Alzheimers Dement. 2017 (Oct 19) [Epub] ~ FULL TEXT
Yang An, Vijay R. Varma, Sudhir Varma, Ramon Casanova, Eric Dammer et al.
Laboratory of Behavioral Neuroscience,
National Institute on Aging (NIA),
National Institutes of Health (NIH),
Baltimore, MD, USA.
INTRODUCTION: It is unclear whether abnormalities in brain glucose homeostasis are associated with Alzheimer's disease (AD) pathogenesis.
METHODS: Within the autopsy cohort of the Baltimore Longitudinal Study of Aging, we measured brain glucose concentration and assessed the ratios of the glycolytic amino acids, serine, glycine, and alanine to glucose. We also quantified protein levels of the neuronal (GLUT3) and astrocytic (GLUT1) glucose transporters. Finally, we assessed the relationships between plasma glucose measured before death and brain tissue glucose.
RESULTS: Higher brain tissue glucose concentration, reduced glycolytic flux, and lower GLUT3 are related to severity of AD pathology and the expression of AD symptoms. Longitudinal increases in fasting plasma glucose levels are associated with higher brain tissue glucose concentrations.
DISCUSSION: Impaired glucose metabolism due to reduced glycolytic flux may be intrinsic to AD pathogenesis. Abnormalities in brain glucose homeostasis may begin several years before the onset of clinical symptoms.
Keywords: Glucose; Insulin resistance; Alzheimer’s disease; GLUT3; GLUT1; Neuritic plaque; Neurofibrillary tangles; Mass
From the Full Text Article:
Although numerous epidemiological studies indicate that
peripheral insulin resistance and diabetes are risk factors for
Alzheimer’s disease (AD) [1–3], it is not known whether
brain glucose dysregulation is a key feature of AD and is
related to severity of AD pathology or symptom expression. [4, 5] Previous studies have shown that several
components of the insulin signaling pathway are abnormal
in AD brains relative to controls, including genes encoding
insulin, IGF-1, and IGF-2 peptides and their receptors.
[6–10] Because these abnormalities appear to be a
common feature of both type-1 and type-2 diabetes, the
term “type-3 diabetes” was proposed to describe brainspecific
abnormalities in insulin signaling associated with
AD. [11, 12] Taken together, the large body of evidence
implicating abnormal insulin signaling in AD has led to
clinical trials targeting these abnormalities in patients with
mild cognitive impairment and AD. [13–15] However, it is
well recognized that glucose transport from the peripheral
circulation across the blood-brain barrier and capillary endothelial
cells into the interstitial fluid and brain tissue are
largely insulin-independent processes. [16, 17] Similarly,
the transport of glucose across the cell membrane into
neurons is largely independent of insulin . Although
18F-deoxyglucose positron emission tomography (18FDGPET)
studies have shown reduced brain glucose uptake in regions
vulnerable to AD pathology [19–22], it is unclear
whether an overall failure of regulation of brain glucose
metabolism is a key etiopathogenic factor in AD and
whether abnormalities of brain glucose homeostasis in AD
are related to peripheral glucose concentration. Answering
these questions is critical to establishing whether central
glucose homeostasis is a potential target for diseasemodifying
treatments in AD.
In this study, we asked the following main questions:
Is brain tissue glucose concentration altered in AD?
What is the relationship between brain tissue glucose
concentration and severity of AD pathology?
What are plausible molecular mechanisms underlying
abnormalities of brain glucose homeostasis in AD?
What is the relationship between trajectories of blood
glucose concentration during life and brain tissue
glucose levels measured at death?
Our results provide the first evidence for brain glucose
dysregulation as a critical event in AD pathogenesis that
closely reflects both severity of AD pathology and the
expression of symptoms.
The Baltimore Longitudinal Study of Aging (BLSA) is a
prospective, ongoing cohort study of community-dwelling
volunteer participants in Baltimore that began in 1958 and
has been described in detail previously [23, 24].
Historically, participants underwent extensive biomedical
examination and neuropsychological testing every 2 years.
From 2003, participants under age 60 years are assessed
every 4 years; those aged 60 to 79 years every 2 years and
participants aged 80 years and older are assessed annually.
Written informed consent was obtained at each visit, and
the study was approved by the local institutional review
board and the National Institute on Aging. The participants
in this report were from the autopsy program of the BLSA
that was initiated in 1986 and has been described
previously . They provided data on concentrations of
brain tissue glucose and the glycolytic amino acids, serine,
glycine, and alanine (N = 43; from the middle frontal gyrus
[MFG], inferior temporal gyrus [ITG], and cerebellum), as
well as proteomic data from the MFG (N = 47). The mean
age at death in the sample was 86.6 ± 9.5 years (range
62.9–99.2). As reported previously, the autopsy subsample
is not significantly different from the BLSA cohort as a
whole in terms of the rates of dementia and clinical stroke
At each assessment, participants underwent a battery of
neuropsychological testing. Clinical and neuropsychological
data were reviewed at consensus case conferences if
they made four or more errors on the Blessed Information
Memory Concentration test, if their Clinical Dementia Rating
score was equal to or greater than 0.5, or if concerns were
raised about their cognitive status. In addition, all participants
were evaluated by case conference on death or withdrawal.
