TIPS-1217; No.

of Pages 13

Review

Amyloid biomarkers in Alzheimers

disease
Kaj Blennow1,2, Niklas Mattsson1,3,4, Michael Scholl5,6, Oskar Hansson7,8, and
Henrik Zetterberg1,9
1

Clinical Neurochemistry Laboratory, Institute of Neuroscience and Physiology, The Sahlgrenska Academy at University of
Gothenburg, Molndal, Sweden
2
The Torsten Soderberg Professorship at the Royal Swedish Academy of Sciences
3
Department of Veterans Affairs Medical Center, Center for Imaging of Neurodegenerative Diseases, San Francisco, CA, USA
4
Department of Radiology and Biomedical Imaging, University of California, San Francisco, CA, USA
5
Helen Wills Neuroscience Institute, University of California, Berkeley, CA, USA
6
Department of Clinical Neuroscience and Rehabilitation, University of Gothenburg, Gothenburg, Sweden
7
Department of Clinical Sciences, Lund University, Lund, Sweden
8
Clinical Memory Research unit, Clinical Sciences, Lund University, Lund, Sweden
9
UCL Institute of Neurology, Queen Square, London, UK

Aggregation of amyloid-b (Ab) into oligomers, fibrils,

and plaques is central in the molecular pathogenesis of
Alzheimers disease (AD), and is the main focus of AD
drug development. Biomarkers to monitor Ab metabolism and aggregation directly in patients are important
for further detailed study of the involvement of Ab in
disease pathogenesis and to monitor the biochemical
effect of drugs targeting Ab in clinical trials. Furthermore, if anti-Ab disease-modifying drugs prove to be
effective clinically, amyloid biomarkers will be of special
value in the clinic to identify patients with brain amyloid
deposition at risk for progression to AD dementia, to
enable initiation of treatment before neurodegeneration
is too severe, and to monitor drug effects on Ab metabolism or pathology to guide dosage. Two types of amyloid biomarker have been developed: Ab-binding ligands
for use in positron emission tomography (PET) and
assays to measure Ab42 in cerebrospinal fluid (CSF).
In this review, we present the rationales behind these
biomarkers and compare their ability to measure Ab
plaque load in the brain. We also review possible shortcomings and the need of standardization of both biomarkers, as well as their implementation in the clinic.
Amyloid in Alzheimers disease
Amyloid is the term used for proteins that are misfolded
into a cross b-sheet structure and thereby bind dyes such
as Congo Red and Thioflavin T [1]. AD is one of the major
amyloidoses, with two types of amyloid deposited in the
brain: (i) Ab forming aggregates in the form of plaques and
cerebrovascular amyloid angiopathy (CAA); and (ii) tau
protein, which forms neurofibrillary tangles, dystrophic
neurites, and neuropil threads (reviewed in [2]). When
Corresponding author: Blennow, K. (kaj.blennow@neuro.gu.se).
Keywords: Alzheimers disease; biomarker; b-amyloid (Ab); cerebrospinal fluid;
positron emission tomography (PET).
0165-6147/
2015 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.tips.2015.03.002

not specified otherwise, we use the term amyloid here

to refer to Ab pathology rather than tau pathology.
Research advances during the past two decades have
resulted in detailed knowledge on disease mechanisms. Ab
is produced by the sequential cleavage of amyloid precursor protein (APP) by two enzymes, b-site APP-cleaving
enzyme 1 (BACE1), also called b-secretase, and the gsecretase complex (Figure 1). The prevailing hypothesis
for AD pathogenesis is called the amyloid cascade hypothesis (Figure 2), posing that Ab aggregation is the initiating
mechanistic event, in which the different stages of aggregates, from soluble oligomers to insoluble fibrils in plaques,
are believed to impair synaptic function and ultimately
damage neurons, resulting in chronic neurodegeneration
leading to cognitive impairment and finally dementia
[3]. AD research advances have also generated a large
number of drug candidates with potential disease-modifying effects. Based on the strong belief in the amyloid
cascade hypothesis, researchers have placed an overwhelming focus on molecules targeting Ab production
and aggregation in AD drug development, and most drug
candidates tested aim to inhibit Ab toxicity by reducing
further Ab aggregation and plaque formation. These drug
candidates include secretase inhibitors to lower Ab production from APP, Ab aggregation inhibitors to inhibit Ab
oligomerization or fibrillization, as well as active and
passive Ab immunotherapies designed to capture either
soluble or aggregated Ab, or both, which will be either
degraded or cleared from the brain (reviewed in [4]).
Alarmingly, an increasing number of large Phase III
clinical trials on Ab targeting drugs have reported no
beneficial effects on cognitive symptoms in patients with
sporadic AD [57]. These discouraging reports have caused
increasing concern in the AD research community that the
amyloid cascade hypothesis eventually will be falsified,
that is, that Ab aggregation is just a bystander, and not
the cause, of neurodegeneration in AD [8]. A more optimistic viewpoint is that there are several logical explanations
for the trial failures, including that the trials enrolled
Trends in Pharmacological Sciences xx (2015) 113

TIPS-1217; No. of Pages 13

Review

Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x

Full-length APP

770
-CTF (C99)
sAPP

671

BACE1

AICD
-amyloid

-secretase

42
TRENDS in Pharmacological Sciences

Figure 1. Generation of b-amyloid (Ab) by metabolism of amyloid precursor protein (APP). APP is a transmembrane protein with a large extracellular N terminus. The Ab
domain is partly embedded in the plasma membrane, with 28 amino acids outside the membrane and 14 amino acids embedded in the membrane. APP is cleaved by b-site
APP-cleaving enzyme 1 (BACE1), also called b-secretase, and a large soluble part (sAPPb) is released. The remaining C-terminal fragment, called b-CTF or C99, is then
cleaved by g-secretase, releasing soluble Ab. g-secretase is an intramembranous protease complex, with four components, the active enzyme presenilin, together with
nicastrin, presenilin enhancer (Pen-2), and anterior pharynx-defective (Aph-1).

patients with AD and dementia, which is probably too

advanced a stage of the disease to enable this type of drug
to show any effect on clinical symptoms, and that trial
patients have been diagnosed based on purely clinical
criteria, which are too nonspecific; thus, trials will comprise a cohort with only approximately 80% of enrolled
patients having genuine AD pathology [9]. Both of these
shortcomings call for diagnostic biomarkers to aid clinicians in making an early and accurate diagnosis. In future
clinical trials on Ab-targeting drugs, amyloid biomarkers
would be especially valuable to confirm that enrolled
patients do have Ab pathology and, thus, the disease for
which the drug is intended, which would increase the
possibility of identifying a positive clinical effect of the
drug [10]. If drug effects will be seen only in subjects with
biomarker evidence of pathology, such biomarkers would
also be useful to guide clinical decisions on whether to
prescribe Ab-targeting drugs, once these are available.
Another possible explanation for some of the trial failures is that poor drug candidates have been taken to Phase
II and III clinical trials based on promising, but misleading data from preclinical drug development [2]. It has been
common in AD drug development to test whether novel Ab
drug candidates reduce the Ab plaque load in AD transgenic mice and, if so, take the drug into large and expensive clinical trials without examining whether target
engagement can be verified in humans. There are numerous examples of failed trials (e.g., tarenflurbil and phenserine), which probably are due to the poor predictive
power of these disease models [2,9]. For this reason, it
is becoming increasingly common to apply theragnostic
biomarkers during the early stages of AD drug development [5,1113]. In this context, amyloid biomarkers
applied in short small-scale trials to prove target engagement in Phase I proof-of-principle studies on healthy
volunteers [9] or Phase II proof-of-concept studies in
patients with AD [11] may be valuable in the selection
of drug candidates and may improve success rates in latestage clinical trials.
2

In contrast to most other neurodegenerative brain disorders, a set of biomarkers has been developed for the
different pathogenic processes in AD and examined in a
large number of clinical studies. These AD biomarkers
include magnetic resonance imaging (MRI) of hippocampal
or whole-brain atrophy, PET evaluation of glucose metabolism in cortical neurons and glial cells, CSF assays to
measure tau protein, reflecting the intensity of the neuronal degeneration, and phosphorylated tau, reflecting the
presence of tangles, and the two amyloid biomarkers
amyloid PET and CSF Ab42 (reviewed in [14]). In this
review, we focus on the amyloid biomarkers, which have
been much examined and reviewed individually, while an
objective head-to-head comparison on their performance to
measure Ab plaque load or Ab metabolism in the brain
is lacking. We also discuss mechanistic differences between
the amyloid biomarkers and their implementation in
clinical trials and in the clinical routine management of
patients with cognitive symptoms.
Biomarkers for AD
According to the National Institutes of Health (NIH) Biomarkers Definitions Working Group, a biomarker is defined as a characteristic that is objectively measured and
evaluated as an indicator of normal biological processes,
pathogenic processes, or pharmacologic responses to a
therapeutic intervention [15]. While the National Cancer
Institute at the NIH defines a biomarker as a biological
molecule found in blood, other body fluids, or tissues that is
a sign of a normal or abnormal process, or of a condition or
disease (http://www.cancer.gov/dictionary?cdrid=45618),
the World Health Organization (WHO) has a broader
definition of biomarkers, which includes almost any measurement reflecting an interaction between a biological
system and a potential hazard, which may be chemical,
physical, or biological. The measured response may be
functional and physiological, biochemical at the cellular
level, or a molecular interaction. (http://www.inchem.org/
documents/ehc/ehc/ehc155.htm).

