG-PET had greater accuracy than MRS for discriminating
AD patients from controls.
Key words: positron-emission tomography, spectrum analysis, magnetic resonance imaging, mild cognitive impairment,
Alzheimer’s disease.

ANÁLISE DO GIRO DO CÍNGULO POSTERIOR COM [18F]FDG-PET E RELAÇÃO NAA/MI NO COMPROMETIMENTO LEVE E NA
DOENÇA DE ALZHEIMER: CORRELAÇÕES E DIFERENÇAS ENTRE OS MÉTODOS
RESUMO. Redução do metabolismo cerebral regional glicolítico (MRG) medido pela PET- 18FDG no giro do cíngulo posterior
(GCP) está relacionada a maior conversão para doença de Alzheimer (DA) em sujeitos com comprometimento cognitivo
leve (CCL). Espectroscopia por ressonância magnética (MRS), um biomarcador promissor, demonstra redução de Naa/mI
no GCP na DA. Raros estudos avaliam relações entre Naa/mI e MRG. Objetivo: Avaliar diferenças e possíveis correlações
entre MRG com PET-18FDG e Naa/mI por MRS no GCP de sujeitos com DA, CCL e voluntários normais. Métodos: Sujeitos
com DA (N=32), CCL amnéstico (N=27) e voluntários idosos normais (GC, N=28), foram submetidos a PET-18FDG e
análise de Naa/mI no GCP. A performance de ambos os métodos foi então comparada e verificou-se a existência de
correlações entre os achados da PET e da MRS. Resultados: Observou-se hipometabolismo glicolítico nos pacientes
com DA no GCP em relação ao GC, porém não no CCL. A MRS demonstrou valores menores de Naa/mI no CP do grupo

This study was conducted at the Centro de Medicina Nuclear, Instituto e Departamento de Radiologia, Hospital das Clínicas da Faculdade de Medicina da Univer-
sidade de São Paulo (HC/FMUSP) and at the Centro de Referência em Distúrbios Cognitivos, HC/FMUSP.
1
Centro de Medicina Nuclear, Instituto e Departamento de Radiologia, HC/FMUSP, LIM 43. 2Serviço de Ressonância Magnética, Instituto e Departamento de
Radiologia, HC/FMUSP, LIM 44. 3Centro de Referência em Distúrbios Cognitivos (CEREDIC) do HC/FMUSP.

Arthur M.N. Coutinho. Department of Radiology/Nuclear Medicine Center/LIM43 / HC/FMUSP – Trav. Ovídio Pires de Campos S/N – Prédio do Centro de
Medicina Nuclear, 2º andar – LIM43 – 05403-010 São Paulo SP – Brazil. E-mail: artur.coutinho@hc.fm.usp.br

Disclosure: The authors report no conflits of interest.

Received September 03, 2015. Accepted in final form November 03, 2015.

Coutinho et al.    FDG-PET and MRS in AD 385

 Dement Neuropsychol 2015 December;9(4):385-393

DA em relação ao GC, porém também sem diferenças entre CCL e GC. A área sob a curva ROC demonstrou valor de
0,70 para MRS e 0,93 para o MRG no GCP para diferenciar DA do GC. Houve correlação positiva entre o MRG e o Naa/
mI no GCP. Conclusão: Os valores de metabolismo de glicose à PET e de Naa/mI à MRS no giro do cíngulo posterior
apresentaram correlação positiva estatisticamente significante na presente amostra. Houve ainda superioridade da
PET-18FDG para diferenciar DA do GC.
Palavras-chave: tomografia por emissão de pósitrons, análise espectral, imagem por ressonância magnética,
comprometimento cognitivo leve, doença de Alzheimer.

