Brain volume abnormalities and neurocognitive deficits in diabetes mellitus: Points of pathophysiological commonality with mood disorders?

Mood Disorders/Psychopharmacology Unit, University Health Network, University of Toronto, 399 Bathurst Street, Toronto, Ontario, Canada.
Advances in Therapy (Impact Factor: 2.27). 04/2010; 27(2):63-80. DOI: 10.1007/s12325-010-0011-z
Source: PubMed
ABSTRACT
It is hypothesized that diabetes mellitus (DM) and mood disorders share points of pathophysiological commonality in the central nervous system.
A PubMed search of all English-language articles published between 1966 and March 2009 was performed with the following search terms: depression, mood disorders, hippocampus, amygdala, central nervous system, brain, neuroimaging, volumetric, morphometric, and neurocognitive deficits, cross-referenced with DM. Articles selected for review were based on adequacy of sample size, the use of standardized experimental procedures, validated assessment measures, and overall manuscript quality. The primary author was principally responsible for adjudicating the merit of articles that were included.
Volumetric studies indicate that individuals with Type 1/2 DM exhibit regional abnormalities in both cortical and subcortical (e.g., hippocampus, amygdala) brain structures. The pattern of neurocognitive deficits documented in individuals with Type 1 DM overlap with Type 2 populations, with suggestions of discrete abnormalities unique to each phenotype. The pattern of volumetric and neurocognitive deficits in diabetic populations are highly similar to that reported in populations of individuals with major depressive disorder.
The prevailing models of disease pathophysiology in DM and major depressive disorder are distinct. Notwithstanding, the common abnormalities observed in disparate effector systems (e.g., insulin resistance, immunoinflammatory activation) as well as brain volume and neurocognitive performance provide the nexus for hypothesizing that both conditions are subserved by overlapping pathophysiology. This conception provides a novel framework for disease modeling and treatment development in mood disorder.

Full-text

Available from: Sidney H Kennedy
Roger S. McIntyre ()
Associate Professor of Psychiatry and Pharmacology,
University of Toronto, Head of Mood Disorders
Psychopharmacology Unit, University Health Network,
399 Bathurst Street, Toronto, Ontario, Canada, M5T 2S8.
Email: roger.mcintyre@uhn.on.ca
Ha T. Nguyen · Candy W. Y. Law · Farah Sultan · Hanna
O. Woldeyohannes · Mohammad T. Alsuwaidan ·
Joanna K. Soczynska · Amanda K. Adams · Jenny S. H.
Cheng · Maria Lourenco · Sidney H. Kennedy
Mood Disorders Psychopharmacology Unit, University
Health Network, Toronto, Ontario, Canada
Heather A. Kenna · Natalie L. Rasgon
Stanford University, Palo Alto, California, USA
Adv Ther (2010) 27(2):1-18.
DOI 10.1007/s12325-010-0011-z
REVIEW
Brain Volume Abnormalities and Neurocognitive Deficits
in Diabetes Mellitus: Points of Pathophysiological
Commonality with Mood Disorders?
Roger S. McIntyre · Heather A. Kenna · Ha T. Nguyen · Candy W. Y. Law · Farah Sultan ·
Hanna O. Woldeyohannes · Mohammad T. Alsuwaidan · Joanna K. Soczynska · Amanda K. Adams ·
Jenny S. H. Cheng · Maria Lourenco · Sidney H. Kennedy · Natalie L. Rasgon
Received: February 4, 2010 / Published online: April 8, 2010
© Springer Healthcare 2010
ABSTRACT
Background: It is hypothesized that diabetes
mellitus (DM) and mood disorders share points
of pathophysiological commonality in the
central nervous system. Methods: A PubMed
search of all English-language articles published
between 1966 and March 2009 was performed
with the following search terms: depression,
mood disorders, hippocampus, amygdala,
central nervous system, brain, neuroimaging,
volumetric, morphometric, and neurocognitive
deficits, cross-referenced with DM. Articles
selected for review were based on adequacy
of sample size, the use of standardized
0011‑z
2
1
experimental procedures, validated assessment
measures, and overall manuscript quality. The
primary author was principally responsible
for adjudicating the merit of articles that
were included. Results: Volumetric studies
indicate that individuals with Type 1/2 DM
exhibit regional abnormalities in both cortical
and subcortical (eg, hippocampus, amygdala)
brain structures. The pattern of neurocognitive
deficits documented in individuals with Type 1
DM overlap with Type 2 populations, with
suggestions of discrete abnormalities unique to
each phenotype. The pattern of volumetric and
neurocognitive deficits in diabetic populations
are highly similar to that reported in populations
of individuals with major depressive disorder.
Conclusion: The prevailing models of disease
pathophysiology in DM and major depressive
disorder are distinct. Notwithstanding, the
common abnormalities observed in disparate
effector systems (eg, insulin resistance,
immunoinflammatory activation) as well as
brain volume and neurocognitive performance
provide the nexus for hypothesizing that
both conditions are subserved by overlapping
pathophysiology. This conception provides
a novel framework for disease modeling and
treatment development in mood disorder.
UNAUTHORIZED USE PROHIBITED
UNAUTHORIZED USE PROHIBITED
Page 1
2 Adv Ther (2010) 27(2):1-18.
Keywords: amygdala; brain; diabetes mellitus;
hippocampus; mood disorders; morphometric;
MRI; neurocognitive deficits
INTRODUCTION
Diabetes mellitus (DM) is associated with
an increased risk of stroke, vascular dementia,
and mild cognitive impairment, as well as
Alzheimer’s disease.
1
The appellation diabetic
encephalopathy, although not rigorously
defined, is an erstwhile notion referring to
the consequences of abnormal glucose-insulin
homeostasis on brain structure and function.
2
During the past decade, intensified research
efforts have begun to parse out insulin’s salience
to both physiological and pathophysiological
brain function.