TIPS-1217; No. of Pages 13

Review

Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x

Familial AD

Sporadic AD

APP mutaons
APP gene duplicaon
PSEN mutaons

Major eects: aging, APOE 4 allele

Minor eects: other genes, environmental risk factors
Unknown factors

Life-long disturbance in A producon:

increase in total A
absolute or relave increase in A42
A pepdes with increased tendency for aggregaon

Failure of A clearance with

gradually increasing A levels in brain

Conformaonal change in A
with formaon of soluble oligomers

Impaired LTP with

synapc dysfuncon

?
Gradual decrease
in CSF A42

Gradual increase
in amyloid PET
ligand retenon

Aggregaon of A into diuse plaques

Microglial and astrocyc acvaon,

inammatory response, and
oxidave stress

Further aggregaon resulng in brillar

A deposits, with tau-posive dystrophic neurites
and glial response as neuric plaques

Synapc and neuronal degeneraon

with failure in neurotransmission

Tau pathology

Progressive cognive dysfuncon and demena

TRENDS in Pharmacological Sciences

Figure 2. The amyloid cascade hypothesis. This is the lead hypothesis for Alzheimers disease (AD) pathogenesis, which posits that the central event is an imbalance
between b-amyloid (Ab) production and clearance. In familial AD, genetic alterations cause a life-long disturbance in Ab production or generate Ab peptides that are more
prone to aggregation. In sporadic AD, advanced age and possession of the apolipoprotein E (ApoE) e4 allele have major effects on the risk for developing AD. The common
denominator in the pathogenesis is a conformational change in Ab, which makes it prone to aggregation, with the initial formation of soluble oligomers, followed by larger
fibrils that accumulate into diffuse plaques and, at a later stage, neuritic plaques. Cognitive impairment is believed to be due to Ab oligomers inhibiting hippocampal longterm potentiation and impaired synaptic function, as well as an inflammatory response, oxidative stress, and synaptic and neuronal degeneration with neurotransmitter
deficits. Tau pathology with tangle formation is regarded a downstream event that contributes to cognitive symptoms. Adapted from [3] with minor modifications,
including the relation between the cascade of pathogenic events and the amyloid biomarkers. Abbreviations: APP, amyloid precursor protein; CSF, cerebrospinal fluid; LTP,
long-term potentiation; PET, positron emission tomography; PSEN, presenilin.

In 1998, the Ronald and Nancy Reagan Research Institute of the Alzheimers Association and the National
Institute of Aging Working Group set up criteria for an
ideal AD diagnostic biomarker. These recommendations
stated that the biomarker should be: able to detect a
fundamental feature of Alzheimers neuropathology, validated in neuropathologically confirmed AD cases, precise
(able to detect AD early in its course and distinguish it
from other dementias), reliable, non-invasive, simple to
perform, and inexpensive [16]. It may be argued that there
are several clinically useful biomarkers that do not detect
a fundamental feature of disease pathology; for example,
prostate-specific antigen (PSA) is an excellent biomarker
for prostate cancer even if not involved in pathophysiology.
Nevertheless, as becomes clear below, several of the biomarkers for Ab pathology fulfill most, if not all, of these
requirements.

Amyloid biomarkers in AD
There are two principal AD amyloid biomarkers: amyloid
PET, which measures the amount of Ab aggregates in the
brain parenchyma, and biochemical analyses of Ab species
and APP-processing products in CSF. Several peptides,
proteins, and enzymes involved in the amyloidogenic APPprocessing can be measured in CSF. The latter includes
differentially truncated species of Ab, including the longer
Ab17 up to Ab42 species generated by b- and g-secretase
cleavage of APP, and the shorter Ab14 to Ab16 species
generated by b- and a-secretase cleavage [17], soluble bsecretase cleaved APP (sAPPb) [18], and Ab oligomers
[19,20]. Except for CSF Ab42, these markers are less well
established or show little or no association with AD in
cross-sectional or longitudinal biomarker studies, but
may still be useful when assessing target engagement
for some of the drug candidates that are being evaluated
3

TIPS-1217; No. of Pages 13

Review
as potential anti-Ab drugs; see, for example, [13]. Here, we
focus on the first biomarker class, but it is important to
remember the additional information that can be gained
from the second biomarker class in drug development, and,
in the future, possibly also to facilitate accurate dosefinding in individual patients.
CSF Ab42
The first report showing that Ab is secreted to the CSF
came in 1992 [21], which made measurement of Ab in
CSF an important candidate biomarker for AD. Disappointingly, initial reports on CSF Ab showed no clear
change in AD [22,23]. These initial reports were based
on enzyme-linked immunosorbent assays (ELISA) methods that did not discriminate between truncated Ab isoforms, that is, the total CSF level of Ab was measured.
The discovery that the 42 amino acid form of Ab, Ab42, is
prone to aggregate and the earliest Ab species deposited in
plaques [24,25] set focus on immunoassays specific for
Ab42.
The first paper using an ELISA method specific for Ab42
was published in 1995 and showed a marked reduction in
CSF in patients with AD [26]. This finding has been
reproduced in numerous studies (reviewed in [27]) and
using many different assay formats, including ELISA
[28], Luminex xMAP technology [29], electrochemiluminescence immunoassay [18], urea-based SDS-PAGE combined with Western blot [30], and antibody-free singlereaction monitoring (SRM) mass spectrometry [31]. In
AD dementia, the CSF level of Ab42 is typically decreased
to approximately 50% of the levels in age-matched cognitively normal individuals [32].
CSF contains many different Ab isoforms, of which
Ab40 is around ten times more abundant than Ab42
[33]. CSF levels of Ab40 are unchanged in AD, but there
is a reduction in the CSF Ab42:Ab40 ratio in AD that is
more marked than the reduction in Ab42 alone [34
37]. The CSF Ab42:Ab40 ratio has been suggested to
further improve diagnostic accuracy, because it can balance interindividual variations in total Ab production
(of all Ab isoforms); some non-AD cases that are low
producers may have false positive Ab42 tests (just above
the cut-off) and some AD cases that are high producers
may have false negative Ab42 tests, while this will be
compensated for by the Ab42:Ab40 ratio [35,37].
Amyloid PET
The first study on amyloid PET used an 11C-labeled modified
derivative of the amyloid-binding histological dye Thioflavin-T called Pittsburgh Compound-B (PiB) and showed
increased ligand retention in cortical brain regions in AD
dementia cases as compared with controls, using cerebellum
as a reference region [38]. Increased retention of PiB in
cortical brain regions in AD has since been verified in many
scientific reports (for review see [39]). The change in global
cortical PiB in cases with AD dementia as compared with
cognitively normal older individuals is typically a 5070%
increased retention (range 2080%) [40].
The short half-life of 11C hinders the use of 11C-PiB
outside expert research centers, unless there is access to
an on-site cyclotron and radiochemistry expertise. Thus,
4

Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x

efforts were initiated to make 18F-labeled tracers, and

currently four such amyloid ligands have been developed,
which have a half-life of approximately 110 min; this
enables centralized production and regional distribution
to centers having a PET camera. These ligands include
18
F-florbetapir (Amyvid), also called AV-45 [41], 18F-flutemetamol (Vizamyl), 30 F-PiB or GE-067) [42], 18F-florbetaben (Neuraseq), also called BAY94-9172 or AV-1 [43], and
18
F-NAV4694, formerly known as AZD4694 [44]. Using
these amyloid tracers, a 4070% increase in cortical ligand
retention is found, but some have higher nonspecific white
matter binding than 11C-PiB, while others show lower
cortical binding in patients with AD [44].
Relation to amyloid metabolism and pathology for
amyloid biomarkers
PiB and other amyloid ligands were developed to bind to
aggregated Ab. Binding of the PiB ligand to b-sheetfolded Ab has also been shown in several post-mortem
studies, with preference for compact plaques and vascular deposits called CAA, while diffuse plaques are less
prominently labeled and amorphous plaques comprising
loosely aggregated Ab containing little bsheet structure
do not bind PiB [4547]. Furthermore, PiB does not bind
aggregated Ab in cases with the arctic APP gene (E693G)
mutation [48]. By contrast, human subjects with CAA
may have positive PiB amyloid PET scans, even in the
absence of amyloid plaques [49] and even if the distribution of PiB binding differs between patients with clinically diagnosed AD and CAA [50]. Several studies have also
shown good agreement between 18F-florbetapir amyloid
PET and post-mortem Ab pathology evaluated in hospice
patients near the end of their lives [51,52] and between
18
F-flutemetamol amyloid PET imaging and cortical fibrillar Ab pathology in biopsies from living patients with
hydrocephalus [53,54].
By contrast, it was less obvious how the reduced CSF
level of Ab42 would be interpreted. Possible explanations
range from decreased Ab production, decreased Ab clearance (active export from the brain to the blood), aggregation and deposition of Ab in brain tissue, with lower
amounts diffusing to the extracellular space and CSF,
increased proteolytic degradation of Ab, and possibly
more. The first indication that the lowering of CSF
Ab42 is due to aggregation of the peptide in the brain
came in 2003 in a study showing that low levels of Ab42 in
post-mortem ventricular CSF showed an inverse correlation with plaque load in cortical regions [55]. This finding
has been replicated in several studies [56,57]. Further
support came from a study in 2006 showing an inverse
relation between CSF Ab42 and PiB binding in cortical
brain regions [58], a finding that has been replicated in
many successive papers, as reviewed below. Taken together, these data indicate that the reduced CSF level of Ab42
in patients with AD is due to Ab deposition and, thus,
reflects plaque (or fibrillar Ab) load in the brain. However,
CSF Ab levels may change dynamically in response to
pharmacological interventions. For example, a transient
lowering of CSF Ab42 is found in response to reduced Ab
production following BACE1 inhibitor treatment, also in
healthy volunteers [13].

