Journal of Internal Medicine - 2018 - Blennow - Biomarkers For Alzheimer S Disease Current Status and Prospects For The

Review
doi: 10.1111/joim.12816
Biomarkers for Alzheimer’s disease: current status and

prospects for the future
K. Blennow & H. Zetterberg
From the Clinical Neurochemistry Laboratory, Institute of Neuroscience and Physiology, The Sahlgrenska Academy at University of
Gothenburg, M€olndal, Sweden
Abstract. Blennow K, Zetterberg H (The Sahlgrenska biological definition of AD. Taken together, this
Academy at University of Gothenburg, M€ olndal, progress will likely serve as the basis for a more
Sweden). Biomarkers for Alzheimer’s disease: general introduction of these diagnostic tests in
current status and prospects for the future clinical routine practice. However, the heterogeneity
(Review). J Intern Med 2018; 284: 643–663. of pathology in late-onset AD calls for an expansion
of the AD CSF biomarker toolbox with additional
Accumulating data from the clinical research sup- biomarkers reflecting additional aspects of AD
port that the core Alzheimer’s disease (AD) cere- pathophysiology. One promising candidate is the
brospinal fluid (CSF) biomarkers amyloid-b (Ab42), synaptic protein neurogranin that seems specific for
total tau (T-tau), and phosphorylated tau (P-tau) AD and predicts future rate of cognitive deteriora-
reflect key elements of AD pathophysiology. Impor- tion. Further, recent studies bring hope for easily
tantly, a large number of clinical studies very accessible and cost-effective screening tools in the
consistently show that these biomarkers contribute early diagnostic evaluation of patients with cognitive
with diagnostically relevant information, also in the problems (and suspected AD) in primary care. In this
early disease stages. Recent technical developments respect, technical developments with ultrasensitive
have made it possible to measure these biomarkers immunoassays and novel mass spectrometry tech-
using fully automated assays with high precision niques give promise of biomarkers to monitor brain
and stability. Standardization efforts have given amyloidosis (the Ab42/40 or APP669-711/Ab42
certified reference materials for CSF Ab42, with the ratios) and neurodegeneration (tau and neurofila-
aim to harmonize results between assay formats ment light proteins) in plasma samples, but future
that would allow for uniform global reference limits studies are warranted to validate these promising
and cut-off values. These encouraging develop- results further.
ments have led to that the core AD CSF biomarkers
have a central position in the novel diagnostic Keywords: Alzheimer’s disease, biomarkers, blood,
criteria for the disease and in the recent National cerebrospinal fluid, diagnosis.
Institute on Aging and Alzheimer’s Association
This review paper gives an update on the research instruments and extensive clinical validation of
and development of cerebrospinal fluid (CSF) diagnostic performance. Recent developments have
biomarkers for Alzheimer’s disease (AD) with focus also given some novel candidate biomarkers for
on diagnostic applications. It is more than two synaptic degeneration, another key aspect of AD
decades since the most commonly used ELISA pathophysiology. Last, technical developments of
methods, the INNOTEST assays, for quantification novel ultrasensitive immunoassay and mass spec-
of total tau (T-tau), phosphorylated tau (P-tau) and trometry methods show promise for blood biomark-
amyloid-b (Ab42) in CSF were published [1–3], ers with potential applications as screening tools
showing increased levels of T-tau and P-tau for neurodegeneration and brain amyloidosis.
together with decreased Ab42, a biomarker pattern
often called the ‘Alzheimer’s CSF profile’. These core
Molecular neuropathology of Alzheimer’s disease
AD CSF biomarkers have made a journey from the
first studies based on research grade assays to their Ever since the first case report on AD, plaques and
current status with assays on fully automated tangles, the neuropathological hallmarks of AD,
ª 2018 The Association for the Publication of the Journal of Internal Medicine 643
13652796, 2018, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/joim.12816 by Readcube (Labtiva Inc.), Wiley Online Library on [25/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Biomarkers for AD: prospects for the future / K. Blennow & H. Zetterberg
were known to be composed of ‘amyloid’, that is the same type of pathology (amyloid plaques and
proteinaceous deposits that can be stained with tangles) as those with presenile AD [11, 12]. After
dyes such as Congo red. However, efforts to identify this, senile dementia in older patients was called
their protein composition were for a long time senile dementia of the Alzheimer’s type (SDAT) or
unsuccessful, largely due to the lack of methods to late-onset AD (LOAD), whilst patients with onset
purify plaques and tangles and the insolubility of before 65 years of age were diagnosed with early-
the amyloid fibrils. After the development of a onset AD (EOAD). However, with time, these bor-
method to purify amyloid plaques [4], a paper in ders bleached out, and all cases was termed AD,
1985 published the full sequence of a 40-amino irrespective of age (for review, see [13]).
acid (4 kDa) protein purified from plaques in AD
brain tissue [5]. Based on its molecular weight, the However, several studies have shown that the
protein was called amyloid A4 protein but today we severity of AD (i.e. plaque and tangle) pathology is
know it as b-amyloid or Ab. The identification of Ab greatest in early-onset AD patients, whilst in late-
(specifically its amino acid sequence) facilitated the onset cases the severity of neuropathological
cloning of the amyloid precursor protein (APP) gene changes varies considerably between patients,
[6]. Importantly, the Ab domain of 42–43 amino and in higher ages, the level of changes overlaps
acids within the APP protein (a transmembrane with that found in cognitively unimpaired elderly
protein with a single transmembrane domain) was [14–18]. Further, after the discoveries of other
predicted to be partially embedded in the plasma types of age-related proteinopathies than tau
membrane [6], and thus, the cleavage of APP by two pathology and Ab plaques, studies have shown
putative enzymes, called b-secretase (today identi- that most late-onset clinically diagnosed AD cases
fied as BACE1) and c-secretase (today identified as do not have pure AD (plaque and tangle) pathology,
the presenilin complex), was needed to generate Ab but also varying severities and combinations of a-
[7]. synuclein and TDP-43 deposits and additional
microvascular changes and hippocampal sclerosis
Tau protein is microtubule-associated protein pri- [19, 20]. This pathophysiological heterogeneity also
marily located in the neuronal axons, which due to makes clinical symptomatology of late-onset AD
alternative splicing has 6 isoforms, with 352–441 variable and unspecific [13] and introduces diag-
amino acids and with molecular weights of 50– nostic difficulties. Indeed, several papers have
65 kDa [8]. In 1986, it was shown that tangles are shown that also in cases that have undergone
composed of abnormally hyperphosphorylated tau evaluations at expert research centres and followed
protein [9] with around three times more phospho- clinically for years, the purely clinical diagnostic
rylated sites than normal tau. The hyperphospho- criteria for AD have poor accuracy, with sensitivity
rylation is believed to abrogate the normal function and specificity figures of around 70–80% when
of tau to bind to and stabilize the microtubules in related to neuropathology [21, 22].
the neurons [10], leading to disruption of the
microtubules and impairment of axoplasmic flow For these reasons, there is a large need of diag-
and loss of neuronal connectivity. Except for neu- nostic tools to support the clinical diagnosis of AD.
rofibrillary tangles, aggregated hyperphosphory- A correct clinical diagnosis of AD is currently
lated tau is also found in neuropil threads and in important to initiate treatment with symptomatic
the dystrophic neurites surrounding amyloid pla- drugs, including acetylcholine esterase (AChE)
ques (Tables 1 and 2). inhibitors and NMDA-receptor antagonists [23],
and will be even more necessary the day disease-
modifying drugs, such as secretase inhibitors or Ab
Pathophysiological rationale for Alzheimer’s disease fluid
immunotherapies, hopefully will be available. In
biomarkers
clinical medicine, fluid laboratory medicine
After the first case description of the disease in biomarkers have a central position and influence
1906, AD was regarded as a ‘presenile’ dementia, up to 70% of clinical medical decisions [24]. For
affecting people between 50 and 65 years of age, brain disorders, including AD, development of fluid
whilst older individuals with dementia were biomarkers was initiated using CSF as the matrix,
believed to have ‘senile dementia’, that is cognitive which compared with blood has the advantage of
deterioration as a part of more or less normal its proximity to the brain parenchyma, with brain
ageing. Around 1970, a number of publications proteins being secreted from the brain extracellular
showed that people with ‘senile dementia’ exhibited space to the CSF, which is accessible for CSF
644 ª 2018 The Association for the Publication of the Journal of Internal Medicine
Journal of Internal Medicine, 2018, 284; 643–663
Table 1 Possible applications of Alzheimer’s disease fluid biomarkers for diagnostics and screening
Application Application Principle Setting Assets Interpretation

Clinical Identification of AD CSF tests as an Specialist clinic/ Ab42 CSF Ab42 and the
diagnosis and differential integral part of the memory clinic Ab42/40 Ab42/40 ratio are
diagnosis in clinical diagnostic patients with SCI, ratio the AD biomarkers
clinical routine work-up MCI or mild-to- T-tau that become positive
Prediction of moderate dementia P-tau earliest during the
progression in Neurogranin clinical course of
patients with MCI disease
or SCI High CSF T-tau and
Enrichment of AD CSF sample taken Phase II and III trials P-tau predict
patients in clinical during the on early AD progression of
trials screening period, dementia or MCI cognitive symptoms
before enrolment cohorts better than Ab42
into a trial during a clinically
relevant time period
(1–2 years).
High CSF
neurogranin is
seemingly specific
for AD and does not
change in the
majority of other
neurodegenerative
disorders
Screening Initial evaluation of Blood tests (plasma) Primary care or Ab42/Ab40 The blood tests may
patients with SCI to identify or rule nonspecialist clinic ratio in future prove
and MCI out brain NFL valuable for
amyloidosis and screening, to select
neurodegeneration patients for
admission to the
specialist clinical for
a detailed diagnostic
evaluation
Ab, amyloid-b; AD, Alzheimer’s disease; CSF, cerebrospinal fluid; NFL, neurofilament light; P-tau, phosphorylated tau; T-
tau, total tau.
collection by lumbar puncture (for details, see

CSF amyloid-b as an Alzheimer’s biomarker
[25]). However, for possible applications as screen-
ing tools in primary care, or for longitudinal In 1992, it was shown that Ab is secreted to the
evaluations with repeated sampling, blood CSF [26], a finding that set the stage for developing
biomarkers would be preferable, as blood is more quantitative immunoassays for Ab in CSF. How-
accessible than CSF. Last, considering that elderly ever, initial papers on ‘total’ Ab in CSF were
AD patients have multiple pathologies, a broader disappointing, showing no or minor differences
panel of fluid biomarkers reflecting not only amy- between AD patients and controls [27]. Guided by
loid and tau pathology would be needed. the finding that Ab deposited in both diffuse and
646
Table 2 Applications for cerebrospinal fluid and blood biomarkers for Alzheimer’s disease
Application Fluid Biomarker Change in AD Stage of development Interpretation

Clinical CSF Ab42 Reduced Ab42 is characteristic Several immunoassays (ELISA, The reduction in CSF Ab42 in
diagnostics of AD dementia and MSD) commercially available AD reflects brain amyloidosis
prodromal AD Fully automated methods are and shows high concordance
Mean change around 50% of available in Europe with amyloid PET
levels in cognitively Mass spectrometry-based
unimpaired elderly reference measurement
procedures (RMP) are

approved
CSF Ab42/Ab40 Reduced Ab42/Ab40 ratio is Several immunoassays (ELISA, The Ab42/Ab40 ratio is
ratio characteristic of AD dementia MSD) are commercially believed to control for
and prodromal AD available between-individual variations
Mean change around 50% of Fully automated methods will in ‘total’ Ab production or
levels in cognitively soon be commercially secretion from neurons,
ª 2018 The Association for the Publication of the Journal of Internal Medicine
unimpaired elderly available variations in CSF dynamics
and/or correct for
preanalytical confounders
CSF T-tau Increased T-tau is Several immunoassays (ELISA, Increased CSF T-tau in AD
characteristic of AD dementia MSD) are commercially probably reflects the intensity
and prodromal AD available of neurodegeneration
Mean change around 250% of Fully automated methods are T-tau is normal in many
levels in cognitively available in Europe neurodegenerative disorders
unimpaired elderly (e.g. FTD and PSP).
A very marked increase in CSF
T-tau is found in CJD, where
P-tau is normal or mildly

increased
A temporary increase in T-tau
is found in acute brain
disorders (e.g. stroke) with
levels depending on the
severity of the lesion
Table 2 (Continued )
Application Fluid Biomarker Change in AD Stage of development Interpretation