The diagnoses of dementia and AD were based on
the Diagnostic and Statistical Manual-III-R  and the National
Institute of Neurological and Communication Disorders
and Stroke–Alzheimer’s Disease and Related
Disorders Association criteria , respectively.
Postmortem brain examinations were performed by an
experienced neuropathologist (J.C.T.). Assessment of
neuritic plaques and neurofibrillary tangles using Consortiumto
Establish a Registry for Alzheimer’s Disease (CERAD)
 and Braak criteria , respectively, have been
described previously . We have previously described
the clinico-pathological features of BLSA participants categorized
as “asymptomatic Alzheimer’s disease (ASYMAD)”
after neuropathological assessment at death .
Briefly, these individuals have significant AD neuropathology
at autopsy, but without evidence for cognitive
impairment during life, as assessed by longitudinal cognitive
evaluations during their BLSA research visits.
Plasma glucose measurement
Plasma glucose measurements were obtained from
venous blood samples after an overnight fast by the
glucose-oxidase method as described previously . We
used all available longitudinal plasma glucose data (445 observations,
mean follow-up interval. 19.1 years). We
excluded 10 data points where fasting plasma glucose values
were beyond three standard deviations from the mean value.
Quantitative metabolomics assays of brain tissue glucose and
glycolytic amino acids (serine, glycine, and
Glucose concentration was measured in frozen brain
tissue samples from 43 BLSA participants (N = 14 AD;
N = 14 control; and N = 15 “ASYMAD”) on the BIOCRATES
P180 platform. Brain tissue regions for glucose
assays were selected a priori in the MFG and ITG to represent
brain regions vulnerable to amyloid and tau deposition,
respectively. The cerebellum was sampled to
represent an additional brain region resistant to classical
AD pathology .
A sterile 4-mm diameter tissue punch was extracted from
the cortical surface of the three brain regions, that is, MFG,
ITG, and the cerebellum from brain tissue samples stored at
–80°C. To extract metabolites from the brain tissue, samples
were homogenized using Precellys with ethanol phosphate
buffer. Samples were centrifuged, and the
supernatant was used for analysis. The BIOCRATES P180
platform was used for the quantification of glucose and the
three glycolytic amino acids, serine, glycine, and alanine.
The fully automated assay was based on phenylisothiocyanate
derivatization in the presence of internal standards followed
by flow injection analysis–mass spectrometry (for
glucose) and liquid chromatography–tandem mass spectrometry
(for glycolytic amino acids) using a SCIEX 4000
QTrap mass spectrometer (SCIEX, Darmstadt, Germany)
with electrospray ionization. Brain tissue glucose concentration
was derived as the sum of hexoses detected and absolute
concentration expressed as nmol/mg tissue.
To assess activities of the three rate-controlling enzymes
of glycolysis (i.e., hexokinase [HK], phosphofructokinase
[PFK], and pyruvate kinase [PK]), we calculated the ratios
of the concentrations of the glycolytic amino acids serine,
glycine, and alanine to glucose. These amino acids are
biosynthetic derivatives of intermediate metabolites in
glycolysis and thus their concentrations relative to glucose
are an indirect measure of the net enzymatic activities catalyzing
the three sequential, rate-controlling, irreversible
steps of glycolysis . As shown in Fig. 1, the amino
acid serine is synthesized from the glycolytic intermediate
3-phosphoglycerate and can be converted to glycine after
transfer of a methyl group from its side chain. The ratio of
the concentrations of serine and glycine to glucose (serine
1 glycine:glucose) therefore reflects the net activities of
the two irreversible enzymatic reactions in glycolysis that
are upstream of 3-phosphoglycerate:
the first reaction in glycolysis, that is, the phosphorylation
of glucose to glucose 6-phosphate, catalyzed by
the inducible enzyme, hexokinase (abbreviated here
as HK) and
the phosphorylation of fructose 6-phosphate to fructose
1, 6-bisphosphate in the presence of phosphofructokinase
(abbreviated here as PFK).
The conversion of phosphoenolpyruvate to pyruvate is
the last step of glycolysis and is an irreversible reaction catalyzed
by the enzyme pyruvate kinase (abbreviated here as
PK) (Figure 1). The amino acid alanine is synthesized from pyruvate
by a transamination reaction catalyzed by alanine
amino transferase. Therefore, the ratio of serine 1 glycine
1 alanine to glucose provides an indirect assessment of
the net activity of pyruvate kinase.
Quantification of brain tissue glucose transporter proteins
We examined protein levels of the two principal glucose
transporters in the brain, that is, neuronal glucose
transporter-3 (GLUT3) and glucose transporter-1
(GLUT1); the main glucose transporter in astrocytes and
within vascular endothelial cells of the blood-brain barrier
Label-free quantification (LFQ) of the brain proteome by
LC-MS/MS was performed in the MFG using a Thermo-
Fisher Scientific (San Jose, CA) Q-Exactive Plus mass spectrometer.
Analysis of raw data was performed using
MaxQuant 18.104.22.168 software (Max Planck Institute of
Biochemistry, Martinsried, Germany) and bootstrap regressed
against age and postmortem interval of each individual.
In addition to protein levels of GLUT3 and GLUT1, we
also quantified levels of nesprin-1, a neuronal nuclear protein.
Additional details on collection of proteomic data,
raw data availability, and analyses parameters are provided
through the Synapse AMP-AD portal (https://www.
synapse.org/#!Synapse:syn3606086) and have been reported
previously . Additional details on the specificity
of the LFQ workflow to identify GLUT3 and GLUT1 proteins
are provided in Supplementary Materials (Table-S1).