TIPS-1217; No. of Pages 13

Review

Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x

Concordance between amyloid biomarkers

After the first study in 2006 showing an inverse correlation
between global cortical amyloid PET ligand retention and
CSF Ab42 levels [58], several studies confirmed this association [5968]. These ten publications included more than
1000 patients and controls who underwent both amyloid
PET and CSF Ab42 examinations (Table 1). Most subjects
(88%) had concordant amyloid biomarker results, with
either negative or positive amyloid PET scans and CSF
Ab42 levels (Table 1). A few had discordant amyloid biomarkers results, with either positive (low) CSF Ab42 but
normal amyloid PET (6.6%), or normal CSF Ab42 levels
but positive amyloid PET scans (5.4%).
Ab plaques in the AD brain comprise several N-terminally truncated species, such as Ab342, Ab442, and
pyroglutamate derivatives of Ab (e.g., pGluAb342)
[69]. These Ab species are not captured by the most commonly used immunoassay for CSF Ab42, which measures
Ab1-42 (i.e., peptides that have the first amino acid preserved) [28,29]. However, the correlation between amyloid
PET positivity and low CSF Ab42 is similar for AbX42
and for Ab142 [64]. Thus, amyloid PET correlates with
the total amount of the aggregation-prone Ab42 peptide in
CSF. By contrast, no correlations are found for Ab species
other than Ab42, such as Ab40 or Ab38 [60,64]. These data
further underline that the correlation reflects the deposition of the aggregation-prone Ab42 peptide in the brain,
and not a general change in the metabolism of Ab.
In addition to the amyloid biomarkers CSF Ab42 and
amyloid PET, the tau biomarkers CSF T-tau and P-tau also
show marked changes in patients with AD dementia and
prodromal AD as compared with control populations
[70]. Thus, in studies examining mixed AD and control
populations, significant correlations are to be expected for
all parameters that show a marked difference between
patients and controls. Indeed, some studies have shown

that patients who are amyloid PET positive have higher

CSF T-tau and P-tau levels, but the correlations with CSF
T-tau and P-tau are weaker than the strong inverse correlation found between amyloid ligand retention and CSF
Ab42 [60,62,63,67], and do not reach statistical significance in smaller patient series [58,61,64]. These findings
indicate that the relation between CSF Ab42 and amyloid
PET is specifically related to Ab42 deposition in the brain,
and not merely to the neurodegenerative disease process in
the AD brain.
Characteristics of discordant cases
The existence of cases that are discordant for CSF and PET
amyloid biomarkers has been a subject for discussion,
especially the group with low CSF Ab42 but normal amyloid PET. Early case reports suggested that CSF Ab42
decreases before brain Ab accumulation is great enough to
enable detection by PET, or that early diffuse Ab deposits,
which bind amyloid ligands poorly, occur before fibrillar
plaques [71]. A large study comparing amyloid PET and
CSF biomarkers and containing a large enough number of
discordant cases to allow subanalyses in these groups,
showed that discordance (mainly isolated CSF Ab positivity) was clearly dependent on disease stage, and found 21%
of cognitively normal older subjects, but only 6% of AD
dementia cases, with intermediate frequencies in patients
with mild cognitive impairment (MCI) [72]. These findings
clearly show that CSF-positive/PET-negative biomarker
discordance is not driven by technical or methodological
variations. Furthermore, CSF Ab42 was more strongly
related to possession of the apolipoprotein E (APOE) e4
allele than was amyloid PET, while amyloid PET amyloid
was more strongly related to CSF tau and cognitive decline
than was CSF Ab42 [72]. Taken together, these findings
indicate that CSF-positive/PET-negative biomarker mismatches may be due to CSF Ab42 being more sensitive in

Table 1. Agreement between CSF Ab and amyloid PETa,b

PET ligand

CSF assay

Cases

Florbetapir

Ab1-42 (Luminex)

Flutemetamol

Ab1-42 (ELISA)

PiB

Ab1-42 (ELISA)

Total = 374 (103 normal, 187 early MCI,

62 late MCI, 22 AD)
Total original cohort = 118
Total validation cohort = 38 c
Total = 50 (43 CDR = 0, 4 CDR = 0.5,
3 CDR = 1)
Total = 16 (6 MCI converters,
11 nonconverters)
Total = 189 (189 controls with CDR = 0)
Total = 55 (11 controls, 34 MCI, 10 AD)
Total = 37 (10 controls, 12 MCI, 15 AD)
Total = 10 (10 probable AD)
Total = 41 (11 controls, 34 MCI, 10 AD)
Total = 136 (64 AD, 34 non-AD, 22 MCI,
16 controls)
1064

Ab1-42 (ELISA)
Ab1-42
Ab1-42
Ab1-42
Ab1-42
Ab1-42
Ab1-42

(ELISA)
(Luminex)
(ELISA)
(ELISA)
(Luminex)
(ELISA)

Total

Concordant
CSF/PET
or CSF+/PET+
322 (86.1%)

Discordant
CSF+/PET

Discordant
CSF/PET+

Refs

21 (5.6%)

31 (8.3%)

[67]

109 (92.4%)
37 (97.4%)
50 (100%)

6 (5.1%)
0
0

3 (2.5%)
1 (2.6%)
0

[68]

16 (100%)

[61]

157 (83%)
50 (91%)
31 (84%)
10 (100%)
41 (100%)
114 (84%)

28 (15%)
2 (4%)
6 (16%)
0
0
7 (5%)

4 (2%)
3 (5%)
0
0
0
15 (11%)

[60]
[62]
[63]
[64]
[65]
[66]

937 (88.0%)

70 (6.6%)

57 (5.4%)

[59] d

Abbreviations: AD, Alzheimers disease; CDR, clinical dementia rating; CSF, cerebrospinal fluid; MCI, mild cognitive impairment.

CSF, normal (= high) CSF Ab42 levels; CSF+ low (below the cut-off in the study) CSF Ab42 levels; PETnormal (= low) amyloid ligand retention; PET+, high (above the cutoff in the study) amyloid ligand retention.

The cut-off for both amyloid PET and CSF Ab1-42 for the validation cohort was established in the original cohort.

Includes the cohort reported in [58].

TIPS-1217; No. of Pages 13

Review

Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x

the early stages of AD, while PET amyloid may still change
dynamically during later stages of disease. This interpretation is corroborated by findings in a recent study showing the cortical flutemetamol retention levels correlate
with disease stage in patients with MCI, while CSF
Ab42 levels do not [68]. Large studies with longitudinal
data sampling are needed to better understand the temporal relation between CSF Ab42 and amyloid PET.
Global or regional amyloid PET versus CSF Ab42
In many studies comparing PET and CSF as amyloid
biomarkers, the uptake of the amyloid PET tracer in
regional neocortical standardized uptake value ratio
(SUVR) is averaged into a single composite SUVR, and
normalized to the uptake in a reference region, most often
the cerebellum (Table 2). This approach is feasible because
the correlation between low CSF Ab42 levels and high
retention of amyloid PET tracer is high in almost all
cortical regions, with similar correlation coefficients for
global (or composite) PET tracer retention as for selected
regional cortical (frontal, temporal, posterior cingulate, or
parietal) retention [63]. This was verified in a study comparing flutemetamol amyloid PET and CSF Ab42, showing
that the best correlation is found in brain regions with the
highest ligand retention, suggesting that CSF Ab42 levels
reflect total cortical Ab deposition [68]. In agreement, a
recent study showed that, due to the high correlations
between regional and global cortical PiB binding potential,
many cortical brain regions are comparably useful for
classifying a scan as positive or negative, although with
different thresholds [73].