3,4
For example, brain insulin
receptors, as well as insulin-sensitive glucose
transporters, are regionally distributed
throughout the central nervous system (CNS)
with differential expression in brain regions
subserving affective and cognitive function
(eg, anterior cingulate cortex, prefrontal
cortex, hippocampus).
4
The consequences of DM on CNS structure
and function have both clinical and research
implications. For example, individuals with
DM are differentially affected by psychiatric
syndromes (eg, mood disorders) that pose a
hazard for the course and outcome of DM (and
vice versa).
5
Moreover, DM is an independent
risk factor for incident mood disorders and
Alzheimer’s disease, conditions characterized by
progressive neurocognitive decline.
6
Identifying
points of pathophysiological commonality
between DM and mood disorders may provide
an opportunity to refine models of disease
pathophysiology for both conditions.
3
In keeping with this view, postmortem studies
indicate that mood disorders are associated with
regional and layer-specific alterations in the
size, shape, and density of neurons and glia.
7,8
Volumetric imaging and neuropsychological
studies have provided correlative findings
indicating that constituents of the anterior
limbic circuit (eg, hippocampus) are abnormal
in structure and function, respectively, in
individuals with mood disorders.
9
Moreover,
emerging evidence also indicates that
individuals with DM exhibit similar volumetric
and neurocognitive deficits to persons with
mood disorders. Although the pathophysiology
of mood disorders and DM are distinct, there
appears to be several points of commonality in
the CNS.
3,10
The objective of this review is to summarize
the evidentiary base reporting on brain
volumetric abnormalities and neurocognitive
deficits in individuals with DM. The
encompassing aim of this endeavor is to reify
the conception that mood disorders and DM
may share pathophysiological substrates and/or
consequences in the CNS. This paper does not
review the neuroanatomical and neurocognitive
deficits in individuals with mood disorders, as
they are reviewed elsewhere.
9
METHODS
A PubMed search of all English-language
articles published between 1966 and March 2009
was performed with the following search terms:
depression, mood disorders, hippocampus,
amygdala, central nervous system, brain,
neuroimaging, volumetric, morphometric, and
neurocognitive decits, cross-referenced with
DM. Articles selected for review were based on
adequacy of sample size, the use of standardized
experimental procedures, validated assessment
measures, and overall manuscript quality. The
primary author was principally responsible
for adjudicating the merit of the articles that
were included.
UNAUTHORIZED USE PROHIBITED
UNAUTHORIZED USE PROHIBITED
Page 2
Adv Ther (2010) 27(2):1-18. 3
RESULTS
Several investigations have reported on
brain volume and neurocognitive deficits
in mixed populations with DM (Table 1
contains detailed information regarding each
of these studies).
11-27
Soininen et al. evaluated
neurocognitive performance and computed
tomography-measured abnormalities in
three discrete groups: non-diabetic (n=59;
mean age=74.0±6.4 years), diet-treated non-
insulin-dependent diabetics (n =13; mean
age=76.0±8.3 years) and medication-treated
non-insulin-dependent diabetics (n=12; mean
age=77.7.4 years). There were no significant
between-group differences in measures of
neurocognitive performance. Nevertheless,
medication-treated diabetic patients exhibited
more pronounced central temporal atrophy
as evidenced by a significantly wider right
temporal horn compared with that in the non-
diabetic group. Fasting blood glucose positively
correlated with the width of the right temporal
horn in the two diabetic groups.
11
Araki et al. evaluated and compared
magnetic resonance imaging (MRI)-measured
brain volume amongst individuals (n=159;
mean age=60.4 years) with DM (disease
duration=3-30 years; mean duration=13.5 years)
to age-matched individuals without DM (n=2566).
Most individuals in the diabetic group were
non-insulin-dependent (n=144). A significantly
higher frequency of cerebral atrophy was
observed in the diabetic group when compared
with the control group. Cerebral atrophy
increased as a function of age in both groups,
with more pronounced abnormalities noted in
the diabetic group (eg, 41.2% vs. 19.8%, 60% vs.
38.9%, 92.3% vs. 56.8% in the sixth, seventh,
and eighth decade of life, respectively).
12
Convit et al. reported that non-diabetic, non-
demented subjects (n=27; mean age=69 years)
with abnormal glucose tolerance exhibited
smaller hippocampal volumes, which were
associated with impairment in memory
(ie, immediate and delayed) performance.
Delayed paragraph recall was also significantly
correlated with hippocampal volume. No
further brain volumetric abnormalities were
noted in other brain regions of interest
(eg, parahippocampal gyrus, the superior
temporal gyrus).
13
Perros et al. aimed to determine the effect
of insulin-dependent DM (IDDM) on MRI-
measured brain volumes. Neurometabolic
parameters were also evaluated with magnetic
resonance spectroscopy (MRS) an association
with neurocognitive function was evaluated.
Eleven patients with IDDM and no history of
severe hypoglycemia were compared with eleven
IDDM patients with a history of five or more
episodes of severe hypoglycemia. Of the twenty-
two IDDM patients, leukoaraiosis, manifesting
as white matter hyperintensities (WMH) with
T2-weighted MRI, was present in four patients
(18.2%) while cortical atrophy was noted in five
patients (22.7%).
14
There were no significant
differences between groups in the prevalence
of leukoaraiosis, although numerically more
individuals with recurrent hypoglycemia
exhibited cortical atrophy. Individuals with
recurrent hypoglycemia had a lower IQ score;
no other significant between-group differences
were reported on any neurocognitive measure.
Moreover, there was no association between
the presence of MRI-measured cortical atrophy
and cognitive function, although trends for
diminished psychomotor speed were noted in
patients with cortical atrophy.
14
Den Heijer et al. examined the association
between Type 2 DM, insulin resistance, and
hippocampal and amygdala atrophy as part of
the Rotterdam Study. The Rotterdam study was a
large population-based cohort study conducted
UNAUTHORIZED USE PROHIBITED
UNAUTHORIZED USE PROHIBITED
Page 3
4 Adv Ther (2010) 27(2):1-18.
Table 1. Brain volume and neurocognitive abnormalities in diabetic populations.