The proteomic analyses were performed on the MFG in
47 BLSA participants (N = 20 AD; N = 13 control; and
N = 14 “ASYMAD”) and included 40 individuals in
whom brain tissue glucose and glycolytic amino acids
concentrations were also estimated by quantitative metabolomics
as described in the previous section.
Differences in severity of AD pathology between groups
were examined using the Mantel-Haenszel chi-square test of
correlation. Proportional odds ordinal logistic models 
are a generalization of the Wilcoxon and Kruskal-Wallis
tests that allow for covariates adjustment. We used these
tests to compare group differences (i.e., AD, control, and
ASYMAD) in brain tissue glucose concentrations as well
as ratios of glycolytic amino acids to glucose within each
of the three brain regions (i.e., ITG, MFG, and cerebellum).
Sensitivity analyses were conducted using sex and age at
death as covariates.
A similar approach was taken for analyses of group differences
in protein levels of the glucose transporters, GLUT3 and GLUT1. To confirm that observed changes in
levels of the neuronal glucose transporter, GLUT3, were
not driven primarily by neuronal loss, LFQ-derived protein
levels of the neuronal nuclear protein, nesprin-1 , were
used as an additional covariate in addition to sex and age
at death in sensitivity analyses comparing GLUT3 protein
levels between groups.
Associations between brain tissue glucose concentration,
ratios of the glycolytic amino acids to glucose, and AD pathology
were examined using Spearman’s rank correlation
before and after adjusting for sex and age at death. Associations
between protein levels of the glucose transporters and
AD pathology were examined similarly with protein levels
of nesprin-1 as an additional covariate in analyses involving
To investigate the associations between brain tissue
glucose concentrations and AD pathology with longitudinal
fasting plasma glucose concentrations measured during the
BLSA research visits, separate linear mixed-effects models
were fit with longitudinal glucose measures as the outcome
and each pathology variable (i.e., CERAD and Braak score)
and brain glucose concentration as the main predictors. The
time of follow-up was anchored at the last measurement of
fasting plasma glucose (using time of last measurement as
the time origin, time = 0), with all previous longitudinal observations
considered negative relative to the last
measurement. This recentering of the time variable allows us
to test the effects of the last plasma fasting glucose levels as
well as the rates of change in fasting plasma glucose concentrations
on pathology and brain tissue glucose concentration
simultaneously in a single model. Other covariates included
age at measurement of the last plasma fasting glucose concentration
and sex. Brain tissue glucose concentration was
natural log transformed and z-scored, age was mean
centered and sex coded as 20.5 for female and 0.5 for male.
All the analyses were conducted in SAS 9.4 (Cary, NC).
Brain tissue glucose concentration in AD
The demographic characteristics of BLSA participants
who contributed data to the measurements of brain tissue
glucose concentration are shown in Table 1.
The three groups, that is, AD, controls, and “ASYMAD” did not differ
significantly in age at death or the postmortem interval to autopsy.
There were also no group differences in sex or APOE ε4 carrier status.
Fig. 2A summarizes results of analyses comparing brain
tissue glucose concentrations between the three groups. In
unadjusted models, brain tissue glucose concentrations
were significantly different in the ITG (global P value for
significance across groups = .0019) with tissue glucose concentration
following the pattern AD > ASYMAD > control
(Fig. 2A). Pairwise comparisons between the groups in the
ITG showed significantly higher concentration of glucose
in the AD (P = .0004) and ASYMAD groups (P = .0497)
relative to controls and a trend toward higher concentration
in AD relative to ASYMADs (P = .057). Results were
similar after adjusting for sex and age at death. In the
MFG, group differences from unadjusted models did not
reach significance (global P =.10) (Fig. 2A), but after additionally
adjusting for sex and age at death, results approached
significance (global P value for significance
across groups = .051). Pairwise comparisons between the
groups in the MFG showed significantly higher concentration
of glucose in the AD (P =.032) and ASYMAD groups
(P =.026) relative to controls. There were no group differences
in tissue glucose concentrations within the cerebellum
in either unadjusted (global P value for significance across
groups = .29; Fig. 2A) or adjusted models.
Brain tissue glucose concentration and AD pathology
The three groups differed significantly in the severity of
both neuritic plaque and neurofibrillary tangle pathology
following a linear trend (Mantel-Haenszel chi-square test
of correlation between group and Braak scores: P < .0001;
CERAD scores: P < .0001) with the AD group showing
the highest, ASYMAD intermediate, and controls lowest
levels of pathology.
Table 4 shows correlations between brain tissue glucose
concentration and measures of AD pathology. Brain tissue
glucose concentration in the ITG was significantly associated
with both Braak (Spearman’s r = 0.37; P = .014)
and CERAD scores (Spearman’s r = 0.47; P =.0014). Tissue
glucose concentrations in the MFG were significantly
associated with CERAD scores (Spearman’s r = 0.39;
P = .0096). No significant association was observed
between glucose concentration in the cerebellum and AD
pathology. These results remained similar after adjusting
for sex and age at death.