CSF P-tau High P-tau is characteristic of Several immunoassays (ELISA, P-tau reflects phosphorylation
AD dementia and prodromal MSD) are commercially state of tau and increased P-
AD available tau in AD likely brain tau
Mean change around 250% of Fully automated methods are pathology
levels in cognitively available in Europe Increased CSF P-tau is only
unimpaired elderly found in AD
CSF Neurogranin High neurogranin is found in Immunoassays are High neurogranin reflects
AD and prodromal AD commercially available synaptic (dendritic)
degeneration
Increased neurogranin seems
specific for AD
Screening Blood Ab42/Ab40 Recent studies show that Commercial immunoassays Further studies needed to
test and reduced Ab42/Ab40 ratio, are available examine the performance of
APP669-711/ and high APP669-711/Ab42 Immunoprecipitation – mass plasma Ab42/Ab40 ratio, and
Ab42 ratio, correlates with brain spectrometry methods are APP669-711/Ab42 ratio, as a
amyloid positivity used in the research setting screening tool for brain
The ratios show a change of amyloidosis
around 15–30% in AD and in Method comparison studies
amyloid-positive cases needed to compare the
performance of
immunoassays vs mass
spectrometry methods for
plasma Ab
Blood Tau Plasma tau is increased in AD, Commercial immunoassays Plasma tau is a biomarker for
but with large overlap with are available acute brain damage and
control levels Several research-grade assays neurodegeneration and is not

have been published specific for AD
Plasma tau works well as a
biomarker for acute neuronal
damage
Method developments
warranted to improve

ª 2018 The Association for the Publication of the Journal of Internal Medicine
performance to identify
neurodegeneration in AD
647
Ab, amyloid-b; AD, Alzheimer’s disease; APP, amyloid precursor protein; CJD, Creutzfeldt–Jakob disease; CSF, cerebrospinal fluid; ELISA, enzyme-linked
immunosorbent assay; FTD, frontotemporal dementia; MSD, Meso Scale Discovery; NFL, neurofilament light; PET, positron emission tomography; PSP,
core amyloid plaques extends to position 42 (Ab42),
neurodegeneration, but is not

which also is the Ab species that is earliest
Plasma NFL is a sensitive deposited in plaques [28], the first ELISA method
marker for acute brain for Ab42 was published in 1995, showing a marked
reduction in CSF samples from AD patients [29].
The decrease in CSF Ab42 in AD dementia has been
specific for AD
validated in numerous subsequent papers; a meta-
Interpretation
damage and
analysis showed very consistent findings across

131 studies with a mean fold change of 0.56 for
CSF Ab42 compared with cognitively unimpaired
elderly [30].
The basis for the decrease in CSF Ab42 was

unresolved for long, but in 2003, an autopsy study
Commercial immunoassay is
found an association between low postmortem

Research use-only (RUO)
assay is used in clinical
ventricular CSF Ab42 levels and high plaque

counts [31], a finding validated in a study revealing
Stage of development
a correlation between reduced CSF Ab42 measured

in antemortem lumbar CSF samples and amyloid
plaque counts measured autopsy [32]. These
results are hampered by analysis of postmortem
available
studies
CSF samples or by the latency between ante-

mortem CSF samples and autopsy measures at
autopsy. However, the introduction of positron
emission tomography (PET) ligands binding to
fibrillar Ab in the brain enabled studies on the
with overlap between groups
relation between these amyloid biomarkers. In

High plasma NFL is found in
progressive supranuclear palsy; P-tau, phosphorylated tau; T-tau, total tau.

AD and prodromal AD, but
2006, a paper presented the findings that elderly

people, regardless of whether they had clinical AD
or were cognitively unimpaired, who had low CSF
Ab42 levels also had positive amyloid PET scans
and vice versa [33]. This matches the hypothesis
Change in AD
that the pathophysiological basis for the reduction

of CSF Ab42 in AD is that this hydrophobic peptide
aggregates and becomes sequestered in plaques,
with lower amounts remaining to be secreted to the
extracellular space and the CSF, resulting in lower
CSF levels of Ab42 [1]. Indeed, even if hampered by
uncertainties, the difference in CSF Ab42 concen-
trations between AD patients and controls (in
pg mL 1) matches the amount of Ab deposited in
Biomarker
plaques in the AD brain when considering the

extent and weight of affected tissue, a duration
NFL
(preclinical and clinical) of AD of around 30 years

and the average CSF production per day and year
[34].
Blood
Fluid
In agreement, several papers have consistently

Table 2 (Continued )
found a high concordance between CSF Ab42 and

amyloid PET status [35]. Except for cross-sec-
Application
tional studies, a high concordance between CSF

Ab42 and amyloid PET status was also shown in
a large prospective and longitudinal clinical study
on consecutive memory clinic patients [36]. The
figure for the concordance between CSF Ab42 the CSF Ab42/Ab40 ratio has diagnostic value in
and amyloid PET status is around 90% in most the clinical setting [50]. The reason for the
studies [35], independent of which assay is used improved performance of the Ab42/Ab40 ratio is
for measurement of CSF Ab42 and whether unclear, but it has been hypothesized that CSF
comparisons are made with mean cortical SUVr Ab40 can serve as a proxy for ‘total’ Ab levels and
or visual read of PET scans [35, 37]. This figure that the ratio normalizes for ‘total’ Ab production
is in the same range as the concordance between level between individuals, meaning that a reduc-
SUVr and visual read of PET scans and between tion in CSF Ab42 in individuals with high total Ab
different readers of PET scans to dichotomize into production can be identified more accurately, and
positive and negative scans [37]. A difference marginally low CSF Ab42 in individuals with low
between CSF and PET biomarkers for brain total Ab production will not be misinterpreted as
amyloid deposition is that whilst the CSF Ab42 indicative of brain amyloidosis [51]. However,
concentration is given as digits, amyloid PET alternative explanations may be that the ratio
scans show the regional distribution of ligand normalizes for differences in CSF dynamics (hav-
retention. However, a clinical study comparing ing similar effects on Ab42 and Ab40) or for
the diagnostic accuracy of CSF biomarkers and preanalytical confounders affecting both Ab42
amyloid PET for diagnosing early AD showed no and Ab40.
benefit of regional PET measures as compared
with global neocortical ligand retention, which
CSF tau proteins as Alzheimer’s biomarkers
both were highly concordant with CSF Ab42 [38].
Discordancy in the form of low CSF Ab42 but The finding that phosphorylated tau is the key
negative amyloid PET is mainly found in cogni- component of tangles [9] made tau proteins in CSF
tively unimpaired elderly and early AD patients, candidate biomarkers for AD. Indeed, using the
whilst it is unusual in AD dementia cases [39]. monoclonal antibody Alz-50, which reacts with
One study showed that nondemented individuals PHF-tau and normal tau protein [52], in 1987
with low CSF Ab42 but negative amyloid PET (Fig. 1), tau protein was identified in AD CSF
scans showed increased brain amyloid accumu- samples using Western blot [53], whilst levels were
lation at a follow-up PET scan, similar to those below detection limit in control CSF samples. This
who had positive amyloid PET at baseline, and at called for quantitative immunoassays to measure
a rate three times higher than those with negative tau in CSF.
scans and normal CSF Ab42 levels [40], suggest-
ing that CSF Ab42 may be an earlier biomarker
Total tau
for brain amyloidosis than amyloid PET. To sum
up, the current literature suggests that CSF Ab42 In 1993, the first ELISA method for quantification
and amyloid PET can be used interchangeably in of tau in CSF was published [54]. This ELISA was
the clinic, the choice may be based on availabil- based on the combination of a monoclonal antitau
ity, costs and risk estimations (radiation expo- antibody against the mid-domain combined with a
sure vs. after lumbar puncture headache), polyclonal antitau antiserum in the sandwich
together with both physician and patient prefer- format [54]. Two years later, the first sandwich
ences. ELISA method based only on monoclonal antibod-
ies, known as the ‘Innogenetics’ or “INNOTEST”
Except for Ab42, several other Ab species are assay, was published [2]. This assay is based on
present in human CSF, with Ab1-40 being the midregion monoclonal antibodies recognizing all
most abundant, found at around 10 times higher six tau isoforms irrespective of phosphorylation
concentrations than Ab42 [41, 42]. In 1998, a first state and therefore got the label as a ‘total’ tau (T-
study showed that a combined analysis of Ab42 tau) assay [2]. A marked increase in CSF t-tau was
and Ab40 improved the diagnostic accuracy for AD found in AD dementia patients, a finding that since
[43]. After this, numerous studies have shown that then has been replicated in hundreds of papers,
whilst CSF Ab40 shows no, or minor, change in AD also using several other assay formats [30].
[30], the CSF Ab42/Ab40 ratio has higher perfor-
mance to identify AD than CSF Ab42 as a single CSF T-tau has been proposed as a ‘state marker’,
biomarker [44–46]. Recent studies also indicate reflecting the intensity of neurodegeneration or
that the Ab42/Ab40 ratio shows better concor- severity of acute neuronal damage [55]. Indeed,
dance with amyloid PET positivity [47–49] and that following acute brain damage, CSF T-tau levels are
CSF P-tau
First sandwich correlates with
immunoassays total cortical
for CSF ligand binding
neurogranin on tau PET
2015 2016
First ELISA Start of the
for CSF International
total tau Federation Fully
Low CSF Aβ42 Fully
of Clinical Chemistry Automated
1993 correlates with automated
Working Group methods for
amyloid load on method
Tau for CSF proteins CSF T-tau
PET for CSF Aβ42
protein (IFCC WG-CSF) and P-tau
detected First ELISA 2016
in CSF
2006 2009 2018
for CSF
phospho tau
1987 β-amyloid
detected 1995
CSF P-tau levels Mass spec Certified
in CSF
Synaptic Low CSF Aβ42 correlate with Launch of the Reference reference
1992 proteins correlates with neocortical tangle Alzheimer’s Measuremen materials
First ELISA Association QC for CSF
detected post-mortem and neuritic t Procedures
for CSF program Aβ42
in CSF amyloid plaques plaque counts for CSF Aβ42
Aβ42
1999 2003 2006 2009 2014 2017
1995
1990 1995 2000 2005 2010 2015
Low CSF CSF biomarkers CSF High CSF

High CSF T-tau
Aβ42 in part of biomarkers Neurogranin
in AD
preclincial AD IWG research part of in prodromal
dementia
criteria for AD NIA-AA AD
1993 2000
research
2007 criteria for AD
2015
High T-tau/low Aβ42

2011
Low CSF Aβ42 in CSF predicts Meta-analysis show
Study with extended
in AD dementia progression in MCI consistent high
clinical follow-up
High CSF performance of the AD
1995 1999 show high that the AD
Neurogranin
CSF biomarkers CSF biomarkers
in AD
identify prodromal AD
dementia 2016
with high accuracy
High CSF P-tau Combined analysis of
CSF Aβ42 and Aβ40 2006 2010
in AD dementia
improves diagnostic CSF biomarkers
1995 accuracy for AD part of the A/T/N classification
system for AD pathophysiology
1998
2016
CSF biomarkers included in the