Furthermore, a large study comparing the diagnostic

performance of CSF Ab42 and florbetapir PET in the
ADNI-2 cohort found similar diagnostic accuracies for
the amyloid biomarkers, and also when comparing CSF
Ab42 with either global or regional (temporal, frontal,
parietal, and cingulate) amyloid PET measurements
[74]. These studies do not support that measurement of
cortical amyloid ligand retention in a certain region, such
as posterior cingulate cortex, results in superior diagnostic
accuracy for AD.
Influence of methodology and the need of
standardization of amyloid biomarkers
Even if the amyloid biomarkers (CSF Ab42 and amyloid
PET) are highly correlated, there are also differences
depending on the methodology for quantification of
Ab42 in CSF and for evaluation of PET scans. In the
studies evaluating concordance, CSF Ab42 has been analyzed using either ELISA [28] or Luminex xMAP [29]
methodologies (Table 2). It is known that the two techniques give different absolute levels for CSF Ab42, but show
a tight linear correlation [29]. Thus, these methods give
different cut-offs for CSF amyloid positivity, approximately 450650 pg/mL in the papers using the ELISA (cut-offs
defined in different ways), while a cut-off at 192 pg/mL has
been used in all papers using the Luminex xMAP technique (Table 2). The 192 pg/mL cut-off for the Luminex
assay was originally defined as the point maximizing the
diagnostic accuracy in a cohort of autopsy-confirmed AD
cases (where pre-mortem CSF was taken several years
before autopsy) and age-matched living healthy controls

Table 2. Methods and cut-offs for CSF b-amyloid and amyloid PETa
CSF b-amyloid (Ab42)

Amyloid PET

CSF assay Cut-off for CSF positive PET ligand

ELISA b

<457 pg/mL
<500 pg/mL
<650 pg/m d

<550 pg/m d
<450 pg/mL
<550 pg/mL

Luminex c

<647 pg/mL
<192 pg/mL

<192 pg/mL

<192 pg/mL

11

C PiB

Cortical and reference regions

Prefrontal, precuneus, lateral temporal and

gyrus rectus; reference = cerebellum
11
C PiB
Prefrontal, precuneus, lateral temporal, and
gyrus rectus; reference = cerebellum
11
C PiB
Frontal (mean of orbital, medial inferior, and
superior), parietal, temporal (mean of superior
and medial inferior), medial temporal
(entorhinal and hippocampus) and posterior
cingulated; reference = cerebellum
11
C PiB
Posterior cingulum; reference = cerebellum
11
C PiB
Frontal, parietal, temporal, and posterior
cingulum; reference = cerebellum
11
C PiB
Frontal, parietal, temporal, and occipital;
reference = cerebellum
Flutemetamol Reference = cerebellum
11
C PiB
Prefrontal, posterior cingulate/precuneus,
lateral temporal, parietal, anterior, and
cingulate cortex; reference = pons and
cerebellum
11
C PiB
Prefrontal, orbitofrontal, parietal, temporal,
anterior cingulate, and posterior cingulate/
precuneus
Florbetapir
Reference = cerebellum

Method for
evaluation
MCBP (BPND)

Cut-off for
Refs
PET positive
>0.2
[59]

MCBP (BPND)

>0.18

[60]

Global cortical PBND

>0.50 d

[63]

ROI/ref PiB retention

Mean cortical PiB
retention
BPND, visually score

>1.6 d
>1.6

[61]
[64]

Positive

[66]

Mean cortical SUVR

Mean cortical SUVR

>1.42
>1.465

[68]
[62]

Global cortical PiB

retention

>1.5

[65]

Mean cortical retention >1.11

Abbreviations: BPND, nondisplaceable binding potential; MCBP, mean cortical binding potential; PBND, normalized ligand binding potential.

Innotest b-amyloid(1-42) (Fujirebio).

Luminex assay is Alzbio3 (Fujirebio).

Cut-off based from optimal cut-off presented in graph.

[67]

TIPS-1217; No. of Pages 13

Review
[75]. As shown in Table 2, the association between CSF
Ab42 and amyloid PET is high for both analytical CSF
techniques (Table 2).
In addition to the problem that absolute levels for CSF
Ab42 vary depending on the analytical technique used
(Table 2), there are also differences in absolute levels between laboratories, even if the same assay format is used
[76]. In addition, ELISA-based CSF assays are known to
show batch-to-batch variations for the kits [76]. Standardization efforts are ongoing within the Alzheimers Association Global Biomarkers Consortium (GBSC) and the
International Federation of Clinical Chemistry Working
Group for CSF proteins (IFCC WG-CSF), with the aim of
standardizing both pre-analytical and laboratory procedures and to harmonize levels between assay formats
[77]. Protocols for standardization of procedures for CSF
collection and sample handling and shipment have been
proposed, together with laboratory procedures for run acceptance and batch bridging to maintain long-term stability
of ELISA-based CSF biomarkers [10,68]. Within the IFCCWG-CSF, SRM mass spectrometry-based reference measurement procedures (RMP), that is, gold standard methods for CSF Ab42, have been published [31,78], and projects
to develop RMPs for CSF tau are ongoing. These methods for
absolute quantification of CSF Ab42 will be used to set the
exact levels on a certified reference material (CRM), that is,
aliquoted CSF pools with three levels (high, medium, or low)
of Ab42 will be used to normalize levels between assay
formats [77,79]. Between-laboratory and longitudinal
(batch-related) variations are monitored in the Alzheimers
Association Quality Control (QC) program for CSF biomarkers [10,80]. These standardization efforts have also stimulated biotech companies to develop upgraded and validated
versions of assays and to establish CSF biomarker assays on
fully automated laboratory analyzers. These developments
will pave the way for uniform cut-off levels for CSF biomarkers and a more general use of these diagnostic tools.
In studies evaluating CSF and PET concordance, three
different amyloid ligands have been used: PiB, florbetapir,
and flutemetamol. Similar concordance rates have been
found in the eight studies using PiB (84%) as compared
with florbetapir (86%) and flutemetamol (93%). Furthermore, there are differences in the general design of amyloid
PET examinations, such as scan duration, type of scan
(dynamic/static), camera settings, and methodology used
for quantification of amyloid ligand binding (Table 2). A
common measure of amyloid PET ligand retention is the
semiquantitative SUVR, which represents the measured
radioactivity in a volume of interest (VOI) at a certain time
point, or for a certain time window, of scanning in relation
to the injected dose and the individuals body weight [81],
The SUVR is created by normalizing the SUV in a certain
target VOI to the SUV in a reference region that ideally
shows no or little specific binding (SUVVOI/SUVREF). This
method does not require time-consuming dynamic PET
scanning or blood sampling for the modeling of ligand
kinetics, which renders it a simple and practical tool also
for clinical settings. Among the studies comparing 11C-PiB
amyloid PET and CSF Ab42, only one used mean cortical
SUVR, with a cut-off value for a positive amyloid PET
scan of 1.45 [62]. Another commonly used measure is the

Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x

binding potential (nondisplaceable, BPND), which is generally defined as the ratio of receptor density to the dissociation constant (Bmax/KD) [82]. The three studies that used
BPND when comparing amyloid PET with CSF Ab42 had
cut-offs for a positive amyloid PET scan from 0.180.20
[59,60] to 0.50 [63] (Table 2). Lastly, three studies used a
technique in which ligand uptake (raw radioactivity) in
cortical regions was divided by ligand uptake in the reference region. This measure was called ROI/ref PiB retention
[61], mean cortical PiB retention [64], and global cortical
PiB retention [65], and had cut-offs for a positive amyloid
PET scan from 1.5 to 1.6 (Table 2).
The definition of the reference region used to calculate
SUVR or BPND measures also differs between studies, with
cerebellar gray matter, whole cerebellum, brain stem, and
the pons used as reference regions. The choice of reference
region may be optimized for different settings. For example,
the pons is the preferable reference region in patients with
autosomal dominant AD, because these patients may have
amyloid accumulation in the cerebellum [83,84]. Furthermore, variability may occur not only when outlining the
reference region, but also when defining the VOIs that are
summarized and averaged to create a composite VOI when
calculating MCBP or global SUVR. These VOIs can be
drawn manually, using different protocols, or using automated, often MRI segmentation-based methods. Taken together, the fact that values for tracer binding or retention
vary between tracers, depending on which technique used to
calculate uptake, and likely between PET scanners and
centers, have made uniform cut-off values difficult to establish. The lack of generally accepted protocol for acquisition,
processing, and analysis of amyloid PET data also complicates comparability between studies. Thus, there is a need
for standardization efforts for amyloid PET biomarkers.
Importantly, an international working group recently proposed a method for standardizing the quantitative analysis
of amyloid PET images across different tracers to enable
better comparability of amyloid PET data in the context of
cross-center, multicenter, and longitudinal studies, as well
as the definition of cut-offs [85]. According to that proposal,
the original PET data are transformed into so-called Centiloids (CL), the CL-scale comprising units where 0 represents uptake in an amyloid-negative brain and 100 the brain
of a patient with typical mild-moderate AD. Awaiting this,
visual evaluation of scans in a binary fashion as amyloid
positive or negative has been accepted as the standard,
and this has been validated by post-mortem assessment of
AD pathology [52].
Amyloid biomarkers in AD and other brain disorders
Both CSF Ab42 and amyloid PET have in numerous
studies been shown to have a high diagnostic performance
for AD. Approximately 8590% of clinically diagnosed
cases with AD dementia have a positive amyloid PET scan
(reviewed in [86]). Approximately 90% of PiB-positive MCI
cases progress to AD dementia during clinical follow-up,
while most PiB negative cases show stable cognition
[61,87,88]. Similarly, most clinically diagnosed AD cases
show low CSF levels of Ab42 [32]. Furthermore, clinical
follow-up studies have shown that more than 90% of those
progressing to AD dementia have low CSF levels of Ab42
7

