imaging amnestic mild cognitive impairment ......imaging amnestic mild cognitive impairment:...
TRANSCRIPT
Imaging Amnestic Mild Cognitive Impairment:
Neuroinflammation, Beta-Amyloid and Glutathione
by
Dunja Knezevic
A thesis submitted in conformity with the requirements for the degree of Master of Science
Institute of Medical Science University of Toronto
© Copyright by Dunja Knezevic 2016
ii
Imaging Amnestic Mild Cognitive Impairment:
Neuroinflammation, Beta-Amyloid and Glutathione
Dunja Knezevic
Master of Science
Institute of Medical Science University of Toronto
2016
Abstract
Amnestic mild cognitive impairment (aMCI) is defined as a transitional state between normal
aging and Alzheimer’s disease (AD). Neuroinflammation, amyloid accumulation and
oxidative stress have been shown to contribute to the pathology of AD, however whether
these processes occur in prodromal individuals is still not well understood. The main aim
was to investigate, with the use of positron emission tomography, whether
neuroinflammation is increased in aMCI patients and if it relates to amyloid burden in-vivo.
Our exploratory aim was to measure glutathione, the brain’s major antioxidant, in-vivo with
the use of magnetic resonance spectroscopy. No significant differences in [18F]-FEPPA
binding and glutathione levels were observed between healthy volunteers and aMCI patients.
In contrast, aMCI patients were found to have significantly more amyloid in the cortical
regions. Our findings suggest that amyloid pathology is an early event, whereas
neuroinflammation and alterations in glutathione may occur only after conversion to AD.
iii
Acknowledgments
The past two years at the University of Toronto would not have been possible without
the numerous people that have helped me along the way.
I would first like to thank my supervisor, Dr. Romina Mizrahi, for giving me the
opportunity to work in her lab and for believing in my abilities to handle such a complex
study. I am extremely grateful to have conducted my MSc thesis under the supervision of
such a hardworking and inspiring clinician-scientist. Thank you for your guidance and
mentorship. I would like to thank my PAC members Dr. Aristotle N. Voineskos and Dr.
Tarek Rajji for their support, time and scholarly inputs during committee meetings.
Additionally, I want to thank Dr. Rajji for assistance in recruitment of aMCI patients. I
would also like to thank Dr. Pablo M. Rusjan for always making time to provide his expertise
with image analysis. I am extremely grateful for all the time he spent with me and feel
honored to have learned for him. I would like to thank Dr. Nicolaas Paul L.G Verhoeff for
his tremendous help with recruitment. I greatly appreciate all the time he spent looking for
potential participants and meeting with me despite his very busy schedule.
I would like to acknowledge the assistance and support of all members of the
Research Imaging Centre. I would like to thank Alvina Ng and Laura Nguyen for their
patience, cooperation and support during PET scans. I would also like to thank all the
research participants and their family members for their commitment and enthusiasm during
the study. I am extremely thankful for their participation.
iv
I would like to thank all my lab members for creating such a welcoming and
supportive environment. To Huai-Hsuan Tseng, thank you for your mentorship and for
always making yourself available to provide advice and share your knowledge with me. To
Sina Hafizi, thank you for your help with study execution, data analysis and for always being
there to answer any question. To Ivana Prce, thank you for your continuous support, tea and
friendship. Thank you to Lauren Drvaric, Jeremy Watts, Efren Navas, Abanti Tagore and
Alex Koppel for your helpfulness, kindness, and friendship.
I would like acknowledge all of my friends for always being there to help or make me
laugh. To Susana Da Silva, Tania Da Silva, David Rocco, Sarah Coakeley and Samantha
Fernandes, thank you for your support, advice and friendships. To Anton Rogachov, thank
you for your love, motivation and constant support along the way.
I would like to thank my family for their unconditional love, encouragement and
support. To my parents, thank you for being my backbone and for supporting me in all of my
decisions. Thank you for encouraging me that I can do anything I set my mind to. To my
brother, Luka, thank you for being the best brother a girl can ask for. Thank you for your
constant encouragement, positive attitude and humor.
v
Contributions
Dunja Knezevic (author): was the lead study staff; performed recruitment of all participants,
execution of study visits, image analysis, results interpretation, and write up of thesis
Dr. Romina Mizrahi (supervisor): study design and concept; provided mentorship throughout
the study; guidance in study execution, results interpretation and thesis write up
Dr. Tarek Rajji: provided guidance with interpretation of results; determined diagnosis
eligibility of aMCI patients and assisted in recruitment of this population
Dr. Aristotle Voineskos: provided guidance with interpretation of results
Dr. Nicolaas Paul LG Verhoeff: determined diagnosis eligibility of aMCI patients and
assisted in recruitment of this population
Dr. Pablo Rusjan: provided expertise and assistance with image analysis
Dr. Sina Hafizi: provided assistance with study execution; guidance with interpretation of
results
vi
Table of Contents
ABSTRACT ii
Acknowledgements iii
Contributions v
List of Tables x
List of Figures xi
List of abbreviations xiii
1. INTRODUCTION 1
1.1 Statement of problem 1
1.2 Purpose 1
1.3 Mild Cognitive Impairment 2
1.3.1 Classification and prevalence 2
1.3.2 Pathology of mild cognitive impairment 3
1.3.2.1 Aβ formation/accumulation 5
1.4 Neuroinflammation 7
1.4.1 Overview of neuroinflammation: the role of microglia 7
1.4.2 Evidence of inflammation: microglia and peripheral markers 9
1.4.3 Microglial activation by Aβ accumulation 10
1.5 In-vivo human studies: Positron emission tomography 13
1.5.1 Imaging amyloid 14
1.5.1.1 Existing amyloid radioligands 14
1.5.1.2 The use of [11C]-PIB in mild cognitive impairment 16
vii
1.5.2 Quantifying neuroinflammation in-vivo 18
1.5.2.1 Targeting microglia: Translocator protein 18kDa 18
1.5.2.2 TSPO radioligands for imaging neuroinflammation 20
1.5.2.3 [18F]-FEPPA as an imaging biomarker of neuroinflammation 23
1.5.2.4 PET imaging of mild cognitive impairment with TSPO radioligands
24
1.6 Oxidative stress 29
1.6.1 Glutathione: The most abundant brain antioxidant 29
1.6.2 Oxidative stress and glutathione in AD and MCI 31
1.6.3 In-vivo quantification of GSH 32
1.6.3.1 Magnetic resonance spectroscopy 32
1.6.3.2 Measurement of GSH using MRS in AD and MCI 33
2. AIMS AND HYPOTHESES 36
2.1 Aims 36
2.2 Hypotheses 38
3. METHODS 39
3.1 Participants 39
3.2 Neuropsychological assessments 40
3.3 PET measures of neuroinflammation and β-amyloid plaques 41
3.3.1 PET image acquisition 41
3.3.2 Regions of interest (ROI)-based PET image analysis 42
3.3.2.1 Measurement of TSPO expression with [18F]-FEPPA 45
3.3.2.2 Measurement of amyloid plaques with [11C]-PIB 47
viii
3.3.3 Voxel-based PET image analysis 48
3.4 MRI acquisition 48
3.4.1 Structural images 49
3.4.2 Spectroscopy 49
3.5 Genotyping 50
3.6 Statistical analysis 51
4. RESULTS 52
4.1 Neuroinflammation and beta-amyloid 52
4.1.1 Participants 52
4.1.2 Increased [11C]-PIB binding in the GM of aMCI patients 54
4.1.3 [18F]-FEPPA in GM of aMCI and healthy volunteers 56
4.1.4 Exploratory correlations between [11C]-PIB and [18F]-FEPPA 62
4.1.5 Exploratory correlations between [11C]-PIB/[18F]-FEPPA and cognition
66
4.2 Glutathione 76
4.2.1 No differences in GSH levels in the LDLPFC 76
4.2.2 Exploratory correlations between GSH and [11C]-PIB, [18F]-FEPPA, and cognition
78
5. DISCUSSION 81
5.1 Increased amyloid in aMCI patients 81
5.2 No differences in [18F]-FEPPA VT 83
5.3 Exploratory correlations between amyloid and neuroinflammation 85
5.4 Neuroinflammation, but not amyloid, may correlate with cognition 86
5.5 No differences in GSH levels 88
ix
5.6 GSH levels correlate with microglia but not amyloid 90
5.7 GSH and performance on neuropsychological tests 93
6. STRENGTHS 94
7. LIMITATIONS 95
8. CONCLUSION 98
9. FUTURE DIRECTIONS 99
10. REFERENCES 102
x
List of Tables
Table 1. A summary of TSPO radioligands used in-vivo
Table 2. A summary of PET studies measuring neuroinflammation in AD
Table 3. A summary of PET studies measuring neuroinflammation in MCI
Table 4. A summary of GSH MRS studies in AD and MCI populations
Table 5. Participant demographics and PET parameters
Table 6. [18F]-FEPPA Standardized Uptake Value Ratios (SUVR)
Table 7. Regional [18F]-FEPPA VT for HV and aMCI participants when stratified by PIB
status
Table 8. Positive correlations between [18F]-FEPPA and [11C]-PIB binding
Table 9. Correlation analyses between [11C]-PIB DVR and neuropsychological scores
Table 10. Correlation analyses between [18F]-FEPPA VT and neuropsychological scores
Table 11. [11C]-PIB binding and GSH levels in aMCI participants
Table 12. [18F]-FEPPA binding and GSH levels in aMCI participants
Table 13. Exploratory correlations between GSH levels in the LDLPFC and performance on
cognitive scales
xi
List of Figures
Figure 1. Proposed pathway of pathological events.
Figure 2. The glutathione reduction-oxidation cycle
Figure 3. Step-by-step procedures for automatic delineation of ROIs using ROMI.
Figure 4. Two-tissue compartment model
Figure 5. Voxel placement in the left dorsolateral prefrontal cortex.
Figure 6. [11C]-PIB Distribution Volume Ratios (DVR) in regions of interest.
Figure 7. [18F]-FEPPA Total Volume Distribution (VT) in regions of interest
Figure 8.T-statistical map of parametric images of HV vs MCI overlaid on a T1 weighted
magnetic resonance imaging (MRI) template.
Figure 9. [18F]-FEPPA VT in regions of interest after characterizing participants based on
amyloid status
Figure 10. Regional associations between [18F]-FEPPA and [11C]-PIB binding
Figure 11. Regional associations between [18F]-FEPPA and [11C]-PIB binding after partial
volume correction
Figure 12. Correlations between [11C]-PIB binding and cognition
Figure 13. Correlations between [11C]-PIB binding and score on the logical memory delayed
task, after partial volume correction
Figure 14. Correlations between [18F]-FEPPA binding and cognition
xii
Figure 15. Correlations between [18F]-FEPPA binding in the hippocampus and cognition,
after partial volume correction
Figure 16. No differences in GSH/H2O levels in the left dorsolateral prefrontal cortex
Figure 17. Positive correlations between [18F]-FEPPA binding and GSH/ H2O levels
Figure 18. Higher GSH levels correlated with a higher score on the Stroop Color Word score
Figure 19. Model of AD pathology demonstrating possible correlations between beta-
amyloid, neuroinflammation and oxidative stress
xiii
List of abbreviations
2TCM Two-tissue compartment model
Aβ Amyloid beta
AD Alzheimer’s disease
aMCI Amnestic mild cognitive impairment
APP Amyloid precursor protein
CAMH Centre for Addiction and Mental Health
CND Concentration of non-displaceable radioligand in tissue
CP Concentration of radioactivity in plasma
CSF Cerebrospinal fluid
DLPFC Dorsolateral prefrontal cortex
DVR Distribution volume ratio
FAS Letter fluency test
GM Grey matter
GSH Glutathione
HAB High affinity binder
HRRT High resolution research tomograph
HV Healthy volunteer
LAB Low affinity binder
LDLPFC Left dorsolateral prefrontal cortex
MAB Mixed affinity binder
MCI Mild cognitive impairment
MMSE Mini-mental state examination
MoCA Montreal cognitive assessment
MRI Magnetic resonance imaging
xiv
MRS Magnetic resonance spectroscopy
NART North American reading test
NFT Neurofibrillary tangle
PET Positron emission tomography
PHF-tau Hyperphosphorylated tau
PVEC Partial volume error correction
ROI Region of interest
ROMI Regions of mental interest
RBANS Repeatable battery for the assessment of neuropsychological status
SUV Standardized uptake value
TAC Time activity curve
TMT Trail making test
TSPO Translocator Protein 18kDa
VT Total distribution volume
WM White matter
1
1 INTRODUCTION
1.1 Statement of Problem
As our population ages, dementia is becoming an emerging public health challenge with
over 35 million people affected (Heppner et al. 2015). Not only are there substantial
emotional burdens for the individuals affected and their families, there are also significant
financial costs. In 2010, the global estimated financial cost of dementia was in excess of
US$600 billion (Heppner et al. 2015). The most prevalent type of dementia is Alzheimer’s
disease (AD), accounting for nearly 60-80% of cases (Krstic and Knuesel 2013). AD is a
progressive disease that causes neural (e.g. accumulation of protein aggregates) changes in
an individual over time. It has been well established that individuals who eventually develop
degenerative dementia such as AD, are likely to go through a period of mild cognitive
impairment (MCI). The substantial number of people living with AD, and the growing
number of cases developing, highlights an important area of research. In order to advance
treatment for AD, it is paramount that research focuses on the mechanisms behind prognosis
and early detection of this debilitating disease.
1.2 Purpose
The aim of this study was to improve the understanding of the neuropathology of the
MCI state. Although neuroinflammation is thought to play a role in the pathology of AD,
whether neuroinflammation occurs early at the MCI stage is still unknown. This study aimed
2
to use well-validated imaging biomarkers of amyloid plaques and neuroinflammation to
examine whether microglial activation is detectable in the brain of aMCI patients, and to
elucidate whether neuroinflammation is related to amyloid burden in-vivo. The second
portion of the study involves the in-vivo quantification of the brain antioxidant glutathione
(GSH) and investigation of possible correlations with amyloid burden and
neuroinflammation.
1.3 Mild Cognitive Impairment
1.3.1 Classification & Prevalence
Mild cognitive impairment (MCI) has been defined as a transitional stage between
normal aging and dementia in which individuals experience some degree of memory
impairment that falls outside of the predicted age and education norms, but is not sufficiently
severe to meet the severity level for diagnosis of clinical probable Alzheimer’s disease
(Petersen et al. 2001). Individuals with MCI may take more time, be less efficient or make
more errors at performing complex functional tasks which they used to perform previously,
such as paying bills, preparing a meal or shopping (Albert et al. 2011). Nevertheless, they are
generally able to maintain their independence of function in daily life with minimal aid or
assistance (Petersen 2004;Albert et al. 2011;Risacher and Saykin 2013). The prevalence of
MCI is 19% among persons younger than 75 years of age and 29% among those 85 years of
age or older (Small et al. 2006). MCI can be divided into two main subtypes: amnestic
(aMCI) which encompasses memory impairments and non-amnestic (naMCI) which is
3
comprised of impairments in non-memory domains such as executive function, language or
visuospatial ability. The categorization is not this simplistic, cross over between the two
broad types can occur, meaning that some individuals can have an impairment in memory
and a deficit in another cognitive domain. Thus, the subtypes can be further characterized
based on whether a single domain or multiple domains are impaired (Reinlieb et al. 2014).
Between the two subtypes, aMCI is more common and is thought to constitute the prodromal
stage of AD, as individuals are at a higher risk of progression to AD (Gauthier et al.
2006;Petersen et al. 2006;Reinlieb et al. 2014).
The progression from MCI to dementia is 10-15% per year, which is significantly
higher compared to the general population, with an estimated conversion rate of 1-2% per
year (Small et al. 2006). According to a population-based study, approximately 25% of MCI
patients convert to AD within an average of 2.5 years (Boyle et al. 2006). In particular those
with aMCI decline considerably more rapidly each year on global cognitive function than
those without cognitive impairment (Petersen et al. 2001;Boyle et al. 2006). However, not all
individuals with MCI convert to dementia, some remain at this stage whereas others have
been found to convert back to normal cognition (Petersen et al. 2001;Jack et al. 2010).
1.3.2 Pathology of Mild Cognitive Impairment
Alzheimer’s disease is characterized by the presence of senile plaques composed of
β-amyloid aggregates and neurofibrillary tangles of hyperphosphorylated tau (Small et al.
2006;Chien et al. 2013;Maruyama et al. 2013;Okamura et al. 2014). As mentioned, it has
been well established that individuals who eventually develop degenerative dementia such as
4
AD, are likely to transition through a period of MCI. Although MCI may be due to a variety
of possible etiologies from degenerative, vascular, metabolic or traumatic causes (Petersen
2004), aMCI in particular is thought to be predominately degenerative in nature and thus
may be referred as the prodromal stage of AD (Palmer et al. 2008). Degeneration includes
the loss of neurons, and accumulation of protein aggregates (Stephan et al. 2012). The
characteristic protein aggregates of AD, amyloid and tau, have been found to accumulate
decades before the onset of dementia, including in the transitional stage of MCI (Small et al.
2006). The deposition and spread of these protein aggregates (Braak staging) has been
characterized in post-mortem brain samples of nearly 3000 individuals(Braak and Braak
1997).The development of cortical Aβ deposits was distinguished in three stages (stages A-
C). In the initial stage plaques can be observed in the basal neocortex, most likely in the
poorly myelinated temporal areas (stage A). As more deposits develop, they spread into
adjacent neocortical areas and the hippocampal formation (stage B). In the last stage,
deposits can be observed in all areas of the cortex (stage C). The spread of neurofibrillary
tangles was explained in 6 stages(Braak and Braak 1997). In stage I, the lesions can be
observed only in specific projection cells in the transentorhinal region. The pathological
process then spreads into the entorhinal region (stage II), the hippocampus and the temporal
proneocortex (stage III) and then the association areas of the adjoining neocortex (stage IV).
In the last stages, the lesions spread superlaterally (stage V) and into the primary areas of the
neocortex (stage VI).
Most studies suggest that Aβ deposition is an early event likely to occur prior to
demonstrable cognitive impairment (Rowe et al. 2007). Evidence has demonstrated that MCI
patients with significant amyloid accumulation are at a higher risk of future conversion to
5
AD (Forsberg et al. 2008;Risacher and Saykin 2013). Although the presence of amyloid
plaques has been widely confirmed in individuals with AD, not all individuals with MCI
have amyloid accumulation, some are classified as having an “AD-like” profile of amyloid
accumulation, whereas others have very low levels and are characterized as having a
“normal” profile (Rowe et al. 2007). Given that fact that amyloid plaques are one of the
hallmarks of AD, it is important to look into the pathology behind the accumulation of these
proteins.
1.3.2.1 Aβ formation and accumulation
Although amyloid plaques were discovered by Alois Alzheimer in 1907, it took
nearly 75 years to identify the major constituent of these plaques as a 40-42 amino acid
peptide, Aβ (Agostinho et al. 2015). This peptide is produced by the cleavage of a
transmembrane protein, amyloid precursor protein (APP). The exact function of APP is still
unclear, however a number of physiological roles have been attributed to this molecule. The
extra- and intracellular regions of APP are thought to be involved with metal (copper and
zinc) binding, extracellular matrix components (heparin, collagen, and laminin), neurotrophic
and adhesion domains and protease inhibition (Thinakaran and Koo 2008). In addition, the
large extracellular domain functions as a cell adhesion motif mediating cell-cell adhesion and
migration (Agostinho et al. 2015). Furthermore, it has been suggested that APP is involved in
synaptogenesis, dendritic spine formation, synaptic vesical and transmitter release, and
synaptic plasticity and behavior (Agostinho et al. 2015).
6
The cleavage of this transmembrane protein occurs either in a non-amyloidogenic
(“physiological”) or in an amyloidogenic (“pathological”) fashion. Under the non-
amyloidogenic pathway, APP is first cleaved by α-secretase and then followed by γ-
secretase. Conversely, in the amyloidogenic pathway, β-secretase cleaves APP first rather
than α-secretase. The different secretases cleave the protein at different sites, which
determines whether the predominant Aβ40 or the more aggregation-prone and neurotoxic
Aβ42 species is generated (Heppner et al. 2015). The type of APP processing that occurs is
dependent on the spatial proximity of this protein with the different secretases in diverse
stages of life and in pathological and non-pathological conditions (Agostinho et al. 2015).
These Aβ peptides are produced in normal conditions as well but are in much smaller
quantities compared to AD conditions. The levels of Aβ peptide production in AD patients
are approximately equivalent to 7 years of total Aβ production in healthy individuals
(Agostinho et al. 2015).
It has been shown that Aβ peptides cause neuronal damage, synaptotoxicity and
affect the function of other brain cells, such as glia and endothelial cells. The increased
accumulation of neurotoxic Aβ peptides has led to one of the leading hypotheses of AD
pathology, the amyloid cascade hypothesis. This hypothesis postulates that the Aβ deposits
are the main causative agent of AD, and are the proximal cause of synaptic dysfunction and
early cognitive impairment in this disease. Although amyloid plaques have been widely
confirmed to be present in AD, we can be certain that this accumulation is not the only driver
of disease progression as it has even been detected in cognitively normal individuals
(Bennett et al. 2006;Jack et al. 2010). It is becoming evident that AD pathogenesis is much
more complex involving more factors than simply Aβ production/accumulation. Another
7
feature that has been postulated for a long time, but only recently appreciated, is the role of
neuroinflammation in AD.
1.4 Neuroinflammation
1.4.1 Overview of neuroinflammation: the role of microglia
As in other systems of the body, the immune response plays an active role in the central
nervous system (CNS). In contrast to peripheral inflammation, neuroinflammation does not
involve conventional inflammatory cells such as white blood cells. Neuroinflammation is
largely driven by microglial activation and/or reactive astrocytosis (Kreisl et al. 2013). The
first line of defense against invading pathogens and other harmful agents are microglial cells.
Microglia are mononuclear phagocytes that are ubiquitously distributed in the brain and
represent approximately 10% of the total cell population in the CNS (Heneka et al. 2015).
Microglia are assumed to exist in two states: resting and activated. In the resting state,
microglia continuously extend and retract their processes as they sample their environment
for pathogens (McGeer and McGeer 2010;Krabbe et al. 2013). During this process, microglia
also provide factors that support tissue maintenance (Heneka et al. 2015). In addition, they
may be neuroprotective as they secrete various growth factors (Schwab and McGeer 2008).
Microglia are particularly sensitive to changes in their microenvironment and can very
quickly be activated in response to infection or injury (Liu and Hong 2003). Upon activation,
8
microglia proliferate and migrate to the site of injury, and adopt a set of morphological and
functional attributes (Heppner et al. 2015). The morphological changes include shortening of
processes and hypertrophy of the cell body (Perry et al. 2010), while the functional attributes
include the upregulation of a variety of surface receptors and the release of several pro-
inflammatory factors, including the cytokines tumor necrosis factor-α (TNFα) and
interleukin-1β (IL-1β), free radicals such as nitric oxide (NO) and superoxide, fatty acid
metabolites such as eicosanoids, and quinolinic acid (Liu and Hong 2003). Although the
release of these pro-inflammatory factors is a defense mechanism of the immune system, it
may ultimately have detrimental effects. Cell culture studies have demonstrated that the
supernatants of highly activated microglia are toxic to neurons (Boje and Arora 1992;Chao et
al. 1992;McGuire et al. 2001;McGeer and McGeer 2010). Overall it can be seen that the role
of microglia in the brain is complex, as their involvement may be both beneficial and
detrimental.
Reactive microglia have been observed during most neuropathological conditions in
addition to AD, such as Parkinson’s disease (McGeer et al. 1988;Sanchez-Guajardo et al.
2013), amylotrophic lateral sclerosis (Brites and Vaz 2015), multiple sclerosis (Benveniste
1997), and the acquired immunodeficiency syndrome (Dickson et al. 1993). Though several
lines of evidence have demonstrated the involvement of microglia in these diseases, it still
remains to be determined at what stage microglia are involved in the disease progression. In
particular interest to this thesis, is whether microglia are involved in the very early stages of
AD development, specifically in the aMCI stage.
9
1.4.2 Evidence of inflammation: microglia and peripheral markers
Ample evidence has demonstrated that microglia are present in pathologically
affected areas of AD. Furthermore, the density of the activated microglia were found to
correlate with the severity of the inflammatory response (Carpenter et al. 1993). In addition
to the finding of microglia in the brains of AD patients, immunohistochemical studies have
also measured inflammatory markers in brain samples. Increased levels of IL-6, IL-8 and
MCP-1 were reported in the inferior temporal cortex of AD patients compared to healthy
volunteers (Sokolova et al. 2009). Furthermore, marked elevations in IL-1β, IL-2, and IL-3
were measured in the hippocampi of AD patients (Araujo and Lapchak 1994).