3
Author n Population
Method (including
neuroimaging and
neurocognitive testing)
MRI
parameters
(Tesla)
Method
(psychometric
measures) Aim
Soininen H et al.
(1992)
11
84 Non-DM (n=59).
NIDDM treated with diet.
(n=13).
NIDDM treated with drugs
(n=12). 68-84 years.
CT measures for overall brain
analysis ECG, chest X-ray,
electroencephalogram.
Not
indicated
WA I S
Compilation of
neuropsychological test
presented as one SD score
To evaluate cognitive performance in DM
and non-DM patients using CT.
Results: No dierence between groups in cognitive performance. Drug-treated DM exhibited greater central temporal atrophy and wider ontal horns (all women). CT measures
were comparable.
Araki Y et al.
(1994)
12
2725 DM (n=159; NIDDM, 144;
IDDM, 15).
Healthy controls (n=2566).
MRI-assessed frequency of
cerebral infarcts hemorrhages,
atrophy, and subcortical
arteriosclerotic encephalopathy.
1.5 None To assess the central eects of DM
with MRI.
Results: Cerebral atrophy was signicantly more equent in DM group than controls om the 6th to 8th decade of life. No signicant dierences in occurrences of cerebroascular
diseases at any age.
Convit A et al.
(1997)
13
76 Normal elderly (n=27).
Minimal cognitive impairment
non-DAT (n=22), DAT
(n=27).
MRI-derived volumes assessing
the temporal lobe.
1.5 None To evaluate the involvement of the
temporal lobe in the preclinical stages
of DAT.
Results: Hippocampal olumes were reduced for the MCI and DAT groups compared with normal elderly.
Perros P et al.
(1997)
14
22 IDDM with no history of
hypoglycemia (n=11). IDDM
with history of hypoglycemia
(n=11).
MRI and MRS evaluated
overall brain structure.
Not
indicated
WA I S
NART
AV LT
IT
CRT
PA S AT
RVIP
To assess IDDM for brain lesions with
and without a history of hypoglycemia
and the relationship of any cognitive
impairments.
Results: Abnormalities were observed in the periventricular WM and cortical atrophy was found in IDDM with history of hypoglycemia. MRS scans showed no dierences.
No signicant relations were found in psychometric measures.
(continued on next page)
UNAUTHORIZED USE PROHIBITED
UNAUTHORIZED USE PROHIBITED
Page 4
Adv Ther (2010) 27(2):1-18. 5
Table 1. Brain volume and neurocognitive abnormalities in diabetic populations.
3
(Continued)
Author n Population
Method (including
neuroimaging and
neurocognitive testing)
MRI
parameters
(Tesla)
Method
(psychometric
measures) Aim
den Heijer T et al.
(2003)
15
506 Non-DM (n=465).
Type 2 DM controls (n=41).
Assessed degree of
hippocampal and amygdala
atrophy using MRI. Marked
for WML and infarcts present.
1.5 15 word learning test
Overall z-score for cognitive
function
To investigate the association between DM, IDDM,
and the degree of hippocampal and amygdala
atrophy. To investigate whether DM increases the
development of DAT through neuropathy.
Results: Signicant negative relation in men but not women between BMI and GMV. VBM showed that GMV in bilateral medial temporal lobes, anterior lobes of cerebellum, occipital lobe, ontal lobe,
and mid brain has negative relation to BMI in men. GMV in bilateral inferior ontal gyri, posterior lobe of cerebellum, ontal lobes temporal lobes thalami, and caudate shows positive relation to BMI in
men.
Ferguson SC et al.
(2003)
16
74 Type 1 DM youth (n=74)
with sucient exposure severe
hypoglycemia.
MRI assessed for TBV, CSF,
and RBV. VBM assessed
temporal lobe and amygdala-
hippocampal areas.
1.0 HADS
WA I S - R
NART
IT
PASAT
To investigate cognitive performance and brain
structure in individuals with Type 1 DM with
exposure to severe hypoglycemia.
Results: Severe hypoglycemia did not inuence cognitive ability or brain structure. Background retinopathy was associated with a signicant cognitive disadvantage in uid intelligence, information
processing, attention, and concentration abilities.
Ferguson SC et al.
(2005)
17
71 Early-onset Type 1 DM before
7 years (n=26). Later-onset Type 1
DM between 7-17 years (n=45).
MRI assessed overall brain
structure.
1.0 HADS
WA I S - R
NART
IT
BVFT
PASAT
To evaluate the eects of early-onset Type 1 DM
in youth, on cognitive performance and brain
structure.
Results: Intellectual ability and information processing ability was inferior in the early-onset DM group; LV olumes were 37% greater and ventricular atrophy was more prevalent in the early-onset
group that those with later-onset.
Lobnig BM et al.
(2006)
18
26 Type 1 DM with 10 years duration
(n=13). Non-DM (n=13).
30-50 years of age.
MRI scans to determine
hippocampal volume and CSF.
1.0 Paired association
SCWT
DS
TMT
15-item CESDS
To examine hippocampal volume and memory
performance in individuals with Type 1 DM.
Results: Hippocampal olumes and memory performance did not dier between subjects and controls. However, signicant increase in CSF olume suggests mild cerebral atrophy. Also found impaired
psychomotor speed and selective attention.
(continued on next page)
UNAUTHORIZED USE PROHIBITED
UNAUTHORIZED USE PROHIBITED
Page 5
6 Adv Ther (2010) 27(2):1-18.
Table 1. Brain volume and neurocognitive abnormalities in diabetic populations.
3
(Continued)
Author n Population
Method (including
neuroimaging and
neurocognitive testing)
MRI
parameters
(Tesla)
Method
(psychometric
measures) Aim
Musen G et al.
(2006)
19
118 Type 1 DM (n=82).
Healthy controls (n=36).