NIA-AA biological definition of AD
2018
Fig. 1 Timeline for the evolution of the core cerebrospinal fluid Alzheimer’s disease biomarkers. Yellow boxes, technical
developments; green boxes, pathophysiological findings; purple boxes, clinical findings; brown boxes, evolution of clinical
diagnostic criteria and pathogenic classifications. A, amyloid; AD, Alzheimer’s disease; CSF, cerebrospinal fluid; MCI, mild
cognitive impairment; N, neurodegeneration; PET, positron emission tomography; P-tau, phosphorylated tau; T, tau
pathology; T-tau, total tau.
dynamic and increase within days following injury The first assay for CSF P-tau ELISA method for
and then stay elevated for weeks until normaliza- quantification of tau phosphorylated at threonine
tion [56, 57], and in chronic neurodegenerative 181 + 231 showed a marked increase in AD [2]. A
disorders, the highest CSF T-tau levels are found in study using a modified assay, specific for tau
disorders with the most intense neurodegenera- phosphorylated at threonine 181 (P-tau181), con-
tion, especially in Creutzfeldt–Jakob disease, firmed this finding [3]. In addition, a marked
where levels are 10- to 20-fold higher than in AD increase in CSF P-tau in AD has been shown for
[58, 59]. In the AD spectrum, higher CSF T-tau and several other mid-domain phosphorylated tau
P-tau predict a more rapid clinical disease pro- residues, including threonine 231 and serine 235
gression [37, 60–62], supporting CSF T-tau as a as well as serine 199 [63], threonine 231 [64], and
biomarker for intensity of neurodegeneration. also for the C-terminal residues serine 396 and 404
[65]. Studies specifically comparing different CSF
P-tau assays as biomarkers for AD are few, but one
Phosphorylated tau
report showed that CSF levels of P-Tau181, P-
Measurement of tau protein in CSF that is phos- Tau199 and P-Tau231 correlate tightly and had a
phorylated at residues that is known to have this similar performance to discriminate AD from other
post-translational modification at the same resi- neurodegenerative disorders and nondemented
dues in AD brain tissue may reflect tau pathology. controls [66].
Some studies have examined the relationship biomarker of ‘disease stage’, correlating with stage
between CSF tau levels and neuropathology mea- of brain atrophy and severity of cognitive deficits.
sures of tau pathology. Some studies report corre- Importantly, in a study enrolling also non-AD
lations between CSF levels of P-tau and neocortical neurodegenerative disorders (progressive supranu-
tangle counts [67], but such associations may clear palsy, nonfluent primary progressive apha-
depend on the patient cohort examined and will sia, corticobasal syndrome and frontotemporal
rely on whether cases with both low (stage 1–2) and dementia), CSF tau proteins and tau PET both
high Braak stages are included in the correlations showed high differential diagnostic value [70]. This
[32]. In addition, several studies also show corre- suggests that it may be important to develop
lations between neuritic plaque counts and higher specific biomarkers for non-AD tau pathology.
CSF T-tau [32] and P-tau [67] levels. The findings
that CSF P-tau levels do not change with acute
CSF biomarkers for diagnosis of early Alzheimer’s disease
brain damage such as acute ischaemic stroke [56]
and are normal or only marginally increased It is logical to assume that disease-modifying drug
(whilst CSF T-tau shows a massive increase) in candidates have a larger chance to show effective-
neurodegenerative disorders with marked neu- ness if treatment can be initiated before neurode-
rodegeneration but no tangles, such as generation is too severe, that is the earliest
Creutzfeldt–Jakob disease [58, 59], support that symptomatic stages of the disease [72]. In 1999, a
the CSF level of P-tau probably reflects the phos- first paper showed that MCI patients whose clinical
phorylation state of tau and not simply neuronal follow-up investigations had deteriorated and
damage or degeneration. Indeed, at the group level, developed AD dementia had high CSF levels of T-
high CSF P-tau is only found in AD and not in other tau and low Ab42 at the baseline investigation [73],
neurodegenerative disorders. but this paper did not include any group with MCI
patients who did not deteriorate clinically. An
Recently, PET has been developed to visualize tau extended clinical follow-up period is important to
pathology directly in patients. In 2016, two studies ascertain that the so-called ‘stable’ MCI patients
examined the relationships between tau proteins will not show cognitive worsening or develop
measured in CSF and tau deposits evaluated by dementia. In 2006, an article presented data on
PET scans, demonstrating that CSF and PET tau the core AD CSF biomarkers on a large prospective
biomarkers show weak global correlations [68, 69]. cohort of MCI patients with an extended (4–
One study on cognitively normal elderly found that 7 years) clinical follow-up period [74]. This study
total cortical tau ligand binding correlates mod- showed a very high (95%) diagnostic sensitivity for
estly with CSF P-tau, but not with T-tau [68], whilst the combination of low Ab42 and high CSF T-tau/
the correlations were stronger with tau PET SUVRs P-tau to predict AD in the prodromal stage of the
in medial temporal lobe structures, the areas disease, together with a high specificity to differ-
known to be affected earliest in AD. The correlation entiate AD from stable MCI cases and those devel-
coefficients between the CSF biomarker and global oping other dementias, such as frontotemporal
tau PET metrics are stronger when including both dementia and Lewy body dementia [74]. High
controls and AD patients [70], apparently due to performance of the core AD CSF biomarkers for
the large difference in both CSF tau levels and tau prodromal AD was later verified in several large
ligand retention between controls and AD patients. multicentre studies such as the DESCRIPA study
A recent study examining the relations between [75], the ADNI study [76], and the Swedish Brain
CSF tau levels, MRI measures of atrophy and tau Power study [77].
PET showed that whilst tau PET correlated with
degree of atrophy on MRI and severity of cognitive Some studies have examined whether CSF
impairment, CSF T-tau and P-tau were highly biomarkers may aid in predicting AD pathology
correlated, with high levels found in preclinical already in the preclinical stage of the disease. In
AD, despite normal tau PET scans [71]. These 2003, a population-based study showed that CSF
findings support the concept that CSF T-tau and P- Ab42 is lowered in cognitively unimpaired 85-year-
tau mainly are biomarkers of ‘disease state’, neu- olds who later developed AD dementia, whilst there
rodegeneration and tau phosphorylation state, was no significant change in CSF Ab40 [78], or CSF
respectively, and are increased also in earlier T-tau or P-tau levels. The finding that low CSF
disease stages, before tau aggregates can be iden- Ab42, but not T-tau, predicts future cognitive
tified on PET scans. In contrast, tau PET is a decline and dementia in unimpaired elderly as long
as 8 years in advance was verified in a subsequent employed biomarkers (one or more of CSF Ab42
paper [79]. A clinical cohort study also showed that and tau proteins, volumetric MRI and amyloid
CSF Ab42, but not T-tau and P-tau, predicts future PET) [86]. Criteria based on similar principles for
cognitive decline in the elderly [80]. In a similar MCI due to AD [87] and dementia due to AD [88]
way, asymptomatic FAD mutation carriers FAD were also published by the National Institute on
mutations have low CSF Ab42 levels [81], whilst Aging and Alzheimer’s Association (NIA-AA) work-
one study also found the AD CSF profile of low CSF ing group on diagnostic guidelines for AD.
Ab42 and high T-tau and P-tau in asymptomatic Recently, an update of the IWG criteria was
mutation carriers [82]. These findings support that published [89], in which CSF biomarkers (low
lowering of CSF Ab42 is a very early indicator of Ab42 combined with high either T-tau or P-tau)
clinically silent brain amyloidosis. together with amyloid PET had a more central
role, whilst topographical biomarkers (volumetric
MRI and FDG-PET) were assigned as tools to
The core AD CSF biomarkers enter diagnostic criteria
monitor neurodegeneration and the disease
Historically, AD is classified into the group of course in the disease. This year, the National
brain disorders called ‘dementias’, which refers to Institute on Aging and Alzheimer’s Association
cognitive symptoms severe enough to interfere (NIA-AA) working group has now defined AD as a
with social or occupational activities. Following pathologic process that is identified primarily by
the dementia concept, clinical diagnostic criteria biomarkers (REF – Jack et al.). In this framework,
for AD were published by the Neurological and biomarkers are grouped into b-amyloid deposition,
Communicative Disorders and Stroke and the tau pathology and neurodegeneration, following
Alzheimer’s Disease and Related Disorders Asso- the A/T/N classification [90]. In the new NIA-AA
ciation (NINCDS-ADRDA) in 1984 [83]. These were definition of AD, the clinical consequences of the
clinical exclusion criteria for AD in the dementia disease (i.e. cognitive symptoms) will only be used
stage warranting a neuropathological examination for staging purposes [91].
after death to make a definite diagnosis. A diag-
nosis of ‘probable AD’ could be set in patients
Efforts to make the AD biomarker assays reach the Clinical
aged 40–90 years, with progressive dementia after
Chemistry grade
exclusion of other disorders than AD that could
account for the deficits in memory and cognition, Although numerous clinical studies reported excel-
whilst a diagnosis of ‘definite AD’ could only be set lent diagnostic performance of core AD CSF
after death and autopsy investigation [83]. It biomarkers measured by ELISA methods [1–3] or
should be noted that an exclusion diagnosis was by the Luminex technology [92], it became evident
the only way possible, given that no fluid or that there was a marked difference in absolute
imaging biomarkers were available at that time- levels reported between studies, also when using
point. In the 1990s, the earlier clinical stages of the same ELISA variant [77], with between-labora-
AD gained increasing attention in clinical AD tory variability being more pronounced for CSF
research, and the new term mild cognitive impair- Ab42 than for T-tau or P-tau [93]. These types of
ment (MCI) was introduced, referring to the tran- differences in absolute levels may be caused by
sitional state between normal ageing and different preanalytical procedures (e.g. type of test
dementia [84]. Similar to dementia, MCI is an tube for CSF collection or freeze–thaw schedule)
aetiologically heterogeneous syndrome that can be across clinics and laboratories [94]. In addition,
due to many underlying diseases, with only between-laboratory variability may also be due to
around half of cases having AD pathology [85]. discrepancies in analytical procedures between
laboratories, or in the manufacturing procedures
An important first step towards a biological defi- for the immunoassays, resulting in batch-to-batch
nition of AD came in 2007, when the International variation.
Working Group (IWG) led by Bruno Dubois pub-
lished the first research criteria for the diagnosis
The Alzheimer’s Association quality control programme
of prodromal AD [86]. First, these criteria provided
the conceptual framework to allow a diagnosis of Strict quality control procedures in a single clinical
AD in patients with a very mild clinical phenotype laboratory can assure correct measurements over
of episodic memory disturbances, that is before time [36] and be the basis for the implementation of
the dementia stage. Second, these criteria CSF biomarker assays in clinical routine
diagnostics [95]. However, differences in absolute which isotope-labelled Ab42 spiked into to the CSF
levels between laboratories preclude the introduc- sample as an internal standard prior to sample
tion of uniform global cut-off levels and hinder a work-up, were developed [99–101]. These candi-
widespread implementation of CSF biomarkers in date methods were then fully validated and
clinical routine. Therefore, the Alzheimer’s Associ- received approval by the Joint Committee for
ation quality control (QC) programme for CSF Traceability in Laboratory Medicine (JCTLM)
biomarkers [93] was launched in 2009, with the [102], which is the regulatory body for reference
aim to establish a platform to monitor the perfor- methods, as RMPs for CSF Ab42 (Nos. C11RMP9
mance of the CSF biomarker measurements and C11RMP9).
between laboratories and between batches of
reagents. Between-laboratory variability and For the development of the Ab42 CRM, com-
between-batch variability are not unique for the mutability studies showed that only native human
AD CSF biomarkers, and the design of the QC CSF would work, as different variants of artificial
programme is also similar to other proficiency CSFs with spiked Ab42 did not behave like the
programmes for routine biomarker assays in clin- native peptide when present in human CSF [103].
ical chemistry, with aliquots from the same pools In collaboration with the Joint Research Centre
being sent out to participants, and biomarker (JRC) European Commission Science Hub, see
reported back, and summary forms presented by https://ec.europa.eu/jrc/en/reference-materials,
the organizers. Disappointingly, the between- the IFCC WG-CSF produced three CRMs with low,
laboratory CVs have consistently been 15–25% medium and high CSF Ab42 levels, with certified
since start, despite standardization and training concentrations assigned using the RMPs after the
efforts, including the introduction of standard evaluation of concordance in round robin studies
operating procedures (SOPs) for the ELISA methods [49]. These three CRMs (meaning large sets of CSF
[96], denoting the need for more precise and auto- aliquots) have also been evaluated for quality
mated analytical techniques. measures such as homogeneity and long-term
stability. The CRMs will serve at the top of the
calibration hierarchy, with the aim to calibrate
Reference methods and materials
commercial assays to these reference aliquots, and
The highest level of standardization in clinical make absolute levels obtained using different
chemistry is through a certified reference material assays comparable.
(CRM), meaning (in our case) a ‘gold standard’ CSF
pool with certified biomarker levels, from which
CSF biomarkers on fully automated instruments
aliquots are distributed to biotech companies for
harmonization of biomarker levels between assay As reviewed above, the between-laboratory vari-
formats and to assure low batch-to-batch variabil- ability for the AD CSF biomarkers seen in the
ity for the assays. In 2009, the International Alzheimer’s Association QC programme showed
Federation of Clinical Chemistry and Laboratory the need to develop assays for these proteins on
Medicine Working Group for CSF proteins (IFCC fully automated laboratory instruments, just like
WG-CSF), see http://www.ifcc.org/ifcc-scientific- for many protein biomarkers (e.g. troponin T for
division/sd-working-groups/csf-proteins-wg-csf/, myocardial infarction and TSH for thyroid disor-
was formed (Fig. 1). Knowing that Ab42 was the ders). These instruments involve no manual steps,
most problematic AD CSF biomarker, the WG and assays generally have superior performance as
started working on standardization of this biomar- compared with ELISA methods.
ker [97] in close conjunction with the Alzheimer’s
Association Global Biomarker Standardization In 2016, the first paper on the full validation and
Consortium (GBSC) [98]. analytical performance of the Cobas Elecsys b-
amyloid [1–42] assay was published [104]. This
To allow the measurement of the absolute concen- assay showed excellent performance and very low
tration of Ab42 in CSF (Fig. 2), an analytical batch-to-batch variability and has also been run-
method capable of absolute quantification without ning in the QC programme since 2014, with
matrix effects was needed. Within the IFCC WG- between-laboratory coefficients of variations (CVs)
CSF, two similar, but not identical, reference of a few percentage as compared with 15–20% for
measurement procedures (RMP) based on selected the ELISA methods. In addition, novel assays for T-
reaction monitoring (SRM) mass spectrometry, in tau and P-tau on the Elecsys instrument were also
Total tau (T-tau)