Although the presence of microglia and inflammatory markers has been studied in the
AD brain, whether these inflammatory processes occur in the prodromal phases of AD is still
not known. Thus far, there have not been any reported post-mortem studies on microglia in
the brains of aMCI patients. However, some groups have studied microglia in post-mortem
brain samples of individuals at very early stages of AD. One study looking at AD patients
with varying degrees of severity and normal individuals reported that microglia and amyloid
deposits appeared at an early stage of the pathological cascade (Arends et al. 2000). Another
study looking at 9 possible AD cases found a significant increase in activated microglia in
the neocortex compared to healthy volunteers. Furthermore, they observed that microglia
were present within the Aβ plaques, and were particularly associated with the core of the
senile plaques (Vehmas et al. 2003). Levels of inflammatory markers in the brain have also
only been measured in AD brain samples but not in the prodromal stage of MCI. However
several studies sought to measure inflammatory markers in the blood and CSF of MCI
patients. Conflicting results have been reported in the levels of cytokines such as IL-1β, IL-6,
10
TNF-α, IFN-α and MCP-1, whereby some groups have reported an elevation while others
demonstrate a decrease (Galimberti et al. 2006;Guerreiro et al. 2007;Bermejo et al.
2008;Forlenza et al. 2009). A recent meta-analysis did not find support for the elevation of
peripheral inflammatory markers in MCI (Saleem et al. 2015). Overall, it is quite evident that
the underlying pathology of this transitional stage is still not well understood.
1.4.3 Microglial activation by Aβ accumulation
The emerging role of microglia and neuroinflammation in AD has become of great
interest in recent decades. One of the major pathogenic concepts in the field of AD has been
the amyloid cascade hypothesis. The accumulation of Aβ is believed to be a key factor that
drives the neuroinflammatory responses in AD (Schwab and McGeer 2008), and ample
evidence has demonstrated that Aβ plaques can activate microglia (McGeer and McGeer
2010). Aβ plaques are recognized as pathological and are targeted by microglia. Post-mortem
immunohistochemical examinations of brain slices have revealed the presence of activated
microglia surrounding Aβ plaques in AD (Rogers et al. 1988;Itagaki et al. 1989;McGeer et
al. 1989). Additionally, microglia were reported to be distributed in graded concentrations in
relation to their distance from Aβ deposits in transgenic mouse models of AD (Frautschy et
al. 1998). The activation of microglia by amyloid β is induced by the binding between Aβ
fibrils and microglial cell surface receptors such as SCARA1, CD36, CD14, a6β1 integrin,
CD47, and Toll-like receptors (TLR2, TLR4, TLR6, and TLR9) (Heneka et al. 2015). Upon
activation, microglia attempt to clear the brain of amyloid through the phagocytosis of Aβ
and by secreting proteolytic enzymes that degrade Aβ, such as IDE (insulin-degrading
enzyme), neprilysin, matrix metalloproteinase 9 (MMP9), and plasminogen (Leissring et al.
11
2003;Yan et al. 2006). In addition to phagocytosis, and as previously mentioned, upon
activation microglia subsequently begin to produce pro-inflammatory cytokines, chemokines,
complement proteins, and may even release toxic free radicals, upon stronger activation
(McGeer and McGeer 2010;Krabbe et al. 2013). Several studies have confirmed that Aβ can
induce microglia to release these pro-inflammatory proteins. When microglia co-cultured
with hippocampal brain slices were treated with aggregated Aβ, there was an upregulation of
various pro-inflammatory molecules and neuronal death (Butovsky et al. 2005).
The relationship between microglia and Aβ is more complex than simply a phagocytic
interaction. The secretion of pro-inflammatory proteins by microglia leads to not only
downstream inflammatory pathways, but also feeds back to modulate the microglia
themselves (Figure 1). The impairment of microglia by inflammatory mediators has been
evidenced at plaque sites (Streit et al. 2009;Krabbe et al. 2013). An in vitro study
demonstrated that the pro-inflammatory cytokine TNFα led to a downregulation of receptors
and enzymes involved in Aβ binding and degradation, and subsequently to an impairment in
phagocytosis (Hickman et al. 2008). The ongoing formation of Aβ and positive feedback
loops between inflammation and APP processing, compromise cessation of inflammation
and lead to a chronic inflammatory state in AD. Although various lines of evidence have
supported the amyloid cascade hypothesis, it still remains controversial. The exact
mechanisms that relate Aβ and microglia, and the overall pathology of AD are still unknown.
12
Figure 1. Proposed pathway of pathological events. In the postmortem AD brain, the
presence of Aβ plaques has been shown to activate microglial cells which in response
phagocytose the protein aggregates and secrete inflammatory cytokines and chemokines. In
AD, Aβ sustains chronic activation of microglia (A), which may lead to a constant
production of inflammatory proteins. In this potential chronic inflammatory state, the pro-
inflammatory proteins feedback to modulate the microglia (B), leading to their impairment
(C).
13
1.5 In-vivo Human Studies: Positron Emission Tomography
The rapid development of non-invasive tools for the imaging of human brains has had
a great effect on our ability to investigate and understand brain function. PET is an analytical
imaging technology used to image and measure biochemical processes of mammalian
biology in-vivo (Phelps 2000). PET requires the design of a ligand that binds with high
specificity to a desired target, but with minimal nonspecific binding to other structures
(Owen and Matthews 2011). The molecular probe or ligand is labeled with a positron-
emitting radioisotope (e.g., 18F, 11C, 15O). The radioligand is then administered intravenously
at a tracer dose, such that the radioligand should only occupy a negligible amount of target
sites (typically defined as <5% of the total available target in the brain). Following the
intravenous administration, the radioligand emits positrons (β+) from its nucleus as it decays.
The emitted positron collides with a nearby electron in the tissue, and annihilation occurs
with their masses converted into their energy equivalent through emission of two 511-keV
photons 180° apart (Phelps 2000). Scintillation detectors surrounding the participant detect
these photons. To determine the spatial distribution of the radioligand in the brain, a
mathematical reconstruction algorithm is then applied during which the dynamic PET data is
corrected for radioactive decay, photon attenuation, photon scattering, and PET detector
dead-time.
14
1.5.1 Imaging amyloid
1.5.1.1 Existing amyloid radioligands
Until relatively recently, plaques could only be assessed at autopsy. With the advent
of PET and specific amyloid radiotracers, amyloid has been able to be quantified in-vivo. The
development of plaque-binding compounds started with monoclonal antibodies against Aβ
and peptide fragments, which were followed by small radiolabeled analogues of Congo red,
Chrysamine-G, and Thioflavin applicable for SPECT and PET, and molecules targeting
amyloid plaques suitable for MRI (Nordberg 2004).The first amyloid-β imaging agent was a
derivative of the beta-sheet dye, Thioflavin-T, carbon-11 labeled Pittsburgh compound-B.
Since its development, [11C]-PIB has been the most extensively used radiotracer for both
preclinical and clinical applications, and has played an immense role in the understanding of
AD pathology (Landau et al. 2013). In-vitro studies have demonstrated that the levels of
[11C]-PIB binding correlate strongly with the total insoluble Aβ and Aβ1-42 levels (Klunk et
al. 2003;Klunk et al. 2005). Furthermore, [11C]-PIB binding was found to correlate with the
topographical distribution of amyloid aggregates in post-mortem brain slices (Bacskai et al.
2007;Ikonomovic et al. 2008;Leinonen et al. 2008), thus providing evidence that [11C]-PIB is
a valid marker of amyloid plaques. In-vivo imaging studies have shown that [11C]-PIB has a
rapid uptake in the brain, particularly in the cortical regions of AD patients (Klunk et al.
2004). Numerous PIB imaging studies have demonstrated clear differences in [11C]-PIB
retention between AD patients and healthy controls (Klunk et al. 2004;Kemppainen et al.
2006;Mintun et al. 2006;Edison et al. 2007). Similar binding patterns of [11C]-PIB in the
cortical regions of AD patients have also been demonstrated by numerous PET studies
(Klunk et al. 2004;Rowe et al. 2007). Other carbon-11 labeled tracers have been developed
15
for imaging amyloid, including [11C]-BTA-1(Neumaier et al. 2007) and [11C]-SB-13
(Verhoeff et al. 2004). [11C]-BTA-1 is a derivative of [11C]-PIB that lacks the 6-hydroxy
group, and has only been used in an autoradiography binding study with post-mortem brain
samples and not in-vivo. The results demonstrated that there was an elevation of binding in
the cortical areas of an AD subject compared to a healthy control (Neumaier et al. 2007).
[11C]-SB-13 was used in a study of five AD subjects and five controls (Verhoeff et al. 2004).
The participants underwent a [11C]-PIB scan as well, in order to compare the binding pattern
to [11C]-SB-13. The group found that the pattern of [11C]-SB-13 was very similar to the
binding observed with [11C]-PIB in the frontal, posterior temporal, and inferior parietal
cortices.
Although [11C]-PIB has been shown to have a high affinity for amyloid, and has been
validated in numerous in-vivo imaging studies, an inherent limitation of this tracer and other
carbon-11 tracers led to a pursuit of new radioligands. The radioactive decay half-life of
carbon-11 (20 minutes) limits the use to centres with on-site cyclotrons and carbon-11
radiochemistry expertise. To overcome this limitation, several tracers labeled with fluorine-
18 (half-life 110 minutes) were developed. Imaging agents with longer half-lives do not
restrict use to centres with an on-site cyclotron, and thus can facilitate wider availability.
Some of the fluorinated compounds include the fluorine-18 analogue of PIB ([18F]-
flutemetamol), and fluorinated compounds of other chemical classes such as ([18F]-
florbetapir, [18F]-florbetaben) and benzofurans ([18F]-NAV4694). These compounds were
also found to bind to fibrillary amyloid- β with high affinity, and have demonstrated in-vivo
detection of amyloid-β pathology in symptomatic patients and in animal models. Given that
fluorine-18 labeled compounds were found to show similar patterns of amyloid deposition
16
(Rowe et al. 2010;Fleisher et al. 2011;Landau et al. 2013), studies sought to compare these
compounds with [11C]-PIB in the same individuals. A group investigating the comparison
between [18F]-florbetaben (or FBB) and [11C]-PIB in AD patients, found an almost identical
cortical distribution of the two tracers, however the degree of retention was lower for FBB.
They reported a 75% increase in the global SUVR for PIB in AD patients versus healthy
controls, compared to a 56% global SUVR increase for FBB (Villemagne et al. 2012).
Similarly, a study comparing [11C]-PIB and [18F]-florbetapir in the same AD individuals,
revealed that the cortical retention ratios of the two tracers were highly correlated. However,
the florbetapir retention values were reduced compared with PIB (Landau et al. 2013). The
reduction has been evidenced with other fluorine-18 labeled alternatives and is thought to be
related to their greater lipophilicity, which leads to greater non-specific binding (Pike
2009;Landau et al. 2013). Thus although these fluorine-18 alternatives allow for a greater
availability of use, [11C]-PIB still remains the most extensively used tracer for imaging
fibrillary amyloid. Given that [11C]-PIB has been widely confirmed to be a valid tracer for
amyloid, we chose to use this radioligand to quantify amyloid burden in aMCI.
1.5.1.2 The use of [11C]-PIB in mild cognitive impairment
With the advent of PET imaging, amyloid has been quantified in-vivo in AD and
other diseases involving amyloid pathology. Studies looking at amyloid in MCI have
commonly found that not all individuals with MCI have an accumulation of Aβ plaques. In
particular, PET studies have shown that about half to two-thirds of aMCI patients display
elevated amyloid deposition (Huijbers et al. 2015). PIB studies have classified aMCI patients
17
into either the “normal range” or “AD range” of amyloid deposition, with the average of this
group lying in an intermediate position (Price et al. 2005;Rowe et al. 2007;Forsberg et al.
2008;Jack et al. 2008;Wiley et al. 2009). Although not all individuals with MCI have
amyloid deposition, the presence of amyloid is thought to increase ones risk of progression to
AD (Huijbers et al. 2015). Forsberg and colleagues reported that 11 out of 21 MCI patients
displayed high [11C]-PIB retention. Of these 11, 7 converted to AD at later follow-ups 2-16
months after their initial PET scan. The converters showed significantly higher PIB retention
in the frontal, parietal, and temporal cortices compared to controls, whereas the non-
converters did not differ from controls in any region. Interestingly, all of the converters were
of the amnestic MCI subtype, clearly demonstrating that this subtype has an increased risk of
progression to AD (Forsberg et al. 2008). Similarly, Okello and colleagues studied 14 aMCI
patients and found that 50% of them had increased amyloid deposition. The PIB-positive
MCI patients had comparable uptake ratios to the AD group, with two-fold increases in the
cingulate and frontal regions. Furthermore during a 2-3 year follow-up, 3 of their PIB-
positive patients converted to AD whereas none of the PIB negative patients with MCI
converted (Okello et al. 2009), thus clearly providing support for the increased risk
associated with amyloid deposition. Overall the reported rates of conversion to AD from
MCI have been as high as 48% within 3 years (Okello et al. 2009;Wolk et al. 2009).
Correlations between amyloid and performance on neuropsychological assessments
have also been performed. Post-mortem studies of AD have demonstrated that amyloid does
not correlate strongly with symptom severity (Arriagada et al. 1992;Bierer et al.
1995;Vehmas et al. 2003). In contrast, in-vivo PET studies have reported inconsistent results,
some groups demonstrated no correlations between retention and performance on cognitive
18
scales (Edison et al. 2008;Okello et al. 2009), whereas others reported some correlations with
episodic memory tests (Forsberg et al. 2008;Villemagne et al. 2011).
1.5.2 Quantifying neuroinflammation in-vivo
1.5.2.1 Targeting microglia: Translocator protein 18kDa
The activation of microglia has been evidenced by the concomitant upregulation or de
novo synthesis of a variety of cell-surface and cytoplasmic molecules (Perry et al. 2010).
One particular protein that has been of great interest in quantifying neuroinflammation in-
vivo is Translocator Protein 18kDa (TSPO). TSPO was identified during central
benzodiazepine receptor (CBR) binding studies (Braestrup and Squires 1977). It was initially
termed as a peripheral benzodiazepine receptor (PBR) as it was shown to be abundantly
distributed in peripheral tissues (Chen and Guilarte 2008). It was subsequently discovered
that PBR is also present in glial and ependymal cells of the brain in addition to peripheral
tissues. PBR was determined to be pharmacologically, anatomically, structurally, and
physiologically distinct from the CBR, and was thus renamed to Translocator Protein 18kDa
(Chen and Guilarte 2008).
TSPO can form a multimeric complex with the 32 kDa voltage-dependent anion
channel (VDAC) also called mitochondrial porin and the 30 kDa adenine nucleotide carrier
(ANC) in the outer mitochondrial membrane (Papadopoulos et al. 2006;Chen and Guilarte
2008). The ratio of TSPO to VDAC and ANT appears to be tissue- and treatment-dependent
(Veenman et al. 2007). Free TSPO, not in complex with VDAC and ANT, has also been
19
suggested to be present in mitochondrial membranes (Veenman et al. 2007). While TSPO
can be found in different parts of the cell, the primary intracellular location is the outer
mitochondrial membrane (Veenman et al. 2007).The exact physiological functions of TSPO
have still not been elucidated, nevertheless it is thought to participate in a variety of functions
including cell growth and proliferation, steroidogenesis, bile acid synthesis, calcium flow,
cholesterol transport, apoptosis and neuroinflammation (Papadopoulos et al. 2006;Veenman
et al. 2007;Chen and Guilarte 2008;Gulyás et al. 2009).
In the brain, under normal physiological conditions, the levels of TSPO are low and
limited to glial cells. However, during brain insults when microglia become activated, TSPO
levels are dramatically upregulated (Chen and Guilarte 2008;Gulyás et al. 2009). The
increased expression of TSPO has been observed in both normal aging and in diseases
involving the CNS such as AD, stroke, and multiple sclerosis (Gulyás et al. 2009).
Microautoradiography and immunohistochemistry studies have confirmed that areas with
increased TSPO levels coincide with the same areas in which there is an increase in
microglia. (Banati et al. 2000;Kuhlmann and Guilarte 2000).
Although upregulation of TSPO has been widely confirmed, the functional
significance of increased TSPO is still unknown. It has been hypothesized that increased
TSPO may be associated with the proliferative, migratory and phagocytic capacity of
microglia or it may be related to the secretion of inflammatory cytokines (Chen and Guilarte
2008). The fact that TSPO levels are low in the brain parenchyma and regionally increased
during brain insults makes TSPO an ideal marker for brain imaging studies.
20
1.5.2.2 TSPO radioligands for imaging neuroinflammation
There are a variety of TSPO radioligands available in PET imaging to quantify
neuroinflammation in-vivo, some of which are summarized in Table 1. The most widely used
radioligand is [11C]-PK11195, which is a selective antagonist for TSPO. [11C]-PK11195 was
initially used as a racemate, but later studies found that the R-enantiomer has a 2-fold greater
affinity for TSPO than the S-enantiomer and thus subsequent studies used [11C]-(R)-
PK11195 to investigate neuroinflammation in-vivo (Vivash and O'Brien 2015). [11C]-
PK11195 has been used for almost three decades in a variety of neurological disorders such
as multiple sclerosis (Politis et al. 2012), Parkinson’s disease (Gerhard et al. 2006),
Huntington’s disease (Pavese et al. 2006), ischemic stroke (Gerhard et al. 2000), and AD
(Cagnin et al. 2001). Although it has been widely used, there are some inherent limitations of
this radioligand. First, it has a relatively low signal-to-noise ratio due to its high nonspecific
binding and low brain permeability, and high plasma protein binding, all of which limit the
accurate quantification of TSPO in-vivo (Chauveau et al. 2008). Second, the short half-life of
carbon-11 restricts the use of the radiotracer to PET centres with an on-site cyclotron. The
limitations of [11C]-PK11195 have led to considerable efforts to develop improved
radioligands, including [11C]-PBR28, [18F]-PBR06, [11C]-DAA1106, [18F]-FEDAA1106
[11C]-DPA713, [18F]-PBR11, [18F]-FEMPA and [18F]-FEPPA.
These radioligands generally have higher affinity and brain uptake, and an improved
signal-to-noise ratio compared with [11C]-PK11195 (Table 1). However their binding affinity
is affected by a single nucleotide polymorphism (SNP) rs6971 in exon 4 of the TSPO gene,
which causes an alanine-to-threonine substitution (Owen et al. 2012). Based on this
polymorphism, individuals can be classified into one of the following three affinity patterns:
21
high-affinity binders (HABs), mixed-affinity binders (MABs) and low-affinity binders
(LABs). HABs and LABs express a single binding site for TSPO with either high or low
affinity, whereas MABs express approximately equal numbers of high and low binding sites
(Owen et al. 2012). The above mentioned polymorphism in TSPO affects binding of all
second-generation radioligands, and thus given the inter-subject variability that it causes, it is
imperative that participants are genotyped to allow for the accurate quantification of TSPO
availability. The differential affinity between genotypes depends on the radioligand, for some
radioligands the difference is very large, while for others it is small (Kreisl et al. 2013). The
proportion of the three affinity groups, or the prevalence of the major (Ala147) and minor
(Thr147) alleles varies across different ethnic groups. In the Hapmap database, the
prevalence of the Thr147 (low-binding allele) is 30% in Caucasians, 25% in Africans, 2% in
Han Chinese and 4% in Japanese (http://hapmap.ncbi.nlm.nih.gov/cgi-
perl/snp_details_phase3?name=rs6971&source=hapmap28_B36&tmpl=snp_details_phase3).
Another improvement to TSPO radioligands has been the development of fluorine-18
labelled radioligand including [18F]-FEDAA1106, [18F]-PBR11, [18F]-FEMPA and [18F]-
FEPPA. Fluorine-18 labelled radioligands have certain advantages over carbon-11 tracers,
such as a longer half-life (110 mins vs 20 min) and lower positron energy (650 keV vs 960
keV), which allow for longer storage and long-distance transportation and may give a higher
quality of images with greater spatial resolution (Zhang et al. 2003). A summary of some
TSPO radioligands that have been used in humans, along with their improvements and
limitations, is listed in Table 1.
22
Table 1. A summary of TSPO radioligands used in-vivo. Improvements of each
radiotracer in comparison to the prototypical radioligand, [11C]-PK11195, are listed, along
with the limitations of each one.
Radioligand Improvement/Strength Limitations
[11C]-PK11195 First and most widely used radiotracer High nonspecific binding, low
brain penetration, high plasma
protein binding, a difficult
synthesis and short half-life
[11C]-DAA1106 Higher affinity (10-fold), higher brain uptake,
higher specific to nonspecific binding,
showed greater contrast between lesioned and
non-lesioned areas
Affinity affected by rs6971
polymorphism, short half-life
[11C]-PBR28 Higher affinity (2-5 fold), higher brain
uptake, lower lipophilicity, higher specific to
nonspecific binding
Plasma radioactivity accounted
mainly by radiometabolite,
affinity affected by rs6971
polymorphism, short half-life
[18F]-FEDAA1106 Higher brain uptake, higher specific to
nonspecific binding, longer half-life
Very lipophilic, binding kinetics
are very slow, large variability in
data
[18F]-PBR06 Good brain penetration, appropriate kinetics,
higher specific to nonspecific binding, ease of
preparation, longer half-life
Suboptimal level of lipophilicity,
produces a brain-penetrant
radiolabeled metabolite, affinity
affected by rs6971
[18F]-FEPPA Rapid brain penetration, appropriate regional
distribution, 3-fold more potent than PBR28
and an order of magnitude more potent than
PK11195, lower metabolization than PBR28,
ease of preparation, lower lipophilicity,
longer half-life
Variability in data - affinity
affected by rs6971 polymorphism
23
1.5.2.3 [18F]-FEPPA as an imaging biomarker for neuroinflammation in-vivo
Although [11C]-PBR28 is a significant improvement to [11C]-PK11195, its short half-life
and radiometabolite, impelled researchers to develop a flurorine-18 radiolabeled analogue,
[18F]-FEPPA. The affinity of [18F]-FEPPA for PBRs was measured by competition with
[3H]-PK11195 in rat mitochondrial membranes. In addition, three other compounds with
affinity for PBR were tested for comparison purposes (Wilson et al. 2008). It was found that
[18F]-FEPPA was threefold more potent than [11C]-PBR28 and an order of magnitude more
potent than the prototypical radioligand, [11C]-PK11195. The extent of metabolism of [18F]-
FEPPA in comparison to [11C]-PBR28 was examined in rats. Following [11C]-PBR28
injection, 10-15% of radioactivity in the rat brain was due to radioactive metabolites,
whereas less than 5% of the radioactivity was due to [18F]-FEPPA metabolites (Wilson et al.
2008). The biodistribution of [18F]-FEPPA was also studied in rats, whereby it was found
that [18F]-FEPPA has a reasonable blood-brain barrier penetration. In pig, [18F]-FEPPA
showed rapid brain penetration with an appropriate regional distribution for binding to TSPO
(Bennacef et al. 2008). The first in-vivo human study demonstrated that the highest uptake
was in the thalamus followed by the cerebellum, temporal cortex, occipital cortex, frontal
cortex, putamen, caudate, insula and cingulate (Rusjan et al. 2011). As with other second-
generation radioligands, [18F]-FEPPA binding is affected by the single nucleotide (rs6971)
polymorphism in the TSPO gene. The variation in [18F]-FEPPA binding was investigated in
a sample of 19 healthy subjects, from which it was reported that the total distribution volume
of [18F]-FEPPA was 30% higher in HABs in comparison to MABs throughout the grey
matter (Mizrahi et al. 2012). Given the fact that [18F]-FEPPA has a high affinity for TSPO,
24
good brain penetration, and few radioactive metabolites, we chose to use this radioligand to
quantify neuroinflammation in-vivo.
1.5.2.4 PET imaging of mild cognitive impairment with TSPO radioligands
In-vivo AD studies have predominately demonstrated that there is an elevation in
TSPO binding, suggesting increased neuroinflammation (Table 2). However, the exact
timing of neuroinflammation is still unknown. In order to gain a better understanding, in-vivo
studies have been performed in individuals at earlier stages of cognitive impairment, such as
the MCI stage. Similarly to AD, different TSPO radioligands have been used to quantify
neuroinflammation in-vivo in MCI. Conflicting reports have been reported using the
protypical radioligand, [11C]-PK11195. Two groups demonstrated no differences in [11C]-
PK11195 binding between healthy volunteers (HV) and MCI (Wiley et al.
2009;Schuitemaker et al. 2013). In contrast, one study with 14 amnestic MCI patients found
increased [11C]-PK11195 uptake in 38% of the patients (Okello et al. 2009). Increased
binding was found in the anterior cingulate, posterior cingulate and frontal cortex, however
after correction for multiple comparisons, only the frontal [11C]-PK11195 binding remained
significantly elevated.