TIPS-1217; No. of Pages 13

Review
at baseline [28], while stable MCI cases have normal CSF
Ab42 [89], findings that have been verified in several large
clinical multicenter studies [75,76,90].
The availability of two amyloid biomarker modalities
provides a solid ground to understand the specificity and
mechanisms for biomarker results found in brain disorders
other than AD. An overview of changes in the amyloid
biomarkers in other disorders than AD is given below.
Other clinical entities with amyloid pathology
Dementia with Lewy bodies (DLB) is characterized by
cortical and subcortical Lewy bodies comprising aggregated a-synuclein together with preferentially diffuse cortical
amyloid (Ab) plaques [91,92]. PET studies in DLB show
high cortical PiB retention in most patients with DLB [93],
to a degree similar to that found in AD [94]. Studies on CSF
Ab42 levels have consistently also shown decreased levels
in DLB [9597]. Some studies have examined whether it
may be possible to differentiate other neurodegenerative
disorders with b-amyloid pathology based on the regional
cortical amyloid PET pattern. Amyloid PET-positive DLB
cases show a cortical PiB pattern that is comparable with
that found in patients with AD [96]. The pattern of amyloid
PET signal is similar between the different clinical phenotypes of AD (early-onset AD, logopenic variant primary
progressive aphasia, and posterior cortical atrophy) [98],
and CSF Ab42 levels are reduced independent of phenotype [99101]. As mentioned above, cases with CAA have
positive PiB amyloid PET scans, even in the absence of
amyloid plaques [49].
Creutzfeldt-Jakob disease
Studies on CSF Ab42 in Creutzfeldt-Jakob disease (CJD)
show contradictory results. Some papers have reported
reduced CSF Ab42 levels [57,102], and it has been debated whether this would indicate that CSF Ab42 may drop
due to reasons other than Ab deposition in the brain, in
some neurodegenerative disorders. However, other CSF
studies found no change Ab42 levels [103,104], and amyloid PET studies on CJD show no cortical PiB retention,
with scans indistinguishable from control subjects
[105,106], although few cases have been examined in
these studies.
Infections and inflammatory brain disorders
Some reports have shown reduced CSF Ab42 levels in
infectious and inflammatory central nervous system
(CNS) disorders. Patients with acute bacterial meningitis
show a marked reduction in CSF Ab42, with normalization
after proper antibiotic treatment and clinical recovery,
while no change is found in cases with viral meningitis
[107,108]. In bacterial meningitis, there is a breakdown of
the bloodbrain barrier with high protein levels in CSF,
but this was shown not to interfere with the assay for Ab42
measurement [107,108]. Instead, the reduction in CSF
Ab42 may be due to Ab degradation by proteases released
in response to the acute inflammatory process [109].
Furthermore, some studies report low CSF Ab42, often
accompanied by low CSF levels of sAPP, in inflammatory
CNS disorders such as HIV dementia [107,110], systemic
lupus erythematosus (SLE) with CNS engagement [111],
8

Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x

Lyme neuroborreliosis (LNB) [109], and multiple sclerosis

(MS) [112114]. These findings suggest that CNS inflammation results in a general inhibition of APP metabolism
with lower production of Ab. This hypothesis is supported
by the finding that the reduction in CSF Ab is related to a
decrease in BACE1 [112], and that the reduction in both
sAPP and Ab species is normalized by treatment of
patients with MS patients with natalizumab, which inhibits transmission of immune cells into the CNS [114], and
after treatment of LNB with doxycycline [109]. However,
other studies found normal CSF Ab42 in HIV dementia
[115], SLE with CNS engagement [112], LNB [107], and
MS [116,117]. Thus, further studies are needed on CSF
Ab42 in inflammatory CNS disorders to resolve whether
an inflammatory process inhibits APP metabolism and Ab
generation.
Brain trauma
Diffuse axonal injury (DAI) is the central neuropathological change in acute traumatic brain injury (TBI; reviewed
in [118]). Postmortem studies on patients who suffered TBI
show that APP rapidly accumulates together with Ab in
damaged axons after trauma, [119] followed by release of
Ab into tissue and amyloid plaque formation [120
122]. The presence of Ab plaques after TBI has also been
verified in studies on fresh brain tissue samples excised
during surgery [123]. A recent PET study also showed
increased cortical PiB binding following severe TBI
[124]. It is difficult to perform lumbar puncture after severe
TBI, but studies on ventricular CSF samples show a
marked increase in Ab42, up to ten times of baseline levels,
during the days following trauma [125].
Concluding remarks
The amyloid biomarkers CSF Ab42 and amyloid PET both
show a high diagnostic ability to identify AD, also during
the earlier stages of the disease. These biomarkers also
show a high concordance, with approximately 90% of cases
being either positive or negative for both biomarkers. Both
amyloid biomarkers have a central position in the newly
revised criteria for prodromal AD outlined by Dubois and
coworkers in the International Working Group (IWG) for
the diagnosis of AD [126] and in the National Institute of
Aging and Alzheimers Association criteria for dementia
due to AD [127] and MCI due to AD [128]. The high
concordance for the amyloid biomarkers suggests that they
may be used interchangeably to aid clinical diagnosis
work-up or patient enrichment in clinical trials. As an
example, several clinical trials, such as the Phase III
solanezumab and gantenerumab trials, are using either
CSF Ab42 or amyloid PET for enrichment of prodromal AD
cases in the screening of MCI cases.
Thus, the choice between amyloid PET and CSF biomarkers as diagnostic tools in the clinic will not depend on
different performance of these biomarkers to identify AD,
but on other factors, such availability, adverse effects,
training status, and willingness among clinicians to perform lumbar puncture, availability of and distance to PET
scanners and cyclotrons, and finally financial considerations that payers have to make (Table 3). Health economic
studies have shown that the use of CSF biomarkers to

TIPS-1217; No. of Pages 13

Review

Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x

Table 3. Amyloid biomarkers: implementation in the clinica

Feature
Invasiveness

Amyloid PET
Radioactive molecule introduced into bloodstream by
intravenous injection
Injected PET ligand crosses bloodbrain barrier and
binds to amyloid deposits in brain
Total radiation exposure per examination is 9 mSv
when CT scan is performed for reconstruction of
imagesb (for comparison, annual background radiation
in USA is  3.1 mSv) c
PET instrument needed at hospital

Availability

CSF Ab42
Lumbar puncture, with introduction of needle into
subarachnoid space, performed for CSF collection

Physicians need experience to perform, or training to

start with, CSF collection by lumbar puncture.
Laboratory need strict internal quality control
systems to assure reproducible results, to both verify
individual runs and maintain long-term stability of
measurements [68]
Headache is most common adverse event (13%)
[130133], but some studies report higher incidence
[134]
Other adverse effects: back pain, dizziness, and
nausea
Reports on CSF Ab42 given as absolute levels, most
often as pg/mL (= ng/L)
Each laboratory has to establish a cut-off level,
because absolute levels vary between laboratories
[76]
Cost of lumbar puncture in Europe is estimated at
s150 (US$180) [129], but varies between countries,
and is integrated in general budget for clinics in some
countries (i.e., no extra charge)
Cost of CSF Ab42 test is s3560 (US$4070) in Europe

Cyclotron needed nearby, because half-life of 18F ligands

is 110 min b

Headache is most common adverse event (2%) b

Adverse effects

Other adverse effects: claustrophobia, anxiety, dizziness,

neck pain, nausea, and fatigue b
PET images labeled as positive or negative by visual
evaluation, by comparing radioactivity in cortical gray
matter with that in adjacent white matter

Interpretation
and reports

Cost of PET scan is s20002500 (US$23002850) in

Europe and US$30004500 in the USA

Health economics
(price per clinical
biomarker analysis)

Abbreviations: CT, computerized tomography.

Examples of radiation and adverse effects based on Amyvid (http://pi.lilly.com/us/amyvid-uspi.pdf).

Source for background radiation: http://www.epa.gov/radiation/understand/perspective.html.

identify AD among MCI patients is cost-effective, even

today when only symptomatic treatments are available
[129]. Except for financial considerations, the risk of complications after lumbar puncture is a common theme when
discussing implementation of the amyloid biomarkers in
the clinic. Several studies have showed that the main

(B)

1.0

Negave

1.2

1.4

Posive

1.6

1.8

2.0

Cut-o >1.11 SUVR

0.8

Low

Normal

100 150 200 250 300 350

Cut-o <192 pg/mL

50

CSF Ab42 (ng/L)

(A)

complication after LP, post-lumbar puncture headache,

is usually mild and of short duration (one or a few days),
and has an incidence of 13% [130133], which is in the
same range as the risk of headache after amyloid PET
(Table 3), although some studies report a higher frequency
[134]. Furthermore, to be implemented in clinical routine

PET Florbetapir (SUVr)

CN

AD

sMCI

pMCI

CN

AD

sMCI

pMCI

TRENDS in Pharmacological Sciences

Figure 3. Raw values for cerebrospinal fluid (CSF) and positron emission tomography (PET) amyloid biomarkers. (A) CSF b-amyloid (Ab) 42 levels in pg/mL (analyzed using
the AlzBio3 Luminex assay) and (B) florbetapir amyloid PET as global standardized uptake value ratio (SUVR). Lines represent established cut-offs cutoffs for Alzheimers
disease (AD), with a CSF Ab42 level below 192 pg/mL and a global florbetapir SUVR above 1.11 (normalized to whole cerebellum) indicative of AD. Diagnostic groups: C,
controls; AD, AD dementia; sMCI, stable mild cognitive impairment; pMCI, progressive MCI. Based on data from the Alzheimers Disease Neuroimaging Initiative (ADNI)
study. Reprinted from [74].