Brain tissue ratios of glycolytic amino acids to glucose in AD
Fig 2B and 2C summarize results of analyses comparing
brain tissue ratios of the glycolytic amino acids, serine 1
glycine to glucose; assessing hexokinase and phosphofructokinase
activities (“HK_PFK”) and serine + glycine + alanine to glucose; assessing pyruvate kinase activity
(“PK”) between the three groups, respectively. In
unadjusted models, brain tissue HK_PFK activities were
significantly different in the ITG (global P value for
significance across groups = .0007) following the pattern
AD < ASYMAD < control (Fig. 2B). Pairwise comparisons
between the groups in the ITG showed significantly
lower HK_PFK activity in the AD group relative to controls
(P 5.0001) as well as lower activity in AD relative
to ASYMADs (P 5.014). Similarly, in unadjusted models
within the ITG, we found that brain tissue PK activity was
also significantly different between the groups (ITG;
global P value for significance across groups 5 .0074)
with the pattern, AD , ASYMAD , control (Fig. 2C).
Pairwise comparisons between the groups in the ITG
showed significantly lower PK activity in the AD group
relative to controls (P = .0021) as well as lower activity
in AD relative to ASYMADs (P = .045). These results
were similar in models additionally adjusted for sex and
age at death. In the MFG, group difference in brain tissue
HK_PFK activities did not reach statistical significance in
the unadjusted model (Fig. 2B) but showed a trend toward
significance in adjusted models (HK_PFK; global P value
for significance across groups = .058). There were no significant
group differences in PK activity in the MFG in
either unadjusted (PK; global P value for significance
across groups = .368; Fig. 2C) or adjusted models. There
were no group differences in brain tissue activities of
either HK_PFK or PK within the cerebellum in either unadjusted
(HK_PFK; global P value for significance across
groups = .394; PK; global P value for significance across
groups - .410, Fig 2B and 2C) or adjusted models.
Brain tissue ratios of glycolytic amino acid to glucose and AD pathology
Table 4 shows correlations between ratios of the glycolytic
amino acids, serine + glycine to glucose; assessing
hexokinase and phosphofructokinase activities (i.e.,
“HK_PFK”) and serine + glycine + alanine to glucose; assessing
pyruvate kinase activity (i.e., “PK”) against measures
of AD pathology. Brain tissue HK_PFK activities in
the ITG were significantly associated with both Braak
(Spearman’s p = –0.37; P = .016) and CERAD (Spearman’s
p = –0.52; P = .0004) scores. Similarly, PK activity
in the ITG was also significantly associated with both Braak
(Spearman’s p = –0.46; P = .0093) and CERAD (Spearman’s
p = –0.55; P = .0010) scores. In the MFG, HK_PFK
activities were significantly associated with the CERAD
(Spearman’s p = –0.38; P = .013) score. These results remained
similar after adjusting for sex and age at death.
Brain tissue protein levels of glucose transporters (GLUT3 and GLUT1)
The demographic characteristics of BLSA participants
who contributed data to the proteomic analyses of brain tissue
glucose transporter levels are shown in Table 2. The
three groups, that is, AD, controls, and “ASYMAD,” did
not differ significantly in age at death or the postmortem interval
to autopsy. There were also no group differences in sex
or APOE ε4 carrier status.
Fig. 3 summarizes results of analyses comparing brain
tissue glucose transporter protein levels between the three
groups in the MFG. In unadjusted models, brain tissue
protein levels of the neuronal glucose transporter GLUT3
were significantly different across the groups (global P value
for significance = .0027) (Fig. 3). Pairwise comparisons
showed significantly lower protein levels of GLUT3 in the
AD (P = .0011) and ASYMAD (P = .0043) groups relative
to controls. These results remained similar in models
adjusted for sex, age at death, and levels of the neuronal nuclear
protein, nesprin-1. No significant differences were
observed in protein levels of the astrocytic glucose transporter
Brain tissue protein levels of glucose transporters and AD pathology
Protein levels of GLUT3 were significantly associated
with both Braak (Spearman’s p = –0.40; P = .0051) and
CERAD (Spearman’s p = –0.42; P = .0033) scores in the
MFG. There were no significant associations between protein
levels of GLUT1 and Braak (Spearman’s p = 0.038;
P = .80) or CERAD (Spearman’s p = –0.21; P - .16)
scores. These results remained unchanged in models
adjusted for sex, age at death, and protein levels of the
neuronal nuclear protein, nesprin-1 (Table 4).
Associations between plasma fasting glucose and brain tissue glucose concentrations
Table 3 summarizes the plasma fasting glucose measurements
that were used in the analyses examining their relationship
with brain tissue glucose concentrations. A total
of 445 plasma fasting glucose measures were available in
the 43 BLSA participants who also contributed measures
of brain tissue glucose concentration described previously.
The average follow-up interval for the plasma glucose measurements
was 19.1 years, and the average interval between
the last available fasting plasma glucose assessment and
death was 5 years.
In linear mixed models adjusted for age at the last
available plasma glucose measurement and sex, we
found that both the last plasma fasting glucose concentration
and longitudinal trajectories of change in plasma
fasting glucose were associated with brain tissue glucose
concentrations in all three regions examined, that is,
ITG, MFG, and cerebellum (Table 5). The direction of
this association indicates that higher levels (i.e., last
measured fasting plasma glucose concentration) as well
as greater increases over time in plasma fasting glucose
are associated with higher brain tissue glucose concentrations.