CSF T-tau gives a measure of the
intensity of neurodegeneration in AD,
Phosphorylated tau (P-tau) but is not a disease-specific marker
CSF P-tau is a measure of the amount
of tau that is phosphorylated, the
variant of tau found in tangles
Neurogranin
Neurogranin is a synaptic protein in
the dendritic spines, with CSF levels
reflecting synaptic dysfunction and
degeneration. High CSF neurogranin is
seemingly specific for AD
Aβ42 and Aβ42/40 ratio

CSF Aβ42 is lowered in AD, reflecting the aggregation
and deposition of the protein in the brain
Aβ40 is the most abundant variant of Aβ in CSF and thus
the CSF Aβ42/40 ratio compensates for between
individual differeces among Aβ isoforms
Fig. 2 Schematic diagram of a neuron with intracellular neurofibrillary tangles and extracellular neuritic amyloid plaques.
The core cerebrospinal fluid Alzheimer’s disease biomarkers and the novel synaptic biomarker candidate neurogranin are
indicated in the boxes. AD, Alzheimer’s disease; CSF, cerebrospinal fluid.
recently published [37] and have performed two

Synaptic biomarkers for AD
rounds in the QC programme, showing excellent
CVs of 2.7% (T-tau) and 1.8% (P-tau). Other Although the core CSF AD biomarkers reflect
companies, such as Fujirebio and Euroimmun, central pathogenic mechanisms of the disease,
have followed and built assays on fully automated novel biomarkers to monitor additional important
laboratory platforms, also showing highly molecular mechanisms in AD are constantly
improved performance in the QC programme. sought. One important component of AD patholog-
ical change and pathophysiology is synaptic dys-
Future developments likely include that fully function and degeneration. Synapses are the
automated laboratory analyser assays will prove central communication units in the neuronal net-
to give stable and precise results for the AD CSF works of the brain. Synapses consist of a presy-
biomarkers between laboratories, which, together naptic domain, where synaptic vesicles that
with CRMs, will allow for the establishment of contain the neurotransmitters that are released
uniform worldwide cut-off levels. This will be upon activation are located. Neurotransmitter
important both in the routine diagnostic evalua- release is a process regulated by a delicate machin-
tion of patients with suspected AD and in the ery of specific presynaptic proteins [105]. After
diagnostic procedure for inclusion in clinical trials release to the synaptic cleft, neurotransmitters
in novel disease-modifying drugs. Further, stable bind to postsynaptic receptors at the dendritic
and exact CSF biomarker levels will allow for spines and activate a cascade of molecular events
merging the data from clinical research studies to advance the neuronal signal [106]. Synaptic
worldwide, in clinical studies on disease patho- dysfunction and degeneration are likely the direct
genesis. cause of the cognitive deterioration in AD.
A large body of the literature supports a marked [122, 123]. Neurogranin expression is highest in
degeneration and loss of synapses in grey matter associative cortical areas, but levels are markedly
regions in AD, also in the early disease stages [107, reduced in the hippocampus and the frontal cortex
108]. Importantly, severity of synaptic loss is more in AD, indicating loss of postsynaptic elements
tightly correlated with degree of cognitive impair- [124, 125]. Thus, measurement of neurogranin in
ment than either plaque or tangle counts [109– CSF may serve as a biomarker for dendritic insta-
111], and synaptic degeneration has been sug- bility and synaptic degeneration.
gested as the best anatomical correlate of cognitive
deficits in AD [109, 112]. Further, experimental After developing novel monoclonal antibodies to
animal studies suggest that both Ab fibrils [113] measure neurogranin by ELISA, high CSF levels
and diffusible Ab oligomers [114] may disturb were found to predict prodromal AD in MCI [126].
dendritic spines by distinct mechanisms. In addi- High CSF neurogranin in AD dementia and pro-
tion, tau hyperphosphorylation and microglia acti- dromal AD has been confirmed in several subse-
vation may also contribute to spine loss [115, 116]. quent papers [127, 128], including in the ADNI
Thus, synaptic biomarkers in CSF may serve as study [129], and using immunoassays recognizing
tools to explore this important aspect of AD patho- different variants of neurogranin, including the
physiology in man and to examine the link between full-length protein [130] and C-terminal peptides
effects on AD molecular pathology and cognitive ending at position 75 [131], as well as mass
symptoms by novel drug candidates with disease- spectrometry quantification of a series of C-term-
modifying potential. Synapses are plastic struc- inal peptides [126]. High CSF neurogranin corre-
tures in the brain, and potentially, synaptic mark- lates with future rate of hippocampal trophy
ers would change rapidly in response to successful measured by MRI and rate of metabolic reductions
treatment. on FDG-PET [129]. Interestingly, a recent study
suggests that high CSF neurogranin may be speci-
fic for AD and not found in other neurodegenerative
Early search for synaptic proteins in CSF
disorders such as frontotemporal dementia, Lewy
Based on the semipreparative scale chromato- body dementia, Parkinson’s disease, progressive
graphic and gel electrophoretic protein separation supranuclear palsy or multiple system atrophy
combined with Western blotting and mass spec- [132].
trometric identification, we were in the late 1990s
able to identify synaptic proteins in CSF from the
Presynaptic biomarkers
key synaptic compartments, including the presy-
naptic vesicle proteins synaptotagmin and rab3a, In the presynaptic terminal, the SNARE complex
the presynaptic membrane protein SNAP-25 and proteins, including synaptosomal-associated pro-
the dendritic protein neurogranin [117, 118]. tein 25 (SNAP-25), syntaxin-1 and vesicle-asso-
These discoveries served as the motivation to ciated membrane protein (VAMP)/synaptobrevin,
initiate a project on the production of novel anti- are key components of the molecular machinery
bodies and detailed mass spectrometric character- that drives fusion of membranes in neurotransmit-
ization of synaptic proteins in human CSF aiming ter exocytosis [133]. Whilst SNAP-25 is located at
at developing quantitative immunoassays for reli- the synaptic vesicles, synaptotagmin-1 (SYT1) is
able quantification in individual samples. A first found in the presynaptic plasma membrane and is
pilot study in 2010, based on the semiquantitative essential for synaptic vesicle exocytosis and thus
immunoprecipitation combined with Western blot- neurotransmitter release [134].
ting, showed promising results with a marked
increase in CSF neurogranin in AD [119]. The levels of both SNAP-25 and SYT1 are reduced
in cortical areas in the AD brain [124, 135],
reflecting the synaptic degeneration and loss in
Dendritic proteins – neurogranin
AD. Interestingly, using immunoprecipitation
Dendritic spines are specialized protrusions on the mass spectrometry methods, a marked increase
dendrites, the point where neurons receive and in the CSF levels of both SNAP-25 and SYT1 was
integrate information. Neurogranin is a dendritic found in AD dementia and prodromal AD cases
protein, expressed in the cortex and hippocampus [135, 136]. These promising results need validation
by excitatory neurons [120, 121], and is known to in future studies, but suggest that a set of synaptic
play an important role in long-term potentiation proteins covering different components of the
synaptic unit (dendrites – neurogranin, presynap- epitope masking by hydrophobic Ab peptides