25-40 years, 5-10 years disease
duration.
MRI screened for brain
structural abnormalities.
VBM assessed GMV,
WMV, and CSF.
1.5 WAI S
D-KEFS
WMS
DS
Grooved peg board
To examine DM and metabolic disturbances
with changes in GMV and CSF.
Results: DM showed lower GMD in regions for language processing and memory.
Wessels AM et al.
(2006)
20
52 Type 1 DM without
microvascular complications
(n=18). Type 1 DM with a
microvascular complication
(n=13). Healthy controls
(n=21).
VBM comparing GMD
between groups.
1.5 None To investigate whether long-term
hyperglycemia, resulting in advanced
retinopathy, contributes to structural changes
in GMD.
Results: Patients with diabetic retinopathy exhibited smaller GMD in the right inferior ontal gyrus and right occipital lobe compared with those without diabetic retinopathyand
healthy controls.
Wessels AM et al.
(2007)
21
34 Type 1 DM (n=25).
Healthy controls (n=9).
Comparing fractional brain
tissue volumes with VBM.
1.5 DS forward and backward
15 word test
ROCF test
Delayed recall condition
WAIS-Symbol
Substitution
Learning test
TMT (A&B)
SCWT (I, II & III)
GIT sorting
WCST
WISC-Mazes
CWF task
WAIS-block design
To assess cognitive performance in Type 1
DM patients who may be compromised due to
chronic hyperglycemia, associated with GMV
and WMV.
Results: Type 1 DM patients exhibited inferior performance on measures of speed of information processing and visuoconstruction. Patients with microascular complication had a
signicantly smaller WMV than non-diabetic controls, also associated with lower performance on the domains of speed of information processing, attention, and executive functioning.
(continued on next page)
UNAUTHORIZED USE PROHIBITED
UNAUTHORIZED USE PROHIBITED
Page 6
Adv Ther (2010) 27(2):1-18. 7
Table 1. Brain volume and neurocognitive abnormalities in diabetic populations.
3
(Continued)
Author n Population
Method (including
neuroimaging and
neurocognitive testing)
MRI
parameters
(Tesla)
Method
(psychometric measures) Aim
Perantie DC et al.
(2007)
22
159 Type 1 DM youth (n=108).
Healthy control siblings
(n=51). Age 7-17 years.
MRI. VBM determined
relationship between
prior hypo-hyperglycemia
to regions of GMV and
WMV.
1.5 None To quantify RBV dierences. Structural
TBV in Type 1 DM youth not previously
studied.
Results: No signicant dierence reported between DM and healthy controls. Severe hypoglycemia was associated with smaller GMV in the le superior temporal region. Exposure
to hyperglycemia was associated with smaller GMV in the right cuneus and precuneus, smaller WMV in the right posterior parietal region, and larger GMV in the right preontal
region.
Jongen C et al.
(2007)
23
145 Type 2 DM (n=99; 56-
80 years). Healthy controls
(n=46; 55-78 years).
Automated segmentation
technique associated with
Type 2 DM, related DM
risk factors, and cognition
with WML volumes
MRI assessed WMV,
GMV, LV, CSF, and WML
1.5 An overall z-score was
acquired for cognition
composite that included 11
dierent tests addressing
cognitive domains of visuo-
construction, attention
and executive function,
information processing
speed, memory, and abstract
reasoning
Type 2 DM is known to be associated
with brain atrophy and cognitive decline;
association of WML is unclear.
Results: Signicantly smaller GMV and signicantly larger lateral ventricle olumes than controls. History of macroascular disease was associated with larger total CSF. DM
patients with lower composite cognitive performance showed smaller TBV.
Kumar R et al.
(2008)
24
478 DM (n=39), no DM (n=428).
60-64 years.
MRI 1.5 MMSE
Spot-the Word Test Version A
SDMT
Immediate and delayed recall
Purdue Pegboard Test (both
hands)
Reaction time (simple and
choice)
Goldberg Scale (for
depression)
To examine the neuroanatomical and
neurocognitive dierences in diabetic
participants (60-64 years)
with depression.
Results: No dierence in WMV and GMV. DM subjects have greater brain atrophy and CSF olume.
(continued on next page)
UNAUTHORIZED USE PROHIBITED
UNAUTHORIZED USE PROHIBITED
Page 7
8 Adv Ther (2010) 27(2):1-18.
Table 1. Brain volume and neurocognitive abnormalities in diabetic populations.
3
(Continued)
Author n Population
Method (including
neuroimaging and
neurocognitive testing)
MRI
parameters
(Tesla)
Method
(psychometric
measures) Aim
Kodl CT et al.
(2008)
25
50 Type 1 DM for 15 years
(n=25), age-and sex-matched
non-diabetic (n=25).
DTI for assessment of
WM microstructure.
3.0 WA S I
PA SAT
DVT
Trails A and B
ROCF
Grooved Pegboard Test
CPT-II
To assess the validity of using DTI for
identifying dierences in the brain of
patients with chronic Type 1 DM and its
possible association with decits identied
by neurocognitive tests.
Results: e posterior corona radiate and optic radiation of subjects with diabetes showed a decreased mean actional anisotropy than non-diabetic controls.
Haroon E et al.
(2009)
26
57 Subjects with DM & MDD
(n=18). Subjects with DM
but no MDD (n=20).
Controls, not depressed or
diabetic (n=19).
MRS to measure levels
of mI
1.5 ROCF (ROCF-Recall
& ROCF-Recognition)
To determine whether visuospatial
decits were attributable to elevations in
dorsolateral mI in patients with DM &
MDD.
Results: No association reported between dorsolateral mI levels and visuospatial decits in patients with DM and MDD.
Northam EA et al.
(2009)
27
181 Type 1 DM (n=106), control
subjects (n=75).
MRS, MRI: volumetry WASI: FSIQ, VIQ
and PIQ
To examine brain functioning in youths
12 years aer diagnosis Type 1 DM.