Second generation radioligands have been used to study neuroinflammation in MCI
populations as well (Yasuno et al. 2012;Kreisl et al. 2013).Significant differences in [11C]-
DAA1106 binding were obtained in the cerebellum, medial prefrontal cortex, parietal cortex,
lateral temporal cortex, anterior cingulate cortex and striatum (Yasuno et al. 2012).
Interestingly, individuals whose [11C]-DAA1106 binding was higher than the control mean ±
25
0.5 S.D developed dementia within 5 years. In contrast, no differences were found between
MCI patients and controls using [11C]-PBR28. This study was the first to control for the
effect of TSPO genotype (rs6971). Out of 10 MCI patients that were included, 4 were HABs
and 6 were MABs (Kreisl et al. 2013).
Given that immunohistochemistry studies have found activated microglial cells
clustered around neuritic plaques, PET studies sought to examine if this relationship can be
observed in-vivo using radioligands that target amyloid, such as [11C]-PIB, and TSPO
radioligands (Edison et al. 2008;Okello et al. 2009;Kreisl et al. 2013). Two studies with
[11C]-PIB and [11C]-PK11195 did not find any regional correlations between the two
radioligands in MCI and AD populations (Edison et al. 2008;Okello et al. 2009). However, a
more recent study using the second-generation radioligand, [11C]-PBR28, found a regional
correlation between the binding of the two radioligands in the inferior parietal lobule,
superior temporal cortex, precuneus, hippocampus, and parahippocampal gyrus. However,
these correlations were obtained only with partial volume corrected data (Kreisl et al. 2013).
Correlations of neuroinflammation with symptom severity and cognition have also
been investigated in PET studies. Earlier studies correlated binding with the most extensively
used scale measuring overall cognitive function, the Mini-Mental State Examination
(MMSE). Some studies found no correlations between the two in either AD and/or MCI
patients (Yasuno et al. 2008;Okello et al. 2009), while others found significant negative
associations (Edison et al. 2008;Yokokura et al. 2011). Two more recent studies performed a
battery of cognitive tests on AD and/or MCI populations (Kreisl et al. 2013;Suridjan et al.
2015). Using [11C]-PBR28 in 19 AD and 10 MCI patients, the researchers found a significant
correlation between binding and impaired performance on the MMSE, CDR, Logical
26
Memory Immediate, Block Design, and Trail Making part B tasks, with the strongest
correlations between [11C]-PBR28 binding in the inferior parietal lobule and CDR score and
performance on Block Design (Kreisl et al. 2013). A more recent study also reported a
correlation between binding and performance on visuospatial tasks. The researchers reported
strong negative correlations between [18F]-FEPPA binding in the parietal and prefrontal
cortices of AD patients and visuospatial performance, as assessed by the repeatable battery
for the assessments of neuropsychological status (RBANS) (Suridjan et al. 2015).
Overall, from the previous 5 TSPO PET studies (as summarized in Table 3) in this
population, it is not clear if neuroinflammation is present, as conflicting results have been
reported. There are certain methodological differences between these previous studies that
may have contributed to the conflicting results such as differences in MCI diagnosis criteria,
use of different TSPO radioligands, and differences in analysis of the results. For my thesis, I
sought to measure neuroinflammation with a novel TSPO radioligand, [18F]-FEPPA, in
amnestic MCI patients as this tracer has been shown to be sensitive enough to detect
neuroinflammation in AD and MDE (Setiawan et al. 2015;Suridjan et al. 2015). Furthermore,
[18F]-FEPPA has numerous improvements over the prototypical TSPO tracer, [11C]-PK11195
(as summarized in Table 1). Given that there is high intersubject variability in binding, only
high-affinity binders (HABs) were included in the study.
27
Table 2. A summary of PET studies measuring neuroinflammation in AD. Groups that
looked into correlations with amyloid, [11C]-PIB binding, and cognition have been specified.
Study Radioligand Sample Neuroinflammation Amyloid Cognition Cagnin et al., 2001
[11C]-PK11195 8 AD, 15 HV, 1 MCI ↑inferior and middle temporal gyri, posterior
cingulate gyrus, inferior parietal cortex
N/A N/A
Edison et al., 2008
[11C]-PK11195 13 AD, 14 HV ↑frontal, temporal, parietal and occipital cortices, anterior and posterior cingulate, striatum No difference in hippocampus
First to study relationship ÆNo correlations found
Negative correlation with MMSE score
Yasuno et al., 2008
[11C]-DAA1106 10 AD, 10 HV ↑dorsal and medial prefrontal, lateral temporal, parietal and occipital cortices, anterior cingulate cortex, striatum, cerebellum No difference in posterior cingulate, medial temporal and thalamus
N/A No correlations found
Wiley et al., 2009
[11C]-PK11195 6 AD, 6 MCI, 5 HV
No difference between diagnostic groups No correlations found N/A
Yokokura et al., 2011
[11C]-PK11195 11 AD, 11 HV ↑medial frontal, parietal, and left temporal cortices, anterior and posterior cingulate No difference in hippocampus
Negative correlation in posterior cingulate
Negative correlation with MMSE
Kreisl et al., 2013
[11C]-PBR28 19 AD, 10 MCI, 13 HV ↑prefrontal, temporal, inferior parietal, and
occipital cortices, posterior cingulate, hippocampus No difference in thalamus, striatum and cerebellum
Positive correlation in inferior parietal lobule, superior temporal cortex, precuneus, hippocampus and parahippocampal gyrus *only with PVEC
Negative correlation with MMSE, immediate memory, trail making and block design task
Varrone et al., 2013
[18F]-FEDAA1106
9 AD, 7 HV No difference between diagnostic groups N/A N/A
Suridjan et al., 2015
[18F]-FEPPA 18 AD, 21 HV ↑prefrontal, temporal, parietal, and occipital cortices, hippocampus, posterior limb of the internal capsule, cingulum bundle
N/A Negative correlation with visuospatial function.
Varrone et al., 2015
[18F]-FEMPA 10 AD, 7 HV ↑medial and lateral temporal cortices, posterior cingulate, caudate, putamen, thalamus, cerebellum
N/A No correlation found
28
Table 3. A summary of PET studies measuring neuroinflammation in MCI. Groups that
looked into correlations with amyloid, [11C]-PIB binding, and cognition have been specified.
The limitations of each study are listed.
Study Radioligand MCI sample
Neuroinflammation Amyloid Cognition Limitation
Wiley et al., 2009
[11C]-PK11195 6 MCI No difference No correlation found
N/A Sample size, MCI participants not characterized by subtype, radioligand, used ratio method for analysis
Okello et al., 2009
[11C]-PK11195 14aMCI ↑anterior and posterior cingulate, frontal cortex
No correlation found
No correlation found
Radioligand, used a simplified reference tissue model (SRTM) to generate parametric images, cerebellum used as reference region, did not perform PVEC
Schuitemaker et al., 2013
[11C]-PK11195 10aMCI No difference N/A No correlation found
Radioligand, used receptor parametric mapping to generate parametric images,
Yasuno et al., 2012
[11C]-DAA1106 5aMCI, 2naMCI ↑medial prefrontal,
parietal, lateral temporal cortices, anterior cingulate, striatum, cerebellum
N/A N/A Sample size, did not genotype participants
Kreisl et al., 2013
[11C]-PBR28 10aMCI; 4HAB, 6MAB
No difference Correlation in inferior parietal, superior temporal, precuneus, hippocampus, and parahippampal gyrus *only with PVEC
Negative correlation with MMSE, immediate memory, block design; positive correlation with CDR, trail making task B
Sample size, for correlations with cognition AD and aMCI patients were grouped
29
1.6 Oxidative stress
Oxidative stress occurs from normal cellular and metabolic activities. However an
unconstrained generation of oxidative species is toxic and thought to be an important
pathogenic factor in schizophrenia, AD, Parkinson’s disease and amyotrophic lateral
sclerosis (Mandal et al. 2012). In particular, in AD it has been suggested that oxidative stress
plays a major role in the pathogenesis and progression of this disease.
1.6.1 Glutathione: The most abundant brain antioxidant
Although the mechanisms underpinning the increase in oxidative stress are unclear,
the antioxidant system is likely to be relevant, particularly glutathione (GSH), the brain’s
major antioxidant (Duffy et al. 2014). GSH is synthesized de novo in the brain and the supply
of GSH from other organs to the brain is restricted (Mandal et al. 2012). GSH plays an
important role in protecting the brain from oxidative damage induced by reactive oxygen
species (ROS). GSH is involved in a number of other essential tasks including DNA
synthesis and repair, protein synthesis, amino acid transport, enhancement of immune
function and enzyme activation (Bermejo et al. 2008). GSH protects cells from ROS damage
both non-enzymatically and enzymatically. For example, GSH reacts with the oxidant
hydrogen peroxide (H2O2) catalyzed by glutathione peroxidase (GPx) and coverts it to H2O
(Bermejo et al. 2008). During this process, GSH is oxidized to glutathione disulphide
(GSSG). The cycle is continued with GSSG being reduced back to GSH by glutathione
reductase (GR). Under physiological conditions, the levels of the reduced form of glutathione
30
are 10-100 fold higher than the oxidized form. Interestingly, the distribution of GSH varies
by neuroanatomical areas (Venkateshappa et al. 2012). During oxidative stress, the ratio of
GSH/GSSG tends to be slightly reduced; nevertheless the cells are able to maintain their
glutathione redox state through different mechanisms (Bermejo et al. 2008;Ansari and Scheff
2010). However, when oxidative stress becomes prolonged, the cellular systems are no
longer able to counteract the ROS-mediated insults, leading to irreversible cell degeneration
and death. The reduction in GSH can be used as a good measure or indicator of oxidative
stress of an organism.
Figure 2. The glutathione reduction-oxidation cycle. Under normal physiological
conditions, GSH converts H2O2 to H2O. In this condition, levels of GSH are 10-100 fold
higher than the oxidized form, GSSG. During oxidative stress, this cycle maintains the redox
state of the cell. However, during pathological conditions such as AD, when oxidative stress
becomes prolonged, this balance can no longer be maintained and the levels of GSSG are
increased. The reduction of the GSH:GSSG ratio can be used as an indicator of the overall
oxidative stress of the animal.
31
1.6.2 Oxidative stress and glutathione in AD and MCI
Oxidative stress has been shown to play a role in the pathogenesis of AD (Bermejo et
al. 2008). Post-mortem studies have shown oxidative modifications of DNA, RNA, lipids
and proteins in the brains of AD patients (Smith et al. 1991;Mecocci et al. 1994;Lovell et al.
1995;Good et al. 1996;Aksenov et al. 2001;Pamplona et al. 2005;Ansari and Scheff 2010).
Peripheral markers of oxidative stress have been studied in AD and MCI patients, in order to
understand the biochemical alterations in these stages. A commonly used marker of protein
oxidation are carbonyls, as these moieties are chemically stable, which is useful for their
detection and storage. Several studies have reported an increase in carbonyl groups in the
plasma of AD and MCI patients (Conrad et al. 2000;Choi et al. 2002;Bermejo et al. 2008).
One group that compared AD, MCI and healthy volunteers, found that the increase in the
level of carbonyls corresponded to the diagnostic group, meaning that AD patients had the
highest increase whereas MCI patients had an intermediate increase (Bermejo et al. 2008). In
addition to protein oxidation, markers of lipid oxidation have also been studied as indicators
of overall oxidative stress. Isoprostanes have been measured in the urine, blood and
cerebrospinal fluid of AD and MCI patients, with AD patients having the largest increase
(Tuppo et al. 2001;Pratico et al. 2002). However the data is not that concrete, as some groups
have reported no differences between patients and healthy volunteers (Montine et al. 2002).
In addition to the oxidation of proteins and lipids, GSH levels have been widely
studied in AD. Animal studies with 3xTg-AD mouse models have demonstrated that GSH
levels are reduced both in-vitro and in-vivo (Ghosh et al. 2012;Ghosh et al. 2014).
Reductions in GSH levels have also been reported in post-mortem human brain slices
32
(Ramassamy et al. 2000;Venkateshappa et al. 2012) and in plasma samples of AD patients
(Bermejo et al. 2008;Puertas et al. 2012). In addition to GSH levels, reductions in the
GSH:GSSG ratio have also been measured (Bermejo et al. 2008;Cristalli et al. 2012).
However, because peripheral changes do not necessarily align with changes in the brain, a
direct association between AD and GSH levels has been elusive due to the limited amount of
in-vivo human studies (Mandal et al. 2015).
1.6.3 In-vivo quantification of GSH
1.6.3.1 Magnetic resonance spectroscopy
Magnetic resonance spectroscopy (MRS) is a non-invasive in-vivo method used to
measure key metabolites in the brain such as N-acetyl aspartate, glutamate (Glu), glutamine,
myo-inositol, choline, creatine (Cr) and glutathione (GSH). It provides a diagnostic tool for
the biochemical characterization of pathophysiological processes in the brain (Gujar et al.
2005). The quantification of these metabolites has been particularly useful in the study of
neurodegenerative disorders. Each metabolite can provide information about the underlying
degenerative process as metabolite levels are sensitive to different in-vivo pathological
processes at the molecular or cellular level (Marjanska et al. 2005). For example, N-acetyl
aspartate is proposed to be a putative marker of neuronal density (Metastasio et al. 2006))
and myo-inositol is thought to be a marker for osmotic stress or astrogliosis (Gujar et al.
2005).
33
MRS is based on the chemical shift properties of a molecule. A variety of nuclei such
as carbon (13C), nitrogen (15N), fluorine (19F), sodium (23Na), phosphorus (31P) and hydrogen
(1H) can be used for the quantifications of metabolites in-vivo. However, only 31P and 1H
exist in high enough concentrations for useful clinical evaluation. 1H-MRS studies have
particularly become of interest due to the natural abundance of protons and their high
absolute sensitivity to magnetic manipulation, better spatial resolution, and relative
simplicity of technique (Gujar et al. 2005;van der Graaf 2010).
1.6.3.2 Measurement of GSH using MRS in AD and MCI
Due to the limited number of in-vivo human studies, there is a lack of understanding
of the impact of GSH reductions in AD (Mandal et al. 2015). Until recently, the
quantification of GSH using MRS has been elusive due to the low levels of brain GSH in
comparison to other brain metabolites. The first GSH MRS study with 45 young and 15 old
healthy volunteers, 11 MCI patients and 14 AD patients, found that GSH levels varied based
on the brain region, gender, age and diagnosis of the individual (Mandal et al. 2012). The
highest levels of GSH were reported in the parietal cortex compared to other brain regions.
When comparing the 4 groups, a trend in GSH levels was reported: young healthy > old
healthy > MCI > AD. However, a significant difference was obtained only between healthy
volunteers and AD patients. Subsequent larger MCI studies have found conflicting results
(Duffy et al. 2014;Mandal et al. 2015). One group found a significant elevation of GSH in
the anterior and posterior cingulate of patients in comparison to healthy volunteers (Duffy et
al. 2014). Furthermore, higher levels of GSH in the anterior cingulate were related to
34
impairments on tests of executive function and elevated GSH in the posterior cingulate was
associated with poorer memory consolidation. The most recent study measured GSH levels
in the hippocampus and frontal cortex of AD and MCI patients, reductions were reported in
both regions of AD patients whereas MCI patients were reported to have a reduction in the
hippocampus only (Mandal et al. 2015). Overall, GSH in MCI is still not well understood
and further investigations of GSH are required to gain a better understanding of the
underlying pathology.
35
Table 4. A summary of GSH MRS studies in AD and MCI populations.
Study Sample Regions Results Limitation
Mandal et al., 2012
11MCI 14AD 45 young HV 15 old HV
Frontal cortex Significant difference between young female HV and female AD patients in right frontal cortex Significant difference Between young male HV and male AD patients in left frontal cortex No difference between HV and MCI Trend in GSH levels Young HV>old HV>MCI>AD
Small sample size of patients
Duffy et al., 2014
54 MCI 42 HV
Anterior and posterior cingulate
MCI patients had significantly elevated GSH in both regions
MCI group includes those with naMCI and aMCI, used single-voxel PRESS sequence instead of MEGA-PRESS, cingulate is not as homogenous of a region as frontal cortex
Mandal et al., 2015
22MCI 21AD 28HV
Frontal cortex and hippocampus
AD patients had significantly reduced GSH levels in both regions MCI patients had significantly reduced GSH levels in hippocampus only
Hippocampal voxel contained some non-hippocampal tissue, frontal voxel contained a substantial fraction of WM
36
2. AIMS AND HYPOTHESES
2.1 Aims
Neuroinflammation, as reviewed earlier, has been shown to play a role in the
pathology of AD. However, the exact timing and progression of this process is still unknown.
We sought to investigate whether neuroinflammation was present at an earlier stage of
cognitive impairment. Since individuals with early signs of cognitive impairment may
progress to different types of dementia, it was important to include patients who have an
underlying AD pathology. The amnestic subtype of MCI has been referred to as a prodromal
stage of AD, thus we pursued to measure neuroinflammation in these at risk individuals.
With the use of PET imaging and a novel TSPO radioligand [18F]-FEPPA, we investigated
whether neuroinflammation was elevated in patients compared to healthy volunteers.
Regions implicated with AD, such as the temporal cortex, inferior parietal cortex, temporal
cortex, occipital cortex, and hippocampus, were chosen to be investigated. As mentioned, the
binding of second-generation radioligands, such as [18F]-FEPPA, is affected by a single
nucleotide rs6971 polymorphism in the TSPO gene. In order to reduce the variability caused
by the differential binding, only HABs were included in the study. Thus far, PET studies
have reported conflicting results about whether neuroinflammation is present in the aMCI
stage. In order to gain a better understanding of the pathology, this study has several
novelties, including the use of a High-Resolution Research Tomograph (HRRT) scanner and
a novel second-generation radioligand, [18F]-FEPPA. In addition, this is the first study to
investigate neuroinflammation in a purely amnestic and HAB population of patients.
37
Amyloid deposition, one of the hallmarks of AD, has been shown to be present in the
brains of aMCI patients as well. As reviewed earlier, several lines of evidence have
demonstrated a relationship between neuroinflammation and amyloid burden. However, the
in-vivo spatial relationship between the two processes still has not been well understood. Our
aim was to quantify amyloid burden in-vivo with the use of a validated radiotracer, [11C]-
PIB, and to explore whether there is a regional association between [18F]-FEPPA and [11C]-
PIB in aMCI patients.
Lastly, oxidative stress has been suggested to play a role in the pathogenesis and
progression of AD. The antioxidant system is particularly relevant when studying oxidative
stress, as disruptions in the balance of the oxidized to reduced form of GSH can be used as
an indicator of the oxidative state of an organism. The quantification of GSH with the use of
MRS has only recently become possible. We sought to explore whether there is an alteration
in GSH levels in the left dorsolateral prefrontal cortex (DLPFC) of aMCI patients compared
to controls. This region was chosen based on previous evidence demonstrating its role in AD.
Additionally, it has a good signal-to-noise ratio for MRS quantification. Given that the
DLPFC is an outcome measure for our PET data, correlations between [18F]-FEPPA, [11C]-
PIB and GSH were explored. This is the first study to quantify GSH in the left DLPFC of
aMCI patients, and to explore correlations with neuroinflammation and amyloid burden.
38
2.2 Hypotheses Primary hypothesis: [18F]-FEPPA uptake will be greater in aMCI patients compared with age-matched HV.
Exploratory hypotheses: i) Regional [18F]-FEPPA uptake will show a positive association with [11C]-PIB
uptake. ii) GSH levels in the L DLPFC will be higher in aMCI patients compared with aged-
matched HV.
39
3. METHODS
3.1 Participants
Individuals between the ages of 45 – 85 years old were recruited to participate in the
study. Participants with aMCI were recruited from the memory clinics at Baycrest Hospital
and the Centre for Addiction and Mental Health (CAMH), Toronto, Ontario. Healthy
volunteers were recruited from the Baycrest Research Participant Database and from local
advertisements. aMCI patients were diagnosed according to the Peterson et al. criteria for
aMCI (Petersen 2004). Briefly, the primary distinction between aMCI and HV participants is
in the area of memory, while other cognitive functions are comparable. The diagnosis of
aMCI was made on a clinical basis, established by a consensus committee comprising of a
neurologist, geriatric psychiatrists, neuropsychologist and other personnel working at the
memory clinics. Over 300 individuals were contacted for participation in the study, 83
individuals were interested and were screened over the phone. A total of 47 individuals met
criteria and were invited for the first visit whereby informed consent was obtained.
Additionally, blood samples were collected for genotyping of TSPO rs6971 polymorphism;
only HABs of both groups were invited to proceed with the study.
Eleven MCI patients and 14 healthy volunteers completed all study procedures: a
[18F]-FEPPA scan, [11C]-PIB scan and an MRI scan. All participants underwent a medical
and psychiatric assessment and a battery of neuropsychological tests. A urine drug screen
was also performed. The exclusion criteria for aMCI and healthy controls included: a current
Axis I disorder, history of closed head injuries with loss of consciousness, strokes, or other
40
neurological disorders with central nervous system involvement. The protocol was approved
by the Research Ethics Boards of CAMH and Baycrest Health Sciences.
3.2 Neuropsychological assessments
During the baseline visit, several neuropsychological assessments were performed on
aMCI and healthy control participants. To evaluate the overall cognitive performance of
participants, the Mini-Mental State Examination (MMSE) was performed (Folstein et al.
1975). Since impairment in episodic memory is most commonly seen in MCI patients that
progress to AD, a variety of episodic memory tests that assess both immediate and delayed
memory recall were performed, including the Logical Memory II subscale from the Wechsler
Memory Scale-Revised (Wechsler 1987) and the Repeatable Battery for the Assessment of
Neuropsychological Status (RBANS) (Randolph et al. 1998). Given that other cognitive
domains may be affected in aMCI, additional tests were performed to assess language,
attention, executive function and visuospatial performance including RBANS, Montreal
Cognitive Assessment (MoCA) (Nasreddine et al. 2005), verbal fluency (Monsch et al.
1992), letter number span (LNS) (Wechsler et al. 2008), Stroop test (Spieler et al. 1996) and
trail-making test (Ashendorf et al. 2008). Premorbid intelligence was also assessed with the
North American Reading Test (NART) (Grober et al. 1991).
41
3.3 PET measures of neuroinflammation and β-amyloid
plaques
The radiosynthesis of [18F]-FEPPA (Wilson et al. 2008) and [11C]-PIB (Mathis et al.
2002) have been described in detail elsewhere, and are currently synthesized at our CAMH-
Research Imaging Centre (CAMH-RIC).
3.3.1 PET image acquisition
PET images were obtained using a 3D High-Resolution Research Tomograph
(HRRT) scanner (CS/Siemens, Knoxvile, TN, USA). Prior to the start of the PET scans, a
custom fitted thermoplastic mask was made for each participant to minimize head
movement.
For [18F]-FEPPA, an intravenous saline solution of 4.91±0.42 mCi was administered
as a 1 minute bolus into the antecubital vein followed by 10 mL of saline. The scan duration
was 125 minutes following injection of the radiotracer. Blood samples were taken throughout
the scan to generate an input function for kinetic analysis (Rusjan et al. 2011). Since there is
no region in the brain that is completely void of TSPO binding, the plasma is used as a
reference. In order to measure radioactivity levels in the plasma, automatic and manual blood
samples were obtained. An automated blood sampling system (ABSS, Model #PBS-101
from Veenstra Instruments, Netherlands) was used to measure arterial blood radioactivity
continuously at a rate of 2.5 mL/minute for the first 22.5 minutes of the PET scan. Manual
arterial samples were obtained at 2.5, 7, 12, 15, 30, 45, 60, 90, and 120 minutes after the
injection of the radiotracer. From the blood samples, the following were determined: the
42
amount of radioactivity in the whole blood, the blood-to-plasma ratio and the amount of
parent radioligand and metabolites. The blood curve was then divided by the bi-exponential
fitting of the blood-to-plasma ratio, and multiplied by the percentage of parent radiotracer to
generate a curve of the amount of parent compound in the plasma, which was then used as
the input function for the kinetic analysis. The images were reconstructed into a series of 34
time frames including 1 frame of variable length, followed by frames comprising 5×30
seconds, 1×45 seconds, 2×60 seconds, 1×90 seconds, 1×120 seconds, 1×210 seconds and
22×300 seconds.
For [11C]-PIB, an intravenous saline solution of 9.64±0.89 mCi was administered as
a 1 min bolus into the antecubital vein followed by 10 mL of saline. The [11C]-PIB scan was
90 minutes long following injection of the radiotracer. In contrast to the [18F]-FEPPA scan,
blood samples were not obtained. Images were reconstructed into a series of 34 frames
comprising 4×15 seconds, 8×30 seconds, 9×60 seconds, 2×180 seconds, 8×300 seconds and
3×600 seconds.