These results were similar in models that were
further adjusted for clinical status, that is, AD/control/
In sensitivity analyses, we excluded 26 plasma fasting
glucose values in the diabetic range, that is, ≥126 mg/dL;
longitudinal associations remained unchanged in the
MFG and ITG; however, cross-sectional associations
only remained significant in the ITG. When we examined
the relationship between plasma fasting glucose
concentrations and AD pathology, there were no
significant associations with either Braak or CERAD
In this study, we have demonstrated that abnormalities
in brain glucose homeostasis are intrinsic to AD pathogenesis
and may begin several years before the onset of
clinical symptoms. We first showed that brain regions
vulnerable to amyloid deposition and neurofibrillary pathology
show significantly higher tissue glucose
concentrations in AD. Moreover, higher concentrations of
brain tissue glucose are associated with greater severity of
both amyloid plaque deposition and neurofibrillary
To further investigate abnormalities in glucose metabolism
associated with higher glucose concentrations in
brain regions vulnerable to AD pathology, we used
concentrations of the glycolytic amino acids, serine,
glycine, and alanine to assess the three rate-controlling,
irreversible steps of glycolysis. Besides the generation of
energy through ATP from the breakdown of glucose, intermediates
of glycolysis serve as biosynthetic precursors of
several metabolites including the nonessential amino
acids, serine, glycine, and alanine (Fig. 1) . The ratios
of these glycolytic amino acids to that of glucose, the primary
substrate of glycolysis, are hence useful as indirect
measures of the enzymatic activities of hexokinase, phosphofructokinase,
and pyruvate kinase. Our results indicate
that the activities of these principal rate-controlling enzymes
of glycolysis are significantly reduced in the ITG
and MFG, but not cerebellum, of individuals with AD.
Moreover, lower activities of these enzymes are associated
with greater severity of both neurofibrillary and amyloid
Next, using mass spectrometry–based proteomics in brain
tissue samples from the same participants, we showed that
protein levels of the neuronal glucose transporter, GLUT3,
are significantly reduced in AD and that lower levels of
GLUT3 protein reflect greater severity of both amyloid plaque
and neurofibrillary tangle pathology. Finally, we demonstrated
that longitudinal increases in fasting plasma glucose
levels, measured up to four decades before death, are
associated with higher brain tissue glucose concentrations
To the best of our knowledge, this report is the first to
measure brain tissue glucose concentrations and assess
glycolytic flux to demonstrate their relationships with
both severity of AD pathology and the expression of AD
symptoms. Including brain tissue samples from “ASYMAD”
individuals who represent an intermediate group
in the gradation of neuropathology from controls to AD patients
in the absence of cognitive impairment during life allowed
us to relate measures of brain glucose concentration
and glycolytic flux to incremental levels of AD pathology
and symptom expression. Equally importantly, by
measuring glucose concentrations in brain regions both
vulnerable to distinct pathological features of AD, that is,
MFG (amyloid deposition) and ITG (tau accumulation)
as well as in a region relatively resistant to AD pathology,
that is, cerebellum , we were able to determine whether
the observed alterations in brain glucose concentrations
and abnormalities in glycolysis were related to ADdefining
Our observation that in the ITG, brain glucose concentrations
are highest in AD, lowest in controls, and
intermediate in ASYMADs suggests that there may be
regionally specific threshold levels of tissue glucose
beyond which the accumulation of AD pathology triggers
the expression of clinical symptoms. Complementing
these observations on brain glucose concentrations
in the ITG, we also observed that activities of the three
rate-controlling enzymes of glycolysis, that is, hexokinase,
phosphofructokinase, and pyruvate kinase, are
lowest in AD, highest in controls, and intermediate in
ASYMADs, suggesting that regionally specific abnormalities
in glycolytic flux may lead to the build-up of
brain glucose levels, evolution of AD pathology, and ultimately
the development of AD symptoms. Furthermore,
the absence of significant group differences in
either brain tissue glucose concentration or glycolytic
flux in the cerebellum, a region that is relatively spared
of classical AD pathology, suggests that these abnormalities
are specifically linked to the defining pathological
hallmarks of AD. This interpretation is further strengthened
by the significant associations between concentrations
of brain tissue glucose and measures of glycolytic
flux with severity of AD pathology in both the ITG and
Our proteomic analysis of brain tissue from the same
groups of participants in the BLSA autopsy sample allowed
us to further explore potential mechanisms underlying
alterations in brain glucose concentration in AD.We found
that protein levels of the neuronal glucose transporter,
GLUT3, are significantly lower in AD and ASYMAD
brains relative to controls. Furthermore, lower protein
levels of GLUT3 are associated with greater severity of
both neuritic plaque and neurofibrillary tangle pathology.
Equally importantly, we find that both the group differences
in GLUT3 protein levels as well as their association with
AD pathology are independent of neuronal loss as they
remain significant even after adjustment for levels of the
neuronal nuclear protein nesprin-1 . This suggests
that lower GLUT3 protein levels in AD are likely to reflect
early changes in the evolution of AD pathophysiology,
rather than downstream consequences of neurodegeneration.
Together with the observation that protein levels of
the astrocytic glucose transporter GLUT1 are unchanged,
our results implicate a specific loss of the neuronal
glucose transporter GLUT3 as an integral feature of AD
Our findings also complement and extend previous
functional neuroimaging studies using 18FDG-PET
showing that decreases in neuronal glucose uptake may
be an early feature of AD pathogenesis [19–22, 40].