TIPS-1217; No. of Pages 13

Review
diagnostics and in clinical trials, standardization efforts
are needed both for CSF Ab42 measurements and amyloid
PET measurements. This is especially important to enable
these amyloid biomarkers to be used interchangeably in
trials, and to allow accurate comparisons between laboratories and in longitudinal follow-up of patients. It should be
noted that, while CSF Ab42 values in clinical routine are
reported as absolute levels in pg/mL as compared with a
cut-off value, amyloid PET scans are reported as either
positive or negative (Table 3). However, for both CSF Ab42
and global amyloid PET SUVR values, there is a continuum of values and an overlap between controls and patients
with AD or MCI, without any distinct cut-off (Figure 3);
thus, values that are close to the cut-off should be interpreted with caution.
Except for use in clinical diagnosis and treatment trials,
amyloid biomarkers have also proven valuable for understanding AD pathogenesis and learning about the time
course and interplay of the different pathogenic mechanisms in AD. The highly cited hypothetical model on the
temporal evolution of AD biomarkers (and pathogenesis)
was built on clinical biomarker studies [135]. In the
updated model, it is hypothesized that an incident Ab
deposition precedes tau pathology and neurodegeneration
[136]. Interestingly, a large report from the Dominantly
Inherited Alzheimers Network (DIAN) study suggested
that mutation carriers show a decrease in CSF Ab42
25 years before expected symptom onset, while amyloid
deposition measured by PET could be detected 15 years
before symptoms, at the same time when CSF levels of tau
protein increase and brain atrophy evaluated by MRI could
be noted [137]. Longitudinal clinical biomarkers studies
using novel biomarker modalities, such as tau PET [138]
and CSF biomarkers, for synaptic function and degeneration, including the presynaptic protein Synaptosomal-associated protein 25 (SNAP-25) [139] and the dendritic
protein neurogranin [140], will add to the understanding
of the evolution of pathogenic processes during the course
of AD.
References
1 Glenner, G.G. (1980) Amyloid deposits and amyloidosis. The betafibrilloses (first of two parts). N. Engl. J. Med. 302, 12831292
2 Blennow, K. et al. (2006) Alzheimers disease. Lancet 368, 387403
3 Hardy, J. (2009) The amyloid hypothesis for Alzheimers disease: a
critical reappraisal. J. Neurochem. 110, 11291134
4 Blennow, K. et al. (2014) Biomarkers in amyloid-beta immunotherapy
trials in Alzheimers disease. Neuropsychopharmacology 39, 189201
5 Salloway, S. et al. (2014) Two phase 3 trials of bapineuzumab in mildto-moderate Alzheimers disease. N. Engl. J. Med. 370, 322333
6 Doody, R.S. et al. (2013) A phase 3 trial of semagacestat for treatment
of Alzheimers disease. N. Engl. J. Med. 369, 341350
7 Doody, R.S. et al. (2014) Phase 3 trials of solanezumab for mild-tomoderate Alzheimers disease. N. Engl. J. Med. 370, 311321
8 Karran, E. and Hardy, J. (2014) Antiamyloid therapy for Alzheimers
diseaseare we on the right road? N. Engl. J. Med. 370, 377378
9 Blennow, K. (2010) Biomarkers in Alzheimers disease drug
development. Nat. Med. 16, 12181222
10 Blennow, K. et al. (2010) Cerebrospinal fluid and plasma biomarkers
in Alzheimer disease. Nat. Rev. Neurol. 6, 131144
11 Lannfelt, L. et al. (2008) Safety, efficacy, and biomarker findings of
PBT2 in targeting Abeta as a modifying therapy for Alzheimers
disease: a phase IIa, double-blind, randomised, placebo-controlled
trial. Lancet Neurol. 7, 779786

10

Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x

12 Winblad, B. et al. (2012) Safety, tolerability, and antibody response of

active Abeta immunotherapy with CAD106 in patients with
Alzheimers disease: randomised, double-blind, placebo-controlled,
first-in-human study. Lancet Neurol. 11, 597604
13 May, P.C. et al. (2011) Robust central reduction of amyloid-beta in
humans with an orally available, non-peptidic beta-secretase
inhibitor. J. Neurosci. 31, 1650716516
14 Hampel, H. et al. (2010) Biomarkers for Alzheimers disease:
academic, industry and regulatory perspectives. Nat. Rev. Drug
Discov. 9, 560574
15 Biomarkers Definitions Working Group (2001) Biomarkers and
surrogate endpoints: preferred definitions and conceptual
framework. Clin. Pharmacol. Ther. 69, 8995
16 Anon (1998) Consensus report of the Working Group on: Molecular
and Biochemical Markers of Alzheimers Disease. The Ronald and
Nancy Reagan Research Institute of the Alzheimers Association and
the National Institute on Aging Working Group. Neurobiol. Aging 19,
109116
17 Portelius, E. et al. (2011) A novel pathway for amyloid precursor
protein processing. Neurobiol. Aging 32, 10901098
18 Zetterberg, H. et al. (2008) Elevated cerebrospinal fluid BACE1
activity in incipient Alzheimer disease. Arch. Neurol. 65, 1102
1107
19 Holtta, M. et al. (2013) Evaluating amyloid-beta oligomers in
cerebrospinal fluid as a biomarker for Alzheimers disease. PLoS
ONE 8, e66381
20 Savage, M.J. et al. (2014) A sensitive abeta oligomer assay
discriminates Alzheimers and aged control cerebrospinal fluid.
J. Neurosci. 34, 28842897
21 Haass, C. et al. (1992) Amyloid beta-peptide is produced by cultured
cells during normal metabolism. Nature 359, 322325
22 Tabaton, M. et al. (1994) Soluble amyloid beta-protein is a marker of
Alzheimer amyloid in brain but not in cerebrospinal fluid. Biochem.
Biophys. Res. Commun. 200, 15981603
23 Van Nostrand, W.E. et al. (1992) Decreased levels of soluble amyloid
beta-protein precursor in cerebrospinal fluid of live Alzheimer disease
patients. Proc. Natl. Acad. Sci. U.S.A. 89, 25512555
24 Iwatsubo, T. et al. (1994) Visualization of A beta 42(43) and A beta
40 in senile plaques with end-specific A beta monoclonals: evidence
that an initially deposited species is A beta 42(43). Neuron 13,
4553
25 Jarrett, J.T. et al. (1993) The carboxy terminus of the beta amyloid
protein is critical for the seeding of amyloid formation: implications
for the pathogenesis of Alzheimers disease. Biochemistry 32, 4693
4697
26 Motter, R. et al. (1995) Reduction of beta-amyloid peptide42 in the
cerebrospinal fluid of patients with Alzheimers disease. Ann. Neurol.
38, 643648
27 Blennow, K. and Hampel, H. (2003) CSF markers for incipient
Alzheimers disease. Lancet Neurol. 2, 605613
28 Andreasen, N. et al. (1999) Cerebrospinal fluid beta-amyloid(1-42) in
Alzheimer disease: differences between early- and late-onset
Alzheimer disease and stability during the course of disease. Arch.
Neurol. 56, 673680
29 Olsson, A. et al. (2005) Simultaneous measurement of beta-amyloid(142), total tau, and phosphorylated tau (Thr181) in cerebrospinal fluid
by the xMAP technology. Clin. Chem. 51, 336345
30 Wiltfang, J. et al. (2002) Highly conserved and disease-specific
patterns of carboxyterminally truncated Abeta peptides 1-37/38/39
in addition to 1-40/42 in Alzheimers disease and in patients with
chronic neuroinflammation. J. Neurochem. 81, 481496
31 Leinenbach, A. et al. (2014) Mass spectrometry-based candidate
reference measurement procedure for quantification of amyloidbeta in cerebrospinal fluid. Clin. Chem. 60, 987994
32 Blennow, K. and Vanmechelen, E. (2003) CSF markers for pathogenic
processes in Alzheimers disease: diagnostic implications and use in
clinical neurochemistry. Brain Res. Bull. 61, 235242
33 Portelius, E. et al. (2006) Determination of beta-amyloid peptide
signatures in cerebrospinal fluid using immunoprecipitation-mass
spectrometry. J. Proteome Res. 5, 10101016
34 Hansson, O. et al. (2007) Prediction of Alzheimers disease using
the CSF Abeta42/Abeta40 ratio in patients with mild cognitive
impairment. Dement. Geriatr. Cogn. Disord. 23, 316320