INTRODUCTION of glial cells, thus its concentration depends on the quan-

A lzheimer’s disease (AD) has become a public health

problem with the rise in life expectancy, since there
is currently no treatment that modifies its progression.1-3
tity of gliosis.16-18 One of the drawbacks of the method,
however, is the need for manually drawn regions of inter-
est (ROI) in different areas of the brain, since values of
Correct diagnosis in the early stages of the disease is Naa/mI may vary among different brain regions, i.e. the
crucial to better understand its pathophysiology and to PCC and the hippocampus. Values measured can also
develop treatments to slow its progression. Mild cogni- vary according to operator experience.16
tive impairment (MCI), especially the amnestic subtype, Although hypometabolism in the PCC measured by
is a symptomatic transitional state from normal aging [18F]FDG-PET is a classical biomarker of disease pro-
to early dementia. MCI is characterized by subjective gression to AD in MCI and some MRS results disclose
memory complaints and objective decline in cognitive early neuronal injury in this area, studies correlating the
performance, with normal or near-normal functional findings of the two modalities are scarce. Given these
activities of daily living.4,5 methods theoretically reflect correlated biologic pro-
Positron emission tomography using [18F]fluorode- cesses, this study sought to investigate whether the two
oxyglucose ([18F]FDG-PET) is a well-established tool for measures are closely related in elderly patients with AD
monitoring regional brain glucose metabolism (rBGM). or amnestic MCI and control subjects without cognitive
A progressive reduction of rBGM in specific areas occurs complaints.
years before the onset of clinical symptoms in patients Thus, the objectives of this study were to assess pos-
with verified AD and during the MCI phase, particularly sible differences in findings on [18F]FDG-PET and in
in the temporoparietal cortex and posterior cingulate NAA/mI ratio (a measure of neuronal injury) assessed
cortex (PCC) association.6-10 Of all the areas, the PCC by MRS in the PCC among patients with AD or MCI and
seems to be the most sensitive marker for predicting cognitively normal volunteers, and also to determine
which patients with MCI will progress to AD.7,10,11 possible correlations between the two methods in the
Magnetic Resonance Spectroscopy (MRS) uses a PCC of these individuals.
standard MRI scanner and acquires a spectrum that
expresses metabolite concentrations in the brain. It is a METHODS
potentially useful noninvasive neuroimaging technique Participants. Older adults (≥60 years old) with subjective
for detecting brain biochemical changes associated with cognitive complaints were recruited from the Cognitive
neurodegenerative diseases.12 MRS has potential utility Disorders Reference Center (CEREDIC) of our hospital.
as a biomarker in MCI and early dementia, helping with Patients had to have reported cognitive complaints,
early (and differential) diagnosis and tracking disease confirmed by a collateral source, usually a relative or
progression surveillance.13-15 spouse. All participants underwent complete neurolog-
Some metabolites commonly studied with MRS and ical and psychiatric evaluation as well as comprehensive
present at high concentrations in the brain are: N-Acetyl neuropsychological tests. The final diagnosis was estab-
Aspartate (NAA), choline (Cho), creatine (Cr), myo-inosi- lished by consensus of at least two physicians (neurolo-
tol (mI), glutamate and glutamine (Glx).16 Each metabo- gists or psychiatrists) with expertise in cognitive and
lite is sensitive to different processes in the brain. MRS behavioral neurology. The healthy older adults without
studies have shown decreased NAA/mI and increased cognitive complaints were recruited in the community
mI/Cr ratios in the brain of subjects with MCI, including or from a pool of cognitive normal older subjects from
in the PCC, which may correspond to neuronal injury.13,14 our Institution to serve as members of the control
NAA is mainly found in neurons, and thus NAA reduc- group. After the initial work-up, participants were clas-
tion reflects neuronal loss or dysfunction; mI is a marker sified into one of three groups: Alzheimer’s disease