3.3.2 Regions of Interest (ROI)-based PET image analysis
A fully automated brain parcelation program, Regions of Mental Interest (ROMI),
was used to delineate regions of interests (ROIs) on each participant’s MRI. The ROIs that
were chosen are regions of the brain that are associated with early emergence of plaques in
AD and higher density of microglial activation in aMCI patients and mild AD cases. The
specific regions were the prefrontal cortex, temporal cortex, inferior parietal cortex, occipital
43
cortex and hippocampus. The reliability in the delineation of ROIs using ROMI was
previously assessed, a brief description of the methods is as follows: (1) a proton density
(PD) MRI template in Montreal Neurologic Institute/International Consortium for Brain
Mapping (MNI/ICBM) with predefined ROIs was transformed to match the participant’s MR
image, (2) the ROIs were refined based on the grey matter probability of voxels in the
individual MR image, (3) the individual MR image was co-registered to the average PET
image, in order to transform the refined ROIs to the PET image space and to generate a time-
activity curve (TAC) which shows the amount of radioactivity in each ROI (Rusjan et al.
2006). Due to the potential confounding effects of age-associated changes in brain volume, a
partial volume error correction (PVEC) was applied to the time activity data of all
participants for both radioligands using the Müller-Gärtner algorithm (Müller-Gärtner et al.
1992).
44
Figure 3. Step-by-step procedures for automatic delineation of ROIs using ROMI. The
same steps are applied to [18F]-FEPPA and [11C]-PIB PET data to generate a TAC file.
45
3.3.2.1 Measurement of TSPO expression with [18F]-FEPPA
To obtain a quantitative measure of [18F]-FEPPA uptake, the data was measured
using a full kinetic model. [18F]-FEPPA TACs were analyzed with a 2-tissue compartment
model (2TCM) and plasma input function using PMOD software (PMOD, Technologies,
Zurich, Switzerland). A previous modeling study of [18F]-FEPPA demonstrated that the
2TCM provides a better fit of the data than a 1TCM (Rusjan et al. 2011). A 2TCM consists
of 3 compartments: the arterial plasma compartment (CP), the non-displaceable compartment
(CND), and the specific compartment (CS) (Innis et al. 2007) . The CND is composed of
radioligand that is free or non-specifically bound to other molecules. In contrast, CS is
composed of radioligand that is bound to a specific target. The kinetic behaviour of
radioligands in compartments can be described with rate constants. K1 describes the transfer
of radioligand from plasma to the non-displaceable compartment, whereas k2 describes the
transfer in the reverse direction. k3 describes the transfer of radioligand from the non-
displaceable compartment to the specific compartment, and k4 describes this transfer in the
reverse direction.
Figure 4. Two-tissue compartment model (Innis et al. 2007). The boxes indicate the
different compartments: plasma (P), non-displaceable which is free (F) + non-specific (NS)
and finally the specific compartment (S).
46
The kinetic modeling of [18F]-FEPPA also demonstrated that the most appropriate
binding outcome was total distribution volume (VT) (Rusjan et al. 2011). VT is equal to the
ratio at equilibrium of the concentration of radioligand in tissue to that in plasma. In
comparison to other outcome measures, such as distribution volume of the specific
compartment (VS) and the binding potential (BPND), VT demonstrated a better level of
identifiability as indicated by the lowest coefficient of variation (COV). Thus [18F]-FEPPA
VT was estimated in selected ROIs such as the prefrontal cortex, temporal cortex, inferior
parietal cortex, occipital cortex, and hippocampus.
In addition to VT, [18F]-FEPPA uptake was measured with a supplementary method
that does not involve the arterial input function. Since there is no true reference region in the
brain (e.g. a region devoid of TSPO), TSPO density is measured in relation to the
concentration of radioligand in arterial blood. This requires arterial catheterization of the
participant, making the scans invasive and more difficult for some participants. Recently, a
group compared the absolute quantification method to the use of the cerebellum as a pseudo-
reference region in a population of AD and MCI patients, and healthy volunteers (Lyoo et al.
2015). The group calculated the standardized uptake volume ratio (SUVR) of [11C]-PBR28
by dividing the SUV between 60-90 minutes of a ROI by cerebellar SUV between 60-90
minutes. They found that the SUVR method correlated with the absolute quantification
(2TCM) and suggested that it may even be more sensitive. Thus, we explored the use of
SUVR as a measure of [18F]-FEPPA uptake in our ROIs. Since our [18F]-FEPPA scan is
longer than the [11C]-PBR28 scan, SUVR was calculated by dividing the SUV between 90-
120 minutes of a ROI by the cerebellar SUV between 90-120 minutes.
47
In summary, VT and SUVR were used as outcome measures for [18F]-FEPPA binding
in the present thesis. Correlations with [11C]-PIB and all neuropsychological assessments
were explored to evaluate if relationships between neuroinflammation and amyloid burden,
and neuroinflammation and cognition are present.
3.3.2.2 Measurement of amyloid plaques with [11C]-PIB
[11C]-PIB retention was measured using the Logan Graphical analysis (Logan et al.
1996). This method has been validated to be the method of choice for [11C]-PIB PET data as
it provides stable and robust ROI results (Lopresti et al. 2005;Price et al. 2005). Regional
distribution volume values were normalized to the reference region (cerebellum) to yield
distribution volume ratio (DVR). The cerebellum has been confirmed to be a valid reference
region for [11C]-PIB, as it is relatively free of fibrillary plaques in AD. Additionally,
participants were classified by either the presence of amyloid “PIB+” or absence of amyloid
“PIB-”. The average DVR of the prefrontal, temporal, inferior parietal and occipital cortices
was calculated, and a cutoff of 1.20 was applied to stratify participants according to PIB
status (Villeneuve et al. 2015).
48
3.3.3 Voxel-based PET image analysis
Parametric images of [18F]-FEPPA VT were generated using the Logan graphical
analysis method by applying a wavelet-based kinetic modeling approach (Banati et al. 2000).
Differences between diagnostic groups were tested using the two-sample t-test in statistical
parametric mapping (SPM8). Significant clusters were thresholded at p<0.001. The family-
wise error rate method was used to correct for multiple comparisons.
3.4 MRI acquisition
3.4.1 Structural images
A 3-Tesla General Electric MR-750 scanner was used for the acquisition of MRI
sequences. PD-weighted images (slice thickness=2.0mm, repetition time=6000ms, echo
time=MinFull, acquisition matrix=256x192, FOV=22cm, Total scan time=2 min 12 sec)
were used for co-registration.. Other structural images such as T1, T2, and FLAIR were also
acquired.
49
3.4.2 Spectroscopy
For the MRS portion of the scan, a voxel was placed in the LDLPFC for
quantification of GSH.
Figure 5. Voxel placement in the left dorsolateral prefrontal cortex. Voxel size 24 cc
(30×20×40 mm3).
As mentioned earlier, GSH is a tripeptide made up of glutamate, cysteine and glycine.
The protons of each of these moieties resonate differently under a magnetic field and are
overlapped by other major brain metabolites (Matsuzawa and Hashimoto 2011). In order to
obtain an uncontaminated GSH resonance, a spectral editing technique is used during
acquisition. Our 1H-MRS data of GSH was acquired using the interleaved J-difference
editing approach, MEshcher-Garwood Point RESolved Spectroscopy (MEGA-PRESS)
(Mescher et al. 1998). This pulse sequence was used as other sequences such as PRESS and
STEAM, cannot differentiate GSH from overlapping brain metabolite as well as MEGA-
PRESS (Mandal et al., 2012). MEGA-PRESS uses additional “selective 180°” editing pulses
to detect GSH, whereby alternating “on” and “off” editing pulses that target the α proton of
the cysteine moiety are applied. During the on pulses, the α-H of the cysteine amino acid
residue is inverted, whereas during the off pulses the selective 180° pulses are not applied.
50
The subtraction of the two scans results in the exclusive detection of the cysteinyl β-
CH2GSH signal. Total experimental time using MEGA-PRESS is 13 minutes and 24
seconds.
Following acquisition, GSH raw data was separated into two spectrums, one in which
water is suppressed and one in which water is unsuppressed. Afterwards, the data from the
combined GSH spectrums and water data was further processed and refined. Water data was
collected as it was used as a reference for GSH, due to the fact that water levels are assumed
to be unchanged in disease. Thus GSH levels are reported as a ratio of GSH to water in the
left DLPFC.
3.5 Genotyping
As previously mentioned, a polymorphism in the TSPO gene affects the binding of
second generation radioligands. In order to correctly quantify the PET data, blood samples
were collected for genetic analysis during the screening visit. Genomic DNA was obtained
from peripheral leukocytes using high salt extraction methods. The TSPO polymorphism
rs6971 was genotyped using a TaqMan assay on demand C_2512465_20
(AppliedBiosystems, CA, USA). The allele T147 was linked to Vic and the allele A147 was
linked to FAM. PCR reactions were performed in a 96-well microtiter-plate on a GeneAmp
PCR System 9700 (Applied Biosystems, CA, USA). After PCR amplification, endpoint plate
read and allele calling was performed using an ABI 7900 HT (Applied Biosystems, CA,
USA) and the corresponding SDS software (v2.2.2). The genotyping was performed at the
51
Neurogenetics Lab at CAMH. Following the results of the genetic test, only HABs were
invited for subsequent visits.
3.6 Statistical Analysis
Statistical analyses were performed using SPSS Statistics (version 22.0, IBM,
Armonk, NY, USA). Demographic variables and [18F]-FEPPA and [11C]-PIB injected
parameters were compared between aMCI and HV using an independent sample t-test.
Regional differences in [18F]-FEPPA and [11C]-PIB binding between the two groups were
compared using an independent samples t-test. Relationships between [18F]-FEPPA, [11C]-
PIB and cognitive scales were explored with non-parametric correlations. GSH levels in
aMCI and healthy participants were compared with an independent samples t-test.
Spearman’s rank correlation coefficient was used to assess the correlation between GSH
levels, [11C]-PIB/[18F]-FEPPA and cognition.
52
4. RESULTS
4.1 Neuroinflammation and beta-amyloid
4.1.1 Participants
Eleven aMCI participants (age range: 60-79 years) and 14 healthy volunteers (age
range: 56-78 years) completed all study procedures. All participants were high-affinity
binders. Descriptive characteristics and PET parameters are presented in Table 5. Diagnostic
groups were matched with regard to gender, age and education. In comparison to healthy
volunteers, aMCI participants had more cardiovascular risk factors such as high blood
pressure and elevated blood cholesterol level. Four aMCI participants were taking anti-
hypertensive drugs and 3 were on statins, whereas only 1 healthy volunteer was taking either
of these medications. Six of the 11 aMCI participants were taking anti-depressants, however
none met criteria for current diagnosis of major depressive disorder. aMCI participants
demonstrated an impairment in overall cognition, immediate and delayed memory,
visuospatial skills and executive function. None of the participants had any history of strokes
or other neurological disorders with central nervous system involvement.
53
a. Sample size varies for cognitive scales. All 11 aMCI participants completed the MMSE, MoCA, Verbal Fluency, TMT and Logical Memory, the remaining cognitive scales were completed by 7 aMCI participants. For HV, MoCA scores are available for 14 participants; MMSE for 10; RBANS, Verbal Fluency and Trail Making for 9; Stroop for 8; Letter Number Span, Logical Memory and NART for 7. *significance is flagged for easy reference
Table 5. Participant demographics and PET parameters (mean±SD).
Descriptives HV (n=14) aMCI (n=11) Age 67.07 ± 6.49 71.91 ± 5.30 t=1.588, p=0.126 Gender Female 9 6 Χ2=0.244, p=0.622
Male 5 5 Education (years) 16.21 ± 3.40 15.82 ± 2.36 t=-0.329, p=0.746 PET Parameters [18F]-FEPPA Amount Injected (mCi) 4.99 ± 0.25 4.81 ± 0.56 t=-1.041, p=0.309
Specific Activity (mCi/μmol) 2868.56 ± 2120.49 1960.49 ± 1481.40 t=-1.205, p=0.240
Mass Injected (μg) 1.22 ± 1.16 1.42 ± 0.83 t=0.481, p=0.635 [11C]-PIB Amount Injected (mCi) 9.47 ± 1.04 9.87 ± 0.66 t=1.176, p=0.252
Specific Activity (mCi/μmol) 1892.86 ± 1029.12 2135.01 ± 1122.85 t=0.561, p=0.580
Mass Injected (μg) 1.90 ± 1.07 1.43 ± 0.59 t=-1.304, p=0.205 Medications Cholinesterase inhibitors 0 2 Anti-hypertensives 1 5 Statins 1 3 Anti-depressants 0 6 Baby aspirin 1 1 Cognitiona MMSE 29.3 ± 1.25 27.3 ± 1.95* t=-2.797, p=0.011 RBANS Immediate Memory 106.6 ±13.6 77.4 ± 14.4* t=-4.148, p=0.001
Visuospatial 105.2 ± 18.9 84.9 ± 16.4* t=-2.258, p=0.04
Language 98.9 ±15.2 83.1 ± 23.1 t=-1.646, p=0.122
Attention 110.8 ±16.8 94.6 ±19.6 t=-1.781, p=0.097
Delayed Memory 101.7 ± 9.5 58.1 ± 17.0* t=-0.653, p<0.001
Total 106.3 ± 12.4 74.1 ± 15.9* t=-4.567, p<0.001 MoCA 27.0 ± 1.4 21.7 ± 3.6* t=-4.630, p=0.001 Letter Number Span 15.3 ± 3.0 11.9 ± 2.7* t=-2.241, p=0.045 Verbal Fluency 47.0 ± 13.7 38.0 ± 13.5 t=-1.459, p=0.162 Trail Making Test Part A (seconds) 44.3 ± 15.0 60.0 ± 32.8 t=1.413, p=0.179
Part B (seconds) 87.8 ± 35.1 157.5 ± 72.8* t=2.800, p=0.013 Stroop Task Color Score 110.8 ± 3.5 110.3 ± 4.6 t=-0.245, p=0.810
Color Word Score 85.9 ± 29.0 56.8 ± 23.7* t=-2.199, p=0.045 Logical Memory Delayed Memory 14.0 ± 2.7 8.8 ± 5.5* t=-2.326, p=0.033 NART 120.6 ± 7.0 100.0 ± 44.7 t=-1.204, p=0.252
54
4.1.2 Increased [11C]-PIB binding in the GM of aMCI patients
Participants with aMCI had significantly more amyloid than healthy volunteers in all
cortical regions of interest (Figure 6). The largest difference between groups was found in
the prefrontal cortex (65%). The order of increased amyloid burden was as follows:
prefrontal cortex>temporal cortex> inferior parietal cortex>occipital cortex>hippocampus.
No significant differences between groups were observed in the hippocampus. The results
survived after correction for partial volume effects.
55
Figure 6. [11C]-PIB Distribution Value Ratios (DVR) in regions of interest. Results are
shown before (A) and after (B) partial volume correction. aMCI participants (n=11) had
higher [11C]-PIB binding in all regions except hippocampus. The cerebellum was used as a
reference region. *p≤0.05, **p<0.01
(A)
45% 65% 37% 27% 6%
* ** * *
* ** * *
69% 105% 62% 47% 13% (B)
56
4.1.3 [18F]-FEPPA in GM of aMCI and healthy volunteers
No significant differences in [18F]-FEPPA binding were observed between aMCI
participants and healthy volunteers in any of the regions of interest (Figure 7). Percent
differences between the two groups were very low, -2.0% to 3.1%, with the exception of the
hippocampus in which a difference of 15.3% was measured. Two HV participants were
observed to have very high VT values, when these two individuals were removed from the
analysis, no significant differences between the two groups was observed. The results
remained after correcting for partial volume effects. Similarly, the supplementary SUVR
method and the parametric analysis did not yield any significant group differences (Table 6,
Figure 8).
After characterizing participants as either PIB+ or PIB-, comparisons between and
within groups were performed (Table 7, Figure 9). Within the healthy volunteers, 11
participants were classified as PIB- and 3 as PIB+. Although not significant, the average
[18F]-FEPPA VT was higher in PIB- healthy volunteers in comparison to PIB+ healthy
volunteers. Within the aMCI group, 8 participants were classified as PIB+ and 3 as PIB-.
Between the two aMCI groups, participants with amyloid pathology had significantly higher
[18F]-FEPPA binding in the temporal, prefrontal, inferior parietal and occipital cortices.
With partial volume correction, [18F]-FEPPA VT was elevated only in the temporal cortex of
PIB+ aMCI participants in comparison to PIB- aMCI participants. Lastly, when comparing
PIB- healthy volunteers and PIB+ aMCI participants, no significant differences were
obtained.
57
Figure 7. [18F]-FEPPA Total Volume Distribution (VT) in regions of interest. No
significant differences were observed between aMCI participants (n=11) and healthy
volunteers (n=14). Results are shown before (A) and after (B) partial volume correction.
(A)
-2.0% 1.7% -0.8% 3.1% 15.3%
(B) 0.5% -0.3% 1.3% 1.8% 20.2%
58
Table 6. [18F]-FEPPA Standardized Uptake Value Ratios (SUVR). The standard uptake
value between 90-120 minutes in an ROI was normalized to the standard uptake value of the
cerebellum between 90-120 minutes. The ratio for each ROI, along with the percent
difference is shown.
A. Before partial volume correction
Grey matter ROI HV (n=14) aMCI (n=11)
mean SD mean SD % Diff t p Temporal Cortex 0.987 0.080 0.968 0.065 -1.915 -0.647 0.524 Prefrontal Cortex 0.918 0.076 0.931 0.080 1.427 0.419 0.679 Inf Parietal Cortex 1.026 0.087 1.012 0.048 -1.365 -0.479 0.636 Occipital Cortex 0.947 0.088 0.972 0.070 2.715 0.793 0.436 Hippocampus 0.739 0.082 0.773 0.113 4.547 0.865 0.396
B. After partial volume correction
Grey matter ROI HV (n=14) aMCI (n=11)
mean SD mean SD % Diff t p Temporal Cortex 1.149 0.135 1.203 0.123 4.771 1.046 0.307 Prefrontal Cortex 1.323 0.149 1.394 0.112 5.405 1.324 0.199 Inf Parietal Cortex 1.410 0.212 1.467 0.139 4.014 0.682 0.453 Occipital Cortex 1.369 0.174 1.487 0.247 8.582 0.176 0.177 Hippocampus 0.782 0.143 0.876 0.194 12.043 0.533 0.174
59
Figure 8. t-statistical map of parametric images of HV vs MCI overlaid on a T1
weighted magnetic resonance imaging (MRI) template. Voxel threshold was set to p<0.05
(FEW corrected). No group differences were detected.
60
Table 7. Regional [18F]-FEPPA VT for HV and aMCI participants when stratified by
PIB status. An average of 1.20 DVR in the cortical regions was used as a cutoff. No
differences were observed between groups, however within the aMCI group those with
amyloid pathology had an elevation in [18F]-FEPPA binding.
A. Before partial volume correction
Grey matter ROI HV (n=14) aMCI (n=11)
% Diff between HV PIB- and aMCI
PIB+ PIB- (n=11) PIB+ (n=3) PIB- (n=3) PIB+ (n=8) mean SD mean SD mean SD mean SD
Temporal Cortex 12.65 3.54 9.58 1.67 8.94** 1.91 12.80** 1.51 1.19 3.58 1.49
Prefrontal Cortex 12.58 3.40 8.73 1.27 9.11* 2.11 13.03* 2.14 Inf Parietal Cortex 13.39 3.70 9.63 1.44 9.54* 2.40 13.59* 2.27 Occipital Cortex 11.77 3.92 8.93 1.50 8.79* 2.62 12.53* 1.97 6.46
Hippocampus 10.19 3.24 7.36 1.10 8.95 5.19 11.83 2.95 16.09 **Significant difference between PIB- and PIB+ aMCI participants, p<0.01 *Significant difference between PIB- and PIB+ aMCI participants, p<0.05
B. After partial volume correction
Grey matter ROI HV (n=14) aMCI (n=11)
% Diff between HV PIB- and aMCI
PIB+ PIB- (n=11) PIB+ (n=3) PIB- (n=3) PIB+ (n=8) mean SD mean SD mean SD mean SD
Temporal Cortex 15.58 4.31 12.16 1.82 10.71* 1.89 16.51* 3.03 5.97 Prefrontal Cortex 18.60 4.78 14.58 2.11 13.17 2.69 19.37 4.56 4.14
5.43 Inf Parietal Cortex 18.98 4.76 14.66 2.06 13.66 3.13 20.01 4.97 Occipital Cortex 17.94 5.85 13.43 2.14 12.67 3.79 19.00 5.06 5.91
Hippocampus 11.46 3.38 8.38 0.88 9.94 5.69 14.12 3.73 23.21 *Significant difference between PIB- and PIB+ aMCI participants, p<0.05
61
Figure 9. [18F]-FEPPA VT in regions of interest after characterizing participants based
on amyloid status. Using a cutoff point of 1.20 DVR in the cortical regions participants
were characterized as either PIB- or PIB+. The empty symbols represent individuals without
amyloid pathology, whereas the filled in symbols represent those with amyloid accumulation.
Significant differences in [18F]-FEPPA VT were observed within the aMCI subgroups before
(A) and after (B) partial volume correction. **p<0.01, *p<0.05
(A)
(B)
* * *
* *
*
62
4.1.4 Exploratory correlations between [11C]-PIB and [18F]-FEPPA
In aMCI participants, [11C]-PIB binding was correlated with [18F]-FEPPA binding in
all regions of interest (Table 8A, Figure 10). The strongest correlation was in the
hippocampus. When the images were corrected for partial volume effects, significant
correlations were observed in the temporal cortex (rho=0.718, p=0.013), prefrontal cortex
(rho=0.609, p=0.047) and hippocampus (rho=0.827, p=0.002) (Table 8B, Figure 11). In
order to correct for multiple comparisons, a p value of 0.01 was applied (0.05/5 regions).
After correcting for multiple comparisons, the regional association remained in the
hippocampus before and after partial volume correction.
63
Table 8. Positive correlations between [18F]-FEPPA and [11C]-PIB binding. Correlations
are demonstrated for aMCI participants (n=11). After correction for multiple comparisons,
the regional association remains significant in the hippocampus (p value after Bonferroni
correction for multiple comparisons is 0.01).
A. Before partial volume correction
Grey matter ROI rho p Temporal Cortex 0.682 0.021 Prefrontal Cortex 0.609 0.047 Inferior Parietal Cortex 0.618 0.043 Occipital Cortex 0.627 0.039 Hippocampus 0.855 0.001
B. After partial volume correction
Grey matter ROI rho p Temporal Cortex 0.718 0.013 Prefrontal Cortex 0.609 0.047 Inferior Parietal Cortex 0.509 0.110 Occipital Cortex 0.373 0.259 Hippocampus 0.827 0.002
64
Figure 10. Regional associations between [18F]-FEPPA and [11C]-PIB binding.
Significant correlations were observed in all regions of interest (A-E). After Bonferroni
correction for multiple comparisons, the correlation survived in the hippocampus (E). aMCI
participants (n=11).
(A)
rho=0.682 p=0.021
(E)
rho=0.609 p=0.047
(B)
(C)
rho=0.618 p=0.043
(C) (D)
rho=0.627 p=0.039
(E)
rho=0.855 p=0.001
65
Figure 11. Regional associations between [18F]-FEPPA and [11C]-PIB binding after
partial volume correction. Significant correlations were observed in the temporal cortex
(A), prefrontal cortex (B) and hippocampus (C). After Bonferroni correction for multiple
comparisons, the correlation remained in the hippocampus. aMCI participants (n=11).
(C)
(A)
rho=0.718 p=0.013
(B)
rho=0.609 p=0.047
(C)
rho=0.827 p=0.002
66
4.1.5 Exploratory correlations between [11C]-PIB/[18F]-FEPPA and
cognition
In aMCI participants, correlations between [11C]-PIB/[18F]-FEPPA and cognition
were explored. Since we do not have the statistical power to test for correlations with
cognition, assessments were grouped by cognitive domain that is evaluated (memory,
executive function, language, attention and overall global cognition), and subsequently
Bonferroni correction was applied.
The Logical Memory Delayed Task, measuring episodic memory, was negatively
correlated with [11C]-PIB binding in the temporal cortex (rho=-0.635, p=0.036) and inferior
parietal cortex (rho=-0.667, p=0.025) (Table 9A, Figure 12). After partial volume correction,
the correlations with Logical Memory were similarly observed, temporal cortex (rho=-0.635,
p=0.036) and inferior parietal cortex (rho=-0.676, p=0.022). An additional correlation was
obtained after partial volume correction between this scale and [11C]-PIB DVR in the
hippocampus (rho=-0.621, p=0.041) (Table 9B, Figure 13). Although it appeared that higher
[11C]-PIB DVR was associated with worse delayed memory recall, none of the correlations
survived after Bonferroni correction for multiple comparisons.