tic plasma membrane – SNAP-25, synaptic vesicles binding to plasma proteins [141], or other inter-
– SYT1) may be valuable tools in clinical studies on ferences that might be mitigated by analytical
the relevance of synaptic dysfunction and degen- improvements.
eration in AD pathogenesis and maybe also in the
clinical evaluation of patients. In 2011, we published a novel method based on the
single-molecule array (Simoa) technique for mea-
surement of Ab42 in plasma [142]. This technique
Blood biomarkers for AD
is based on immunocapture of the protein biomar-
As blood is more accessible than CSF, there is little ker on magnetic beads, which are trapped in
doubt that blood sampling would be preferable to femtolitre volume wells, followed by the addition
CSF when it comes to taking fluid samples to of enzyme-labelled detection antibody and digital
measure AD biomarkers, both for clinical diagnosis quantification that allows for exact quantification
or screening and for repeated sampling in clinical of Ab42 down to subpicogram per mL levels (limit of
trials. However, developing blood biomarkers for quantification of 0.04 pg mL 1). The high analyti-
AD has proven difficult; whilst the CSF is contin- cal sensitivity allows for predilution of samples that
uous with the brain extracellular fluid, with a free may reduce matrix interferences. When evaluating
exchange of molecules from the brain to the CSF, this assay in the large Swedish BioFINDER study
only a fraction of brain proteins enters the blood- cohort, weak but significant correlations were
stream. Further, blood is a more challenging found between both plasma Ab42 and the Ab42/
matrix than CSF for brain biomarkers, for several 40 ratio and the corresponding CSF measures, as
reasons. First, the minute amounts of brain pro- well as to cortical [18F]flutemetamol PET retention
teins entering the blood have to be measured in a [143]. Significantly lower plasma Ab42/40 ratio
matrix containing very high levels of plasma pro- (P < 0.002) was found in both MCI and AD cases as
teins, such as albumin and IgG, introducing a high compared with controls.
risk of interference in analytical methods [137].
Second, in addition to dilution, brain proteins In an attempt to evaluate whether mass spectro-
released into blood may be degraded by proteases, metric analysis may give a more accurate quantifi-
metabolized in the liver or cleared by the kidneys, cation of Ab peptides in plasma, we developed an
which will introduce a variance that is unrelated to immunoprecipitation (IP) mass spectrometry (MS)
brain changes and difficult to control for. This selected reaction monitoring (SRM) method for
limits the potential for finding blood biomarkers for quantification of Ab42 and Ab40, where stable
AD [138]. Nevertheless, technical developments in isotope-labelled Ab peptides are added to the
the field of ultrasensitive immunoassays and mass sample before analysis (and thus processed and
spectrometry have given new hopes [139]. analysed simultaneously with endogenous Ab pep-
tides) and using the detergent octyl glucopyra-
noside to disrupt complexes between Ab and
Ab in plasma
plasma proteins such as albumin [144]. In a small
Whilst numerous papers on CSF Ab42 consis- pilot clinical study based on clinically diagnosed
tently have found a high concordance with amy- cases, we were not able to find any significant
loid PET measures of plaque burden [35], and a change, even if there was an apparent trend for a
marked decrease in AD, studies on plasma Ab42 reduction on both plasma Ab42 and the Ab42/40
as a biomarker reflecting brain amyloid pathology ratio in AD [144]. Interestingly, using a similar IP-
(and thus AD) have been disappointing, with MS method, also involving LysN proteolytic diges-
contradictory results, with no or minor changes tion of Ab peptides before analysis, significantly
and large overlaps in both Ab42 and Ab40 levels lower Ab42 concentration and Ab42/40 ratio were
between patients and controls [30]. This lack of found in amyloid PET-positive compared with PET-
association with disease pathology may be due to negative cases [145]. The Ab42/40 ratio was 14%
the contribution from peripheral tissues to plasma lower in the amyloid PET-positive group, which
Ab, as also evidenced by the lack of correlation gave an impressive ROC value of 0.89 [145].
between plasma and CSF Ab concentrations [140]. Additional MS-based studies suggest that a ratio
The poor disease association might also be related of a certain APP fragment (APP669-711) to Ab42 or
to analytical shortcomings using ELISA methods Ab42/Ab40 in plasma identifies Ab-positive indi-
or other standard immunoassays, for example viduals with high sensitivity and specificity [146,
147]. Plasma APP669-711-to-Ab42 ratio was 20– between plasma and CSF levels of NFL protein is
40% higher and the Ab40/Ab42 ratio was 17–32% tight [153].
higher in Ab-positive than in Ab-negative individu-
als, which gave sensitivity and specificity figures of A recent study on the ADNI cohort showed a
predicting 11CPiB in AD and MCI patients, as well marked increase in plasma NFL in AD cases
as in cognitively normal individuals, of 91% and (149% of control levels), with a receiver operating
87%, respectively [147]. These promising results characteristic (ROC) area under the curve (AUC)
call for further studies to evaluate plasma Ab as a value of 0.87, which is comparable to the core AD
screening tool for brain amyloidosis and AD, also CSF biomarkers [155]. While the change in the MCI
including larger clinical cohorts and comparisons group was less pronounced, plasma NFL was
of different analytical platforms for measurement. highest MCI cases with positive amyloid PET
scans, and predicted faster cognitive deterioration,
higher rate of future both brain atrophy (measured
Tau protein in plasma
by MRI) and hypometabolism as measured by
Ultrasensitive immunoassay techniques also allow FDG-PET [155]. Importantly, in a study on 48
for measurement of tau protein in blood samples familial AD (FAD) mutation carriers and noncarri-
[139], with increased tau levels in plasma in AD ers, blood NFL was increased in symptomatic FAD
found using both the immunomagnetic reduction cases, but also in presymptomatic mutation carri-
(IMR) [148] and Simoa [149] methods. A large study ers, with levels correlating with expected estimated
on both the ADNI and BioFINDER cohorts could year of symptom onset as well as both cognitive
confirm an increase in plasma tau concentrations and MRI measures of disease stage [156]. These
in AD dementia, although with a substantial over- results indicate that blood NFL detects neurode-
lap in levels with controls [150]. Interestingly, generation also in the preclinical stage of AD.
longitudinal data showed significant correlations
between plasma tau levels and future cognitive In this context, an important piece of knowledge is
decline, as well as increases in atrophy measured that high plasma (or CSF) NFL is not a feature that
by MRI and in hypometabolism measured by FDG- is specific for AD. Instead, increased levels are
PET during the follow-up [150]. Thus, current data found in many neurodegenerative disorders, such
suggest a minor increase in plasma tau in AD, as frontotemporal dementia, progressive supranu-
although with too large overlap with controls to be clear palsy and corticobasal syndrome [157, 158].
diagnostically useful. Tau protein in CSF has been Thus, a possible future application for plasma NFL
found to be present as truncated fragments [151], is as a screening test at the first clinical evaluation
and it is possible that development of assays based of patients with cognitive disturbances, for example
on antibodies for specific tau fragments will at the primary care unit. Here, plasma NFL might
improve performance. Alternatively, measurement serve as simple, noninvasive and cheap screening
of T-tau or P-tau in neuron-enriched exosome tool, primarily to rule out neurodegeneration.
preparations may improve performance for tau as
a blood biomarker [152], but further studies are
Concluding remarks
needed to validate this finding.
The last 20 years have seen an enormous expan-
sion in research on fluid biomarkers for AD. The
Neurofilament light in plasma
core AD CSF biomarkers T-tau, P-tau and Ab42
Recently, the first Simoa method for quantification (and the Ab42/40 ratio) have been evaluated in
of the axonal neurofilament light (NFL) protein in hundreds of clinical neurochemical studies with
blood samples was published [153]. This assay has extraordinary consistent results, showing high
many-fold higher analytical sensitivity than assays diagnostic accuracy not only for AD dementia,
using the same anti-NFL antibodies based on the but importantly also for prodromal AD. These
electrochemiluminescence (ECL) Meso Scale Diag- biomarkers have undergone a phase of standard-
nostics (MSD) technique or standard ELISA [154], ization, and new assay versions on fully automated
meaning that NFL can be measured also in blood instruments show excellent analytical performance
samples from normal individuals who have plasma and low intra- and interlaboratory variation. The
NFL concentrations that are below the level for core AD biomarkers are today part of research
accurate quantification when using ECL-MSD or diagnostic criteria, and we foresee an increased use
ELISA. In contrast to tau protein, the correlation of these diagnostic tests in clinical routine practice.
The AD CSF biomarker toolbox has been expanded 4 Merz PA, Wisniewski HM, Somerville RA, Bobin SA, Masters
with novel biomarkers reflecting additional aspects CL, Iqbal K. Ultrastructural morphology of amyloid fibrils
from neuritic and amyloid plaques. Acta Neuropathol 1983;
of AD pathology, such as synaptic dysfunction.
60: 113–24.
5 Masters CL, Simms G, Weinman NA, Multhaup G, McDonald
We envision further development and validation of BL, Beyreuther K. Amyloid plaque core protein in Alzheimer
assays reflecting other pathologies common in age- disease and Down syndrome. Proc Natl Acad Sci USA 1985;
related neurodegenerative disorders, including 82: 4245–9.
Lewy body and TDP-43 pathology, reaching the 6 Kang J, Lemaire HG, Unterbeck A et al. The precursor of
Alzheimer’s disease amyloid A4 protein resembles a cell-
stage of clinical applications in the coming years,
surface receptor. Nature 1987; 325: 733–6.
so that CSF biomarkers can be part in a person- 7 Masters CL, Bateman R, Blennow K, Rowe CC, Sperling RA,
alized medicine approach to improve the evalua- Cummings JL. Alzheimer’s disease. Nat Rev Dis Primers
tion of patients with cognitive disturbances. Last, 2015; 1: 15056.
we hope that blood biomarkers may be imple- 8 Goedert M, Spillantini MG, Potier MC, Ulrich J, Crowther
mented as screening tools in the first-in-line clin- RA. Cloning and sequencing of the cDNA encoding an
ical evaluation of this group of patients, once we isoform of microtubule-associated protein tau containing
four tandem repeats: differential expression of tau protein
have disease-modifying drugs at hand.
mRNAs in human brain. EMBO J 1989; 8: 393–9.
9 Grundke-Iqbal I, Iqbal K, Tung YC, Quinlan M, Wisniewski
HM, Binder LI. Abnormal phosphorylation of the micro-
Acknowledgements
tubule-associated protein tau (tau) in Alzheimer cytoskeletal
Dr. Blennow holds the Torsten S€ oderberg Profes- pathology. Proc Natl Acad Sci USA 1986; 83: 4913–7.
sorship at the Royal Swedish Academy of Sciences. 10 Iqbal K, Alonso AD, Gondal JA et al. Mechanism of neu-
rofibrillary degeneration and pharmacologic therapeutic
Dr. Zetterberg is a Wallenberg Academy Fellow.
approach. J Neural Transm Suppl 2000; 59: 213–22.
11 Tomlinson BE, Blessed G, Roth M. Observations on the
Conflict of interest statement brains of non-demented old people. J Neurol Sci 1968; 7:
331–56.
KB has served as a consultant or at advisory 12 Tomlinson BE, Blessed G, Roth M. Observations on the brains
boards for Alzheon, BioArctic, Biogen, Eli Lilly, of demented old people. J Neurol Sci 1970; 11: 205–42.
13 Blennow K, Wallin A. Clinical heterogeneity of probable
Fujirebio Europe, IBL International, Merck, Novar-
Alzheimer’s disease. J Geriatr Psychiatry Neurol 1992; 5:
tis, Pfizer and Roche Diagnostics and is a cofoun- 106–13.
der of Brain Biomarker Solutions in Gothenburg 14 Dayan AD. Quantitative histological studies on the aged
AB, a GU Venture-based platform company at the human brain. II. Senile plaques and neurofibrillary tangles
University of Gothenburg. HZ has served at advi- in senile dementia (with an appendix on their occurrence in
sory boards for Eli Lilly, Roche Diagnostics and cases of carcinoma). Acta Neuropathol 1970; 16: 95–102.
Wave, has received travel support from Teva and is 15 Dayan AD. Quantitative histological studies on the aged
human brain. I. Senile plaques and neurofibrillary tangles in
a cofounder of Brain Biomarker Solutions in
“normal” patients. Acta Neuropathol 1970; 16: 85–94.
Gothenburg AB, a GU Venture-based platform 16 Mann DM, Yates PO, Marcyniuk B. Alzheimer’s presenile
company at the University of Gothenburg. dementia, senile dementia of Alzheimer type and Down’s
syndrome in middle age form an age related continuum of
pathological changes. Neuropathol Appl Neurobiol 1984; 10:
185–207.
References 17 Price JL, McKeel DW Jr, Buckles VD et al. Neuropathology of
nondemented aging: presumptive evidence for preclinical
1 Andreasen N, Hesse C, Davidsson P et al. Cerebrospinal
Alzheimer disease. Neurobiol Aging 2009; 30: 1026–36.
fluid beta-amyloid(1-42) in Alzheimer disease: differences
18 Price JL, Morris JC. Tangles and plaques in nondemented
between early- and late-onset Alzheimer disease and stabil-
aging and “preclinical” Alzheimer’s disease. Ann Neurol