386 FDG-PET and MRS in AD     Coutinho et al.

 Dement Neuropsychol 2015 December;9(4):385-393

group (AD), mild cognitive impairment group (MCI) or with the provisions of the Declaration of Helsinki. All
control group (CG). subjects signed a consent form.
Patients from the AD group were diagnosed according
to the DSM-IV and the NINCDS-ADRDA criteria.19 The Magnetic resonance imaging acquisition. All patients
revised Petersen criteria were used to diagnose individu- underwent a standard brain MRI scan to exclude the
als with MCI.4,5 Only patients with amnestic MCI were presence of significant lesions and for co-registration
included. Severity of the cognitive complaints was mea- with [18F]FDG-PET images.
sured by the Clinical Dementia Rating (CDR) scale.20 Only Brain MRI exams were performed on a 3.0T magnetic
individuals with a score of 1.0 on the Clinical Dementia resonance scanner (Intera Achieva, PHILIPS Healthcare,
Rating were included in the AD group (defined as early Best, The Netherlands) with an 8-channel head coil and
AD). All subjects from the MCI and Control groups had the imaging protocol included the following sequences:
CDR=0.5 (MCI) and CDR=0 (CG), respectively. 3D-T1 Fast Field Echo (3D-T1 FFE), axial T2-weighted
All subjects were submitted to the Mini-Mental State fast spin echo (FSE), axial fluid-attenuated inversion
Examination,21 the Brief Cognitive Screening Battery recovery (FLAIR), coronal T2- weighted fast spin echo
(BCSB),22 the Dementia Rating Scale23,24 and to a compre- (FSE) with fat saturation (SPIR), and diffusion. Finally,
hensive neuropsychological evaluation, which included a single-voxel 1H-MRS was obtained from the PCC using
the following tests: Visual Reproduction subtest of the the PRESS sequence with 128 averages, TR of 1500 ms
Wechsler Memory Scale – Revised (WMS-R),25 Rey Com- and TE of 35 ms. Voxel size was 2×2×2 cm3 and placed
plex Figure - delayed recall,26 Logical Memory subtest of in the PCC (Figure 1). NAA and mI concentrations were
the Wechsler Memory Scale – Revised (WMS-R),25 Selec- quantified relative to an internal water reference using
tive Reminding Test,27 Block Design subtest – Wechsler LCModel.29
Adult Intelligence Scale (WAIS),28 Rey Complex Figure
copy,26 attention/executive functions (Trail Making Test Positron emission tomography imaging acquisition. Patients
A and B),26 and phonemic verbal fluency (F.A.S.),26 and with blood glucose levels lower than 180 mg/mL and at
language (semantic verbal fluency – supermarket).23,24 least 4 hours of fasting received an intravenous injection
The application, scoring and interpretation of the results of 370 MBq of [18F]FDG in a peripheral vein, and rested
obtained for all tests were performed according to their with eyes open and ears unplugged for 60 minutes in
respective reference guides. All brain-imaging procedures a calm, silent and slightly darkened room. Images were
were performed within 2 weeks of the clinical examina- acquired using a Siemens Biograph PET-CT scanner
tions and neuropsychological testing. (CTI/ Siemens, Knoxville, TN, USA).
Exclusion criteria included: [1] volunteers with clini- PET data was analyzed on a voxel-by-voxel basis using
cally relevant psychiatric symptoms meeting DSM-IV cri- the SPM8 software program (Wellcome Department of
teria; [2] any uncompensated clinical comorbidity, such Cognitive Neurology, Functional Imaging Laboratory,
as cardiac failure or anemia; [3] history or presence of London, UK) in conjunction with MATLAB R2009a
signs of other neurologic diseases, such as Parkinson’s (The Mathworks Inc., U.S.A.). Each PET study was co-
disease, epilepsy, inflammatory disease or stroke, with registered with the individuals’ respective MRI images
the exception of migraine; [4] presence of any drug abuse (volumetric T1) and spatially normalized in SPM8 into
(especially alcoholism); [5] patients with diabetes mel- a standard stereotactic space, based on the SPM8/Mon-
litus without adequate glycemic control in the last two treal Neurologic Institute (MNI) space. Global uptake
weeks; [6] demented subjects with CDR >1.0; [7] pres- differences between brain scans were adjusted using the
ence of neoplastic or significant vascular lesions on the “proportional scaling” SPM option. The relevant peak
MRI, according to the judgment of an assistant neuro- voxels were identified in terms of coordinates accord-
radiologist and of the authors (AMNC); [8] contrain- ing to Talairach and Tournoux with the aid of the Talai-
dication of the MRI exam. Antidepressant use was not rach Client software, and after conversion from the
strictly exclusionary; individuals using antidepressants SPM/MNI space. Complete details of the [18F]FDG-PET
were allowed to participate if on a stable dose for at least acquisition and imaging processing have been described
three months and without symptoms of an active psy- previously.30,31
chiatric disease at the time of screening.
This research project was approved by the ethics Statistical analysis and [18F]FDG-PET ROI definition. An
committee of the Hospital das Clínicas da Faculdade de analysis of variance (ANOVA) test was used to search
Medicina da Universidade de São Paulo, and complied for regional brain glucose metabolism (rBGM) differ-

Coutinho et al.    FDG-PET and MRS in AD 387

 Dement Neuropsychol 2015 December;9(4):385-393

A1 A2

B
Figure 1. Illustration of the regions of inter-
est on MRS and [18F]FDG-PET. [A1] ROI in
posterior cingulate drawn in the FLAIR se-
quence of MRI (red square); [A2] different
peaks calculated on MRS; [B] (lower row):
ROI in PCC of [18F]FDG-PET images, drawn
with the SPM8 MarsBar toolbox.