Correlations with [18F]-FEPPA and cognition were also explored in aMCI
participants (Table 10A and B). Negative correlations were obtained between scores on the
Logical Memory Delayed Task and [18F]-FEPPA VT in the temporal cortex (rho=-0.607,
p=0.048), occipital cortex (rho=-0.639, p=0.034) and hippocampus (rho=-0.699, p=0.017)
(Figure 14). However, these correlations did not survive Bonferroni correction. After partial
volume correction, higher [18F]-FEPPA VT in the hippocampus was correlated with lower
67
scores on the Logical Memory Delayed Task (rho=-0.886, p=0.0003) and the Stroop Color
Word Task (rho=-0.714, p=0.047) (Figure 15). After Bonferroni correction, the correlation
between Logical Memory Delayed and [18F]-FEPPA VT in the hippocampus survived.
68
Table 9. Correlation analyses between [11C]-PIB DVR and neuropsychological score. Spearman’s rank coefficient (rho).
Table 9A. Correlations before partial volume error correction.
MEMORY LANGUAGE
Cognitive Measure MCI (n=7)
RBANS Immediate Memory
RBANS Delayed Memory
Logical Memory Delayed (n=11)
RBANS Language
Verbal Fluency (n=11)
DVR rho p rho p rho p rho p rho p Temporal Cortex Prefrontal Cortex Inferior Parietal Cortex Occipital Cortex Hippocampus
0.071 0.000 0.286 0.393 -0.071
0.879 1.000 0.535 0.383 0.879
-0.054 0.432 -0.108 -0.180 0.162
0.908 0.333 0.818 0.699 0.728
-0.635 -0.584 -0.667 -0.584 -0.557
0.036 0.059 0.025 0.059 0.075
-0.143 0.143 -0.036 -0.071 -0.071
0.760 0.760 0.939 0.879 0.879
0.369 0.205 0.446 0.310 -0.005
0.264 0.545 0.169 0.354 0.989
GLOBAL COGNITION ATTENTION VISUOSPATIAL
Cognitive Measure MCI (n=11) MMSE MoCA
RBANS Attention
(n=7)
TMT Task A
RBANS Visuospatial
(n=7) DVR rho p rho p rho p rho p rho p Temporal Cortex Prefrontal Cortex Inferior Parietal Cortex Occipital Cortex Hippocampus
-0.302 -0.181 -0.149 -0.306 -0.167
0.367 0.594 0.663 0.360 0.623
-0.169 -0.330 -0.046 -0.078 -0.549
0.619 0.322 0.894 0.820 0.080
0.055 0.109 0.218 0.273 -0.582
0.908 0.816 0.638 0.554 0.170
0.018 -0.127 -0.018 0.045 0.009
0.958 0.709 0.958 0.894 0.979
0.342 0.252 0.523 0.414 -0.198
0.452 0.585 0.229 0.355 0.670
EXECUTIVE FUNCTION PREMORBID
INTELLIGENCE Cognitive Measure
MCI (n=11) TMT
Task B
Stroop Test Color Task
(n=8)
Stroop Test Color Word Task
(n=8)
Letter Number Span (n=7)
NART (n=6)
DVR rho p rho p rho p rho p rho p Temporal Cortex Prefrontal Cortex Inferior Parietal Cortex Occipital Cortex Hippocampus
-0.191 -0.218 -0.282 -0.127 -0.045
0.574 0.519 0.401 0.709 0.894
0.109 0.109 0.218 0.218 -0.343
0.797 0.797 0.604 0.604 0.406
-0.333 -0.167 -0.357 -0.190 -0.310
0.420 0.693 0.385 0.651 0.456
0.315 0.315 0.315 0.512 -0.335
0.491 0.491 0.491 0.240 0.463
0.086 -0.314 0.371 0.314 0.029
0.871 0.544 0.468 0.544 0.957
69
Table 9B. Correlations after partial volume correction
MEMORY LANGUAGE Cognitive Measure
MCI (n=7)
RBANS Immediate Memory
RBANS Delayed Memory
Logical Memory Delayed (n=11)
RBANS Language
Verbal Fluency (n=11)
DVR rho p rho p rho p rho p rho p Temporal Cortex Prefrontal Cortex Inferior Parietal Cortex Occipital Cortex Hippocampus
0.071 0.000 0.071 0.393 -0.321
0.879 1.000 0.879 0.383 0.482
-0.054 0.432 -0.054 -0.180 0.414
0.908 0.333 0.908 0.699 0.355
-0.635 -0.584 -0.676 -0.525 -0.621
0.036 0.059 0.022 0.097 0.041
-0.143 0.143 -0.143 -0.071 0.000
0.760 0.760 0.760 0.879 1.000
0.369 0.205 0.456 0.287 0.118
0.264 0.545 0.159 0.392 0.729
GLOBAL COGNITION ATTENTION VISUOSPATIAL Cognitive Measure
MCI (n=11) MMSE MoCA RBANS Attention
(n=7)
TMT Task A
RBANS Visuospatial
(n=7) DVR rho p rho p rho p rho p rho p Temporal Cortex Prefrontal Cortex Inferior Parietal Cortex Occipital Cortex Hippocampus
-0.302 -0.181 -0.204 -0.330 -0.176
0.367 0.594 0.547 0.322 0.604
-0.169 -0.330 -0.124 -0.027 -0.572
0.619 0.322 0.717 0.936 0.066
0.055 0.109 0.055 0.273 -0.346
0.908 0.816 0.908 0.554 0.448
0.018 -0.127 0.000 0.055 -0.100
0.958 0.709 1.000 0.873 0.770
0.342 0.252 0.342 0.414 0.072
0.452 0.585 0.452 0.355 0.878
EXECUTIVE FUNCTION PREMORBID INTELLIGENCE
Cognitive Measure MCI (n=11)
TMT Task B
Stroop Test Color Task
(n=8)
Stroop Test Color Word Task
(n=8)
Letter Number Span (n=7)
NART (n=6)
DVR rho p rho p rho p rho p rho p Temporal Cortex Prefrontal Cortex Inferior Parietal Cortex Occipital Cortex Hippocampus
-0.191 -0.218 -0.273 -0.118 -0.191
0.574 0.519 0.417 0.729 0.574
0.109 0.109 0.109 0.218 -0.171
0.797 0.797 0.797 0.604 0.685
-0.333 -0.167 -0.405 -0.190 -0.333
0.420 0.693 0.320 0.651 0.420
0.315 0.315 0.315 0.512 -0.256
0.491 0.491 0.491 0.240 0.579
0.086 -0.314 0.086 0.314 -0.314
0.872 0.544 0.872 0.544 0.544
70
Table 10. Correlation analyses between [18F]-FEPPA VT and neuropsychological score. Spearman’s rank coefficient (rho).
Table 10A. Correlations before partial volume error correction.
MEMORY LANGUAGE
Cognitive Measure MCI (n=7)
RBANS Immediate Memory
RBANS Delayed Memory
Logical Memory Delayed (n=11)
RBANS Language
Verbal Fluency (n=11)
VT rho p rho p rho p rho p rho p Temporal Cortex Prefrontal Cortex Inferior Parietal Cortex Occipital Cortex Hippocampus
-0.464 -0.143 -0.071 -0.393 0.250
0.294 0.760 0.879 0.383 0.589
0.342 0.450 0.631 0.252 0.162
0.452 0.310 0.129 0.585 0.728
-0.607 -0.502 -0.493 -0.639 -0.699
0.048 0.115 0.123 0.034 0.017
0.214 0.571 0.714 0.179 0.250
0.645 0.180 0.071 0.702 0.589
0.551 0.565 0.551 0.533 -0.014
0.079 0.070 0.079 0.091 0.968
GLOBAL COGNITION ATTENTION VISUOSPATIAL Cognitive Measure
MCI (n=11) MMSE MoCA RBANS Attention
(n=7)
TMT Task A
RBANS Visuospatial
(n=7) VT rho p rho p rho p rho p rho p Temporal Cortex Prefrontal Cortex Inferior Parietal Cortex Occipital Cortex Hippocampus
-0.097 0.014 0.088 -0.395 -0.242
0.776 0.968 0.797 0.292 0.530
-0.293 -0.192 -0.192 -0.548 -0.572
0.382 0.571 0.571 0.127 0.107
0.055 0.273 0.273 0.560 -0.297
0.908 0.554 0.554 0.326 0.627
-0.245 -0.291 -0.300 -0.373 0.100
0.467 0.385 0.370 0.259 0.770
0.324 0.450 0.450 0.468 0.090
0.478 0.310 0.310 0.289 0.848
EXECUTIVE FUNCTION PREMORBID INTELLIGENCE
Cognitive Measure MCI (n=11)
TMT Task B
Stroop Test Color Task
(n=8)
Stroop Test Color Word
Task (n=8)
Letter Number Span (n=7)
NART (n=6)
VT rho p rho p rho p rho p rho p Temporal Cortex Prefrontal Cortex Inferior Parietal Cortex Occipital Cortex Hippocampus
-0.255 -0.282 -0.300 -0.318 0.073
0.450 0.401 0.370 0.340 0.832
0.327 0.546 0.546 0.327 -0.296
0.429 0.162 0.162 0.429 0.476
-0.048 0.238 0.262 -0.190 -0.452
0.911 0.570 0.531 0.651 0.260
0.020 0.118 0.118 -0.079 -0.571
0.967 0.801 0.801 0.867 0.180
-0.257 -0.029 -0.086 -0.029 0.429
0.623 0.957 0.872 0.957 0.397
71
Table 10B. Correlations after partial volume error correction.
MEMORY LANGUAGE
Cognitive Measure MCI (n=7)
RBANS Immediate Memory
RBANS Delayed Memory
Logical Memory Delayed (n=11)
RBANS Language
Verbal Fluency (n=11)
VT rho p rho p rho p rho p rho p Temporal Cortex Prefrontal Cortex Inferior Parietal Cortex Occipital Cortex Hippocampus
-0.607 -0.643 -0.643 -0.607 -0.607
0.148 0.119 0.119 0.148 0.148
0.342 0.631 0.631 0.342 0.450
0.452 0.129 0.129 0.452 0.310
-0.539 -0.411 -0.429 -0.406 -0.886
0.087 0.209 0.188 0.215 0.0003
-0.036 0.214 0.214 -0.036 -0.107
0.939 0.645 0.645 0.939 0.819
0.510 0.474 0.428 0.387 0.178
0.109 0.141 0.189 0.239 0.601
GLOBAL COGNITION ATTENTION VISUOSPATIAL
Cognitive Measure MCI (n=11) MMSE MoCA
RBANS Attention
(n=7)
TMT Task A
RBANS Visuospatial
(n=7) VT rho p rho p rho p rho p rho p Temporal Cortex Prefrontal Cortex Inferior Parietal Cortex Occipital Cortex Hippocampus
-0.297 -0.097 -0.074 -0.186 -0.070
0.375 0.776 0.828 0.585 0.839
-0.435 -0.481 -0.517 -0.503 -0.586
0.181 0.135 0.103 0.114 0.058
-0.109 -0.200 -0.200 -0.109 -0.491
0.816 0.667 0.667 0.816 0.263
-0.145 -0.300 -0.318 -0.482 0.173
0.670 0.370 0.340 0.133 0.612
0.090 0.018 0.018 0.090 -0.090
0.848 0.969 0.969 0.848 0.848
EXECUTIVE FUNCTION PREMORBID
INTELLIGENCE Cognitive Measure
MCI (n=11) TMT
Task B
Stroop Test Color Task
(n=8)
Stroop Test Color Word Task
(n=8)
Letter Number Span (n=7)
NART (n=6)
VT rho p rho p rho p rho p rho p Temporal Cortex Prefrontal Cortex Inferior Parietal Cortex Occipital Cortex Hippocampus
-0.318 -0.345 -0.273 -0.345 0.064
0.340 0.298 0.417 0.298 0.853
0.109 0.078 0.078 0.109 -0.452
0.797 0.854 0.854 0.797 0.261
-0.119 -0.048 -0.048 -0.119 -0.714
0.779 0.911 0.911 0.779 0.047
0.118 -0.118 -0.118 0.118 -0.493
0.801 0.801 0.801 0.801 0.261
-0.600 -0.543 -0.543 -0.600 -0.371
0.208 0.266 0.266 0.208 0.468
72
Figure 12. Correlations between [11C]-PIB binding and cognition. Spearman correlations
between Logical Memory Delayed Task and [11C]-PIB binding in the temporal (A) and
inferior parietal (B) cortices. Correlations did not survive correction for multiple
comparisons.
rho=-0.635 p=0.036
rho=-0.667 p=0.025
(A)
(B)
73
Figure 13. Correlations between [11C]-PIB binding and score on the logical memory
delayed task, after partial volume correction. Correlations were observed in the temporal
cortex (A), inferior parietal cortex (B) and hippocampus (C), and did not survive correction
for multiple comparisons.
(A) rho=-0.635 p=0.036
(B) rho=-0.676 p=0.022
(C)
rho=-0.621 p=0.041
74
Figure 14. Correlations between [18F]-FEPPA binding and cognition. Spearman
correlations between Logical Memory Delayed Task and [18F]-FEPPA VT in the temporal
cortex (A), occipital cortex (B) and hippocampus (C). Correlations did not survive after
correction for multiple comparisons.
(A) rho=-0.607 p=0.048
(B) rho=-0.639 p=0.034
(C) rho=-0.699 p=0.017
75
Figure 15. Correlations between [18F]-FEPPA binding in the hippocampus and
cognition, after partial volume correction. Scores for 11 aMCI participants are included
for the Logical Memory Delayed Task (A), while for the Stroop Test Color Word Task
scores for 8 aMCI participants are included (B). After Bonferroni correction for multiple
comparisons (0.05/3 for memory), the correlation with Logical Memory Delayed survived.
rho=-0.886 p=0.0003
rho=-0.714 p=0.047
(A)
(B)
rho=-0.886 p=0.0003
rho=-0.714 p=0.047
76
4.2 Glutathione
4.2.1 No differences in GSH levels in the LDLPFC
Nine healthy volunteers and 11 aMCI participants completed a GSH MRS scan;
however 1 HV and 2 aMCI participants had to be removed from the analysis due to noise in
the data. Those included in the analysis did not differ in age, HV (67.13±10.66) and aMCI
(71.67±5.61 years). An independent sample t-test showed no significant differences between
the two groups, t(15)=-0.277, p=0.785 (Figure 16). The mean GSH levels were very similar
between groups, 0.00147±0.00062 for HV and 0.00146±0.00059 for aMCI participants. A
large amount of variability is present in both of the groups.
77
Figure 16. No differences in GSH/H2O levels in the left dorsolateral prefrontal cortex.
No significant differences were observed between aMCI participants (n=9) and healthy
volunteers (n=8). Within both groups there is a large variability in GSH/H2O levels.
78
4.2.2 Exploratory correlations between GSH and amyloid, neuroinflammation, and cognition
From our pilot data, amyloid, as reflected by [11C]-PIB DVR, was not significantly
correlated with GSH levels in aMCI participants (Table 11). In contrast, positive correlations
were found between neuroinflammation, as reflected by [18F]-FEPPA VT, and GSH levels in
the left DLPFC and full DLPFC of aMCI participants (Table 12, Figure 17). Correlations
were observed only before correction for partial volume effects. No correlations were
observed between [18F]-FEPPA in the right DLPFC and GSH levels in the left DLPFC. With
regards to cognition, higher GSH levels in the LDLPFC correlated with better performance
on the Stroop Test Color Word Task (Table 13, Figure 18). The correlation with the Stroop
Test Color Word Task survived Bonferroni correction.
Table 11. [11C]-PIB binding and GSH levels in aMCI participants. No regional
association between amyloid burden and GSH were found. Spearman’s rank coefficient was
used to test for correlations.
Region aMCI (n=9) r p L DLPFC 0.450 0.224 R DLPFC 0.383 0.308 DLPFC 0.383 0.308 With partial volume correction
L DLPFC 0.417 0.265 R DLPFC 0.567 0.112 DLPFC 0.583 0.099
79
Table 12. [18F]-FEPPA binding and GSH levels in aMCI participants. Higher
neuroinflammation correlated with a higher amount of GSH in the LDLPFC and full DLPFC.
The correlations were not observed after correction for partial volume effects. Spearman’s
rank coefficient was used to test for correlations.
Region aMCI (n=9) rho p L DLPFC 0.750 0.020 R DLPFC 0.600 0.088 DLPFC 0.767 0.016 With partial volume correction
L DLPFC 0.500 0.170 R DLPFC 0.450 0.224 DLPFC 0.450 0.224
Figure 17. Positive correlations between [18F]-FEPPA binding and GSH/H2O levels.
Higher levels of GSH were correlated with higher [18F]-FEPPA binding in the LDLPFC (A)
and the full DLPFC (B) in aMCI participants (n=9). Correlations are without partial volume
correction.
(B) (A) rho=0.750 p=0.020
rho=0.767 p=0.016
80
Table 13. Exploratory correlations between GSH levels in the LDLPFC and
performance on cognitive scales. Correlation with Stroop task survived after Bonferroni
correction (executive function scales 0.05/4).
Cognitive Measure rho p aMCI (n=6) Memory
RBANS Immediate RBANS Delayed Logical Memory (n=9)
0.257 0.029 -0.301
0.623 0.957 0.431
Language RBANS Language Verbal Fluency (n=9)
0.371 0.377
0.468 0.318
Global Cognition MMSE (n=9) MoCA (n=9)
-0.222 -0.203
0.565 0.601
Attention RBANS Attention 0.353 0.492
Visuospatial RBANS Visuospatial TMT – Task A (n=9)
-0.116 0.233
0.827 0.546
Executive Function TMT – Task B (n=9) Stroop Task – Color Score Stroop Task – Color Word Score
-0.050 0.655 0.943
0.898 0.158 0.005
Letter Number Span 0.507 0.305 Premorbid Intelligence
NART 0.086 0.872
Figure 18. Higher GSH levels correlated with a higher score on the Stroop Color Word
score. aMCI participants (n=6).
rho=0.943 p=0.005
81
5. DISCUSSION
5.1 Increased amyloid in aMCI patients
Patients with aMCI had significantly more amyloid in the cortical regions, with the
highest amount in the prefrontal cortex followed by the temporal cortex, inferior parietal
cortex and occipital cortex. The hippocampus was the only region in which aMCI patients
did not have a significant increase of [11C]-PIB retention. Our observed distribution pattern
of [11C]-PIB retention is consistent with the pattern of Aβ plaque deposition observed in
post-mortem studies of the AD brain (Arnold et al. 1991;Thal et al. 2002), whereby studies
have shown large increases in neuritic plaques in cortical regions and low levels in the
medial temporal cortex, which includes the hippocampus. Furthermore PET imaging studies
of MCI and AD patients have similarly reported increases in [11C]-PIB retention in the
cortical regions and lower levels in the hippocampus (Klunk et al. 2004;Lopresti et al.
2005;Mintun et al. 2006;Kemppainen et al. 2007;Rowe et al. 2007;Edison et al.
2008;Forsberg et al. 2008;Okello et al. 2009;Wiley et al. 2009). In our aMCI sample, the
percent difference in [11C]-PIB retention between the two groups was 5% for the
hippocampus and 43% for the cortical regions. In an AD study, the difference in [11C]-PIB
retention in the hippocampus and cortical regions was 14% and 70-80%, respectively (Rowe
et al. 2007). Our lower uptake in comparison to AD studies supports the idea that PIB
retention in MCI patients is intermediate between healthy controls and AD patients (Forsberg
et al. 2008).
82
The idea that aMCI patients fall into the intermediate range of PIB retention is further
supported by reports of patients being characterized as “PIB+” or “PIB-”. In our sample, 8/11
aMCI patients were characterized as PIB+ as indicated by an average cortical [11C]-PIB
DVR >1.20, whereas the remaining 3 aMCI patients were characterized as being PIB-. We
can speculate that these 3 PIB- patients may be on a different disease trajectory and may
develop another type of dementia other than AD. The presence of amyloid in the majority of
our aMCI patients (73%) is supported by other PET studies. Okello and colleagues, reported
7/14 aMCI participants as PIB+, with nearly two-fold increased uptake in the cingulate and
frontal regions (Okello et al. 2009). Likewise, Wiley and colleagues, reported 4/6 MCI
patients as being PIB+ and two-fold increases in the parietal, frontal, posterior cingulate, and
precuneus regions (Wiley et al. 2009). A larger study with 24 aMCI patients found that 18/24
patients (or 75%) were PIB+ (Rowe et al. 2007). Conversely, only 3/14 (or 21%) of our
healthy volunteers were characterized as PIB+, with the majority being classified as PIB-.
Increased amyloid deposition has been reported in up to 1/3 of cognitively normal elderly
participants (Mintun et al. 2006;Jack et al. 2008), which explains the finding of increased
[11C]-PIB binding in 3 of our healthy volunteers. In comparison to PIB- healthy volunteers,
those characterized as PIB+ did not demonstrate an impairment on any cognitive assessment.
Overall, the presence of amyloid in the majority of aMCI patients supports the idea that
amyloid deposition is an early event.
83
5.2 No differences in [18F]-FEPPA VT
In contrast to amyloid pathology, we did not observe a significant difference in [18F]-
FEPPA binding between aMCI and healthy volunteers in any region of interest with either
the absolute quantification method (2TCM) or the supplementary SUVR method. PET
studies investigating neuroinflammation in MCI participants have observed conflicting
results, with some reporting no differences from controls (Wiley et al. 2009;Kreisl et al.
2013;Schuitemaker et al. 2013), while others reported increases in certain regions of interest
(Okello et al. 2009;Yasuno et al. 2012). It is important to note that there are notable
differences in the methodology and demographic characteristics between studies. Three of
the studies have used the prototypical radioligand, [11C]-PK11195, which as reviewed earlier
has several limitations including a short half-life, high nonspecific binding, low brain
penetration and high plasma protein binding (Okello et al. 2009;Wiley et al.
2009;Schuitemaker et al. 2013). Furthermore, two of the studies did not use a purely
amnestic sample of MCI patients (Wiley et al. 2009;Yasuno et al. 2012), which is a problem
as MCI is a broad category that may encompass a multitude of underlying causes. Thus these
two studies may not be a good comparative for us to use. In terms of studies that have
included aMCI patients specifically, one group observed an elevation in [11C]-PK11195
binding in patients with increased PIB retention (Okello et al. 2009). Similarly, in our aMCI
sample, those that were PIB+ had significantly higher [18F]-FEPPA binding in the prefrontal,
temporal, inferior parietal and occipital cortices compared to aMCI patients classified as
PIB-. However, when PIB+ aMCI were compared to PIB- healthy volunteers, no significant
differences were obtained. This may in part be explained by the variability in the arterial
input function. Two healthy volunteers had higher than expected K1 values (ratio of delivery)
84
and ultimately higher VTs. Thus although the whole aMCI group did not have a significant
difference in [18F]-FEPPA binding, we can speculate that amyloid pathology may play a role
in the activation of microglia. This idea is consistent with immunohistochemical
examinations of brain slices that have revealed the presence of microglia surrounding Aβ
plaques in AD (Rogers et al. 1988;Itagaki et al. 1989;McGeer et al. 1989). However, a larger
cohort of patients is required to confirm whether those with increased amyloid have a parallel
increase in neuroinflammation. The most recent PET study with 10 aMCI patients, reported
similar group results as us, whereby they observed increases in [11C]-PIB retention but no
differences in [11C]-PBR28 binding between MCI patients and controls (Kreisl et al. 2013).
Thus from this second generation study and ours we can speculate that neuroinflammation
may only occur after conversion to AD. Another interesting speculation that arises from our
results is the relationship between tau and microglial activation. Intriguingly, the largest
difference in [18F]-FEPPA binding between aMCI and HV participants was in the
hippocampus, a region with earliest signs of tau accumulation (Small et al. 2006;Chien et al.
2013). Recently, in a mouse model it was shown that microglia play an important role in tau
propagation (Asai et al. 2015).Thus we can hypothesize that our finding of increased
microglial activation in the hippocampus is due to an increase in tau accumulation in this
region. This is further supported by the fact that the percent differences in [18F]-FEPPA
binding are low in the cortical regions, areas in which tau accumulates later on in disease
(Braak and Braak 1995).
85
5.3 Exploratory correlations between amyloid and
neuroinflammation
In order to further investigate whether amyloid and neuroinflammation are regionally
associated, correlations between [11C]-PIB and [18F]-FEPPA binding were explored. In aMCI
patients, significant correlations between [11C]-PIB and [18F]-FEPPA binding were found in
the temporal cortex, prefrontal cortex, inferior parietal cortex, occipital cortex and
hippocampus. After correction for partial volume effects, the correlations remained in the
temporal cortex, prefrontal cortex and hippocampus. Our results are consistent with
histopathological studies that have demonstrated a colocalization of activated microglia with
Aβ-containing neuritic plaques in the AD brain (Rogers et al. 1988;Itagaki et al.