These studies show a characteristic pattern of
hypometabolism affecting primarily the parietotemporal,
posterior cingulate, and frontal cortices with relative
sparing of the cerebellum, thalamus, and basal ganglia
. This pattern of lower cerebral glucose uptake on
FDG-PET imaging therefore appears consistent with our
current observation of higher brain tissue glucose levels in
the frontal and temporal cortices in AD but sparing the
cerebellum. Previous studies have shown that the density
of the neuronal GLUT3 transporter in the brain is coupled
to local cerebral glucose utilization . Our current results
indicate that lower cerebral glucose utilization, as reflected
in lower rates of glycolysis and higher brain tissue
glucose concentrations, may thus downregulate GLUT3
protein levels, especially in brain regions vulnerable to
Previous studies of GLUT3 and GLUT1 protein levels
in the AD brain have relied on small sample sizes and
important methodological differences with our current
report merit consideration. Simpson and colleagues
measured GLUT3 and GLUT1 protein levels using
immunoblotting and autoradiography from a sample of
12 AD and 12 control brains . The mean age of the
control group (56 ± 22 years) in their report was considerably
lower than that of the AD group (76 ± 5 years).
Although they observed overall reductions in GLUT1
and GLUT3 protein levels averaged across several regions,
considerable interindividual variability made comparisons
of regional differences in glucose transporter
levels difficult. Using cytochalasin B binding assays,
Kalaria and colleagues demonstrated lower glucose
transporter levels in AD brains compared to controls
. However, as cytochalasin B is a nonspecific blocker
of both GLUT1 and GLUT3 transporters , a reliable
assessment of differences in neuronal versus astrocytic
glucose transporter levels was not feasible. Moreover,
in both these studies, the absence of “high-pathology
controls” or “ASYMADs” as in our current report precludes
assessment of changes in glucose transporter
levels in relation to severity of AD pathology and expression
of AD symptoms. A definitive assessment of
regional differences in GLUT3 and GLUT1 levels in
AD may require a larger sample size and proteomic
profiling across several brain regions.
It is important to note that our measure of brain tissue
glucose is based on mass spectrometric quantification of
total hexose levels transported from the circulation.
Glucose is the predominant hexose in blood and is present
in far greater concentrations (5450.0 ± 1200 µM) than
other hexoses such as fructose (46.26 ± 25.22 µM), galactose
(88.3 ± 34.7 µM), and mannose (64.0 ± 12.0 µM)
[45–47]. Therefore, the total concentration of brain
hexoses measured in our samples is likely to comprise
predominantly of glucose.
Given previous epidemiological evidence that insulin
resistance and diabetes are associated with increased
risk of AD [1–3], we also examined whether peripheral
glucose levels measured decades before death may
influence brain tissue glucose concentration. We have
previously shown that this relationship does not appear
to be mediated by a direct effect of blood glucose
concentrations in either brain amyloid deposition or
neurofibrillary pathology . In the current report, we
found that higher concentrations of plasma fasting
glucose measured before death as well as greater increases
in fasting plasma glucose over time were associated
with higher brain tissue glucose concentrations in
all three regions examined. In additional sensitivity analyses,
we further confirmed that the longitudinal associations
between fasting plasma glucose and brain tissue
glucose concentrations are not driven by individuals
with diabetes. These findings open up new lines of investigation
in appropriate animal models of AD where longitudinal
analyses of brain tissue and blood glucose
concentrations in relation to accumulating AD pathology
may provide important insights into the role of glucose
dysregulation in early stages of AD pathogenesis. Such
experimental studies are essential to test both causal relationships
confirm the temporal sequence of molecular
events related to glucose dysregulation and AD pathogenesis
in humans. It would also be important to test
whether abnormalities of brain glucose utilization are a
specific feature of AD pathogenesis or may be shared
by other neurodegenerative diseases. The lack of brain
and blood tissue samples from well-characterized cohorts
representing other non-AD neurological diseases
is a limitation of our present report.
Considering our results together, we propose that failure
of neuronal glucose utilization due to impaired glycolysis is
a fundamental feature of AD. At regionally specific
threshold levels of brain glucose and impaired glycolytic
flux, accumulating pathology in vulnerable brain regions
triggers the onset of AD symptoms. Our results have important
translational implications and set the stage for future
studies that may uncover therapeutic interventions targeting
brain glucose dysregulation in AD.
The authors are grateful to participants in the Baltimore
Longitudinal Study of Aging for their invaluable contribution.
This research was supported in part by the Intramural
Research Program of the NIH, National Institute on Aging.
Support was provided by the Accelerating Medicine Partnership
AD grant (grant number: U01AG046161-02), the
NINDS Emory Neuroscience Core (grant number:
P30NS055077), and the Emory Alzheimer’s Disease
Research Center (grant number: P50AG025688). The
Intramural Research Program of the National Institute on
Aging (NIH) had no role in study design; in the collection,
analysis, and interpretation of data; in the writing of
the report; and in the decision to submit the article for
RESEARCH IN CONTEXT
We reviewed (using PubMed) all
publications describing abnormalities of glucose
metabolism in Alzheimer’s disease (AD). It is unclear
whether glucose dysregulation in the brain is
related to severity of AD pathology and symptom
expression. Similarly, the relationships between longitudinal
changes in blood glucose concentration and
brain tissue glucose levels are unclear. We measured
brain tissue concentrations of glucose and the glycolytic
amino acids, serine, glycine, and alanine as well
as protein levels of the neuronal (GLUT3) and astrocytic
(GLUT1) glucose transporters in the autopsy
sample of the Baltimore Longitudinal Study of Aging.