Figure 2. Illustrative anatomic location of

peak voxels of rBGM reductions as mea-
sured by [18F]FDG-PET. [A] (upper row): areas B
of rBGM reduction in the AD group versus the
CG in large areas of the temporoparietal cor-
tex, posterior cingulate and prefrontal cortex;
[B] (middle row): areas of rBGM reduction in
the AD group versus the MCI group, show-
ing hypometabolism in similar areas seen in C
A, albeit with lesser extension and intensity;
[C] areas of rBGM reduction in the MCI group
versus CG, restricted to the temporal lobes
and temporo-parietal association cortex,
without significant changes in the PCC.

ences across the groups (AD, MCI and CG) using the In order to obtain values of the radioactive counts
SPM software. Post-hoc analyses with unpaired T-tests related to the rBGM in the PCC as measured with [18F]
were used to examine differences between each pair of FDG-PET, a direct analysis of this region was per-
groups. SPM8 maps were generated with a visualization formed with SPM, adopting the small volume correc-
threshold of p<0.001 and the threshold for significance tion approach (SVC). After identifying the cluster with
at the voxel level was set at p=0.001 (Z score=3.09) with rBGM reduction in the PCC in the AD group, a volumet-
a minimum extension of 10 voxels in the corresponding ric region of interest (ROI) of this cluster was generated
cluster. The initial exploratory analyses with SPM maps (Figure 1). In order to increase the specificity of this anal-
generated a t statistic for each voxel, thus constituting ysis, the statistical cutoff was set at p<0.05, corrected for
statistical parametric maps. multiple comparisons with the familywise error method

388 FDG-PET and MRS in AD     Coutinho et al.

 Dement Neuropsychol 2015 December;9(4):385-393

(pFWE), with a minimum extension of 20 voxels in the sion. The majority of these areas persisted after correc-
corresponding cluster. Subsequently, numeric values rep- tions for multiple comparisons using the FWE method
resenting [18F]FDG uptake measures in that cluster for (pFWE<0.001). The MCI individuals showed rBGM
each individual in all groups (after the whole normaliza- reduction in the temporal association cortex in relation
tion process) were extracted with the toolbox MarsBar to CG (p<0.001)(not surviving correction for multiple
for SPM (http://marsbar.sourceforge.net/) under the comparisons) that was more restricted to the temporal
option “explore design/files and factors”.32 lobes compared to the hypometabolism seen in the AD
Demographic data and the values of the MRS NAA/ group. The SVC analysis of the PCC depicted no differ-
mI ratio in the PCC were compared across groups by an ences between the MCI group and the CG after correc-
ANOVA analysis with the aid of SPSS software version tion for multiple comparisons (pFWE<0.05). The areas of
17.0 (SPSS Inc., Chicago IL). metabolic reduction are illustrated in Figure 2 (complete
After obtaining the average radioactive counts in statistical results of the SPM8 analysis are beyond the
the PCC and the NAA/mI ratio of all subjects, numeric scope of the present work and are not provided).
data were assessed with the SPSS software to identify MRS analysis showed lower NAA/mI values in the
possible correlations among the data. Sensitivity and AD group compared to the CG (p=0.024). A tendency for
specificity curves for each method were also generated lower NAA/mI peak in the PCC was found in the MCI
in order to compare the diagnostic performance of the group compared to the CG (p=0.06). This data is also
two approaches. shown in Table 2 and illustrated in Figure 3.
A positive correlation between rBGM and NAA/mI
RESULTS peak in the PCC was found (r= 0.317; p= 0.012) (Table
Eighty-seven (87) individuals were included and clas- 2 and Figure 3). Lower NAA/mI in the PCC voxel differ-
sified into one of the three groups: AD (n=32), MCI entiated AD patients from control subjects, with an area
(n=27) and CG (n=28). Demographic data for the under the receiver operating characteristic curve of 0.70
sample is shown in Table 1. Subjects included in the (CI=0.57-0.84, p =0.006), while the ROI analysis of the
CG were younger (p<0.001) than those from the AD PET data yielded a value of 0.93 (CI=0.88-0.99, p<0.001)
group, had more years of education than both patient (Figure 3).
groups (p=0.001 for MCI and p<0.001 for AD) and also
higher Mini-Mental State Examination (MMSE) scores DISCUSSION
than both the MCI (p=0.031) and AD (p<0.001) groups. Hypometabolism in the PCC showed good correlation
Performance on the MMSE was also higher in the MCI with clinical measures of cognitive impairment such
group than in the AD group (p<0.001). as the CDR sum of boxes.33 This reduction is classi-
The AD group exhibited rBGM reduction in large cally related to conversion from MCI to AD and is also
areas of the PCC and temporoparietal cortex compared considered a standard biomarker for differentiating AD
to the CG, but also in less evident areas of the frontal from non-demented subjects.7,10,34,35 The areas of rBGM
cortex. This metabolism reduction existed in similar reduction seen in the temporoparietal cortex of AD and
areas among the MCI patients, albeit with lesser exten- MCI groups in comparison to the CG were also previ-

Table 1. Demographic data for the sample.