ity during the course of disease. Arch Neurol 1999; 56: 673–
1999; 45: 358–68.
80.
19 James BD, Wilson RS, Boyle PA, Trojanowski JQ, Bennett
2 Blennow K, Wallin A, Agren H, Spenger C, Siegfried J,
DA, Schneider JA. TDP-43 stage, mixed pathologies, and
Vanmechelen E. Tau protein in cerebrospinal fluid: a
clinical Alzheimer’s-type dementia. Brain 2016; 139: 2983–
biochemical marker for axonal degeneration in Alzheimer
93.
disease? Mol Chem Neuropathol 1995; 26: 231–45.
20 Kovacs GG, Milenkovic I, Wohrer A et al. Non-Alzheimer
3 Vanmechelen E, Vanderstichele H, Davidsson P et al. Quan-
neurodegenerative pathologies and their combinations are
tification of tau phosphorylated at threonine 181 in human
more frequent than commonly believed in the elderly brain: a
cerebrospinal fluid: a sandwich ELISA with a synthetic
community-based autopsy series. Acta Neuropathol 2013;
phosphopeptide for standardization. Neurosci Lett 2000;
126: 365–84.
285: 49–52.
21 Beach TG, Monsell SE, Phillips LE, Kukull W. Accuracy of identifying early Alzheimer disease. Neurology. 2015; 85:
the clinical diagnosis of Alzheimer disease at National 1240–9.
Institute on Aging Alzheimer Disease Centers, 2005-2010. 39 Mattsson N, Insel PS, Donohue M et al. Independent infor-
J Neuropathol Exp Neurol 2012; 71: 266–73. mation from cerebrospinal fluid amyloid-beta and florbe-
22 Knopman DS, DeKosky ST, Cummings JL et al. Practice tapir imaging in Alzheimer’s disease. Brain 2015; 138: 772–
parameter: diagnosis of dementia (an evidence-based review). 83.
Report of the Quality Standards Subcommittee of the Amer- 40 Palmqvist S, Mattsson N, Hansson O; Alzheimer’s Disease
ican Academy of Neurology. Neurology 2001; 56: 1143–53. Neuroimaging I. Cerebrospinal fluid analysis detects cere-
23 Blennow K, de Leon MJ, Zetterberg H. Alzheimer’s disease. bral amyloid-beta accumulation earlier than positron emis-
Lancet 2006; 368: 387–403. sion tomography. Brain 2016; 139: 1226–36.
24 Beastall GH, Watson ID. Clinical Chemistry and Laboratory 41 Portelius E, Tran AJ, Andreasson U et al. Characterization of
Medicine: an appreciation. Clin Chem Lab Med 2013; 51: 3–4. amyloid beta peptides in cerebrospinal fluid by an auto-
25 Blennow K, Hampel H, Weiner M, Zetterberg H. Cere- mated immunoprecipitation procedure followed by mass
brospinal fluid and plasma biomarkers in Alzheimer disease. spectrometry. J Proteome Res 2007; 6: 4433–9.
Nat Rev Neurol 2010; 6: 131–44. 42 Portelius E, Westman-Brinkmalm A, Zetterberg H, Blennow
26 Seubert P, Vigo-Pelfrey C, Esch F et al. Isolation and K. Determination of beta-amyloid peptide signatures in
quantification of soluble Alzheimer’s beta-peptide from bio- cerebrospinal fluid using immunoprecipitation-mass spec-
logical fluids. Nature 1992; 359: 325–7. trometry. J Proteome Res 2006; 5: 1010–6.
27 Nakamura T, Shoji M, Harigaya Y et al. Amyloid beta protein 43 Shoji M, Matsubara E, Kanai M et al. Combination assay of
levels in cerebrospinal fluid are elevated in early-onset CSF tau, A beta 1-40 and A beta 1-42(43) as a biochemical
Alzheimer’s disease. Ann Neurol 1994; 36: 903–11. marker of Alzheimer’s disease. J Neurol Sci 1998; 158: 134–
28 Iwatsubo T, Odaka A, Suzuki N, Mizusawa H, Nukina N, Ihara 40.
Y. Visualization of A beta 42(43) and A beta 40 in senile plaques 44 Hansson O, Zetterberg H, Buchhave P et al. Prediction of
with end-specific A beta monoclonals: evidence that an initially Alzheimer’s disease using the CSF Abeta42/Abeta40 ratio in
deposited species is A beta 42(43). Neuron 1994; 13: 45–53. patients with mild cognitive impairment. Dement Geriatr
29 Motter R, Vigo-Pelfrey C, Kholodenko D et al. Reduction of Cogn Disord 2007; 23: 316–20.
beta-amyloid peptide42 in the cerebrospinal fluid of patients 45 Lewczuk P, Esselmann H, Otto M et al. Neurochemical
with Alzheimer’s disease. Ann Neurol 1995; 38: 643–8. diagnosis of Alzheimer’s dementia by CSF Abeta42, Abe-
30 Olsson B, Lautner R, Andreasson U et al. CSF and blood ta42/Abeta40 ratio and total tau. Neurobiol Aging 2004; 25:
biomarkers for the diagnosis of Alzheimer’s disease: a 273–81.
systematic review and meta-analysis. Lancet Neurol 2016; 46 Wiltfang J, Esselmann H, Bibl M et al. Amyloid beta peptide
15: 673–84. ratio 42/40 but not A beta 42 correlates with phospho-Tau
31 Strozyk D, Blennow K, White LR, Launer LJ. CSF Abeta 42 in patients with low- and high-CSF A beta 40 load. J
levels correlate with amyloid-neuropathology in a popula- Neurochem 2007; 101: 1053–9.
tion-based autopsy study. Neurology 2003; 60: 652–6. 47 Janelidze S, Zetterberg H, Mattsson N et al. CSF Abeta42/
32 Tapiola T, Alafuzoff I, Herukka SK et al. Cerebrospinal fluid Abeta40 and Abeta42/Abeta38 ratios: better diagnostic
{beta}-amyloid 42 and tau proteins as biomarkers of markers of Alzheimer disease. Ann Clin Transl Neurol 2016;
Alzheimer-type pathologic changes in the brain. Arch Neurol 3: 154–65.
2009; 66: 382–9. 48 Lewczuk P, Matzen A, Blennow K et al. Cerebrospinal Fluid
33 Fagan AM, Mintun MA, Mach RH et al. Inverse relation Abeta42/40 Corresponds Better than Abeta42 to Amyloid
between in vivo amyloid imaging load and cerebrospinal fluid PET in Alzheimer’s Disease. J Alzheimers Dis 2017; 55: 813–
Abeta42 in humans. Ann Neurol 2006; 59: 512–9. 22.
34 Permanne B, Buee L, David JP, Fallet-Bianco C, Di Menza C, 49 Pannee J, Gobom J, Shaw LM et al. Round robin test on
Delacourte A. Quantitation of Alzheimer’s amyloid peptide quantification of amyloid-beta 1-42 in cerebrospinal fluid by
and identification of related amyloid proteins by dot-blot mass spectrometry. Alzheimers Dement 2016; 12: 55–9.
immunoassay. Brain Res 1995; 685: 154–62. 50 Dumurgier J, Schraen S, Gabelle A et al. Cerebrospinal fluid
35 Blennow K, Mattsson N, Scholl M, Hansson O, Zetterberg H. amyloid-beta 42/40 ratio in clinical setting of memory cen-
Amyloid biomarkers in Alzheimer’s disease. Trends Pharma- ters: a multicentric study. Alzheimers Res Ther 2015; 7: 30.
col Sci 2015; 36: 297–309. 51 Lewczuk P, Lelental N, Spitzer P, Maler JM, Kornhuber J.
36 Palmqvist S, Zetterberg H, Blennow K et al. Accuracy of Amyloid-beta 42/40 cerebrospinal fluid concentration ratio
brain amyloid detection in clinical practice using cere- in the diagnostics of Alzheimer’s disease: validation of two
brospinal fluid beta-amyloid 42: a cross-validation study novel assays. J Alzheimers Dis 2015; 43: 183–91.
against amyloid positron emission tomography. JAMA Neu- 52 Ksiezak-Reding H, Binder LI, Yen SH. Immunochemical and
rol 2014; 71: 1282–9. biochemical characterization of tau proteins in normal and
37 Hansson O, Seibyl J, Stomrud E et al. CSF biomarkers of Alzheimer’s disease brains with Alz 50 and Tau-1. J Biol
Alzheimer’s disease concord with amyloid-beta PET and Chem 1988; 263: 7948–53.
predict clinical progression: A study of fully automated 53 Wolozin B, Davies P. Alzheimer-related neuronal protein A68:
immunoassays in BioFINDER and ADNI cohorts. Alzheimers specificity and distribution. Ann Neurol 1987; 22: 521–6.
Dement 2018; https://doi.org/10.1016/j.jalz.2018.01.010. 54 Vandermeeren M, Mercken M, Vanmechelen E et al. Detec-
[Epub ahead of print]. tion of tau proteins in normal and Alzheimer’s disease
38 Palmqvist S, Zetterberg H, Mattsson N et al. Detailed cerebrospinal fluid with a sensitive sandwich enzyme-linked
comparison of amyloid PET and CSF biomarkers for immunosorbent assay. J Neurochem 1993; 61: 1828–34.
55 Blennow K, Hampel H. CSF markers for incipient Alzhei- 72 Blennow K. Biomarkers in Alzheimer’s disease drug devel-
mer’s disease. Lancet Neurol 2003; 2: 605–13. opment. Nat Med 2010; 16: 1218–22.
56 Hesse C, Rosengren L, Andreasen N et al. Transient increase 73 Andreasen N, Minthon L, Vanmechelen E et al. Cere-
in total tau but not phospho-tau in human cerebrospinal brospinal fluid tau and Abeta42 as predictors of develop-
fluid after acute stroke. Neurosci Lett 2001; 297: 187–90. ment of Alzheimer’s disease in patients with mild cognitive
57 Zetterberg H, Hietala MA, Jonsson M et al. Neurochemical impairment. Neurosci Lett 1999; 273: 5–8.
aftermath of amateur boxing. Arch Neurol 2006; 63: 1277– 74 Hansson O, Zetterberg H, Buchhave P, Londos E, Blennow K,
80. Minthon L. Association between CSF biomarkers and incipi-
58 Riemenschneider M, Wagenpfeil S, Vanderstichele H et al. ent Alzheimer’s disease in patients with mild cognitive impair-
Phospho-tau/total tau ratio in cerebrospinal fluid discrim- ment: a follow-up study. Lancet Neurol 2006; 5: 228–34.
inates Creutzfeldt-Jakob disease from other dementias. Mol 75 Visser PJ, Verhey F, Knol DL et al. Prevalence and prognos-
Psychiatry 2003; 8: 343–7. tic value of CSF markers of Alzheimer’s disease pathology in
59 Skillback T, Rosen C, Asztely F, Mattsson N, Blennow K, patients with subjective cognitive impairment or mild cogni-
Zetterberg H. Diagnostic performance of cerebrospinal fluid tive impairment in the DESCRIPA study: a prospective
total tau and phosphorylated tau in Creutzfeldt-Jakob cohort study. Lancet Neurol 2009; 8: 619–27.
disease: results from the Swedish Mortality Registry. JAMA 76 Shaw LM, Vanderstichele H, Knapik-Czajka M et al. Cere-
Neurol 2014; 71: 476–83. brospinal fluid biomarker signature in Alzheimer’s disease
60 Wallin AK, Blennow K, Zetterberg H, Londos E, Minthon L, neuroimaging initiative subjects. Ann Neurol 2009; 65: 403–
Hansson O. CSF biomarkers predict a more malignant 13.
outcome in Alzheimer disease. Neurology 2010; 74: 1531– 77 Mattsson N, Zetterberg H, Hansson O et al. CSF biomarkers
7. and incipient Alzheimer disease in patients with mild cog-
61 Buchhave P, Minthon L, Zetterberg H, Wallin AK, Blennow K, nitive impairment. JAMA 2009; 302: 385–93.
Hansson O. Cerebrospinal fluid levels of beta-amyloid 1-42, 78 Skoog I, Davidsson P, Aevarsson O, Vanderstichele H,
but not of tau, are fully changed already 5 to 10 years before Vanmechelen E, Blennow K. Cerebrospinal fluid beta-
the onset of Alzheimer dementia. Arch Gen Psychiatry 2012; amyloid 42 is reduced before the onset of sporadic dementia:
69: 98–106. a population-based study in 85-year-olds. Dement Geriatr
62 Hertze J, Minthon L, Zetterberg H, Vanmechelen E, Blennow Cogn Disord 2003; 15: 169–76.
K, Hansson O. Evaluation of CSF biomarkers as predictors 79 Gustafson DR, Skoog I, Rosengren L, Zetterberg H, Blennow
of Alzheimer’s disease: a clinical follow-up study of K. Cerebrospinal fluid beta-amyloid 1-42 concentration may
4.7 years. J Alzheimers Dis 2010; 21: 1119–28. predict cognitive decline in older women. J Neurol Neurosurg
63 Ishiguro K, Ohno H, Arai H et al. Phosphorylated tau in Psychiatry 2007; 78: 461–4.
human cerebrospinal fluid is a diagnostic marker for 80 Stomrud E, Hansson O, Blennow K, Minthon L, Londos E.
Alzheimer’s disease. Neurosci Lett 1999; 270: 91–4. Cerebrospinal fluid biomarkers predict decline in subjective
64 Kohnken R, Buerger K, Zinkowski R et al. Detection of tau cognitive function over 3 years in healthy elderly. Dement
phosphorylated at threonine 231 in cerebrospinal fluid of Geriatr Cogn Disord 2007; 24: 118–24.
Alzheimer’s disease patients. Neurosci Lett 2000; 287: 187– 81 Moonis M, Swearer JM, Dayaw MP et al. Familial Alzheimer
90. disease: decreases in CSF Abeta42 levels precede cognitive
65 Hu YY, He SS, Wang X et al. Levels of nonphosphorylated decline. Neurology 2005; 65: 323–5.
and phosphorylated tau in cerebrospinal fluid of Alzheimer’s 82 Ringman JM, Younkin SG, Pratico D et al. Biochemical
disease patients : an ultrasensitive bienzyme-substrate- markers in persons with preclinical familial Alzheimer
recycle enzyme-linked immunosorbent assay. Am J Pathol disease. Neurology 2008; 71: 85–92.
2002; 160: 1269–78. 83 McKhann G, Drachman D, Folstein M, Katzman R, Price D,
66 Hampel H, Buerger K, Zinkowski R et al. Measurement of Stadlan EM. Clinical diagnosis of Alzheimer’s disease: report
phosphorylated tau epitopes in the differential diagnosis of of the NINCDS-ADRDA Work Group under the auspices of
Alzheimer disease: a comparative cerebrospinal fluid study. Department of Health and Human Services Task Force on
Arch Gen Psychiatry 2004; 61: 95–102. Alzheimer’s Disease. Neurology 1984; 34: 939–44.
67 Buerger K, Ewers M, Pirttila T et al. CSF phosphorylated tau 84 Petersen RC. Mild cognitive impairment as a diagnostic
protein correlates with neocortical neurofibrillary pathology entity. J Intern Med 2004; 256: 183–94.
in Alzheimer’s disease. Brain 2006; 129: 3035–41. 85 Winblad B, Palmer K, Kivipelto M et al. Mild cognitive
68 Chhatwal JP, Schultz AP, Marshall GA et al. Temporal T807 impairment–beyond controversies, towards a consensus:
binding correlates with CSF tau and phospho-tau in normal report of the International Working Group on Mild Cognitive
elderly. Neurology 2016; 87: 920–6. Impairment. J Intern Med 2004; 256: 240–6.
69 Gordon BA, Friedrichsen K, Brier M et al. The relationship 86 Dubois B, Feldman HH, Jacova C et al. Research criteria for
between cerebrospinal fluid markers of Alzheimer pathology the diagnosis of Alzheimer’s disease: revising the NINCDS-
and positron emission tomography tau imaging. Brain 2016; ADRDA criteria. Lancet Neurol 2007; 6: 734–46.
139: 2249–60. 87 Albert MS, DeKosky ST, Dickson D et al. The diagnosis of
70 La Joie R, Bejanin A, Fagan AM et al. Associations between mild cognitive impairment due to Alzheimer’s disease: rec-
[(18)F]AV1451 tau PET and CSF measures of tau pathology ommendations from the National Institute on Aging-Alzhei-
in a clinical sample. Neurology 2018; 90: e282–90. mer’s Association workgroups on diagnostic guidelines for
71 Mattsson N, Scholl M, Strandberg O et al. (18)F-AV-1451 Alzheimer’s disease. Alzheimers Dement 2011; 7: 270–9.
and CSF T-tau and P-tau as biomarkers in Alzheimer’s 88 McKhann GM, Knopman DS, Chertkow H et al. The diagno-