Higher brain tissue glucose concentration,
reduced glycolytic flux, and lower levels of GLUT3
are related to severity of AD pathology and the
expression of AD symptoms. Increasing fasting
plasma glucose levels are associated with higher
brain tissue glucose concentrations.
Abnormalities in brain glucose homeostasis
are intrinsic to AD pathogenesis and begin
several years before clinical symptoms.
These results set the stage for
future studies testing brain glucose dysregulation as
a target of disease-modifying interventions in AD.
Leibson CL, Rocca WA, Hanson VA, Cha R, Kokmen E, O’Brien PC,
Risk of dementia among persons with diabetes mellitus: a population-based cohort study.
Am J Epidemiol 1997;145:301–8.
Luchsinger JA, Tang MX, Stern Y, Shea S, Mayeux R.
Diabetes mellitus and risk of Alzheimer’s disease and dementia with stroke in a multiethnic cohort.
Am J Epidemiol 2001;154:635–41.
Ott A, Stolk RP, van Harskamp F, Pols HA, Hofman A, Breteler MM.
Diabetes mellitus and the risk of dementia: The Rotterdam Study.
Chen Z, Zhong C.
Decoding Alzheimer’s disease from perturbed cerebral glucose metabolism: implications for diagnostic and therapeutic strategies.
Prog Neurobiol 2013;108:21–43.
Brain glucose and energy metabolism abnormalities in sporadic Alzheimer disease. Causes and consequences: an update.
Exp Gerontol 2000;35:1363–72.
Ahmed S, Mahmood Z, Zahid S.
Linking insulin with Alzheimer’s disease: emergence as type III diabetes.
Neurol Sci 2015;36:1763–9.
Chami B, Steel AJ, De La Monte SM, Sutherland GT.
The rise and fall of insulin signaling in Alzheimer’s disease.
Metab Brain Dis 2016; 31:497–515.
Freude S, Schilbach K, Schubert M.
The role of IGF-1 receptor and insulin receptor signaling for the pathogenesis of Alzheimer’s disease: from model organisms to human disease.
Curr Alzheimer Res 2009; 6:213–23.
Hokama M, Oka S, Leon J, Ninomiya T, Honda H, Sasaki K, et al.
Altered expression of diabetes-related genes in Alzheimer’s disease brains: the Hisayama study.
Cereb Cortex 2014;24:2476–88.
Talbot K, Wang HY, Kazi H, Han LY, Bakshi KP, Stucky A, et al.
Demonstrated brain insulin resistance in Alzheimer’s disease patients is associated with IGF-1 resistance, IRS-1 dysregulation, and cognitive decline.
J Clin Invest 2012;122:1316–38.
de la Monte SM.
Type 3 diabetes is sporadic Alzheimers disease: minireview.
Eur Neuropsychopharmacol 2014;24:1954–60.
de la Monte SM, Wands JR.
Alzheimer’s disease is type 3 diabetesevidence reviewed.
J Diabetes Sci Technol 2008;2:1101–13.
Claxton A, Baker LD, Hanson A, Trittschuh EH, Cholerton B,
Morgan A, et al.
Long-acting intranasal insulin detemir improves cognition for adults with mild cognitive impairment or earlystage Alzheimer’s disease dementia.
J Alzheimers Dis 2015; 44:897–906.
Claxton A, Baker LD, Wilkinson CW, Trittschuh EH, Chapman D,
Watson GS, et al.
Sex and ApoE genotype differences in treatment response to two doses of intranasal insulin in adults with mild cognitive impairment or Alzheimer’s disease.
J Alzheimers Dis 2013; 35:789–97.
Reger MA, Watson GS, Green PS, Wilkinson CW, Baker LD,
Cholerton B, et al.
Intranasal insulin improves cognition and modulates beta-amyloid in early AD.
Brain glucose transporters: implications for neurologic disease.
Simpson IA, Carruthers A, Vannucci SJ.
Supply and demand in cerebral energy metabolism: the role of nutrient transporters.
J Cereb Blood Flow Metab 2007;27:1766–91.
Joost HG, Thorens B.
The extended GLUT-family of sugar/polyol transport facilitators: nomenclature, sequence characteristics, and potential function of its novel members (review).
Mol Membr Biol 2001; 18:247–56.
Cerebral glucose metabolism in preclinical and prodromal Alzheimer’s disease.
Expert Rev Neurother 2010;10:1667–73.
Hunt A, Schonknecht P, Henze M, Seidl U, Haberkorn U, Schroder J.
Reduced cerebral glucose metabolism in patients at risk for Alzheimer’s disease.
Psychiatry Res 2007;155:147–54.
Jagust WJ, Haan MN, Eberling JL,Wolfe N, Reed BR.
Functional imaging predicts cognitive decline in Alzheimer’s disease.
J Neuroimaging 1996;6:156–60.
Brain glucose metabolism in the early and specific diagnosis of Alzheimer’s disease. FDG-PET studies in MCI and AD.
Eur J Nucl Med Mol Imaging 2005;32:486–510.