CG=28 (CDR=0) MCI=27 (CDR=0.5) AD=32 (CDR=1.0) P (two-tailed)

Mean (SD) Mean (SD) Mean (SD) Multiple comparison

Age (Y)* 69.7 (6.6) 72.7 (6.8) 76.3 (6.7) 0.001

CG × AD

Gender (F/M)** 6 / 22 12 / 15 10 / 22 p>0.05

Education (Y)* 12.8 (5.1) 8.0 (4.9) 7.1 (4.1) <0.001

CG × AD (<0.001) &
CG × MCI (p =0.001)

MMSE* 29.0 (1.0) 27.2 (2.1) 22.9 (3.4) <0.001

CG × AD (<0.001); MCI × AD (p<0.001) & CG × MCI (p =0.031). *ANOVA (Post-hoc test: Bonferroni); **Chi-Square; AD: Alzheimer’s disease; CG: control group; F: female; M: male; MCI: Mild
cognitive impairment; MMSE: Mini-Mental State Examination; SD: Standard deviation; Y: Years.

Coutinho et al.    FDG-PET and MRS in AD 389

 Dement Neuropsychol 2015 December;9(4):385-393

ously described as typical areas of neurodegeneration in the AD group, showing a lower Naa/mI ratio for
in these conditions.6,8,9 The results of the present study the AD group compared to CG. Naa/mI was also
confirmed these findings by showing rBGM reduction in lower in the MCI group, but again did not reach sta-
the temporoparietal cortex of both AD and MCI groups, tistical significance compared to the CG. These results
albeit with lesser extension and intensity in the latter, as failed to corroborate the final results of a related
expected. However, the rBGM reduction in the PCC was meta-analysis, which found lower values of Naa/
not statistically significant in the MCI group, thereby mI in MCI subjects.13 Some of the articles included
failing to corroborate the results of other authors. in this meta-analysis, however, also found no differ-
The MRS Naa/mI analysis revealed similar results ences in the Naa/mI ratio in the PCC of MCI subjects,

Table 2. Summary of key findings of the study.

A – [18F]FDG-PET analysis using SVC of the PCC with SPM8*
Comparisons Cluster size (mm3) pFWE p# Z score
AD × CG
Right posterior cingulate gyrus 129 < 0.001 < 0.00001 7.49

Left posterior cingulate gyrus 241 < 0.001 < 0.00001 7.56
MCI × CG No suprathreshold clusters (p >0.001)
B – Naa/mI MRS ROI**
Comparisons p
AD × CG 0.024
MCI × CG 0.060
C – Correlation analysis
Correlation p
[ F] FDG-PET × Naa/mI
18
0.361 0.001
D – ROC curve analysis of the different PCC ROIs
Area under the ROC curve
[18F] FDG-PET 0.935
Naa/mI 0.708
*Results at the peak voxel level (ANOVA and post-hoc unpaired t-test); **ANOVA and post-hoc unpaired t-test with SPSS; #p value uncorrected for multiple comparisons; pFWE: p value corrected
for multiple comparisons with the familywise error method; MCI: Amnestic MCI; CG: Control group; SPM: statistical parametric mapping; SVC: small volume correction method, directed to the PCC.

Figure 3. [A] Scatter plot of the correlation curve

between values of rBGM (y axis) and Naa/mI (x
axis) in the PCC; [B] ROC curves of the directed
analysis of the PCC with MRS and Naa/mI peak
A B and with the [18F]FDG-PET ROI analysis.