1989;McGeer et al. 1989). With Bonferroni correction for multiple comparisons, the
association between amyloid and neuroinflammation remained significant only in the
hippocampus. The relationship between amyloid burden and neuroinflammation has
previously been investigated in-vivo. The first PET study to investigate the spatial
relationship in AD patients did not find any significant correlations between [11C]-PIB and
[11C]-PK11195 in any region of interest (Edison et al. 2008). Similarly, two PET studies in
MCI patients did not find any regional associations between the two radioligands (Okello et
al. 2009;Wiley et al. 2009). The negative results observed by these previous studies may be
in part due to the use of [11C]-PK11195 and simplified reference tissue models rather than
the 2TCM. Thus far only one other study has evaluated the spatial relationship of amyloid
and neuroinflammation in AD and MCI with the use of a second generation radioligand and
the 2TCM. The study reported a significant correlation between [11C]-PBR28 and [11C]-PIB
86
in the inferior parietal lobule, superior temporal cortex, precuneus, hippocampus and
parahippocampal gyrus (Kreisl et al. 2013). However, the correlations were observed only
after correction for partial volume effects. Overall, it is clear that the in-vivo spatial
relationship between amyloid and microglia is still not well understood, and a larger study is
required to elucidate the relationship.
5.4 Neuroinflammation, but not amyloid, may correlate with
cognition
In attempts to gain a better understanding of the underlying causes of impairment in
AD and MCI, studies have investigated whether pathologies such as amyloid and
neuroinflammation are associated with poorer performance on cognitive scales. Post-mortem
studies of AD have generally demonstrated that amyloid does not correlate well with
symptom severity or cognitive impairment (Arriagada et al. 1992;Bierer et al. 1995;Vehmas
et al. 2003), whereas in-vivo PET studies have reported conflicting results, with some groups
demonstrating no correlations between retention and performance on cognitive scales
(Edison et al. 2007;Jack et al. 2008;Okello et al. 2009), while others report correlations with
an impairment on episodic memory tests (Pike et al. 2007;Forsberg et al. 2008;Villemagne et
al. 2011). From our exploratory analyses, [11C]-PIB binding appeared to be associated with a
measure of episodic memory, the Logical Memory Delayed Task. The correlation suggests
that those with higher amyloid pathology have worse delayed memory recall. However, none
of the correlations between [11C]-PIB and memory survived Bonferroni correction for
87
multiple comparisons, and thus we cannot consider them as true. Overall our result is
consistent with previous studies that have demonstrated that amyloid pathology is generally
not correlated with cognition.
Previous studies have shown conflicting results regarding the relationship between
neuroinflammation and cognition (Edison et al. 2008;Okello et al. 2009;Yokokura et al.
2011;Yasuno et al. 2012;Kreisl et al. 2013;Schuitemaker et al. 2013;Suridjan et al.
2015;Varrone et al. 2015). In MCI studies specifically, only one group has reported a
significant correlation with cognition and neuroinflammation (Kreisl et al. 2013). The study
reported strong correlations between [11C]-PBR28 binding in the inferior parietal lobule and
CDR score and performance on Block Design. However, the researchers did not consider the
results as true as they did not survive correction for partial volume effects. Furthermore, AD
patients were included in the correlation as well. Other MCI studies report no correlations
with MMSE (Okello et al. 2009;Yasuno et al. 2012) and a battery of other
neuropsychological assessments such as New York University Recall Test, Rey’s Auditory
Verbal Learning Test, Trail Making Test, Rey’s complex figure, Boston Naming Test, and
forward and backward condition of the Digit Span (Schuitemaker et al. 2013). In aMCI
patients, we observed negative correlations between Logical Memory Delayed Task and the
Stroop Color Word Task, however the only correlation to survive Bonferroni correction was
between [18F]-FEPPA binding in the hippocampus (after partial volume correction) and the
Logical Memory Delayed Task. Our finding suggests that those with higher microglial
activation in the hippocampus have worse delayed memory recall. Out of all cognitive scales
performed, an association with a measure of delayed memory is supported by the fact that
aMCI is characterized by a decline in episodic memory (Murphy et al. 2008). The conflicting
88
results between cognition and neuroinflammation are in part due to methodological and
demographic differences between all studies. As already mentioned in previous sections,
several MCI TSPO studies have been performed with [11C]-PK11195, which has known
limitations (Okello et al. 2009;Wiley et al. 2009;Schuitemaker et al. 2013). Furthermore, two
of the PET studies only performed the MMSE rather than a battery of neuropsychological
assessments (Okello et al. 2009;Yasuno et al. 2012). Overall, future studies should be
performed with larger sample sizes and with a range of neuropsychological assessments to
better understand the association between neuroinflammation and cognition.
5.5 No differences in GSH levels
Although oxidative stress is thought to be an important feature of AD pathology,
whether GSH levels, the brain’s major antioxidant, are altered in-vivo is still not well
understood. Previous studies measuring GSH in AD and MCI populations have reported
conflicting results (Mandal et al. 2012;Duffy et al. 2014;Mandal et al. 2015). The first study
with young and old HV, MCI, and AD measured no differences between MCI and HV in the
frontal cortex (Mandal et al. 2012). However, they did report a trend in GSH levels: young
HV>old HV>MCI>AD. A subsequent study with a larger MCI sample observed a significant
elevation of GSH in the anterior and posterior cingulate (Duffy et al. 2014). Whereas the
most recent MCI study, observed a significant reduction in GSH in the hippocampus and
no differences in the frontal cortex (Mandal et al. 2015). Our study was the first to evaluate
GSH levels in the LDLPFC of aMCI patients and healthy volunteers. From the individuals
that had analyzable GSH data, no significant differences were observed between aMCI and
89
healthy volunteers. A large variability in GSH levels was found in both groups, which may in
part be explained by a polymorphism in the gene coding for the catalytic (GCLC) subunit of
glutamate-cysteine ligase (GCL), the rate-limiting enzyme for GSH synthesis, which has
been reported to influence GSH concentrations (Xin et al. 2016). Similar to other MRS
studies, it is important to note that our voxel contained white and grey matter, however there
were no significant differences in the composition of the voxel between the two groups.
When comparing our results to other MCI MRS studies, certain caveats need to be
considered. Firstly, relating to the Duffy et al. study, the group did not have a purely
amnestic MCI sample. Additionally, the group used the PRESS pulse sequence which does
not distinguish GSH from overlapping brain metabolites as well as MEGA-PRESS (Mandal
et al. 2012). The latter study used the hippocampus as a region of interest, which is more
difficult to image using MRS because of its close proximity to ventricular cerebrospinal fluid
and low signal-to-noise ratio (Duffy et al. 2014). The lack of significant differences in GSH
levels in the LDLPFC of aMCI and healthy volunteers is supported by two earlier studies
that reported no difference in the full frontal cortex (Mandal et al. 2012;Mandal et al. 2015).
Other studies have suggested that GSH alterations are region-specific, thus we can speculate
that GSH levels in the DLPFC may remain unaffected. However, it is important to note that
we had a very small sample size, thus a larger sample is required to study GSH alterations in
prodromal AD patients. Future studies should include AD patients in order to study GSH on
a continuum from normal to aMCI to AD. Moreover, it would be beneficial to study GSH in
more than one region of interest, in attempts to better characterize the regional pattern of
GSH alterations in AD pathology.
90
5.6 GSH levels correlate with microglia but not amyloid
In order to gain a better understanding of the underlying pathology of aMCI, we
explored potential correlations between GSH and our PET measures, amyloid and
neuroinflammation (Figure 19). This was the first study to investigate correlations between
GSH, amyloid and neuroinflammation in the in-vivo brain.
91
Figure 19. Model of AD pathology demonstrating possible correlations between beta-
amyloid, neuroinflammation and oxidative stress. AD pathology is complex and includes
numerous pathological features; one idea of the interplay between these 3 pathologies is as
follows: Aβ has been shown to reduce GSH levels by modulating the synthesis of the
antioxidant (1). Although the exact mechanism is unknown, possible explanations include
the inhibition of a cysteine transporter and modulation of enzymes involved in GSH
synthesis (Mandal et al. 2015). Microglia, one of the major cellular drivers of
neuroinflammation, may be related to an increase in GSH (2). Chronic activation of
microglia has been shown to result in the secretion of ROS (McGeer and McGeer 2010).
Thus it can be postulated that GSH would be increased to reduce the oxidative species
produced (2).
92
In aMCI patients, [11C]-PIB binding in the right, left and full DLPFC did not correlate
to GSH levels. All 3 regions were explored, as we assumed changes occur symmetrically
between the two hemispheres. No correlations were observed after partial volume correction.
Our results are in contrast to histopathological, animal and cell culture studies which have
demonstrated a correlation between amyloid accumulation and oxidative stress
(Anantharaman et al. 2006;Sultana et al. 2009). In a mouse model of AD, it was shown that
the overexpression of the amyloid precursor protein led to a decreased protein level of
EAAT3, the primary transporter of the rate-limiting amino acid in GSH synthesis, cysteine
(Nieoullon et al. 2006). Moreover, a cell culture study demonstrated the inhibition of
EAAT3 by Aβ oligomers (Hodgson et al. 2013).Thus it is expected that a negative
correlation between amyloid and GSH would have been observed. It is important to note that
our exploratory analyses between amyloid and GSH are underpowered and a larger sample
size is required to better understand whether there is a correlation between the two measures.
One additional possible explanation for the lack of correlation may be due to the fact that
[11C]-PIB binds to fibrillary amyloid beta aggregates and not soluble Aβ oligomers which are
more frequently correlated with GSH. In addition to amyloid, microglial activation can be
related to GSH levels. From our exploratory analyses, increased [18F]-FEPPA binding in the
left DLPFC and full DLPFC of aMCI patients was correlated with higher amounts of GSH.
Cell cultures studies have shown that highly activated microglia release free radicals (Boje
and Arora 1992;Chao et al. 1992;McGuire et al. 2001;McGeer and McGeer 2010). Thus one
plausible explanation for the positive association between [18F]-FEPPA and GSH is that the
brain is utilizing its antioxidant system to defend against free radicals produced by activated
93
microglia. The sample size for these exploratory correlations was small (9 aMCI), thus a
larger cohort of patients is required in order to confirm this relationship.
5.7 GSH and performance on neuropsychological tests
Two MRS studies in AD and MCI populations have reported correlations between
alterations in GSH levels and poorer performance on neuropsychological assessments (Duffy
et al. 2014;Mandal et al. 2015). In our sample of aMCI patients, higher levels of GSH were
correlated with better performance on an executive function task evaluating response
inhibition (Stroop). Our results are in contrast to another group that reported an association
between higher levels of GSH in the anterior cingulate and poorer performance on tests of set
shifting (TMT-B) and response inhibition (Stroop) (Duffy et al. 2014). The discrepancy
between the findings may be explained by the different brain regions assessed or differences
in pulse sequences used during acquisition. Additionally, our sample of patients is purely
amnestic, whereas the other study included both subtypes of MCI. A recent study with AD
and MCI patients reported an association between GSH reduction in the hippocampus and
frontal cortex and decline on global cognitive function, as measured by MMSE and CDR
(Mandal et al. 2015).This study suggests that less GSH is associated with an impairment,
which conversely might mean that more GSH is beneficial for cognition, as observed by our
exploratory correlations. Nevertheless, it should be noted that our sample size for the
correlation obtained was only 6 aMCI participants, thus a larger cohort of patients is required
to investigate correlations between GSH and cognition.
94
6. STRENGTHS
It is important consider some of the strengths and novelties of our study. Firstly, we
were the first to evaluate amyloid, neuroinflammation and glutathione in-vivo. Furthermore
we were the first to include only a HAB population of participants. It has previously been
shown that [18F]-FEPPA VT is 30% higher in HAB healthy participants in comparison to
MABs (Mizrahi et al. 2012). Thus the inclusion of only HABs reduces the variability of the
sample. Recently, a study demonstrated that the pathologic features, clinical phenotypes and
rate of cognitive decline in AD and MCI were similar in all 3 TSPO genotypes (Fan et al.
2015). Additionally, another study reported no significant association of TSPO genotype
with either degree of cerebral amyloid angiopathy or microglial activation (Felsky et al.
2016). Taken together, these findings support our rational for inclusion of only HABs.
Our study also encompasses several methodological strengths. Firstly, we used a
high-resolution research tomograph (HRRT) scanner, which is a dedicated human brain PET
scanner with improved spatial resolution and sensitivity. For the quantification of TSPO, we
used a second-generation radioligand that has several advantages over the prototypical
radioligand [11C]-PK1195 including a longer half-life, higher affinity, lower metabolization
and easier preparation. Additionally, we obtained the arterial input function for absolute
quantification of TSPO binding. Unlike other multi-tracer PET studies, all of our participants
underwent [11C]-PIB scans and were included in the study regardless of their PIB status.
Lastly, with respect to the MRS portion of the study, we used a pulse sequence that has been
reported to be better in the quantification of GSH, MEGA-PRESS, and a region of interest
with a good signal-to-noise ratio
95
7. LIMITATIONS
As with all imaging studies, certain limitations and confounding factors need to be
considered. Firstly, our sample size is small; however the inclusion of only HABs reduces
the variability of the sample. A caveat of TSPO PET studies is that radioligands bind to
TSPO expressed by astrocytes as well (Kreisl et al. 2013). Thus [18F]-FEPPA binding may
also be indicating the presence of reactive astrocytes in the brain, as currently there is no
direct evidence demonstrating that activated microglia are the main cellular source of [18F]-
FEPPA binding. However, a previous post-mortem study demonstrated that [3H]-PK11195
and [3H]-DAA1106 binding corresponded mainly to activated microglia (Venneti et al.
2008). Another consideration is that TSPO radioligands do not differentiate the two
phenotypes of microglia, M1 and M2, which are pro-inflammatory and neuroprotective,
respectively (Heneka et al. 2015). Perhaps in our elderly healthy volunteers, there are more
M2 microglia, whereas in the aMCI participants there are more M1 microglia, leading to an
overall similar level of microglial activation.
Although aMCI patients did not meet criteria for current Axis I disorder, 6 of the 11
participants were taking anti-depressants. There is some evidence that selective serotonin
reuptake inhibitors (SSRIs), selective norepinephrine reuptake inhibitors (SNRIs) and
tricyclic antidepressants (TCAs) can alter the inflammatory potential of microglia. Previous
studies have examined the ability of these anti-depressants to modulate microglial production
of cytokines (including TNF-α, IL-1β, IL-6) and the free radical nitric oxide (NO). Variable
results have been reported whereby some have reported no effects (Horikawa et al. 2010),
whereas others have reported an increase in production (Kubera et al. 2004;Ha et al.
96
2006;Tynan et al. 2012) or a decrease in production (Obuchowicz et al. 2006;Hashioka et al.
2007). Nevertheless, currently there is no evidence on the effect of SSRIs on TSPO
radioligand binding. A recent study investigating [18F]-FEPPA binding in AD, that similarly
included patients taking SSRIs, reported that differences in [18F]-FEPPA binding remained
significant in all GM and WM regions of interest even after excluding those on anti-
depressants (Suridjan et al. 2015). Similarly, within our sample, no differences in [18F]-
FEPPA binding were observed between patients on anti-depressants (n=6) when compared to
those not taking anti-depressants (n=5).
Other caveats should be considered in TSPO PET imaging studies. The lack of a
reference region (e.g. a region that does not have any specific binding) makes imaging TSPO
more difficult as arterial catheterization of the participant is required. Furthermore, arterial
sampling adds a potential source of error that may increase PET data variability (Lyoo et al.
2015). Recently a study investigating [11C]-PBR28 binding in AD and MCI participants
suggested that the cerebellum can be used as a pseudo-reference region and that the simple
ratio method may be more sensitive than the absolute quantitation method (Lyoo et al. 2015).
Our SUVR analysis was congruent with VT, whereby no significant differences were
observed between groups. Regarding the issue of possible errors in the arterial input
function, two of our healthy volunteers had high K1 values (ratio of delivery) and higher than
expected VT’s in regions of interest. The two participants were not taking any medications
and did not have any history of significant illness that may have contributed to the results.
Furthermore, neither of the individuals had an elevation in [11C]-PIB retention, thus we
cannot speculate that the increased microglial activation is due to amyloid pathology. When
97
the two healthy volunteers were removed from the analysis, our results remained the same,
with no significant differences between the two groups.
The in-vivo measurement of GSH has previously been difficult to quantify due to the
overlapping resonances of other brain metabolites. The MRS portion of our study is limited
by the small sample size. Additionally, we did not genotype our participants for the
polymorphism that was recently shown to influence GSH concentrations in-vivo. A GAG
trinucleotide repeat (TNR) polymorphism in the gene coding for the catalytic (GCLC)
subunit of glutamate-cysteine ligase (GCL), the rate-limiting enzyme for GSH synthesis, was
reported to influence GSH concentrations (Xin et al. 2016). This polymorphism may explain
the high variability in the GSH data. Similarly to other MRS studies, it is also to be noted
that our DLPFC voxels contained a fraction of white matter. Nevertheless, there were no
significant differences in the voxel composition between healthy volunteers and aMCI
participants. Finally, a caveat of MRS studies is that the technique does not differentiate
between GSH in neurons, glial cells or in extracellular pools, which may mask certain
differences between the two groups (Xin et al. 2016). Overall, a larger sample is required to
investigate whether GSH alterations are evident in this prodromal state.
.
98
8. CONCLUSION
In summary, this was the first study to investigate amyloid burden,
neuroinflammation and GSH in the in-vivo brain of aMCI and healthy volunteers. Our
findings indicate that amyloid deposition is an early pathological event, as evidenced by
increased [11C]-PIB binding in the cortical regions of aMCI patients. The lack of significant
difference in [18F]-FEPPA binding between aMCI patients and healthy volunteers may
suggest that neuroinflammation occurs later during the progression to AD. On the contrary,
the lack of significant difference in [18F]-FEPPA binding may be due to the fact that
individuals with and without amyloid were included in both groups. This speculation is
supported by our findings of correlations between [18F]-FEPPA and [11C]-PIB, and by the
finding of increased [18F]-FEPPA in the subset of aMCI patients that were characterized to
be PIB+. A larger sample size will be required to confirm our exploratory correlations
between amyloid and neuroinflammation. Furthermore, our results also suggest that
neuroinflammation, but not amyloid, may be related to an impairment in episodic memory.
With respect to GSH, our findings indicate that there are no significant alterations in
GSH levels in aMCI patients. We were the first to explore possible correlations between
GSH, neuroinflammation, and amyloid. Our pilot data suggests that GSH is positively
correlated with neuroinflammation, but not amyloid, in the dorsolateral prefrontal cortex.
Finally, higher GSH levels may be related to better performance on an executive function
task, Stroop Color Word Test.
99
9. FUTURE DIRECTIONS
Although neuroinflammation has been shown to be an important feature of AD
pathology, the timing and onset of this pathology is still not well understood. In order to gain
a better understanding a longitudinal [18F]-FEPPA study with aMCI and AD patients should
be performed to possibly elucidate the relationship between neuroinflammation and disease
progression. Moreover, this longitudinal study can include an additional group of
participants, those with subjective memory impairments but no diagnosis of aMCI, in order
to study the disease progression on a continuum from normal cognition to AD. Although the
inclusion of only one TSPO genetic group reduces the variability of the sample, and is one of
the strengths of our study, it makes recruitment much more difficult. Studies that may not
have access to large samples of patients may want to include participants that are MABs as
well.
To clarify associations between neuroinflammation and cognitive impairment, as
measured by neuropsychological assessments, large sample sizes are required. Furthermore,
future studies should include a battery of assessments, not just MMSE as is commonly seen
in PET studies of AD and MCI. Moreover, it would be interesting to include scores on
cognitive tests from healthy volunteers as well, in order to study correlations between
neuroinflammation and cognition on a continuum.
With respect to amyloid pathology, which has been demonstrated to be an early event
as observed by our PET study and others, it would be interesting to follow aMCI patients that
are both PIB- and PIB+ to investigate differences in neuroinflammation and conversion rates
to AD. Thus far there has only been one PET study that followed 5 MCI patients for 5 years
100
after their initial [11C]-DAA1106 scans (Yasuno et al. 2012). The group reported that all
subjects with initial [11C]-DAA1106 binding higher than control mean ±0.5 SD developed
dementia. This study however included MCI patients of both subtypes and did not evaluate
correlations with amyloid pathology.
Another important pathological feature of AD is the accumulation of neurofibrillary
tangles (NFTs) made up of hyperphospherylated tau (PHF-tau). Unlike amyloid-β plaques,
tau aggregates have been associated with cognitive decline and disease severity (Small et al.
2006;Chien et al. 2013). Tau deposition begins in a very limited area and then spreads as
clinical symptoms of dementia progress (Okamura et al. 2014). Moreover, the formations of
tau aggregates have been documented as preceding the cognitive symptoms of AD,
constituting them as a potentially reliable marker of early AD (Chien et al. 2013). In
comparison to amyloid, there are less PET studies that specifically target PHF-tau or NFTs.
The detection and better quantification of NFT burden may lead to potential therapeutics
down the line. Thus future studies should aim to image tau in aMCI and AD patients to better
characterize the underlying pathologies. Moreover, since neuroinflammation is thought to
aggregate tau pathology, a multi-tracer study can be performed with PET, whereby both
neuroinflammation and tau are quantified in-vivo (Heppner et al. 2015).
Given the complexity of AD, multi-modality imaging studies may be ideal in gaining
a better understanding of the different pathological features of this disease. For example,
future studies may want to investigate cortical thickness and structural atrophy, which can be
quantified with MRI, while imaging neuroinflammation with PET. These disease pathologies
can be measured over time to study progression. With the use of MRS brain metabolites,
indicative of atrophy, can be measured in combination with other PET or MRI measures.
101
Finally, in addition to the peripheral studies measuring markers of inflammation and
oxidative stress in the blood of AD and MCI patients, it would be interesting to correlate
measures in the blood to those in the brain. With regards to inflammation, a study can
investigate relationships between inflammatory cytokines and chemokines in the blood to
[18F]-FEPPA binding in the brain. With respect to oxidative stress, glutathione levels in the
blood can be correlated to glutathione levels in the brain with the use of MRS.
In sum, the use of multi-modality imaging techniques, including PET, MRI and
MRS, will allow for the quantification of different pathologies that characterize AD.
102
10. REFERENCES
Agostinho, P., A. Pliássova, C. R. Oliveira and R. A. Cunha (2015). "Localization and Trafficking of Amyloid-β Protein Precursor and Secretases: Impact on Alzheimer's Disease." Journal of Alzheimer's Disease 45(2): 329-347.
Aksenov, M., M. Aksenova, D. Butterfield, J. Geddes and W. Markesbery (2001). "Protein oxidation in the brain in Alzheimer's disease." Neuroscience 103(2): 373-383.
Albert, M. S., S. T. DeKosky, D. Dickson, B. Dubois, H. H. Feldman, N. C. Fox, A. Gamst, D. M. Holtzman, W. J. Jagust and R. C. Petersen (2011). "The diagnosis of mild cognitive impairment due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer's disease." Alzheimer's & dementia 7(3): 270-279.
Anantharaman, M., J. Tangpong, J. N. Keller, M. P. Murphy, W. R. Markesbery, K. K. Kiningham and D. K. S. Clair (2006). "β-Amyloid mediated nitration of manganese superoxide dismutase: implication for oxidative stress in a APPNLh/NLh X PS-1P264L/P264L double knock-in mouse model of Alzheimer's disease." The American journal of pathology 168(5): 1608-1618.
Ansari, M. A. and S. W. Scheff (2010). "Oxidative stress in the progression of Alzheimer disease in the frontal cortex." Journal of neuropathology and experimental neurology 69(2): 155.
Araujo, D. and P. Lapchak (1994). "Induction of immune system mediators in the hippocampal formation in Alzheimer's and Parkinson's diseases: selective effects on specific interleukins and interleukin receptors." Neuroscience 61(4): 745-754.
Arends, Y., C. Duyckaerts, J. Rozemuller, P. Eikelenboom and J. Hauw (2000). "Microglia, amyloid and dementia in alzheimer disease: a correlative study." Neurobiology of aging 21(1): 39-47.
Arnold, S. E., B. T. Hyman, J. Flory, A. R. Damasio and G. W. Van Hoesen (1991). "The topographical and neuroanatomical distribution of neurofibrillary tangles and neuritic plaques in the cerebral cortex of patients with Alzheimer's disease." Cerebral cortex 1(1): 103-116.
Arriagada, P. V., J. H. Growdon, E. T. Hedley-Whyte and B. T. Hyman (1992). "Neurofibrillary tangles but not senile plaques parallel duration and severity of Alzheimer's disease." Neurology 42(3): 631-631.