disease. EMBO Mol Med 2017; 9: 1212–23. sis of dementia due to Alzheimer’s disease:
recommendations from the National Institute on Aging- 104 Bittner T, Zetterberg H, Teunissen CE et al. Technical
Alzheimer’s Association workgroups on diagnostic guideli- performance of a novel, fully automated electrochemilumi-
nes for Alzheimer’s disease. Alzheimers Dement 2011; 7: nescence immunoassay for the quantitation of beta-amyloid
263–9. (1-42) in human cerebrospinal fluid. Alzheimers Dement
89 Dubois B, Feldman HH, Jacova C et al. Advancing research 2016; 12: 517–26.
diagnostic criteria for Alzheimer’s disease: the IWG-2 crite- 105 Jahn R, Fasshauer D. Molecular machines governing exo-
ria. Lancet Neurol 2014; 13: 614–29. cytosis of synaptic vesicles. Nature 2012; 490: 201–7.
90 Jack CR Jr, Bennett DA, Blennow K et al. A/T/N: An 106 Sudhof TC. Neurotransmitter release: the last millisecond in
unbiased descriptive classification scheme for Alzheimer the life of a synaptic vesicle. Neuron 2013; 80: 675–90.
disease biomarkers. Neurology 2016; 87: 539–47. 107 Masliah E, Mallory M, Alford M et al. Altered expression of
91 Jack CR Jr, Bennett DA, Blennow K et al. NIA-AA Research synaptic proteins occurs early during progression of Alzhei-
Framework: toward a biological definition of Alzheimer’s mer’s disease. Neurology 2001; 56: 127–9.
disease. Alzheimers Dement 2018; 14: 535–62. 108 Scheff SW, Price DA, Schmitt FA, DeKosky ST, Mufson EJ.
92 Olsson A, Vanderstichele H, Andreasen N et al. Simultane- Synaptic alterations in CA1 in mild Alzheimer disease and
ous measurement of beta-amyloid(1-42), total tau, and mild cognitive impairment. Neurology 2007; 68: 1501–8.
phosphorylated tau (Thr181) in cerebrospinal fluid by the 109 DeKosky ST, Scheff SW. Synapse loss in frontal cortex
xMAP technology. Clin Chem 2005; 51: 336–45. biopsies in Alzheimer’s disease: correlation with cognitive
93 Mattsson N, Andreasson U, Persson S et al. The Alzheimer’s severity. Ann Neurol 1990; 27: 457–64.
association external quality control program for cere- 110 Blennow K, Bogdanovic N, Alafuzoff I, Ekman R, Davidsson
brospinal fluid biomarkers. Alzheimers Dement 2011; 7: P. Synaptic pathology in Alzheimer’s disease: relation to
386–95. e6. severity of dementia, but not to senile plaques, neurofibril-
94 Bjerke M, Portelius E, Minthon L et al. Confounding factors lary tangles, or the ApoE4 allele. J Neural Transm 1996;
influencing amyloid Beta concentration in cerebrospinal 103: 603–18.
fluid. Int J Alzheimers Dis 2010; 2010: pii: 986310. 111 Sze CI, Troncoso JC, Kawas C, Mouton P, Price DL, Martin
95 Andreasen N, Minthon L, Davidsson P et al. Evaluation of LJ. Loss of the presynaptic vesicle protein synaptophysin in
CSF-tau and CSF-Abeta42 as diagnostic markers for Alzhei- hippocampus correlates with cognitive decline in Alzheimer
mer disease in clinical practice. Arch Neurol 2001; 58: 373– disease. J Neuropathol Exp Neurol 1997; 56: 933–44.
9. 112 Terry RD, Masliah E, Salmon DP et al. Physical basis of
96 Mattsson N, Andreasson U, Persson S et al. CSF biomarker cognitive alterations in Alzheimer’s disease: synapse loss is
variability in the Alzheimer’s Association quality control the major correlate of cognitive impairment. Ann Neurol
program. Alzheimers Dement 2013; 9: 251–61. 1991; 30: 572–80.
97 Kuhlmann J, Andreasson U, Pannee J et al. CSF Abeta1-42 - 113 Bittner T, Burgold S, Dorostkar MM et al. Amyloid plaque
an excellent but complicated Alzheimer’s biomarker - a route formation precedes dendritic spine loss. Acta Neuropathol
to standardisation. Clin Chim Acta 2017; 467: 27–33. 2012; 124: 797–807.
98 Carrillo MC, Blennow K, Soares H et al. Global standardiza- 114 Shankar GM, Bloodgood BL, Townsend M, Walsh DM,
tion measurement of cerebral spinal fluid for Alzheimer’s Selkoe DJ, Sabatini BL. Natural oligomers of the Alzheimer
disease: an update from the Alzheimer’s Association Global amyloid-beta protein induce reversible synapse loss by
Biomarkers Consortium. Alzheimers Dement 2013; 9: 137– modulating an NMDA-type glutamate receptor-dependent
40. signaling pathway. J Neurosci 2007; 27: 2866–75.
99 Korecka M, Waligorska T, Figurski M et al. Qualification of a 115 Dickstein DL, Brautigam H, Stockton SD Jr, Schmeidler J,
surrogate matrix-based absolute quantification method for Hof PR. Changes in dendritic complexity and spine mor-
amyloid-beta(4)(2) in human cerebrospinal fluid using 2D phology in transgenic mice expressing human wild-type tau.
UPLC-tandem mass spectrometry. J Alzheimers Dis 2014; Brain Struct Funct 2010; 214: 161–79.
41: 441–51. 116 Bellucci A, Westwood AJ, Ingram E, Casamenti F, Goedert
100 Leinenbach A, Pannee J, Dulffer T et al. Mass spectrometry- M, Spillantini MG. Induction of inflammatory mediators and
based candidate reference measurement procedure for microglial activation in mice transgenic for mutant human
quantification of amyloid-beta in cerebrospinal fluid. Clin P301S tau protein. Am J Pathol 2004; 165: 1643–52.
Chem 2014; 60: 987–94. 117 Davidsson P, Jahn R, Bergquist J, Ekman R, Blennow K.
101 Pannee J, Portelius E, Oppermann M et al. A selected Synaptotagmin, a synaptic vesicle protein, is present in
reaction monitoring (SRM)-based method for absolute quan- human cerebrospinal fluid: a new biochemical marker for
tification of Abeta38, Abeta40, and Abeta42 in cerebrospinal synaptic pathology in Alzheimer disease? Mol Chem Neu-
fluid of Alzheimer’s disease patients and healthy controls. J ropathol 1996; 27: 195–210.
Alzheimers Dis 2013; 33: 1021–32. 118 Davidsson P, Puchades M, Blennow K. Identification of
102 Armbruster D, Miller RR. The Joint Committee for Trace- synaptic vesicle, pre- and postsynaptic proteins in human
ability in Laboratory Medicine (JCTLM): a global approach to cerebrospinal fluid using liquid-phase isoelectric focusing.
promote the standardisation of clinical laboratory test Electrophoresis 1999; 20: 431–7.
results. Clin Biochem Rev 2007; 28: 105–13. 119 Thorsell A, Bjerke M, Gobom J et al. Neurogranin in
103 Bjerke M, Andreasson U, Kuhlmann J et al. Assessing the cerebrospinal fluid as a marker of synaptic degeneration in
commutability of reference material formats for the harmo- Alzheimer’s disease. Brain Res 2010; 1362: 13–22.
nization of amyloid-beta measurements. Clin Chem Lab Med 120 Represa A, Deloulme JC, Sensenbrenner M, Ben-Ari Y,
2016; 54: 1177–91. Baudier J. Neurogranin: immunocytochemical localization
of a brain-specific protein kinase C substrate. J Neurosci the way for mechanistic understanding. Nat Med 2015; 21:
1990; 10: 3782–92. 217–9.
121 Guadano-Ferraz A, Vinuela A, Oeding G, Bernal J, Rausell 138 O’Bryant SE, Gupta V, Henriksen K et al. Guidelines for the
E. RC3/neurogranin is expressed in pyramidal neurons of standardization of preanalytic variables for blood-based
motor and somatosensory cortex in normal and denervated biomarker studies in Alzheimer’s disease research. Alzhei-
monkeys. J Comp Neurol 2005; 493: 554–70. mers Dement 2015; 11: 549–60.
122 Huang KP, Huang FL, Jager T, Li J, Reymann KG, Balschun 139 Andreasson U, Blennow K, Zetterberg H. Update on ultra-
D. Neurogranin/RC3 enhances long-term potentiation and sensitive technologies to facilitate research on blood
learning by promoting calcium-mediated signaling. J Neu- biomarkers for central nervous system disorders. Alzhei-
rosci 2004; 24: 10660–9. mers Dement (Amst) 2016; 3: 98–102.
123 Wu J, Li J, Huang KP, Huang FL. Attenuation of protein 140 Hansson O, Zetterberg H, Vanmechelen E et al. Evaluation
kinase C and cAMP-dependent protein kinase signal trans- of plasma Abeta(40) and Abeta(42) as predictors of conver-
duction in the neurogranin knockout mouse. J Biol Chem sion to Alzheimer’s disease in patients with mild cognitive
2002; 277: 19498–505. impairment. Neurobiol Aging 2010; 31: 357–67.
124 Davidsson P, Blennow K. Neurochemical dissection of 141 Kuo YM, Emmerling MR, Lampert HC et al. High levels of
synaptic pathology in Alzheimer’s disease. Int Psychogeriatr circulating Abeta42 are sequestered by plasma proteins in
1998; 10: 11–23. Alzheimer’s disease. Biochem Biophys Res Commun 1999;
125 Reddy PH, Mani G, Park BS et al. Differential loss of 257: 787–91.
synaptic proteins in Alzheimer’s disease: implications for 142 Zetterberg H, Mortberg E, Song L et al. Hypoxia due to
synaptic dysfunction. J Alzheimer’s Dis. 2005; 7: 103–17; cardiac arrest induces a time-dependent increase in serum
discussion 73-80. amyloid beta levels in humans. PLoS ONE 2011; 6: e28263.
126 Kvartsberg H, Duits FH, Ingelsson M et al. Cerebrospinal 143 Janelidze S, Stomrud E, Palmqvist S et al. Plasma beta-
fluid levels of the synaptic protein neurogranin correlates amyloid in Alzheimer’s disease and vascular disease. Sci Rep
with cognitive decline in prodromal Alzheimer’s disease. 2016; 6: 26801.
Alzheimers Dement 2015; 11: 1180–90. 144 Pannee J, Tornqvist U, Westerlund A et al. The amyloid-
127 Hellwig K, Kvartsberg H, Portelius E et al. Neurogranin and beta degradation pattern in plasma–a possible tool for
YKL-40: independent markers of synaptic degeneration and clinical trials in Alzheimer’s disease. Neurosci Lett 2014;
neuroinflammation in Alzheimer’s disease. Alzheimers Res 573: 7–12.
Ther 2015; 7: 74. 145 Ovod V, Ramsey KN, Mawuenyega KG et al. Amyloid beta
128 Kvartsberg H, Portelius E, Andreasson U et al. Characteri- concentrations and stable isotope labeling kinetics of
zation of the postsynaptic protein neurogranin in paired human plasma specific to central nervous system amyloi-
cerebrospinal fluid and plasma samples from Alzheimer’s dosis. Alzheimers Dement 2017; 13: 841–9.
disease patients and healthy controls. Alzheimers Res Ther 146 Kaneko N, Nakamura A, Washimi Y et al. Novel plasma
2015; 7: 40. biomarker surrogating cerebral amyloid deposition. Proc Jpn
129 Portelius E, Zetterberg H, Skillback T et al. Cerebrospinal Acad Ser B Phys Biol Sci 2014; 90: 353–64.
fluid neurogranin: relation to cognition and neurodegener- 147 Nakamura A, Kaneko N, Villemagne VL et al. High perfor-
ation in Alzheimer’s disease. Brain 2015; 138: 3373–85. mance plasma amyloid-beta biomarkers for Alzheimer’s
130 Kester MI, Teunissen CE, Crimmins DL et al. Neurogranin, a disease. Nature 2018; 554: 249–54.
CSF biomarker for synaptic loss, predicts decline to symp- 148 Tzen KY, Yang SY, Chen TF et al. Plasma Abeta but not tau
tomatic Alzheimer disease. JAMA Neurol 2015; 72: 1275– is related to brain PiB retention in early Alzheimer’s disease.
80. ACS Chem Neurosci 2014; 5: 830–6.
131 De Vos A, Jacobs D, Struyfs H et al. C-terminal neurogranin 149 Zetterberg H, Wilson D, Andreasson U et al. Plasma tau
is increased in cerebrospinal fluid but unchanged in plasma levels in Alzheimer’s disease. Alzheimers Res Ther 2013;
in Alzheimer’s disease. Alzheimers Dement 2015; 11: 1461–9. 5: 9.
132 Wellington H, Paterson RW, Portelius E et al. Increased CSF 150 Mattsson N, Zetterberg H, Janelidze S et al. Plasma tau in
neurogranin concentration is specific to Alzheimer disease. Alzheimer disease. Neurology 2016; 87: 1827–35.
Neurology 2016; 86: 829–35. 151 Meredith JE Jr, Sankaranarayanan S, Guss V et al. Char-
133 Sutton RB, Fasshauer D, Jahn R, Brunger AT. Crystal acterization of novel CSF Tau and ptau biomarkers for
structure of a SNARE complex involved in synaptic exocy- Alzheimer’s disease. PLoS ONE 2013; 8: e76523.
tosis at 2.4 A resolution. Nature 1998; 395: 347–53. 152 Fiandaca MS, Kapogiannis D, Mapstone M et al. Identifica-
134 Geppert M, Goda Y, Hammer RE et al. Synaptotagmin I: a tion of preclinical Alzheimer’s disease by a profile of
major Ca2 + sensor for transmitter release at a central pathogenic proteins in neurally derived blood exosomes: a
synapse. Cell 1994; 79: 717–27. case-control study. Alzheimers Dement 2015; 11: 600–7. e1.
135 Brinkmalm A, Brinkmalm G, Honer WG et al. SNAP-25 is a 153 Gisslen M, Price RW, Andreasson U et al. Plasma concen-
promising novel cerebrospinal fluid biomarker for synapse tration of the neurofilament light protein (NFL) is a
degeneration in Alzheimer’s disease. Mol Neurodegener biomarker of CNS injury in HIV infection: a cross-sectional
2014; 9: 53. study. EBioMedicine 2016; 3: 135–40.
136 Ohrfelt A, Brinkmalm A, Dumurgier J et al. The pre- 154 Kuhle J, Barro C, Andreasson U et al. Comparison of three
synaptic vesicle protein synaptotagmin is a novel biomarker analytical platforms for quantification of the neurofilament
for Alzheimer’s disease. Alzheimers Res Ther 2016; 8: 41. light chain in blood samples: ELISA, electrochemilumines-
137 Blennow K, Zetterberg H. Understanding biomarkers of cence immunoassay and Simoa. Clin Chem Lab Med 2016;
neurodegeneration: ultrasensitive detection techniques pave 54: 1655–61.
155 Mattsson N, Andreasson U, Zetterberg H, Blennow K, 158 Rojas JC, Karydas A, Bang J et al. Plasma neurofilament
Alzheimer’s Disease Neuroimaging I. Association of plasma light chain predicts progression in progressive supranuclear
neurofilament light with neurodegeneration in patients with palsy. Ann Clin Transl Neurol 2016; 3: 216–25.
Alzheimer Disease. JAMA Neurol. 2017; 74: 557–66.
156 Weston PSJ, Poole T, Ryan NS et al. Serum neurofilament Correspondence: Kaj Blennow MD, PhD, Clinical Neurochemistry
light in familial Alzheimer disease: a marker of early Laboratory, Institute of Neuroscience and Physiology, The
neurodegeneration. Neurology 2017; 89: 2167–75. Sahlgrenska Academy at University of Gothenburg, M€ olndal
157 Rohrer JD, Woollacott IO, Dick KM et al. Serum neurofila- Campus, Sahlgrenska University Hospital, SE-431 80 M€ olndal,
ment light chain protein is a measure of disease intensity in Sweden.
frontotemporal dementia. Neurology 2016; 87: 1329–36. (fax: + 43 31 3432426; e-mail: kaj.blennow@neuro.gu.se).