390 FDG-PET and MRS in AD     Coutinho et al.

 Dement Neuropsychol 2015 December;9(4):385-393

while the lower number of subjects included in the to the deposition of amyloid in the AD neurodegenera-
present study should also be taken into account.13 tion process.36,42 Our findings of a positive correlation
On the PCC evaluation, the ROC curve analysis of between rBGM and Naa/mI in the PCC of subjects
[18F]FDG was superior than the Naa/mI ratio for dis- exhibiting different stages of cognitive function are in
criminating AD subjects from cognitively normal older line with this hypothesis. This indicates that the hypome-
adults. These results indicate that, although a promising tabolism seen in AD and MCI in the PCC is proportion-
tool for evaluating subjects with cognitive decline, analy- ally accompanied by a reduction in neuronal density as
sis of Naa/mI peak by MRS still lacks the sensitivity of measured by MRS, which likely indicates neuronal injury.
rBGM evaluation with [18F]FDG-PET. The present study has some limitations. First,
With regard to the MCI group, both methods failed to patients from the AD group were older than subjects
detect significant differences between the MCI group and from the CG a factor that may have had some influence
the CG in the PCC. [18F]FDG-PET, however, disclosed dif- on the results. However, age is a major risk factor for
ferences between the MCI and CG groups in other areas. AD and age differences are therefore expected.1,2 Also,
A comprehensive analysis of the whole brain with MRS the present degree of rBGM reduction in the PCC and
was not performed since it is technically difficult, repre- temporoparietal association cortex is not expected in
senting a limitation of the method. normal aging.36 Thus, it is unlikely that the higher age
Which areas first present hypometabolism or atrophy of the subjects included in the AD group influenced the
in AD and normal aging remains unclear and a matter of results of the imaging analysis.
ongoing debate.36 While some authors have found hypo- Second, the CG had more years of education than
metabolism in the PCC before other changes in MCI, the other groups. Bearing in mind the cognitive reserve
others have found that blood flow and rBGM reductions hypothesis, education is probably a protective factor for
in the precuneus and temporoparietal cortex can occur the development of AD and may influence the results
without evident PCC hypometabolism.31,37 of neuropsychological and neuroimaging tests.36,45 How-
Some authors also argue that rBGM reduction in MCI ever, according to this theory, subjects with more years of
could be the indirect result of atrophy and partial volume education would have preserved cognitive performance
effect (PVE), especially in the medial temporal lobes,38 even if presenting some degree of neurodegeneration.46-48
since atrophy in large areas of the temporal lobes occurs Hence, the subjects in the CG should have lower levels of
in early AD.39 Given our data was not corrected for PVE, rBGM in certain areas, with cognitive functioning close
this hypothesis could not be tested here and may be con- to or within the normal range. This was not seen in the
sidered a limitation of the present study. present cohort, where the MCI and AD groups presented
Hinrichs et al (2011),40 using a machine-learning with significant areas of hypometabolism. Therefore, it
multi-modal approach, proposed that the combination cannot be excluded that this factor could have contrib-
of different biomarkers is superior to each individually uted to the lack of differences in the PCC between the CG
for predicting conversion to AD in MCI. However, [18F] and the MCI group seen in both methods. Some of the
FDG-PET tended to be better than other techniques as patients with a mild degree of neurodegeneration and
a single modality although the authors did not include higher educational levels could be classified as normal on
MRS in their analysis. The present study adds informa- the clinical tests, yet harbor some degree of degeneration
tion to the cited study, supporting the notion that [18F] in the PCC. However, this is one the drawbacks of the
FDG as a single modality is superior to others for detect- clinical diagnosis based on neuropsychological testing.
ing neurodegeneration in patients with early AD, espe- This possibility can only be tested by comparing these
cially in the PCC. values in a cohort of MCI and cognitively normal elderly
Brain glucose metabolism is a surrogate marker of subjects paired by age and years of education or after
synaptic activity.41 Accordingly, metabolism should cor- prospective evaluation of the patients.
relate with measures of neuronal activity and density, In summary, rBGM and NAA/mI ratio in the PCC
such as Naa/mI ratio measured with MRS. This hypoth- showed a positive correlation in elderly individuals with
esis was confirmed in the present analysis of the PCC AD, MCI and no cognitive impairment. Thus, hypome-
cortex and is the most remarkable finding of the study. tabolism and neuronal injury are probably directly related
The PCC is a hub of the brain’s default mode network in the different phases of the AD pathologic and normal
and one of the most active parts of the brain in the rest aging process. Both methods proved able to distinguish
state.42-44 According to some theories, this renders the AD patients from controls when evaluating the PCC, with
region particularly vulnerable to neuronal injuries and [18F]FDG-PET providing greater accuracy than Naa/mI.

Coutinho et al.    FDG-PET and MRS in AD 391

 Dement Neuropsychol 2015 December;9(4):385-393

Authors contibution. Coutinho AMN, Leite CC and MCO script critically for important intellectual content and
conceived the study, participated in its design and approved the final version.
coordination, and drafted the manuscript. FHGC, PFZ
and RFN performed statistical analysis, assisted with
imaging process using SP8 and with drafting of the Acknowledgments. This study was funded by grants
manuscript. CMCB and CAB participated in the study from the Fundação de Amparo à Pesquisa do Estado
design and coordination. TLP performed the spec- de São Paulo (FAPESP) numbers 2011/18245-4 and
troscopy analysis. All authors revised the final manu- 2009/17398-1 in Brazil.