TIPS-1217; No. of Pages 13

Review
35 Lewczuk, P. et al. (2015) Amyloid-beta 42/40 cerebrospinal fluid
concentration ratio in the diagnostics of Alzheimers disease:
validation of two novel assays. J. Alzheimers Dis. 43, 183191
36 Mehta, P.D. et al. (2000) Plasma and cerebrospinal fluid levels of
amyloid beta proteins 1-40 and 1-42 in Alzheimer disease. Arch.
Neurol. 57, 100105
37 Wiltfang, J. et al. (2007) Amyloid beta peptide ratio 42/40 but not A
beta 42 correlates with phospho-Tau in patients with low- and highCSF A beta 40 load. J. Neurochem. 101, 10531059
38 Klunk, W.E. et al. (2004) Imaging brain amyloid in Alzheimers
disease with Pittsburgh Compound-B. Ann. Neurol. 55, 306319
39 Jack, C.R., Jr et al. (2013) Cerebral amyloid PET imaging in
Alzheimers disease. Acta Neuropathol. 126, 643657
40 Klunk, W.E. (2011) Amyloid imaging as a biomarker for cerebral betaamyloidosis and risk prediction for Alzheimer dementia. Neurobiol.
Aging 32 (Suppl. 1), S20S36
41 Wong, D.F. et al. (2010) In vivo imaging of amyloid deposition in
Alzheimer disease using the radioligand 18F-AV-45 (florbetapir
[corrected] F 18). J. Nucl. Med. 51, 913920
42 Nelissen, N. et al. (2009) Phase 1 study of the Pittsburgh compound B
derivative 18F-flutemetamol in healthy volunteers and patients with
probable Alzheimer disease. J. Nucl. Med. 50, 12511259
43 Rowe, C.C. et al. (2008) Imaging of amyloid beta in Alzheimers
disease with 18F-BAY94-9172, a novel PET tracer: proof of
mechanism. Lancet Neurol. 7, 129135
44 Rowe, C.C. et al. (2013) Head-to-head comparison of 11C-PiB and 18FAZD4694 (NAV4694) for beta-amyloid imaging in aging and
dementia. J. Nucl. Med. 54, 880886
45 Lockhart, A. et al. (2007) PIB is a non-specific imaging marker of
amyloid-beta (Abeta) peptide-related cerebral amyloidosis. Brain 130,
26072615
46 Ikonomovic, M.D. et al. (2008) Post-mortem correlates of in vivo PiBPET amyloid imaging in a typical case of Alzheimers disease. Brain
131, 16301645
47 Bacskai, B.J. et al. (2007) Molecular imaging with Pittsburgh
Compound B confirmed at autopsy: a case report. Arch. Neurol. 64,
431434
48 Scholl, M. et al. (2012) Low PiB PET retention in presence of
pathologic CSF biomarkers in Arctic APP mutation carriers.
Neurology 79, 229236
49 Greenberg, S.M. et al. (2008) Detection of isolated cerebrovascular
beta-amyloid with Pittsburgh compound B. Ann. Neurol. 64, 587
591
50 Ly, J.V. et al. (2010) 11C-PIB binding is increased in patients
with cerebral amyloid angiopathy-related hemorrhage. Neurology
74, 487493
51 Clark, C.M. et al. (2011) Use of florbetapir-PET for imaging betaamyloid pathology. JAMA 305, 275283
52 Clark, C.M. et al. (2012) Cerebral PET with florbetapir compared
with neuropathology at autopsy for detection of neuritic amyloidbeta plaques: a prospective cohort study. Lancet Neurol. 11,
669678
53 Rinne, J.O. et al. (2012) [(18)F]Flutemetamol PET imaging and
cortical biopsy histopathology for fibrillar amyloid beta detection in
living subjects with normal pressure hydrocephalus: pooled analysis
of four studies. Acta Neuropathol. 124, 833845
54 Leinonen, V. et al. (2014) Diagnostic effectiveness of quantitative
[(1)(8)F]flutemetamol PET imaging for detection of fibrillar amyloid
beta using cortical biopsy histopathology as the standard of truth in
subjects with idiopathic normal pressure hydrocephalus. Acta
Neuropathol. Commun. 2, 46
55 Strozyk, D. et al. (2003) CSF Abeta 42 levels correlate with amyloidneuropathology in a population-based autopsy study. Neurology 60,
652656
56 Tapiola, T. et al. (2009) Cerebrospinal fluid {beta}-amyloid 42 and tau
proteins as biomarkers of Alzheimer-type pathologic changes in the
brain. Arch. Neurol. 66, 382389
57 Mollenhauer, B. et al. (2011) Different CSF beta-amyloid processing in
Alzheimers and Creutzfeldt-Jakob disease. J. Neural Transm. 118,
691697
58 Fagan, A.M. et al. (2006) Inverse relation between in vivo amyloid
imaging load and cerebrospinal fluid Abeta42 in humans. Ann.
Neurol. 59, 512519

Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x

59 Fagan, A.M. et al. (2007) Cerebrospinal fluid tau/beta-amyloid(42)

ratio as a prediction of cognitive decline in nondemented older adults.
Arch. Neurol. 64, 343349
60 Fagan, A.M. et al. (2009) Cerebrospinal fluid tau and ptau(181)
increase with cortical amyloid deposition in cognitively normal
individuals: implications for future clinical trials of Alzheimers
disease. EMBO Mol. Med. 1, 371380
61 Forsberg, A. et al. (2008) PET imaging of amyloid deposition in
patients with mild cognitive impairment. Neurobiol. Aging 29,
14561465
62 Jagust, W.J. et al. (2009) Relationships between biomarkers in aging
and dementia. Neurology 73, 11931199
63 Tolboom, N. et al. (2009) Relationship of cerebrospinal fluid markers
to 11C-PiB and 18F-FDDNP binding. J. Nucl. Med. 50, 14641470
64 Degerman Gunnarsson, M. et al. (2010) Pittsburgh compound-B and
Alzheimers disease biomarkers in CSF, plasma and urine: An
exploratory study. Dement. Geriatr. Cogn. Disord. 29, 204212
65 Weigand, S.D. et al. (2011) Transforming cerebrospinal fluid Abeta42
measures into calculated Pittsburgh Compound B units of brain
Abeta amyloid. Alzheimers Dement. 7, 133141
66 Zwan, M. et al. (2014) Concordance between cerebrospinal fluid
biomarkers and [11C]PIB PET in a memory clinic cohort. J.
Alzheimers Dis. 41, 801807
67 Landau, S.M. et al. (2013) Comparing positron emission tomography
imaging and cerebrospinal fluid measurements of beta-amyloid. Ann.
Neurol. 74, 826836
68 Palmqvist, S. et al. (2014) Accuracy of brain amyloid detection in
clinical practice using cerebrospinal fluid beta-amyloid 42: a crossvalidation study against amyloid positron emission tomography.
JAMA Neurol. 71, 12821289
69 Portelius, E. et al. (2010) Mass spectrometric characterization of brain
amyloid beta isoform signatures in familial and sporadic Alzheimers
disease. Acta Neuropathol. 120, 185193
70 Hampel, H. et al. (2010) Total and phosphorylated tau protein as
biological markers of Alzheimers disease. Exp. Gerontol. 45, 3040
71 Cairns, N.J. et al. (2009) Absence of Pittsburgh compound B detection
of cerebral amyloid beta in a patient with clinical, cognitive, and
cerebrospinal fluid markers of Alzheimer disease: a case report. Arch.
Neurol. 66, 15571562
72 Mattsson, N. et al. (2015) Independent information from cerebrospinal
fluid amyloid-beta and florbetapir imaging in Alzheimers disease.
Brain 138, 772783
73 Su, Y. et al. (2013) Quantitative analysis of PiBPET with FreeSurfer
ROIs. PLoS ONE 8, e73377
74 Mattsson, N. et al. (2014) Diagnostic accuracy of CSF Ab42 and
florbetapir PET for Alzheimers disease. Ann. Clin. Transl. Neurol.
1, 534543
75 Shaw, L.M. et al. (2009) Cerebrospinal fluid biomarker signature in
Alzheimers disease neuroimaging initiative subjects. Ann. Neurol.
65, 403413
76 Mattsson, N. et al. (2009) CSF biomarkers and incipient Alzheimer
disease in patients with mild cognitive impairment. JAMA 302, 385
393
77 Carrillo, M.C. et al. (2013) Global standardization measurement of
cerebral spinal fluid for Alzheimers disease: an update from the
Alzheimers Association Global Biomarkers Consortium. Alzheimers
Dement. 9, 137140
78 Korecka, M. et al. (2014) Qualification of a surrogate matrix-based
absolute quantification method for amyloid-beta(4)(2) in human
cerebrospinal fluid using 2D UPLC-tandem mass spectrometry.
J. Alzheimers Dis. 41, 441451
79 Mattsson, N. et al. (2012) Reference measurement procedures
for Alzheimers disease cerebrospinal fluid biomarkers: definitions
and approaches with focus on amyloid beta42. Biomark. Med. 6, 409
417
80 Mattsson, N. et al. (2011) The Alzheimers Association external
quality control program for cerebrospinal fluid biomarkers.
Alzheimers Dement. 7, 386395
81 Thie, J.A. (2004) Understanding the standardized uptake value, its
methods, and implications for usage. J. Nucl. Med. 45, 14311434
82 Mintun, M.A. et al. (1984) A quantitative model for the in vivo
assessment of drug binding sites with positron emission
tomography. Ann. Neurol. 15, 217327
11