Results: Type 1 DM showed decreased GMV in bilateral thalami and right parahippocampal gyrus and insular cortex. Type 1 DM showed decreased WMV in bilateral
parahippocampi, le temporal lobe, and middle ontal area.
AVLT=Auditory Verbal Learning Test; BMI=body mass index; BVFT=behavioral variant frontotemporal dementia; CESDS=Centre for
Epidemiologic Studies Depression Scale; CPT=Connor’s Continuous Performance Test; CRT=Choice Reaction Time test; CSF=cerebrospinal fluid;
CT=computed tomography; CVLT=California Verbal Learning Test; CWF=Category Word Fluency task; D-KEFS=Delis-Kaplan Executive Function
System; DAT=dementia of Alzheimer’s type; DM=diabetes mellitus; DS=digit span; DTI=diffusion tensor imaging; DVT=Digit Vigilance Test;
ECG=electrocardiograph; FSIQ=full scale IQ; GIT=general information test; GMD=gray matter density; GMV=gray matter volume; HADS=Hospital
Anxiety and Depression Scale; LV=left ventricular; MCI=minimal cognitive impairment; MDD=major depressive disorder; mI=myo-inositol;
MMSE=mini-mental state evaluation; MRI=magnetic resonance imaging; MRS=magnetic resonance spectroscopy; NART=National Adult Reading
Test; N/IDDM=non/insulin-dependent diabetes mellitus; PASAT=Paced Auditory Serial Addition Test; PIQ=performance IQ; RBV=relative blood
volume; ROCF=Rey-Osterrieth Complex Figure; RVIP=Rapid Visual Information Processing; SCWT=Stroop Color-Word Test; SDMT=Symbol
Digit Modalities Test; TBV=total brain volume; TMT=Trail Making Test; VBM=voxel-based morphometry; VIQ=verbal IQ; WAIS=Wechsler Adult
Intelligence Scale; WASI=Wechsler Abbreviated Scale of General Intelligence; WCST=Wisconsin Card Sorting Test; WISC=Wechsler Intelligence Scale
for Children; WML=white matter lesions; WMS=Wechsler Memory Scale; WMV=white matter volume.
UNAUTHORIZED USE PROHIBITED
UNAUTHORIZED USE PROHIBITED
Page 8
Adv Ther (2010) 27(2):1-18. 9
in the Netherlands that aimed to investigate
the prevalence, incidence, and determinants
of chronic disease in the elderly. Baseline
examinations were completed between 1990 and
1993. In 1996, 506 living members (60-90 years
of age) were investigated with MRI to evaluate
age-related brain abnormalities. Type 2 DM
was operationalized as the reported use of oral
antidiabetic treatment, and a plasma glucose
level following a 2-hour glucose tolerance greater
than or equal to 11.1 mmol/L. Insulin resistance
in non-diabetic subjects was assessed by the ratio
of postload insulin levels divided by peripheral
glucose concentration.
15
None of the participants were known to have
a dementing disorder; nevertheless, individuals
with Type 2 DM (n=41; 8.1%) exhibited decreased
performance in memory testing. Individuals
with DM had more atherosclerotic plaques in the
carotid arteries and were 1.7 times more likely to
exhibit cerebral infarctions compared with those
without Type 2 DM. Individuals with Type 2 DM
had smaller bilateral hippocampal and amygdala
volumes after adjusting for body mass index,
pack-years of cigarette smoking, blood pressure,
and cholesterol levels. Exclusion of participants
with infarcts did not change the results, nor did
stratification as a function of APOE (a genetic
vulnerability factor for Alzheimer’s disease)
status. Individuals with high postload insulin
concentrations or insulin resistance also exhibited
smaller amygdala volume, but no difference in
hippocampal volume. Volumetric changes noted
in the insulin-resistant group remained after
multivariate analysis. The association between
insulin resistance and amygdala volume was
statistically significant only in non-carriers of
the APOE ε4 allele.
15
It has been documented that tight
glycemic control reduces the risk of diabetic
microangiopathy and increases the risk
for hypoglycemia. Protracted periods of
hypoglycemia predominantly affect neuronal
function in the frontal lobes and subcortical
grey matter.
28
Repeat exposure to severe
hypoglycemia has been associated with cortical
atrophy.
14
Susceptibility to hypoglycemia-
related cerebral atrophy may be higher in older
populations (ie, over 45 years).
14,29
Available
evidence suggests that both hyperglycemia
and hypoglycemia exert toxic effects on brain
structure and function.
Ferguson et al. cross-sectionally evaluated
individuals (minimum 10 years illness duration)
with Type 1 DM (n=74; age at illness onset <18).
The aim of their investigation was to ascertain
the effect of recurrent severe hypoglycemia and
microvascular disease (ie, detected by digital
retinal imaging defined by the presence of two
or more microaneurysms in one eye-Airlie House
Gradings 1a-1c) on MRI-measured brain volume
and cognitive performance.
16
The Diabetes
Control and Complications Trial defined
hypoglycemia as an episode requiring external
assistance for recovery.
16
Taken together, a history of severe
hypoglycemia did not correlate with measures
of neuropsychological performance. Diabetic
individuals, however, exhibited deficits across
most of the neurocognitive domains examined.
Background retinopathy was associated with
inferior intellectual performance, notably in
spatial ability and mental flexibility/psychomotor
speed, respectively. Information processing
ability was also inferior in those with background
retinopathy, as was sustained attention and
concentration. Measures of verbal fluency
did not differ between groups. No signicant
correlation was identified between any measure
of previous exposure to severe hypoglycemia
and MRI-measured volumetric abnormalities.
Moreover, there was no association between
background retinopathy, cerebral atrophy, or
brain volumetric measurements. Individuals
UNAUTHORIZED USE PROHIBITED
UNAUTHORIZED USE PROHIBITED
Page 9
10 Adv Ther (2010) 27(2):1-18.
with background retinopathy were more likely
to exhibit mild basal ganglia periventricular
small WMH.