REFERENCES
1. World Health Organization and Alzheimer’s Disease International. 20. Morris JC. The Clinical Dementia Rating (CDR) – current version and
Dementia: a public health priority. [Internet]. [cited 2015 jul 3]; Avail- scoring rules. Neurology 1993;43: 2412-2414.
able from: <http://www.who.int/mental_health/publications/dementia_ 21. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”. A practical
report_2012/en/>. method for grading the cognitive state of patients for the clinician. J
2. Harper S. Economic and social implications of aging societies. Science Psychiatr Res 1975;12:189-98.
2014;346:587-591. 22. Nitrini R, Caramelli P, Herrera EJr, et al. Performance of illiterate and
3. Clopton D, Jason DT. Advances in dementia imaging. Semin Roentgenol literate nondemented elderly subjects in two tests of long-term memory.
2014;49:53-63. J Int Neuropsychol Soc 2004;10:634-638.
4. Petersen R, Smith G, Waring S, Ivnik R, Tangalos E, Kokmen E. Mild 23. Mattis S. Dementia Rating Scale. Professional manual. Florida: Psycho-
Cognitive Impairment: Clinical Characterization and Outcome. Arch logical Assessment Resources;1988.
Neurol 1999;56:303-8. 24. Porto CS, Fichman HC, Caramelli P, Bahia VS, Nitrini R. Brazilian version
5. Petersen RC. Mild Cognitive Impairment. NEJM 2011;364:2234. of the Mattis Dementia Rating Scale: Diagnosis of mild dementia in
6. Cerami C, Della Rosa PA, Magnani G, et al. Brain metabolic maps in Mild Alzheimer`s Disease. Arq Neuropsiquiatr 2003;61:339-345.
Cognitive Impairment predict heterogeneity of progression to dementia. 25. Wechsler D. Wechsler Memory Scale. Manual The Psychological Corpo-
NeuroImage Clin 2015;7:187-194. ration Harcourt Brace Jovanovich; 1987.
7. Minoshima S, Giordani B, Berent S, Frey K, Foster N, Kuhl D. Meta- 26. Spreen O, Strauss E. A Compendium of Neuropsychological Tests.
bolic reduction in the posterior cingulate cortex in very early Alzheimer’s Administration, Norms, and Commentary. Second Edition. Oxford
disease. Ann Neurol 1997;42:85-94. University Press; 1998.
8. Jagust W, Gitcho A, Sun F, Kuczynski B, Mungas D, Haan M. Brain 27. Buschke, H. Selective reminding for analysis of memory and learning. J
imaging evidence of preclinical Alzheimer’s disease in normal aging. Ann Verbal Learning Verbal Behav 1973;12:543-550.
Neurol 2006;59:673-681. 28. Nascimento, E. WAIS-III: Escala de Inteligência Wechsler para Adultos:
9. Salmon E, Lekeu F, Garraux G, et al. Metabolic correlates of clinical Manual David Wechsler;Adaptação e padronização de uma amostra
heterogeneity in questionable Alzheimer’s disease. Neurobiol Aging brasileira. 1st ed. São Paulo: Casa do Psicólogo; 2004.
2008;29:1823-1829. 29. Provencher SW. Estimation of metabolite concentrations from localized in
10. Drzezga A, Lautenschlager N, Siebner H, et al. Cerebral metabolic vivo proton NMR spectra. Magn Reson Med 1993;30:672-679.
changes accompanying conversion of mild cognitive impairment into 30. Porto FHG, Coutinho AMN, Pinto ALS, et al. Effects of Aerobic Training
Alzheimer’s disease: a PET follow-up study. Eur J Nucl Med Molr Imaging on Cognition and Brain Glucose Metabolism in Subjects with Mild Cogni-
2003;30:1104-1113. tive Impairment. J Alzheimers Dis 2015;46:747-760.
11. Silverman D. Brain 18F-FDG PET in the diagnosis of neurodegenerative 31. Coutinho AMN, Porto FHG, Duran FLS, et al. Brain metabolism and
dementias: comparison with perfusion SPECT and with clinical evalua- cerebrospinal fluid biomarkers profile of non-amnestic mild cognitive
tions lacking nuclear imaging. J Nucl Med 2004;45:594-607. impairment in comparison to amnestic mild cognitive impairment and
12. Kantarci K. Proton MRS in mild cognitive impairment. J Magn Reson normal older subjects. Alzheimers Res Ther 2015;7:58.
Imaging 2013;37:770-777. 32. Brett M, Anton J, Valabregue R, Poline J. Region of interest analysis
13. Tumati S, Martens S, Aleman A. Magnetic resonance spectroscopy using an SPM toolbox. NeuroImage 2002;16: abstract 497.
in mild cognitive impairment: systematic review and meta-analysis. 33. Perneczky R, Hartmann J, Grimmer T, Drzezga A, Kurz A. Cerebral
Neurosci Biobehav Rev 2013;37:2571-2586. metabolic correlates of the clinical dementia rating scale in mild cogni-
14. Targosz-Gajniak MG, Siuda JS, Wicher MM, et al. Magnetic resonance tive impairment. J Geriatr Psychiatry Neurol 2007;20:84-88.
spectroscopy as a predictor of conversion of mild cognitive impairment 34. Herholz K, Westwood S, Haense C, Dunn G. Evaluation of a calibrated
to dementia. J Neurol Sci 2013;335:58-63. (18)F-FDG PET score as a biomarker for progression in Alzheimer
15. Zhang B, Ferman TJ, Boeve BF, et al. MRS in mild cognitive impair- disease and mild cognitive impairment. J Nucl Med 2011;52:1218-1226.
ment: early differentiation of dementia with Lewy bodies and Alzheimer’s 35. Herholz K. Use of FDG PET as an imaging biomarker in clinical trials of
disease. Neuroimaging 2015;25:269-274. Alzheimer’s disease. Biomark Med 2012;6:431-439.
16. Walecki J, Barcikowska M, Cwikła JB, Gabryelewicz T. N-acetylaspar- 36. Fjell AM, McEvoy L, Holland D, Dale AM, Walhovd KM. Alzheimer’s
tate, choline, myoinositol, glutamine and glutamate (glx) concentration Disease Neuroimaging Initiative. What is normal in normal aging? Effects
changes in proton MRspectroscopy (1H MRS) in patients with mild of aging, amyloid and Alzheimer’s disease on the cerebral cortex and the
cognitive impairment (MCI). Med Sci Monit 2011;17:MT105-111. hippocampus. Prog Neurobiol 2014;117:20-40.
17. Zhu X, Cao L, Hu X, et al. Brain metabolism assessed via proton 37. Johnson K, Moran EK, Becker JA, Blacker D, Fischman AJ, Albert MS.
magnetic resonance spectroscopy in patients with amnestic or vascular Single photon emission computed tomography perfusion differences
mild cognitive impairment. Clin Neurol Neurosurg 2015;130:80-85. in mild cognitive impairment. J Neurol Neurosurg Psychiatry 2007;78:
18. Glodzik L, Sollberger M, Gass A, et al. Global N-acetylaspartate in 240-247.
normal subjects, mild cognitive impairment and Alzheimer’s disease 38. Karow DS, McEvoy LK, Fennema-Notestine C, et al. Relative capa-
patients. J Alzheimers Dis 2015;43:939-947. bility of MR imaging and FDG PET to depict changes associated with
19. McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan prodromal and early Alzheimer disease. Radiology 2010;256: 932-942.
EM. Clinical diagnosis of Alzheimer’s disease: report of the NINCDS- 39. Carmichael O, McLaren DG, Tommet D, Mungas D, Jones RN. Alzheim-
ADRDAWork Group under the auspices of Department of Health and er’s disease neuroimaging initiative. Coevolution of brain structures in
Human Services Task Force on Alzheimer’s Disease. Neurology 1984;34: amnestic mild cognitive impairment. Neuroimage 2013;66:449-456.
939-944. 40. Hinrichs C, Singh V, Xu G, Johnson S, Initiative A. Predictive markers for