The Baltimore Longitudinal Study of Aging (BLSA): a 50- year-long journey and plans for the future.
J Gerontol A Biol Sci Med Sci 2008;63:1416–9.
Shock NW, Gruelich R, Andres R, Arenberg D, Costa PT, Lakatta E,
Normal human aging. The Baltimore Longitudinal Study of Aging.
Washington, DC: US Government Printing Office; 1984.
O’Brien RJ, Resnick SM, Zonderman AB, Ferrucci L, Crain BJ,
Pletnikova O, et al.
Neuropathologic studies of the Baltimore Longitudinal Study of Aging (BLSA).
J Alzheimers Dis 2009;18:665–75.
Gamaldo A, Moghekar A, Kilada S, Resnick SM, Zonderman AB,
Effect of a clinical stroke on the risk of dementia in a prospective cohort.
Diagnostic and Statistical Manual of Mental Disorders, DSM-III-R.
Washington, D.C.: American Psychiatric Association. Q6
McKhann G, Drachman D, Folstein M, Katzman R, Price D,
Clinical diagnosis of Alzheimer’s disease: report of the NINCDS-ADRDAWork Group under the auspices of Department of Health and Human Services Task Force on Alzheimer’s Disease.
Mirra SS, Heyman A, McKeel D, Sumi SM, Crain BJ, Brownlee LM,
he Consortium to Establish a Registry for Alzheimer’s Disease (CERAD). Part II. Standardization of the neuropathologic assessment of Alzheimer’s disease.
Braak H, Braak E.
Neuropathological stageing of Alzheimer-related changes.
Acta neuropathol 1991;82:239–59.
Troncoso JC, Zonderman AB, Resnick SM, Crain B, Pletnikova O,
Effect of infarcts on dementia in the Baltimore longitudinal study of aging.
Ann Neurol 2008;64:168–76.
Iacono D, Resnick SM, O’Brien R, Zonderman AB, An Y, Pletnikova O,
Mild cognitive impairment and asymptomatic Alzheimer disease subjects: equivalent beta-amyloid and tau loads with divergent cognitive outcomes.
J Neuropathol Exp Neurol 2014;73:295–304.
Metter EJ, Windham BG, Maggio M, Simonsick EM, Ling SM,
Egan JM, et al.
Glucose and insulin measurements fromthe oral glucose tolerance test and mortality prediction.
Diabetes care 2008;31:1026–30.
The cerebellum in Alzheimer’s disease.
Dement Geriatr Cogn Disord 1997;8:203–9.
Berg JM, Tymoczko JL, Stryer L, Stryer L.
Biochemistry. 5th ed.
New York: W.H. Freeman; 2002.
Vannucci SJ, Maher F, Simpson IA.
Glucose transporter proteins in brain: delivery of glucose to neurons and glia.
SeyfriedNT,DammerEB, SwarupV,NandakumarD,DuongDM,YinL, et al.
A Multi-network Approach Identifies Protein-Specific Co-expression in Asymptomatic and Symptomatic Alzheimer’s Disease.
Cell Syst 2017;4:60–72.e4.
Regression Modeling Strategies with Applications to Linear Models, Logistic and Ordinal Regression, and Survival Analysis.
Cham: Springer; 2015.
Dammer EB, Duong DM, Diner I, Gearing M, Feng Y, Lah JJ, et al.
Neuron enriched nuclear proteome isolated from human brain.
J Proteome Res 2013;12:3193–206.
Reiman EM, Chen K, Alexander GE, Caselli RJ, Bandy D, Osborne D,
Functional brain abnormalities in young adults at genetic risk for late-onset Alzheimer’s dementia.
Proc Natl Acad Sci U S A 2004; 101:284–9.
Duelli R, Kuschinsky W.
Brain glucose transporters: relationship to local energy demand.
News Physiol Sci 2001;16:71–6.
Simpson IA, Chundu KR, Davies-Hill T, Honer WG, Davies P.
Decreased concentrations of GLUT1 and GLUT3 glucose transporters in the brains of patients with Alzheimer’s disease.
Ann Neurol 1994; 35:546–51.
Kalaria RN, Harik SI.
Reduced glucose transporter at the blood-brain barrier and in cerebral cortex in Alzheimer disease.
J Neurochem 1989;53:1083–8.
Dick AP, Harik SI, Klip A, Walker DM.
Identification and characterization of the glucose transporter of the blood-brain barrier by cytochalasin
B binding and immunological reactivity.
Proc Natl Acad Sci U S A 1984;81:7233–7.
Wahjudi PN, Patterson ME, Lim S, Yee JK, Mao CS, Lee WN.
Measurement of glucose and fructose in clinical samples using gas chromatography/mass spectrometry.
Clin Biochem 2010;43:198–207.
Geigy scientific tables, Vol. 1.
Basle: Ciba-Geigy; 1981.
Tietz clinical guide to laboratory tests.
St. Louis: Saunders Elsevier; 2006.
Thambisetty M, Jeffrey Metter E, Yang A, Dolan H, Marano C,
Zonderman AB, et al.
Glucose intolerance, insulin resistance, and pathological features of Alzheimer disease in the Baltimore Longitudinal Study of Aging.
JAMA Neurol 2013;70:1167–72.
Return to the NUTRITION Section
Return to the OMEGA-3 FATTY ACIDS Page