16
A subsequent study by Ferguson et al.
aimed to determine if cognitive performance
was adversely affected in young adults (n=71)
who had developed DM before the age of 7
(n=26) compared with a “late-onset” diabetes
group (ie, onset between ages 7-17; n=45).
Prior cognitive ability, retinopathy status, and
diabetes onset were identified as independent
predictors of cognitive ability with multivariate
analysis.
17
Results indicated that the “early-
onset” diabetic group exhibited significantly
lower performance on non-verbal IQ as well
as information processing. Lateral ventricular
volume was estimated to be 37% greater in the
“early-onset” subjects. No correlation was noted
between MRI-measured abnormalities (ie, brain
atrophy, WMH) and cognitive or information
processing performance. Larger brain volume
was associated with superior cognitive
performance in sustained attention and
concentration ability, information processing
speed, and performance IQ. Hippocampal
WMHs were observed more frequently in the
“early-onset” group; mesial temporal lobes
sclerosis has been reported in other studies to
be more frequent in individuals who develop
DM in early childhood.
17
Lobnig et al. sought to evaluate associations
between MRI-measured hippocampal volume
and cognitive performance in Type 1 DM (n=13;
age=30-50 years) with a minimum illness duration
of 10 years.
18
Ten individuals also had comorbid
hypertension managed with medication and
lifestyle modification. Ten patients had stable
diabetic retinopathy while five patients had
mild peripheral polyneuropathy. The mean
glycated hemoglobin (HbA
1c
) concentration
was 8.21%; most individuals had experienced
at least one hypoglycemic episode (ie, blood
glucose levels below 3 mmol/L with or without
symptoms or below 3.5 mmol/L with symptoms
of hypoglycemia).
18
Results indicated that diabetic individuals
trended towards slower performance in the Trail-
Making Test and exhibited significantly more
interference in the Stroop test. After controlling
for differences in intracranial vault size, Type 1
DM subjects had a significantly larger amount
of global cerebral spinal fluid (CSF) and smaller
cerebral volume when compared with a gender-
and age-matched control subjects. Hippocampal
volumes, however, did not differ between
patients and control subjects.
18
Musen et al. evaluated the effect of Type 1
DM on grey matter density by comparing
diabetic individuals (n=82; mean age=32 years)
to an age-matched healthy control group. Grey
matter volumes (GMV) were measured with
voxel-based morphometry (VBM) analysis of
MRI data. Decrements in GMV were noted
in the left and right superior temporal gyri
(STG), left angular gyrus, left middle temporal
and middle frontal gyri, and left thalamus in
subjects with Type 1 DM as compared to the
control group. Furthermore, the presence of
Type 1 DM remained a significant predictor of
grey matter STG density loss after controlling for
diabetes status, age, sex, handedness, education,
depression, drug use, and alcohol use.
19
Wessels et al. evaluated whether long-
term hyperglycemia was associated with brain
structural changes in Type 1 DM patients
with proliferative retinopathy (n=13) compared
with individuals with Type 1 DM without
retinopathy (n =18). Both diabetic groups
were compared with healthy controls (n=21).
Reduced grey matter density was noted in the
diabetic retinopathy group compared with
the non-diabetic retinopathy group in the
left middle frontal gyrus, right inferior frontal
gyrus, right occipital lobe, and left cerebellum.
UNAUTHORIZED USE PROHIBITED
UNAUTHORIZED USE PROHIBITED
Page 10
Adv Ther (2010) 27(2):1-18. 11
Differences between healthy controls and both
patient groups were not statistically significant,
although a trend towards smaller grey matter
density was noted in the diabetic retinopathy
group. No significant differences between brain
atrophy and HbA
1c
, onset and duration of DM,
or blood pressure in the Type 1 DM group
was noted.
20
Wessels et al. separately evaluated
neurocognitive performance in an overlapping
sample of individuals with Type 1 DM
to determine the relationship between
microvascular complications or fractional GMV
and white matter volumes (WMV). Twenty-five
patients with Type 1 DM as well as nine non-
diabetic controls were included.
21
Individuals
with DM exhibited a lower performance
on information-processing speed and the
neurocognitive domain of visuoconstruction.
There was a significant difference in fractional
WMV between groups. Individuals with
diabetic retinopathy displayed a significant
reduction in fractional WMV compared with
the non-diabetic controls. No differences
in fractional WMV were noted between the
non-diabetic retinopathy group and the non-
diabetic control group; as well, no differences
were noted in any group on fractional GMV.
WMV positively correlated with performance
on speed of information processing, attention,
and executive functioning. No correlation
was noted between GMV and any of the
neuropsychological measures.
21
Perantie et al. evaluated regional brain volume
differences in youths with Type 1 DM (n=108;
age=7-17 years) compared to an age-matched
healthy control group (n=51). Using VBM, no
significant differences between groups were noted
in GMV or WMV. However, Type 1 DM patients
with a lifetime history of hypoglycemic episodes
exhibited less GMV than the non-hypoglycemic
group in the left superior temporal/occipital
cortex and left inferior occipital cortex. Exposure
to more frequent hyperglycemic episodes
correlated with less GMV in the right cuneus and
precuneus. Measures of hyperglycemic exposure
also correlated with increased GMV in the right
frontal middle gyrus and with smaller WMV in
the right superior parietal matter.
22
Jongen et al. aimed to quantitatively
determine the effects of Type 2 DM on cerebral
volume and WMHs. Subjects were recruited
from the Utrecht Diabetic and Encephalopathy
study, a cross-sectional population-based
study evaluating determinants of impaired
cognition in Type 2 DM. Eligible subjects
(n=99; age=56-80 years) were required to have
a diabetes illness duration of at least 1 year.
23
Diabetic patients exhibited smaller GMV and
signicantly larger lateral ventricular volume
when compared with healthy controls. A
significantly smaller GMV and total brain
volume as well as significantly larger lateral
ventricle, CSF, and total CSF volume were
noted in female, but not in male, Type 2 DM
subjects. WMV were unaffected; nevertheless,
white matter lesion (WML) volume was
significantly larger in Type 2 DM patients.