392 FDG-PET and MRS in AD     Coutinho et al.

 Dement Neuropsychol 2015 December;9(4):385-393

AD in a multi-modality framework: an analysis of MCI progression in the 44. Buckner R. The serendipitous discovery of the brain’s default network.
ADNI population. NeuroImage 2011;55:574-589. Neuroimage 2012;62:1137-1145.
41. Jack CR Jr, Knopman DS, Jagust WJ, et al. Tracking pathophysiological 45. Larson EB, Yaffe K, Langa KM. New Insights into the Dementia Epidemic.
processes in Alzheimer’s disease: An updated hypothetical model of N Engl J Med 2013;369:2275-2277.
dynamic biomarkers. Lancet Neurol 2013;12:207-216. 46. Garibotto V, Borroni B, Kalbe E, et al. Education and occupation as
42. Buckner R, Snyder AZ, Shannon BJ, et al. Molecular, structural, and proxies for reserve in aMCI converters and AD FDG-PET evidence.
functional characterization of Alzheimer’s disease: evidence for a relation- Neurology 2008;71:1342-1349.
ship between default activity, amyloid, and memory. J Neurosci 2005; 47. Tucker AM, Stern Y. Cognitive reserve in aging. Curr Alzheimer Res
25:7709-7717. 2011;8:354-360.
43. Buckner R, Andrews-Hanna J, Schacter D. The brain’s default network: 48. Morbelli S, Perneczky R, Drzezga A, et al. Metabolic networks underlying
anatomy, function, and relevance to disease. Ann N Y Acad Sci 2008; cognitive reserve in prodromal alzheimer disease: a European alzheimer
1124:1-38. disease consortium project. J Nucl Med 2013;54:894-902.

Coutinho et al.    FDG-PET and MRS in AD 393