Asai, H., S. Ikezu, S. Tsunoda, M. Medalla, J. Luebke, T. Haydar, B. Wolozin, O. Butovsky, S. Kügler and T. Ikezu (2015). "Depletion of microglia and inhibition of exosome synthesis halt tau propagation." Nature neuroscience.
103
Ashendorf, L., A. L. Jefferson, M. K. O’Connor, C. Chaisson, R. C. Green and R. A. Stern (2008). "Trail Making Test errors in normal aging, mild cognitive impairment, and dementia." Archives of Clinical Neuropsychology 23(2): 129-137.
Bacskai, B. J., M. P. Frosch, S. H. Freeman, S. B. Raymond, J. C. Augustinack, K. A. Johnson, M. C. Irizarry, W. E. Klunk, C. A. Mathis and S. T. DeKosky (2007). "Molecular imaging with Pittsburgh Compound B confirmed at autopsy: a case report." Archives of neurology 64(3): 431-434.
Banati, R., J. Newcombe, R. Gunn, A. Cagnin, F. Turkheimer, F. Heppner, G. Price, F. Wegner, G. Giovannoni and D. Miller (2000). "The peripheral benzodiazepine binding site in the brain in multiple sclerosis." Brain 123(11): 2321-2337.
Bennacef, I., C. Salinas, G. Horvath, R. Gunn, T. Bonasera, A. Wilson, A. Gee and M. Laruelle (2008). Comparison of [11C] PBR28 and [18F] FEPPA as CNS peripheral benzodiazepine receptor PET ligands in the pig. Society of Nuclear Medicine Annual Meeting Abstracts, Soc Nuclear Med.
Bennett, D., J. Schneider, Z. Arvanitakis, J. Kelly, N. Aggarwal, R. Shah and R. Wilson (2006). "Neuropathology of older persons without cognitive impairment from two community-based studies." Neurology 66(12): 1837-1844.
Benveniste, E. N. (1997). "Role of macrophages/microglia in multiple sclerosis and experimental allergic encephalomyelitis." Journal of molecular medicine 75(3): 165-173.
Bermejo, P., S. Martín-Aragón, J. Benedí, C. Susín, E. Felici, P. Gil, J. Manuel Ribera and Á. M. Villar (2008). "Peripheral levels of glutathione and protein oxidation as markers in the development of Alzheimer's disease from Mild Cognitive Impairment." Free radical research 42(2): 162-170.
Bermejo, P., S. Martín-Aragón, J. Benedí, C. Susín, E. Felici, P. Gil, J. M. Ribera and Á. M. Villar (2008). "Differences of peripheral inflammatory markers between mild cognitive impairment and Alzheimer's disease." Immunology letters 117(2): 198-202.
Bierer, L. M., P. R. Hof, D. P. Purohit, L. Carlin, J. Schmeidler, K. L. Davis and D. P. Perl (1995). "Neocortical neurofibrillary tangles correlate with dementia severity in Alzheimer's disease." Archives of neurology 52(1): 81-88.
Boje, K. M. and P. K. Arora (1992). "Microglial-produced nitric oxide and reactive nitrogen oxides mediate neuronal cell death." Brain research 587(2): 250-256.
Boyle, P., R. Wilson, N. Aggarwal, Y. Tang and D. Bennett (2006). "Mild cognitive impairment Risk of Alzheimer disease and rate of cognitive decline." Neurology 67(3): 441-445.
Braak, H. and E. Braak (1995). "Staging of Alzheimer's disease-related neurofibrillary changes." Neurobiology of aging 16(3): 271-278.
104
Braak, H. and E. Braak (1997). "Frequency of stages of Alzheimer-related lesions in different age categories." Neurobiology of aging 18(4): 351-357.
Braestrup, C. and R. F. Squires (1977). "Specific benzodiazepine receptors in rat brain characterized by high-affinity (3H) diazepam binding." Proceedings of the National Academy of Sciences 74(9): 3805-3809.
Brites, D. and A. R. Vaz (2015). "Microglia centered pathogenesis in ALS: insights in cell interconnectivity." Frontiers in cellular neuroscience 8.
Butovsky, O., A. E. Talpalar, K. Ben-Yaakov and M. Schwartz (2005). "Activation of microglia by aggregated β-amyloid or lipopolysaccharide impairs MHC-II expression and renders them cytotoxic whereas IFN-γ and IL-4 render them protective." Molecular and Cellular Neuroscience 29(3): 381-393.
Cagnin, A., D. J. Brooks, A. M. Kennedy, R. N. Gunn, R. Myers, F. E. Turkheimer, T. Jones and R. B. Banati (2001). "In-vivo measurement of activated microglia in dementia." The Lancet 358(9280): 461-467.
Carpenter, A. F., P. W. Carpenter and W. R. Markesbery (1993). "Morphometric analysis of microglia in Alzheimer's disease." Journal of Neuropathology & Experimental Neurology 52(6): 601-608.
Chao, C., S. Hu, T. Molitor, E. Shaskan and P. Peterson (1992). "Activated microglia mediate neuronal cell injury via a nitric oxide mechanism." The Journal of Immunology 149(8): 2736-2741.
Chauveau, F., H. Boutin, N. Van Camp, F. Dollé and B. Tavitian (2008). "Nuclear imaging of neuroinflammation: a comprehensive review of [11C] PK11195 challengers." European journal of nuclear medicine and molecular imaging 35(12): 2304-2319.
Chen, M.-K. and T. R. Guilarte (2008). "Translocator protein 18 kDa (TSPO): molecular sensor of brain injury and repair." Pharmacology & therapeutics 118(1): 1-17.
Chien, D. T., S. Bahri, A. K. Szardenings, J. C. Walsh, F. Mu, M.-Y. Su, W. R. Shankle, A. Elizarov and H. C. Kolb (2013). "Early clinical PET imaging results with the novel PHF-tau radioligand [F-18]-T807." Journal of Alzheimer's Disease 34(2): 457-468.
Choi, J., C. A. Malakowsky, J. M. Talent, C. C. Conrad and R. W. Gracy (2002). "Identification of oxidized plasma proteins in Alzheimer's disease." Biochemical and biophysical research communications 293(5): 1566-1570.
Conrad, C. C., P. L. Marshall, J. M. Talent, C. A. Malakowsky, J. Choi and R. W. Gracy (2000). "Oxidized proteins in Alzheimer's plasma." Biochemical and Biophysical Research Communications 275(2): 678-681.
105
Cristalli, D. O., N. Arnal, F. A. Marra, M. J. de Alaniz and C. A. Marra (2012). "Peripheral markers in neurodegenerative patients and their first-degree relatives." Journal of the neurological sciences 314(1): 48-56.
Dickson, D. W., S. C. Lee, L. A. Mattiace, S. H. C. Yen and C. Brosnan (1993). "Microglia and cytokines in neurological disease, with special reference to AIDS and Alzheimer's disease." Glia 7(1): 75-83.
Duffy, S. L., J. Lagopoulos, I. B. Hickie, K. Diamond, M. B. Graeber, S. J. Lewis and S. L. Naismith (2014). "Glutathione relates to neuropsychological functioning in mild cognitive impairment." Alzheimer's & Dementia 10(1): 67-75.
Edison, P., H. Archer, R. Hinz, A. Hammers, N. Pavese, Y. Tai, G. Hotton, D. Cutler, N. Fox and A. Kennedy (2007). "Amyloid, hypometabolism, and cognition in Alzheimer disease An [11C] PIB and [18F] FDG PET study." Neurology 68(7): 501-508.
Edison, P., H. A. Archer, A. Gerhard, R. Hinz, N. Pavese, F. E. Turkheimer, A. Hammers, Y. F. Tai, N. Fox and A. Kennedy (2008). "Microglia, amyloid, and cognition in Alzheimer's disease: An [11C](R) PK11195-PET and [11C] PIB-PET study." Neurobiology of disease 32(3): 412-419.
Fan, Z., D. Harold, G. Pasqualetti, J. Williams, D. J. Brooks and P. Edison (2015). "Can studies of neuroinflammation in a TSPO genetic subgroup (HAB or MAB) be applied to the entire AD cohort?" Journal of Nuclear Medicine 56(5): 707-713.
Felsky, D., P. L. De Jager, J. A. Schneider, K. Arfanakis, D. A. Fleischman, Z. Arvanitakis, W. G. Honer, J. G. Pouget, R. Mizrahi and B. G. Pollock (2016). "Cerebrovascular and microglial states are not altered by functional neuroinflammatory gene variant." Journal of Cerebral Blood Flow & Metabolism: 0271678X15626719.
Fleisher, A. S., K. Chen, X. Liu, A. Roontiva, P. Thiyyagura, N. Ayutyanont, A. D. Joshi, C. M. Clark, M. A. Mintun and M. J. Pontecorvo (2011). "Using positron emission tomography and florbetapir F 18 to image cortical amyloid in patients with mild cognitive impairment or dementia due to Alzheimer disease." Archives of neurology 68(11): 1404-1411.
Folstein, M. F., S. E. Folstein and P. R. McHugh (1975). "“Mini-mental state”: a practical method for grading the cognitive state of patients for the clinician." Journal of psychiatric research 12(3): 189-198.
Forlenza, O. V., B. S. Diniz, L. L. Talib, V. A. Mendonça, E. B. Ojopi, W. F. Gattaz and A. L. Teixeira (2009). "Increased serum IL-1β level in Alzheimer’s disease and mild cognitive impairment." Dementia and geriatric cognitive disorders 28(6): 507-512.
Forsberg, A., H. Engler, O. Almkvist, G. Blomquist, G. Hagman, A. Wall, A. Ringheim, B. Långström and A. Nordberg (2008). "PET imaging of amyloid deposition in patients with mild cognitive impairment." Neurobiology of aging 29(10): 1456-1465.
106
Frautschy, S. A., F. Yang, M. Irrizarry, B. Hyman, T. Saido, K. Hsiao and G. M. Cole (1998). "Microglial response to amyloid plaques in APPsw transgenic mice." The American journal of pathology 152(1): 307.
Galimberti, D., C. Fenoglio, C. Lovati, E. Venturelli, I. Guidi, B. Corrà, D. Scalabrini, F. Clerici, C. Mariani and N. Bresolin (2006). "Serum MCP-1 levels are increased in mild cognitive impairment and mild Alzheimer's disease." Neurobiology of aging 27(12): 1763-1768.
Gauthier, S., B. Reisberg, M. Zaudig, R. C. Petersen, K. Ritchie, K. Broich, S. Belleville, H. Brodaty, D. Bennett and H. Chertkow (2006). "Mild cognitive impairment." The Lancet 367(9518): 1262-1270.
Gerhard, A., B. Neumaier, E. Elitok, G. Glatting, V. Ries, R. Tomczak, A. C. Ludolph and S. N. Reske (2000). "In vivo imaging of activated microglia using [11 C] PK11195 and positron emission tomography in patients after ischemic stroke." Neuroreport 11(13): 2957-2960.
Gerhard, A., N. Pavese, G. Hotton, F. Turkheimer, M. Es, A. Hammers, K. Eggert, W. Oertel, R. B. Banati and D. J. Brooks (2006). "In vivo imaging of microglial activation with [11 C](R)-PK11195 PET in idiopathic Parkinson's disease." Neurobiology of disease 21(2): 404-412.
Ghosh, D., K. R. LeVault, A. J. Barnett and G. J. Brewer (2012). "A reversible early oxidized redox state that precedes macromolecular ROS damage in aging nontransgenic and 3xTg-AD mouse neurons." The Journal of Neuroscience 32(17): 5821-5832.
Ghosh, D., K. R. LeVault and G. J. Brewer (2014). "Dual-energy precursor and nuclear erythroid–related factor 2 activator treatment additively improve redox glutathione levels and neuron survival in aging and Alzheimer mouse neurons upstream of reactive oxygen species." Neurobiology of aging 35(1): 179-190.
Good, P. F., P. Werner, A. Hsu, C. W. Olanow and D. P. Perl (1996). "Evidence of neuronal oxidative damage in Alzheimer's disease." The American journal of pathology 149(1): 21.
Grober, E., M. Sliwinsk and S. R. Korey (1991). "Development and validation of a model for estimating premorbid verbal intelligence in the elderly." Journal of Clinical and Experimental Neuropsychology 13(6): 933-949.
Guerreiro, R. J., I. Santana, J. M. Brás, B. Santiago, A. Paiva and C. Oliveira (2007). "Peripheral inflammatory cytokines as biomarkers in Alzheimer’s disease and mild cognitive impairment." Neurodegenerative Diseases 4(6): 406-412.
Gujar, S. K., S. Maheshwari, I. Björkman-Burtscher and P. C. Sundgren (2005). "Magnetic resonance spectroscopy." Journal of neuro-ophthalmology 25(3): 217-226.
Gulyás, B., B. Makkai, P. Kása, K. Gulya, L. Bakota, S. Várszegi, Z. Beliczai, J. Andersson, L. Csiba and A. Thiele (2009). "A comparative autoradiography study in post mortem whole hemisphere human brain slices taken from Alzheimer patients and age-matched controls
107
using two radiolabelled DAA1106 analogues with high affinity to the peripheral benzodiazepine receptor (PBR) system." Neurochemistry international 54(1): 28-36.
Ha, E., K. H. Jung, B.-K. Choe, J.-H. Bae, D.-H. Shin, S.-V. Yim and H. H. Baik (2006). "Fluoxetine increases the nitric oxide production via nuclear factor kappa B-mediated pathway in BV 2 murine microglial cells." Neuroscience letters 397(3): 185-189.
Hashioka, S., A. Klegeris, A. Monji, T. Kato, M. Sawada, P. L. McGeer and S. Kanba (2007). "Antidepressants inhibit interferon-γ-induced microglial production of IL-6 and nitric oxide." Experimental neurology 206(1): 33-42.
Heneka, M. T., M. J. Carson, J. El Khoury, G. E. Landreth, F. Brosseron, D. L. Feinstein, A. H. Jacobs, T. Wyss-Coray, J. Vitorica and R. M. Ransohoff (2015). "Neuroinflammation in Alzheimer's disease." The Lancet Neurology 14(4): 388-405.
Heppner, F. L., R. M. Ransohoff and B. Becher (2015). "Immune attack: the role of inflammation in Alzheimer disease." Nature Reviews Neuroscience 16(6): 358-372.
Hickman, S. E., E. K. Allison and J. El Khoury (2008). "Microglial dysfunction and defective β-amyloid clearance pathways in aging Alzheimer's disease mice." The Journal of neuroscience 28(33): 8354-8360.
Hodgson, N., M. Trivedi, C. Muratore, S. Li and R. Deth (2013). "Soluble oligomers of amyloid-β cause changes in redox state, DNA methylation, and gene transcription by inhibiting EAAT3 mediated cysteine uptake." Journal of Alzheimer's Disease 36(1): 197-209.
Horikawa, H., T. A. Kato, Y. Mizoguchi, A. Monji, Y. Seki, T. Ohkuri, L. Gotoh, M. Yonaha, T. Ueda and S. Hashioka (2010). "Inhibitory effects of SSRIs on IFN-γ induced microglial activation through the regulation of intracellular calcium." Progress in Neuro-Psychopharmacology and Biological Psychiatry 34(7): 1306-1316.
Huijbers, W., E. C. Mormino, A. P. Schultz, S. Wigman, A. M. Ward, M. Larvie, R. E. Amariglio, G. A. Marshall, D. M. Rentz and K. A. Johnson (2015). "Amyloid-β deposition in mild cognitive impairment is associated with increased hippocampal activity, atrophy and clinical progression." Brain: awv007.
Ikonomovic, M. D., W. E. Klunk, E. E. Abrahamson, C. A. Mathis, J. C. Price, N. D. Tsopelas, B. J. Lopresti, S. Ziolko, W. Bi and W. R. Paljug (2008). "Post-mortem correlates of in vivo PiB-PET amyloid imaging in a typical case of Alzheimer's disease." Brain 131(6): 1630-1645.
Innis, R. B., V. J. Cunningham, J. Delforge, M. Fujita, A. Gjedde, R. N. Gunn, J. Holden, S. Houle, S.-C. Huang and M. Ichise (2007). "Consensus nomenclature for in vivo imaging of reversibly binding radioligands." Journal of Cerebral Blood Flow & Metabolism 27(9): 1533-1539.
108
Itagaki, S., P. McGeer, H. Akiyama, S. Zhu and D. Selkoe (1989). "Relationship of microglia and astrocytes to amyloid deposits of Alzheimer disease." Journal of neuroimmunology 24(3): 173-182.
Jack, C. R., V. J. Lowe, M. L. Senjem, S. D. Weigand, B. J. Kemp, M. M. Shiung, D. S. Knopman, B. F. Boeve, W. E. Klunk and C. A. Mathis (2008). "11C PiB and structural MRI provide complementary information in imaging of Alzheimer's disease and amnestic mild cognitive impairment." Brain 131(3): 665-680.
Jack, C. R., H. J. Wiste, P. Vemuri, S. D. Weigand, M. L. Senjem, G. Zeng, M. A. Bernstein, J. L. Gunter, V. S. Pankratz and P. S. Aisen (2010). "Brain beta-amyloid measures and magnetic resonance imaging atrophy both predict time-to-progression from mild cognitive impairment to Alzheimer’s disease." Brain 133(11): 3336-3348.
Kemppainen, N., S. Aalto, I. Wilson, K. Någren, S. Helin, A. Brück, V. Oikonen, M. Kailajärvi, M. Scheinin and M. Viitanen (2006). "Voxel-based analysis of PET amyloid ligand [11C] PIB uptake in Alzheimer disease." Neurology 67(9): 1575-1580.
Kemppainen, N., S. Aalto, I. Wilson, K. Någren, S. Helin, A. Brück, V. Oikonen, M. Kailajärvi, M. Scheinin and M. Viitanen (2007). "PET amyloid ligand [11C] PIB uptake is increased in mild cognitive impairment." Neurology 68(19): 1603-1606.
Klunk, W. E., H. Engler, A. Nordberg, Y. Wang, G. Blomqvist, D. P. Holt, M. Bergström, I. Savitcheva, G. F. Huang and S. Estrada (2004). "Imaging brain amyloid in Alzheimer's disease with Pittsburgh Compound‐B." Annals of neurology 55(3): 306-319.
Klunk, W. E., B. J. Lopresti, M. D. Ikonomovic, I. M. Lefterov, R. P. Koldamova, E. E. Abrahamson, M. L. Debnath, D. P. Holt, G.-f. Huang and L. Shao (2005). "Binding of the positron emission tomography tracer Pittsburgh compound-B reflects the amount of amyloid-β in Alzheimer's disease brain but not in transgenic mouse brain." The Journal of neuroscience 25(46): 10598-10606.
Klunk, W. E., Y. Wang, G.-f. Huang, M. L. Debnath, D. P. Holt, L. Shao, R. L. Hamilton, M. D. Ikonomovic, S. T. DeKosky and C. A. Mathis (2003). "The binding of 2-(4′-methylaminophenyl) benzothiazole to postmortem brain homogenates is dominated by the amyloid component." The Journal of neuroscience 23(6): 2086-2092.
Krabbe, G., A. Halle, V. Matyash, J. L. Rinnenthal, G. D. Eom, U. Bernhardt, K. R. Miller, S. Prokop, H. Kettenmann and F. L. Heppner (2013). "Functional impairment of microglia coincides with Beta-amyloid deposition in mice with Alzheimer-like pathology." PloS one 8(4): e60921.
Kreisl, W. C., K. J. Jenko, C. S. Hines, C. H. Lyoo, W. Corona, C. L. Morse, S. S. Zoghbi, T. Hyde, J. E. Kleinman and V. W. Pike (2013). "A genetic polymorphism for translocator protein 18 kDa affects both in vitro and in vivo radioligand binding in human brain to this putative biomarker of neuroinflammation." Journal of Cerebral Blood Flow & Metabolism 33(1): 53-58.
109
Kreisl, W. C., C. H. Lyoo, M. McGwier, J. Snow, K. J. Jenko, N. Kimura, W. Corona, C. L. Morse, S. S. Zoghbi and V. W. Pike (2013). "In vivo radioligand binding to translocator protein correlates with severity of Alzheimer’s disease." Brain 136(7): 2228-2238.
Krstic, D. and I. Knuesel (2013). "Deciphering the mechanism underlying late-onset Alzheimer disease." Nature Reviews Neurology 9(1): 25-34.
Kubera, M., G. Kenis, E. Bosmans, M. Kajta, A. Basta-Kaim, S. Scharpe, B. Budziszewska and M. Maes (2004). "Stimulatory effect of antidepressants on the production of IL-6." International immunopharmacology 4(2): 185-192.
Kuhlmann, A. C. and T. R. Guilarte (2000). "Cellular and subcellular localization of peripheral benzodiazepine receptors after trimethyltin neurotoxicity." Journal of neurochemistry 74(4): 1694-1704.
Landau, S. M., C. Breault, A. D. Joshi, M. Pontecorvo, C. A. Mathis, W. J. Jagust and M. A. Mintun (2013). "Amyloid-β imaging with Pittsburgh compound B and florbetapir: comparing radiotracers and quantification methods." Journal of Nuclear Medicine 54(1): 70-77.
Leinonen, V., I. Alafuzoff, S. Aalto, T. Suotunen, S. Savolainen, K. Någren, T. Tapiola, T. Pirttilä, J. Rinne and J. E. Jääskeläinen (2008). "Assessment of β-amyloid in a frontal cortical brain biopsy specimen and by positron emission tomography with carbon 11–labeled Pittsburgh Compound B." Archives of Neurology 65(10): 1304-1309.
Leissring, M. A., W. Farris, A. Y. Chang, D. M. Walsh, X. Wu, X. Sun, M. P. Frosch and D. J. Selkoe (2003). "Enhanced proteolysis of β-amyloid in APP transgenic mice prevents plaque formation, secondary pathology, and premature death." Neuron 40(6): 1087-1093.
Liu, B. and J.-S. Hong (2003). "Role of microglia in inflammation-mediated neurodegenerative diseases: mechanisms and strategies for therapeutic intervention." Journal of Pharmacology and Experimental Therapeutics 304(1): 1-7.
Logan, J., J. S. Fowler, N. D. Volkow, G.-J. Wang, Y.-S. Ding and D. L. Alexoff (1996). "Distribution volume ratios without blood sampling from graphical analysis of PET data." Journal of Cerebral Blood Flow & Metabolism 16(5): 834-840.
Lopresti, B. J., W. E. Klunk, C. A. Mathis, J. A. Hoge, S. K. Ziolko, X. Lu, C. C. Meltzer, K. Schimmel, N. D. Tsopelas and S. T. DeKosky (2005). "Simplified quantification of Pittsburgh Compound B amyloid imaging PET studies: a comparative analysis." Journal of Nuclear Medicine 46(12): 1959-1972.
Lovell, M. A., W. D. Ehmann, S. M. Butler and W. R. Markesbery (1995). "Elevated thiobarbituric acid-reactive substances and antioxidant enzyme activity in the brain in Alzheimer's disease." Neurology 45(8): 1594-1601.
Lyoo, C. H., M. Ikawa, J.-S. Liow, S. S. Zoghbi, C. L. Morse, V. W. Pike, M. Fujita, R. B. Innis and W. C. Kreisl (2015). "Cerebellum can serve as a pseudo-reference region in
110
Alzheimer disease to detect neuroinflammation measured with PET radioligand binding to translocator protein." Journal of Nuclear Medicine 56(5): 701-706.
Mandal, P. K., S. Saharan, M. Tripathi and G. Murari (2015). "Brain Glutathione Levels–A Novel Biomarker for Mild Cognitive Impairment and Alzheimer’s Disease." Biological psychiatry 78(10): 702-710.
Mandal, P. K., M. Tripathi and S. Sugunan (2012). "Brain oxidative stress: Detection and mapping of anti-oxidant marker ‘Glutathione’in different brain regions of healthy male/female, MCI and Alzheimer patients using non-invasive magnetic resonance spectroscopy." Biochemical and biophysical research communications 417(1): 43-48.
Marjanska, M., G. L. Curran, T. M. Wengenack, P.-G. Henry, R. L. Bliss, J. F. Poduslo, C. R. Jack, K. Uğurbil and M. Garwood (2005). "Monitoring disease progression in transgenic mouse models of Alzheimer's disease with proton magnetic resonance spectroscopy." Proceedings of the national academy of sciences of the United States of America 102(33): 11906-11910.
Maruyama, M., H. Shimada, T. Suhara, H. Shinotoh, B. Ji, J. Maeda, M.-R. Zhang, J. Q. Trojanowski, V. M.-Y. Lee and M. Ono (2013). "Imaging of tau pathology in a tauopathy mouse model and in Alzheimer patients compared to normal controls." Neuron 79(6): 1094-1108.