Amongst the diabetic patients, lower composite
cognitive performance was associated with
signicantly smaller total brain volume, larger
WML volume, and non-significantly larger
CSF volume.
Kumar et al. examined the relationship
between neurocognitive function, depression,
neuroanatomical variables, and the relationship
with Type 2 DM (n=478; mean age=60-64) in
randomly selected community residents.
24
They
reported that individuals with DM have larger
CSF volumes and more total brain atrophy
than controls. They did not, however, find
a relationship between diabetes status and
hippocampal volume, nor did they find an
association between DM and WMH. Individuals
UNAUTHORIZED USE PROHIBITED
UNAUTHORIZED USE PROHIBITED
Page 11
12 Adv Ther (2010) 27(2):1-18.
with DM exhibited poor performance on
measures of fine motor dexterity.
Kodl et al. aimed to determine if fractional
anisotropy correlated with neurocognitive test
performance in patients with Type 1 DM (n=25)
when compared with controls.
25
They reported
that white matter integrity, as measured by
diffusion tensor imaging, was lower in several
white matter tracts including the posterior
corona radiata and optic radiation in patients
with longstanding Type 1 DM. A correlation
between fractional anisotropy in white matter
tracts and reduced neurocognitive performance
believed to assess white matter function, notably
in visuoconstruction and motor dexterity,
was reported.
Haroon et al. evaluated associations between
visuospatial functioning and MRS-measured
myo-inositol (mI) concentrations in three
matched populations: depressed diabetic patients
(n=18), non-depressed diabetic patients (n=20),
and normal controls (n=19).
26
Higher absolute
and normalized concentrations of mI in right
prefrontal white matter areas were associated
with decreased performance on visuospatial
construction recall and recognition scores in the
healthy control group. Both DM groups did not
exhibit similar associations. A sloping pattern
in the relationship between mI and visuospatial
recognition performance was suggested; the
relationship was strongest in the healthy control
group, intermediary in the non-depressed group,
and weakest in the depressed group.
Northam et al. evaluated patients with
Type 1 DM (n=106) and healthy controls (n=75)
with respect to neurocognitive function and
neuroimaging changes 12 years after illness
onset.
27
Individuals with Type 1 DM had lower
verbal and full scale IQ scores when compared
with healthy controls. The patient group also
exhibited lower N-acetylaspartate levels in
frontal lobes and basal ganglia and higher mI
and choline levels in frontal and temporal
lobes and basal ganglia when compared to
controls. Patients with Type 1 DM also exhibited
decreased GMV in bilateral thalamus and right
parahippocampal gyrus and insular cortex. WMV
was decreased in bilateral parahippocampus,
left temporal lobe, and middle frontal area.
Hypoglycemia was associated with lower verbal
IQ scores and volume reduction in the thalamus;
poor metabolic control predicted elevated mI and
decreased T2 in the thalamus. The altered levels
of mI suggest advanced gliosis and demyelination
processes in the Type 1 DM group.
The foregoing studies are heterogeneous
in methodology, sample composition, aims,
hypothesis, neuroimaging techniques,
neuropsychological measures, and outcomes.
The challenge is to extract coherent, substantive,
and replicated themes. Nevertheless, results
from these studies indicate that both cortical
and subcortical structures are adversely affected
in individuals with DM. The functional correlate
of these volumetric changes is suggested by
neuropsychological measures wherein diverse
deficits in performance are reported.
23
Putative Mediators of Neurotoxicity in
Diabetic Populations
There are several non-mutually exclusive
mechanisms that putatively mediate the
volumetric changes observed in disparate diabetic
populations (Figure 1). Hyperglycemia is associated
with accelerated formation of advanced glycation
end products that may cross-link amyloid and
tau protein, thereby facilitating extracellular
plaque and intracellular neurofibrillary tangle
formation.
30
Conversely, repeated hypoglycemic
events are associated with cerebral atrophy,
WMLs, and persistent cognitive impairment.
16
The association between peripheral
hyperinsulinemia and brain volumetric changes
UNAUTHORIZED USE PROHIBITED
UNAUTHORIZED USE PROHIBITED
Page 12
Adv Ther (2010) 27(2):1-18. 13
in euglycemic patients suggests that insulin
homeostasis directly affects brain volume.
15
After binding to its cell surface receptor, insulin
activates two parallel signal transduction
processes: prosurvival and proapoptotic
pathways. The prosurvival pathway begins with
activation of the insulin receptor and subsequent
tyrosine phosphorylation of the insulin receptor
substrate (IRS) proteins. Activation of IRS proteins
portends the recruitment and activation of
phosphoinositide 3-kinase (PI3K) with resultant
protein kinase B or “Akt” (PKB/Akt) activation.
Activated PKB/Akt inhibits proapoptotic
proteins (ie, Bcl-2-associated death promoter
protein, glycogen synthase kinase-3 [GSK3]
and Forkhead box O family of transcriptional
activators). The inactivation of GSK3 inhibits tau
phosphorylation, a neuropathological hallmark
of Alzheimer’s disease.
Activation of the insulin receptor stimulates
proapoptotic pathways via activation of the
Src homology 2 domain (SHC) protein, growth
factor binding protein 2 (Grb2) and son of
sevenless, a Ras (a signal transduction cascade
protein) guanine exchange factor. The latter
step activates extracellular signal-regulated
kinase-1 and -2 (ERK1/2) which are mediators of
excitotoxic cell death.
31
Taken together, the net effect on neuronal
integrity is contingent on the predominant
intracellular cascade activated (ie, PI3K vs. SHC)
as well as cross-talk between the two parallel
pathways. Chronic central hyperinsulinemia,
however, may result in desensitized antiapoptotic
pathways and resultant amyloid beta (Aβ)
accumulation (ie, decreased insulin degrading
enzyme-mediated Aβ accumulation) indicating
that disturbances in insulin homeostasis rather
than relative excess or deficiency endangers
cellular integrity.