TIPS-1217; No. of Pages 13

Review
83 Koivunen, J. et al. (2008) PET amyloid ligand [11C]PIB uptake
shows predominantly striatal increase in variant Alzheimers
disease. Brain 131, 18451853
84 Edison, P. et al. (2012) Can target-to-pons ratio be used as a reliable
method for the analysis of [11C]PIB brain scans? Neuroimage 60,
17161723
85 Klunk, W.E. et al. (2015) The centiloid project: standardizing
quantitative amyloid plaque estimation by PET. Alzheimers
Dement. 11, 115
86 Frisoni, G.B. et al. (2013) Imaging markers for Alzheimer disease:
which vs how. Neurology 81, 487500
87 Okello, A. et al. (2009) Conversion of amyloid positive and negative
MCI to AD over 3 years: an 11C-PIB PET study. Neurology 73, 754
760
88 Jack, C.R., Jr et al. (2010) Brain beta-amyloid measures and magnetic
resonance imaging atrophy both predict time-to-progression from
mild cognitive impairment to Alzheimers disease. Brain 133,
33363348
89 Hansson, O. et al. (2006) Association between CSF biomarkers and
incipient Alzheimers disease in patients with mild cognitive
impairment: a follow-up study. Lancet Neurol. 5, 228234
90 Visser, P.J. et al. (2009) Prevalence and prognostic value of CSF
markers of Alzheimers disease pathology in patients with
subjective cognitive impairment or mild cognitive impairment in
the DESCRIPA study: a prospective cohort study. Lancet Neurol. 8,
619627
91 Perry, R. et al. (1997) Lewy body dementia: clinical, pathological and
neurochemical interconnections. J. Neural Transm. Suppl. 51, 95
109
92 Ballard, C. et al. (2006) Differences in neuropathologic characteristics
across the Lewy body dementia spectrum. Neurology 67, 1931
1934
93 Edison, P. et al. (2008) Amyloid load in Parkinsons disease dementia
and Lewy body dementia measured with [11C]PIB positron emission
tomography. J. Neurol. Neurosurg. Psychiatry 79, 13311338
94 Gomperts, S.N. et al. (2008) Imaging amyloid deposition in Lewy body
diseases. Neurology 71, 903910
95 Parnetti, L. et al. (2008) Cerebrospinal fluid biomarkers in
Parkinsons disease with dementia and dementia with Lewy
bodies. Biol. Psychiatry 64, 850855
96 Maetzler, W. et al. (2009) Cortical PIB binding in Lewy body disease is
associated with Alzheimer-like characteristics. Neurobiol. Dis. 34,
107112
97 Hall, S. et al. (2012) Accuracy of a panel of 5 cerebrospinal fluid
biomarkers in the differential diagnosis of patients with dementia
and/or parkinsonian disorders. Arch. Neurol. 69, 14451452
98 Lehmann, M. et al. (2013) Diverging patterns of amyloid deposition
and hypometabolism in clinical variants of probable Alzheimers
disease. Brain 136, 844858
99 Baumann, T.P. et al. (2010) CSF-tau and CSF-Abeta(1-42) in
posterior cortical atrophy. Dement. Geriatr. Cogn. Disord. 29, 530
533
100 de Souza, L.C. et al. (2011) Cerebrospinal fluid biomarkers in the
differential diagnosis of Alzheimers disease from other cortical
dementias. J. Neurol. Neurosurg. Psychiatry 82, 240246
101 Seguin, J. et al. (2011) CSF biomarkers in posterior cortical atrophy.
Neurology 76, 17821788
102 Otto, M. et al. (2000) Decreased beta-amyloid1-42 in cerebrospinal
fluid of patients with Creutzfeldt-Jakob disease. Neurology 54, 1099
1102
103 Van Everbroeck, B. et al. (2003) A prospective study of CSF markers
in 250 patients with possible Creutzfeldt-Jakob disease. J. Neurol.
Neurosurg. Psychiatry 74, 12101214
104 Zanusso, G. et al. (2011) Cerebrospinal fluid markers in sporadic
Creutzfeldt-Jakob disease. Int. J. Mol. Sci. 12, 62816292
105 Hyare, H. et al. (2012) 11C-PiB PET does not detect PrP-amyloid
in prion disease patients including variant Creutzfeldt-Jakob disease.
J. Neurol. Neurosurg. Psychiatry 83, 340341
106 Villemagne, V.L. et al. (2009) 11C-PiB PET studies in typical sporadic
Creutzfeldt-Jakob disease. J. Neurol. Neurosurg. Psychiatry 80,
9981001
107 Krut, J.J. et al. (2013) Cerebrospinal fluid Alzheimers biomarker
profiles in CNS infections. J. Neurol. 260, 620626
12

Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x

108 Sjogren, M. et al. (2001) Low cerebrospinal fluid beta-amyloid

42 in patients with acute bacterial meningitis and normalization
after treatment. Neurosci. Lett. 314, 3336
109 Mattsson, N. et al. (2010) Neuroinflammation in Lyme
neuroborreliosis affects amyloid metabolism. BMC Neurol. 10, 51
110 Brew, B.J. et al. (2005) CSF amyloid beta42 and tau levels correlate
with AIDS dementia complex. Neurology 65, 14901492
111 Trysberg, E. et al. (2004) Decreased levels of soluble amyloid betaprotein precursor and beta-amyloid protein in cerebrospinal fluid of
patients with systemic lupus erythematosus. Arthritis Res. Ther. 6,
R129R136
112 Mattsson, N. et al. (2009) Reduced cerebrospinal fluid BACE1 activity
in multiple sclerosis. Mult. Scler. 15, 448454
113 Mori, F. et al. (2011) Cognitive and cortical plasticity deficits correlate
with altered amyloid-beta CSF levels in multiple sclerosis.
Neuropsychopharmacology 36, 559568
114 Augutis, K. et al. (2013) Cerebrospinal fluid biomarkers of betaamyloid metabolism in multiple sclerosis. Mult. Scler. 19, 543552
115 Gisslen, M. et al. (2009) Amyloid and tau cerebrospinal fluid
biomarkers in HIV infection. BMC Neurol. 9, 63
116 Mai, W. et al. (2011) Cerebrospinal fluid levels of soluble amyloid
precursor protein and beta-amyloid 42 in patients with multiple
sclerosis, neuromyelitis optica and clinically isolated syndrome.
J. Int. Med. Res. 39, 24022413
117 Szalardy, L. et al. (2013) Evaluating biomarkers of neuronal
degeneration and neuroinflammation in CSF of patients with
multiple sclerosis-osteopontin as a potential marker of clinical
severity. J. Neurol. Sci. 331, 3842
118 DeKosky, S.T. et al. (2013) Acute and chronic traumatic
encephalopathies: pathogenesis and biomarkers. Nat. Rev. Neurol.
9, 192200
119 Johnson, V.E. et al. (2010) Traumatic brain injury and amyloidbeta pathology: a link to Alzheimers disease? Nat. Rev. Neurosci.
11, 361370
120 Graham, D.I. et al. (1995) Distribution of beta-amyloid protein in
the brain following severe head injury. Neuropathol. Appl. Neurobiol.
21, 2734
121 Roberts, G.W. et al. (1991) beta A4 amyloid protein deposition in brain
after head trauma. Lancet 338, 14221423
122 Smith, D.H. et al. (2003) Amyloid beta accumulation in axons after
traumatic brain injury in humans. J. Neurosurg. 98, 10721077
123 Ikonomovic, M.D. et al. (2004) Alzheimers pathology in human
temporal cortex surgically excised after severe brain injury. Exp.
Neurol. 190, 192203
124 Hong, Y.T. et al. (2014) Amyloid imaging with carbon 11-labeled
Pittsburgh compound B for traumatic brain injury. JAMA Neurol.
71, 2331
125 Olsson, A. et al. (2004) Marked increase of beta-amyloid(1-42) and
amyloid precursor protein in ventricular cerebrospinal fluid after
severe traumatic brain injury. J. Neurol. 251, 870876
126 Dubois, B. et al. (2014) Advancing research diagnostic criteria for
Alzheimers disease: the IWG-2 criteria. Lancet Neurol. 13, 614629
127 McKhann, G.M. et al. (2011) The diagnosis of dementia due to
Alzheimers disease: recommendations from the National Institute
on Aging-Alzheimers Association workgroups on diagnostic
guidelines for Alzheimers disease. Alzheimers Dement. 7, 263269
128 Albert, M.S. et al. (2011) The diagnosis of mild cognitive impairment
due to Alzheimers disease: recommendations from the National
Institute on Aging-Alzheimers Association workgroups on
diagnostic guidelines for Alzheimers disease. Alzheimers Dement.
7, 270279
129 Valcarcel-Nazco, C. et al. (2014) Cost-effectiveness of the use of
biomarkers in cerebrospinal fluid for Alzheimers disease. J.
Alzheimers Dis. 42, 777788
130 Blennow, K. et al. (1993) Low frequency of post-lumbar puncture
headache in demented patients. Acta Neurol. Scand. 88, 221223
131 Peskind, E. et al. (2009) Safety of lumbar puncture procedures in
patients with Alzheimers disease. Curr. Alzheimer Res. 6, 290292
132 Peskind, E.R. et al. (2005) Safety and acceptability of the research
lumbar puncture. Alzheimer Dis. Assoc. Disord. 19, 220225
133 Zetterberg, H. et al. (2010) Low incidence of post-lumbar puncture
headache in 1,089 consecutive memory clinic patients. Eur. Neurol.
63, 326330

TIPS-1217; No. of Pages 13

Review
134 Alcolea, D. et al. (2014) Feasibility of lumbar puncture in the study of
cerebrospinal fluid biomarkers for Alzheimers disease: a multicenter
study in Spain. J. Alzheimers Dis. 39, 719726
135 Jack, C.R., Jr et al. (2010) Hypothetical model of dynamic biomarkers
of the Alzheimers pathological cascade. Lancet Neurol. 9, 119128
136 Jack, C.R., Jr et al. (2013) Tracking pathophysiological processes in
Alzheimers disease: an updated hypothetical model of dynamic
biomarkers. Lancet Neurol. 12, 207216
137 Bateman, R.J. et al. (2012) Clinical and biomarker changes in
dominantly inherited Alzheimers disease. N. Engl. J. Med. 367, 795804

Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x

138 Zimmer, E.R. et al. (2014) Developments in Tau PET Imaging.

Can. J. Neurol. Sci. 41, 547553
139 Brinkmalm, A. et al. (2014) SNAP-25 is a promising novel
cerebrospinal fluid biomarker for synapse degeneration in
Alzheimers disease. Mol. Neurodegener. 9, 53
140 Kvartsberg, H. et al. (2014) Cerebrospinal fluid levels of the synaptic
protein neurogranin correlates with cognitive decline in prodromal
Alzheimers disease. Alzheimers Dement. Published December 19,
2014. http://dx.doi.org/10.1016/j.jalz.2014.10.009

13