Mathis, C. A., B. J. Bacskai, S. T. Kajdasz, M. E. McLellan, M. P. Frosch, B. T. Hyman, D. P. Holt, Y. Wang, G.-F. Huang and M. L. Debnath (2002). "A lipophilic thioflavin-T derivative for positron emission tomography (PET) imaging of amyloid in brain." Bioorganic & medicinal chemistry letters 12(3): 295-298.
Matsuzawa, D. and K. Hashimoto (2011). "Magnetic resonance spectroscopy study of the antioxidant defense system in schizophrenia." Antioxidants & redox signaling 15(7): 2057-2065.
McGeer, E. G. and P. L. McGeer (2010). "Neuroinflammation in Alzheimer's disease and mild cognitive impairment: a field in its infancy." Journal of Alzheimer's Disease 19(1): 355-361.
McGeer, P., H. Akiyama, S. Itagaki and E. McGeer (1989). "Activation of the classical complement pathway in brain tissue of Alzheimer patients." Neuroscience letters 107(1): 341-346.
McGeer, P., S. Itagaki, B. Boyes and E. McGeer (1988). "Reactive microglia are positive for HLA‐DR in the substantia nigra of Parkinson's and Alzheimer's disease brains." Neurology 38(8): 1285-1285.
McGuire, S. O., Z. D. Ling, J. W. Lipton, C. E. Sortwell, T. J. Collier and P. M. Carvey (2001). "Tumor necrosis factor α is toxic to embryonic mesencephalic dopamine neurons." Experimental neurology 169(2): 219-230.
111
Mecocci, P., U. MacGarvey and M. F. Beal (1994). "Oxidative damage to mitochondrial DNA is increased in Alzheimer's disease." Annals of neurology 36(5): 747-751.
Mescher, M., H. Merkle, J. Kirsch, M. Garwood and R. Gruetter (1998). "Simultaneous in vivo spectral editing and water suppression." NMR in Biomedicine 11(EPFL-ARTICLE-177509): 266-272.
Metastasio, A., P. Rinaldi, R. Tarducci, E. Mariani, F. T. Feliziani, A. Cherubini, G. P. Pelliccioli, G. Gobbi, U. Senin and P. Mecocci (2006). "Conversion of MCI to dementia: role of proton magnetic resonance spectroscopy." Neurobiology of aging 27(7): 926-932.
Mintun, M., G. Larossa, Y. Sheline, C. Dence, S. Y. Lee, R. Mach, W. Klunk, C. Mathis, S. DeKosky and J. Morris (2006). "[11C] PIB in a nondemented population Potential antecedent marker of Alzheimer disease." Neurology 67(3): 446-452.
Mizrahi, R., P. M. Rusjan, J. Kennedy, B. Pollock, B. Mulsant, I. Suridjan, V. De Luca, A. A. Wilson and S. Houle (2012). "Translocator protein (18 kDa) polymorphism (rs6971) explains in-vivo brain binding affinity of the PET radioligand [18F]-FEPPA." Journal of Cerebral Blood Flow & Metabolism 32(6): 968-972.
Monsch, A. U., M. W. Bondi, N. Butters, D. P. Salmon, R. Katzman and L. J. Thal (1992). "Comparisons of verbal fluency tasks in the detection of dementia of the Alzheimer type." Archives of Neurology 49(12): 1253-1258.
Montine, T. J., J. F. Quinn, D. Milatovic, L. C. Silbert, T. Dang, S. Sanchez, E. Terry, L. J. Roberts, J. A. Kaye and J. D. Morrow (2002). "Peripheral F2‐isoprostanes and F4‐neuroprostanes are not increased in Alzheimer's disease." Annals of neurology 52(2): 175-179.
Müller-Gärtner, H. W., J. M. Links, J. L. Prince, R. N. Bryan, E. McVeigh, J. P. Leal, C. Davatzikos and J. J. Frost (1992). "Measurement of radiotracer concentration in brain gray matter using positron emission tomography: MRI-based correction for partial volume effects." Journal of Cerebral Blood Flow & Metabolism 12(4): 571-583.
Murphy, K. J., A. K. Troyer, B. Levine and M. Moscovitch (2008). "Episodic, but not semantic, autobiographical memory is reduced in amnestic mild cognitive impairment." Neuropsychologia 46(13): 3116-3123.
Nasreddine, Z. S., N. A. Phillips, V. Bédirian, S. Charbonneau, V. Whitehead, I. Collin, J. L. Cummings and H. Chertkow (2005). "The Montreal Cognitive Assessment, MoCA: a brief screening tool for mild cognitive impairment." Journal of the American Geriatrics Society 53(4): 695-699.
Neumaier, B., S. Deisenhofer, D. Fürst, C. von Arnim, S. Thees, A. Buck, G. Glatting, G. Landwehrmeyer, B. Krause and H. Müller (2007). "Radiosynthesisandevaluationof [11C] BTA-1 and [11C] 3'-Me-BTA-1aspotentialradiotracers forinvivoimagingofβ-amyloidplaques." Nuklearmedizin 46: 271-280.
112
Nieoullon, A., B. Canolle, F. Masmejean, B. Guillet, P. Pisano and S. Lortet (2006). "The neuronal excitatory amino acid transporter EAAC1/EAAT3: does it represent a major actor at the brain excitatory synapse?" Journal of neurochemistry 98(4): 1007-1018.
Nordberg, A. (2004). "PET imaging of amyloid in Alzheimer's disease." The lancet neurology 3(9): 519-527.
Obuchowicz, E., J. Kowalski, K. Labuzek, R. Krysiak, J. Pendzich and Z. S. Herman (2006). "Amitriptyline and nortriptyline inhibit interleukin-1β and tumour necrosis factor-α release by rat mixed glial and microglial cell cultures." The International Journal of Neuropsychopharmacology 9(01): 27-35.
Okamura, N., S. Furumoto, M. T. Fodero-Tavoletti, R. S. Mulligan, R. Harada, P. Yates, S. Pejoska, Y. Kudo, C. L. Masters and K. Yanai (2014). "Non-invasive assessment of Alzheimer’s disease neurofibrillary pathology using 18F-THK5105 PET." Brain: awu064.
Okello, A., P. Edison, H. Archer, F. Turkheimer, J. Kennedy, R. Bullock, Z. Walker, A. Kennedy, N. Fox and M. Rossor (2009). "Microglial activation and amyloid deposition in mild cognitive impairment A PET study." Neurology 72(1): 56-62.
Owen, D. and P. M. Matthews (2011). "Imaging brain microglial activation using positron emission tomography and translocator protein-specific radioligands." Int Rev Neurobiol 101: 19-39.
Owen, D. R., A. J. Yeo, R. N. Gunn, K. Song, G. Wadsworth, A. Lewis, C. Rhodes, D. J. Pulford, I. Bennacef and C. A. Parker (2012). "An 18-kDa translocator protein (TSPO) polymorphism explains differences in binding affinity of the PET radioligand PBR28." Journal of Cerebral Blood Flow & Metabolism 32(1): 1-5.
Palmer, K., L. Bäckman, B. Winblad and L. Fratiglioni (2008). "Mild cognitive impairment in the general population: occurrence and progression to Alzheimer disease." The American Journal of Geriatric Psychiatry 16(7): 603-611.
Pamplona, R., E. Dalfó, V. Ayala, M. J. Bellmunt, J. Prat, I. Ferrer and M. Portero-Otín (2005). "Proteins in human brain cortex are modified by oxidation, glycoxidation, and lipoxidation Effects of Alzheimer disease and identification of lipoxidation targets." Journal of Biological Chemistry 280(22): 21522-21530.
Papadopoulos, V., M. Baraldi, T. R. Guilarte, T. B. Knudsen, J.-J. Lacapère, P. Lindemann, M. D. Norenberg, D. Nutt, A. Weizman and M.-R. Zhang (2006). "Translocator protein (18kDa): new nomenclature for the peripheral-type benzodiazepine receptor based on its structure and molecular function." Trends in pharmacological sciences 27(8): 402-409.
Pavese, N., A. Gerhard, Y. Tai, A. Ho, F. Turkheimer, R. Barker, D. Brooks and P. Piccini (2006). "Microglial activation correlates with severity in Huntington disease A clinical and PET study." Neurology 66(11): 1638-1643.
113
Perry, V. H., J. A. Nicoll and C. Holmes (2010). "Microglia in neurodegenerative disease." Nature Reviews Neurology 6(4): 193-201.
Petersen, R. C. (2004). "Mild cognitive impairment as a diagnostic entity." Journal of internal medicine 256(3): 183-194.
Petersen, R. C., R. Doody, A. Kurz, R. C. Mohs, J. C. Morris, P. V. Rabins, K. Ritchie, M. Rossor, L. Thal and B. Winblad (2001). "Current concepts in mild cognitive impairment." Archives of neurology 58(12): 1985-1992.
Petersen, R. C., J. E. Parisi, D. W. Dickson, K. A. Johnson, D. S. Knopman, B. F. Boeve, G. A. Jicha, R. J. Ivnik, G. E. Smith and E. G. Tangalos (2006). "Neuropathologic features of amnestic mild cognitive impairment." Archives of neurology 63(5): 665-672.
Phelps, M. E. (2000). "Positron emission tomography provides molecular imaging of biological processes." Proceedings of the National Academy of Sciences 97(16): 9226-9233.
Pike, K. E., G. Savage, V. L. Villemagne, S. Ng, S. A. Moss, P. Maruff, C. A. Mathis, W. E. Klunk, C. L. Masters and C. C. Rowe (2007). "β-amyloid imaging and memory in non-demented individuals: evidence for preclinical Alzheimer's disease." Brain 130(11): 2837-2844.
Pike, V. W. (2009). "PET radiotracers: crossing the blood–brain barrier and surviving metabolism." Trends in pharmacological sciences 30(8): 431-440.
Politis, M., P. Giannetti, P. Su, F. Turkheimer, S. Keihaninejad, K. Wu, A. Waldman, O. Malik, P. M. Matthews and R. Reynolds (2012). "Increased PK11195 PET binding in the cortex of patients with MS correlates with disability." Neurology 79(6): 523-530.
Pratico, D., C. M. Clark, F. Liun, V. Y.-M. Lee and J. Q. Trojanowski (2002). "Increase of brain oxidative stress in mild cognitive impairment: a possible predictor of Alzheimer disease." Archives of Neurology 59(6): 972-976.
Price, J. C., W. E. Klunk, B. J. Lopresti, X. Lu, J. A. Hoge, S. K. Ziolko, D. P. Holt, C. C. Meltzer, S. T. DeKosky and C. A. Mathis (2005). "Kinetic modeling of amyloid binding in humans using PET imaging and Pittsburgh Compound-B." Journal of Cerebral Blood Flow & Metabolism 25(11): 1528-1547.
Puertas, M., J. Martinez-Martos, M. Cobo, M. Carrera, M. Mayas and M. Ramirez-Exposito (2012). "Plasma oxidative stress parameters in men and women with early stage Alzheimer type dementia." Experimental gerontology 47(8): 625-630.
Ramassamy, C., D. Averill, U. Beffert, L. Theroux, S. Lussier-Cacan, J. S. Cohn, Y. Christen, A. Schoofs, J. Davignon and J. Poirier (2000). "Oxidative insults are associated with apolipoprotein E genotype in Alzheimer's disease brain." Neurobiology of disease 7(1): 23-37.
114
Randolph, C., M. C. Tierney, E. Mohr and T. N. Chase (1998). "The Repeatable Battery for the Assessment of Neuropsychological Status (RBANS): preliminary clinical validity." Journal of clinical and experimental neuropsychology 20(3): 310-319.
Reinlieb, M., L. M. Ercoli, P. Siddarth, N. S. Cyr and H. Lavretsky (2014). "The patterns of cognitive and functional impairment in amnestic and non-amnestic mild cognitive impairment in geriatric depression." The American Journal of Geriatric Psychiatry 22(12): 1487-1495.
Risacher, S. L. and A. J. Saykin (2013). "Neuroimaging and other biomarkers for Alzheimer's disease: the changing landscape of early detection." Annual review of clinical psychology 9: 621.
Rogers, J., J. Luber-Narod, S. D. Styren and W. H. Civin (1988). "Expression of immune system-associated antigens by cells of the human central nervous system: relationship to the pathology of Alzheimer's disease." Neurobiology of aging 9: 339-349.
Rowe, C., S. Ng, U. Ackermann, S. Gong, K. Pike, G. Savage, T. Cowie, K. Dickinson, P. Maruff and D. Darby (2007). "Imaging β-amyloid burden in aging and dementia." Neurology 68(20): 1718-1725.
Rowe, C. C., K. A. Ellis, M. Rimajova, P. Bourgeat, K. E. Pike, G. Jones, J. Fripp, H. Tochon-Danguy, L. Morandeau and G. O'Keefe (2010). "Amyloid imaging results from the Australian Imaging, Biomarkers and Lifestyle (AIBL) study of aging." Neurobiology of aging 31(8): 1275-1283.
Rusjan, P., D. Mamo, N. Ginovart, D. Hussey, I. Vitcu, F. Yasuno, S. Tetsuya, S. Houle and S. Kapur (2006). "An automated method for the extraction of regional data from PET images." Psychiatry Research: Neuroimaging 147(1): 79-89.
Rusjan, P. M., A. A. Wilson, P. M. Bloomfield, I. Vitcu, J. H. Meyer, S. Houle and R. Mizrahi (2011). "Quantitation of translocator protein binding in human brain with the novel radioligand [ 18F]-FEPPA and positron emission tomography." Journal of Cerebral Blood Flow & Metabolism 31(8): 1807-1816.
Saleem, M., N. Herrmann, W. Swardfager, R. Eisen and K. L. Lanctôt (2015). "Inflammatory markers in mild cognitive impairment: A meta-analysis." J Alzheimers Dis 47: 669-679.
Sanchez-Guajardo, V., C. J. Barnum, M. G. Tansey and M. Romero-Ramos (2013). "Neuroimmunological processes in Parkinson's disease and their relation to α-synuclein: microglia as the referee between neuronal processes and peripheral immunity." ASN neuro 5(2): AN20120066.
Schuitemaker, A., M. A. Kropholler, R. Boellaard, W. M. van der Flier, R. W. Kloet, T. F. van der Doef, D. L. Knol, A. D. Windhorst, G. Luurtsema and F. Barkhof (2013). "Microglial activation in Alzheimer's disease: an (R)-[11 C] PK11195 positron emission tomography study." Neurobiology of aging 34(1): 128-136.
115
Schwab, C. and P. L. McGeer (2008). "Inflammatory aspects of Alzheimer disease and other neurodegenerative disorders." Journal of Alzheimer's Disease 13(4): 359-369.
Setiawan, E., A. A. Wilson, R. Mizrahi, P. M. Rusjan, L. Miler, G. Rajkowska, I. Suridjan, J. L. Kennedy, P. V. Rekkas and S. Houle (2015). "Role of translocator protein density, a marker of neuroinflammation, in the brain during major depressive episodes." JAMA psychiatry 72(3): 268-275.
Small, G. W., V. Kepe, L. M. Ercoli, P. Siddarth, S. Y. Bookheimer, K. J. Miller, H. Lavretsky, A. C. Burggren, G. M. Cole and H. V. Vinters (2006). "PET of brain amyloid and tau in mild cognitive impairment." New England Journal of Medicine 355(25): 2652-2663.
Smith, C., J. M. Carney, P. Starke-Reed, C. Oliver, E. Stadtman, R. Floyd and W. Markesbery (1991). "Excess brain protein oxidation and enzyme dysfunction in normal aging and in Alzheimer disease." Proceedings of the National Academy of Sciences 88(23): 10540-10543.
Sokolova, A., M. D. Hill, F. Rahimi, L. A. Warden, G. M. Halliday and C. E. Shepherd (2009). "Monocyte Chemoattractant Protein‐1 Plays a Dominant Role in the Chronic Inflammation Observed in Alzheimer's Disease." Brain pathology 19(3): 392-398.
Spieler, D. H., D. A. Balota and M. E. Faust (1996). "Stroop performance in healthy younger and older adults and in individuals with dementia of the Alzheimer's type." Journal of Experimental Psychology: Human Perception and Performance 22(2): 461.
Stephan, B., S. Hunter, D. Harris, D. Llewellyn, M. Siervo, F. Matthews and C. Brayne (2012). "The neuropathological profile of mild cognitive impairment (MCI): a systematic review." Molecular psychiatry 17(11): 1056-1076.
Streit, W. J., H. Braak, Q.-S. Xue and I. Bechmann (2009). "Dystrophic (senescent) rather than activated microglial cells are associated with tau pathology and likely precede neurodegeneration in Alzheimer’s disease." Acta neuropathologica 118(4): 475-485.
Sultana, R., M. Perluigi and D. A. Butterfield (2009). "Oxidatively modified proteins in Alzheimer’s disease (AD), mild cognitive impairment and animal models of AD: role of Abeta in pathogenesis." Acta neuropathologica 118(1): 131-150.
Suridjan, I., B. Pollock, N. Verhoeff, A. Voineskos, T. Chow, P. Rusjan, N. Lobaugh, S. Houle, B. Mulsant and R. Mizrahi (2015). "In-vivo imaging of grey and white matter neuroinflammation in Alzheimer’s disease: a positron emission tomography study with a novel radioligand, [18F]-FEPPA." Molecular psychiatry 20(12): 1-9.
Thal, D. R., U. Rüb, M. Orantes and H. Braak (2002). "Phases of Aβ-deposition in the human brain and its relevance for the development of AD." Neurology 58(12): 1791-1800.
Thinakaran, G. and E. H. Koo (2008). "Amyloid precursor protein trafficking, processing, and function." Journal of Biological Chemistry 283(44): 29615-29619.
116
Tuppo, E., L. Forman, B. Spur, R. Chan-Ting, A. Chopra and T. Cavalieri (2001). "Sign of lipid peroxidation as measured in the urine of patients with probable Alzheimer’s disease." Brain research bulletin 54(5): 565-568.
Tynan, R. J., J. Weidenhofer, M. Hinwood, M. J. Cairns, T. A. Day and F. R. Walker (2012). "A comparative examination of the anti-inflammatory effects of SSRI and SNRI antidepressants on LPS stimulated microglia." Brain, behavior, and immunity 26(3): 469-479.
van der Graaf, M. (2010). "In vivo magnetic resonance spectroscopy: basic methodology and clinical applications." European Biophysics Journal 39(4): 527-540.
Varrone, A., V. Oikonen, A. Forsberg, J. Joutsa, A. Takano, O. Solin, M. Haaparanta-Solin, S. Nag, R. Nakao and N. Al-Tawil (2015). "Positron emission tomography imaging of the 18-kDa translocator protein (TSPO) with [18F] FEMPA in Alzheimer’s disease patients and control subjects." European journal of nuclear medicine and molecular imaging 42(3): 438-446.
Veenman, L., V. Papadopoulos and M. Gavish (2007). "Channel-like functions of the 18-kDa translocator protein (TSPO): regulation of apoptosis and steroidogenesis as part of the host-defense response." Current pharmaceutical design 13(23): 2385-2405.
Vehmas, A. K., C. H. Kawas, W. F. Stewart and J. C. Troncoso (2003). "Immune reactive cells in senile plaques and cognitive decline in Alzheimer’s disease." Neurobiology of aging 24(2): 321-331.
Venkateshappa, C., G. Harish, A. Mahadevan, M. S. Bharath and S. Shankar (2012). "Elevated oxidative stress and decreased antioxidant function in the human hippocampus and frontal cortex with increasing age: implications for neurodegeneration in Alzheimer’s disease." Neurochemical research 37(8): 1601-1614.
Venneti, S., G. Wang, J. Nguyen and C. A. Wiley (2008). "The positron emission tomography ligand DAA1106 binds with high affinity to activated microglia in human neurological disorders." Journal of neuropathology and experimental neurology 67(10): 1001.
Verhoeff, N. P., A. A. Wilson, S. Takeshita and L. Trop (2004). "In-Vivo Imaging of Alzheimer Disease B-Amyloid With [^ sup 11^ C] SB-13 PET." The American journal of geriatric psychiatry 12(6): 584.
Villemagne, V. L., R. S. Mulligan, S. Pejoska, K. Ong, G. Jones, G. O’Keefe, J. G. Chan, K. Young, H. Tochon-Danguy and C. L. Masters (2012). "Comparison of 11C-PiB and 18F-florbetaben for Aβ imaging in ageing and Alzheimer’s disease." European journal of nuclear medicine and molecular imaging 39(6): 983-989.
Villemagne, V. L., K. E. Pike, G. Chételat, K. A. Ellis, R. S. Mulligan, P. Bourgeat, U. Ackermann, G. Jones, C. Szoeke and O. Salvado (2011). "Longitudinal assessment of Aβ and cognition in aging and Alzheimer disease." Annals of neurology 69(1): 181-192.
117
Villeneuve, S., G. D. Rabinovici, B. I. Cohn-Sheehy, C. Madison, N. Ayakta, P. M. Ghosh, R. La Joie, S. K. Arthur-Bentil, J. W. Vogel and S. M. Marks (2015). "Existing Pittsburgh Compound-B positron emission tomography thresholds are too high: statistical and pathological evaluation." Brain 138(7): 2020-2033.
Vivash, L. E. and T. J. O'Brien (2015). "Imaging microglial activation with TSPO PET: Lighting up neurological diseases?" Journal of Nuclear Medicine: jnumed. 114.141713.
Wechsler, D. (1987). WMS-R: Wechsler memory scale-revised, Psychological Corporation.
Wechsler, D., D. L. Coalson and S. E. Raiford (2008). WAIS-IV: Wechsler adult intelligence scale, Pearson San Antonio, TX.
Wiley, C. A., B. J. Lopresti, S. Venneti, J. Price, W. E. Klunk, S. T. DeKosky and C. A. Mathis (2009). "Carbon 11–Labeled Pittsburgh Compound B and Carbon 11–Labeled (R)-PK11195 Positron Emission Tomographic Imaging in Alzheimer Disease." Archives of neurology 66(1): 60-67.
Wilson, A. A., A. Garcia, J. Parkes, P. McCormick, K. A. Stephenson, S. Houle and N. Vasdev (2008). "Radiosynthesis and initial evaluation of [18 F]-FEPPA for PET imaging of peripheral benzodiazepine receptors." Nuclear medicine and biology 35(3): 305-314.
Wolk, D. A., J. C. Price, J. A. Saxton, B. E. Snitz, J. A. James, O. L. Lopez, H. J. Aizenstein, A. D. Cohen, L. A. Weissfeld and C. A. Mathis (2009). "Amyloid imaging in mild cognitive impairment subtypes." Annals of neurology 65(5): 557-568.
Xin, L., R. Mekle, M. Fournier, P. Baumann, C. Ferrari, L. Alameda, R. Jenni, H. Lu, B. Schaller and M. Cuenod (2016). "Genetic Polymorphism Associated Prefrontal Glutathione and Its Coupling With Brain Glutamate and Peripheral Redox Status in Early Psychosis." Schizophrenia bulletin.
Yan, P., X. Hu, H. Song, K. Yin, R. J. Bateman, J. R. Cirrito, Q. Xiao, F. F. Hsu, J. W. Turk and J. Xu (2006). "Matrix metalloproteinase-9 degrades amyloid-β fibrils in vitro and compact plaques in situ." Journal of Biological Chemistry 281(34): 24566-24574.
Yasuno, F., J. Kosaka, M. Ota, M. Higuchi, H. Ito, Y. Fujimura, S. Nozaki, S. Takahashi, K. Mizukami and T. Asada (2012). "Increased binding of peripheral benzodiazepine receptor in mild cognitive impairment–dementia converters measured by positron emission tomography with [11 C] DAA1106." Psychiatry Research: Neuroimaging 203(1): 67-74.
Yasuno, F., M. Ota, J. Kosaka, H. Ito, M. Higuchi, T. K. Doronbekov, S. Nozaki, Y. Fujimura, M. Koeda and T. Asada (2008). "Increased binding of peripheral benzodiazepine receptor in Alzheimer's disease measured by positron emission tomography with [11 C] DAA1106." Biological psychiatry 64(10): 835-841.
Yokokura, M., N. Mori, S. Yagi, E. Yoshikawa, M. Kikuchi, Y. Yoshihara, T. Wakuda, G. Sugihara, K. Takebayashi and S. Suda (2011). "In vivo changes in microglial activation and
118
amyloid deposits in brain regions with hypometabolism in Alzheimer’s disease." European journal of nuclear medicine and molecular imaging 38(2): 343-351.
Zhang, M.-R., J. Maeda, K. Furutsuka, Y. Yoshida, M. Ogawa, T. Suhara and K. Suzuki (2003). "[18 F] FMDAA1106 and [18 F] FEDAA1106: two positron-Emitter labeled ligands for peripheral benzodiazepine receptor (PBR)." Bioorganic & medicinal chemistry letters 13(2): 201-204.