A separate hypothesis is that DM may be
associated with abnormal central insulin-like
growth factor activity. In addition to the well-
established effects of insulin growth factor 1
(IGF-1) on somatic growth and metabolic
processes, IGF-1 is also a critical mediator of
brain growth and development, hippocampal
neurogenesis, neuroprotection, and myelination.
32
Compelling evidence indicates that IGF-1, similar
to insulin, exerts a direct effect on the metabolism
and clearance of Aβ.
33
Moreover, processes which
decrease IGF-1 translocation to the CNS are
associated with an accumulation of neurotoxic
Aβ, as well as tau protein.
33
Vasculopathic changes and altered lipid
metabolism may mediate the central effects of DM
on brain volume and function. Individuals with
DM often exhibit cerebral vascular abnormalities
including infarcts and WMLs. Nevertheless,
adjusting for markers of vasculopathy does not
appear to alter the association between DM and
hippocampal/amygdala volumes.
Chronic activation of the immuno-
inflammatory network is associated with a
diminished neurocognitive performance and
abnormal brain activation patterns.
34,35
Figure 1. Mediators of brain volumetric decits in diabetic
populations.
Brain volume decits
NEUROCOGNITIVE DEFICITS
ALTERED INSULIN SENSITIVITY/HOMEOSTASIS
Advanced glycation end-products
Hypoglycemia
Insulin
Insulin-growth factor
Pro-inammatory cytokines
Reactive oxygen species
Glucocorticoids
Vasculopathy
UNAUTHORIZED USE PROHIBITED
UNAUTHORIZED USE PROHIBITED
Page 13
14 Adv Ther (2010) 27(2):1-18.
Moreover, small vessel disease results in the
release of proinflammatory cytokines, such
as interleukin (IL)-1, IL-6, and tumor necrosis
factor alpha (TNF-α).
36
TNF-α, the master
regulatorof the immune response, is the key
initiator of immune-mediated inflammation
in multiple organ systems, including the
brain.
37
Recently, investigators identified a
polymorphism in the promoter region of the
TNF gene that is associated with greater risk
for Alzheimer’s disease.
38
Furthermore, Aβ has
been shown to stimulate secretion of TNF-α.
39
In keeping with the view that abnormal
inflammatory homeostasis is a critical mediator
in the pathophysiology of DM, a viable
hypothesis is that brain volumetric changes in
DM population represent end-organ damage
mediated by neuroinflammation.
The CNS is vulnerable to the effects of
oxidative stress due to its high oxidative
metabolic activity, polyunsaturated fatty
acid content, and relatively low endogenous
antioxidant capacity.
40
The accumulation of
reactive oxygen species (ROS), a by-product
of mitochondrial metabolic processes,
results in oxidative damage, including lipid
peroxidation, protein oxidation, and DNA
damage which can ultimately lead to cell
death.
40,41
Accumulation of oxidative radicals
and resultant somatic toxicity is unequivocally
documented in diabetic patients.
41
It is possible
that the central accumulation of ROS may play
a critical role in altering neuronal and glial
cytoarchitecture and integrity in DM as well
as mood populations.
Disturbances in glucocorticoid signaling are
a highly replicated physiological abnormality in
both mood disorder and DM samples. For example,
approximately half of depressed individuals are
non-suppressors” with the Dexamethasone
Suppression Test.
42
Chronic elevation of cortisol
adversely affects neurotrophism, neuroplasticity,
and cellular resilience directly and via disparate
secondary mechanisms.
7,43
Peripheral abnormalities in neurometabolic,
neuroendocrine, and neuroinflammatory
processes are documented in both DM and mood
disorder populations. These foregoing effector
systems are critical mediators of neuronal (and
glial) changes observed in both populations.
Taken together, end-organ damage in DM
populations includes neurodegenerative changes
in the CNS. Neuronal (and glial) degeneration
and loss of neurotrophic support are also
implicated in mood disorders indicating that
a pathophysiological nexus exists for both DM
and mood disorders.
Limitations
There are several methodological deficiencies
that affect inferences and interpretations
that may be drawn from the extant literature
documenting associations between DM and
brain volume abnormalities and neurocognitive
deficits (Table 2). Two major deficiencies are
heterogeneity in sample composition and
adjusting for the effects of a mental disorder. The
pertinacity of this issue is accentuated by the fact
that Type 1 and Type 2 DM populations are not
identical; discrete physiological processes may
have important differences in CNS pathology
as measured by the degree and progression of
neuropsychological impairment.
Moreover, relatively few studies have
sufficiently screened for and/or adjusted for the
effect of a comorbid psychiatric disorder (ie, mood
disorders) which may mediate (or moderate)
the neuroanatomical and neurocognitive
abnormalities noted. For example, individuals
with major depressive disorder and bipolar
disorder exhibit volumetric changes in brain
regions as well as neurocognitive changes similar
to individuals with DM.
9
UNAUTHORIZED USE PROHIBITED
UNAUTHORIZED USE PROHIBITED
Page 14
Adv Ther (2010) 27(2):1-18. 15
Type 1 and 2 DM are associated with
structural and neurocognitive abnormalities.
Type 1 and Type 2 DM exhibit overlapping
and distinct changes in neurocognitive
function.
44
For example, individuals with
Type 1 DM are more likely to exhibit decits
in psychomotor speed and efficiency, while
individuals with Type 2 DM are more likely to
exhibit abnormalities in psychomotor efficiency,
attention, learning and memory and executive
function.
44
Moreover, individuals with Type 2
DM may exhibit more rapid neurocognitive
decline, which is accompanied by more MRI-
measured abnormalities (ie, WMH).
44
CONCLUSION
Notwithstanding the methodological
limitations, there are important transdisciplinary
themes that emerge fr