Journal of Internal Medicine - 2018 - Blennow - Biomarkers For Alzheimer S Disease Current Status and Prospects For The

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Journal of Internal Medicine - 2018 - Blennow - Biomarkers For Alzheimer S Disease Current Status and Prospects For The

Uploaded by

Copyright:

Available Formats

Review

Biomarkers for Alzheimer’s disease: current status and

Application Application Principle Setting Assets Interpretation

collection by lumbar puncture (for details, see

Application Fluid Biomarker Change in AD Stage of development Interpretation

Journal of Internal Medicine, 2018, 284; 643–663

P-tau is normal or mildly

Application Fluid Biomarker Change in AD Stage of development Interpretation

control levels Several research-grade assays neurodegeneration and is not

Journal of Internal Medicine, 2018, 284; 643–663

neurodegeneration, but is not

analysis showed very consistent findings across

The basis for the decrease in CSF Ab42 was

found an association between low postmortem

ventricular CSF Ab42 levels and high plaque

a correlation between reduced CSF Ab42 measured

CSF samples or by the latency between ante-

relation between these amyloid biomarkers. In

progressive supranuclear palsy; P-tau, phosphorylated tau; T-tau, total tau.

2006, a paper presented the findings that elderly

that the pathophysiological basis for the reduction

plaques in the AD brain when considering the

(preclinical and clinical) of AD of around 30 years

In agreement, several papers have consistently

found a high concordance between CSF Ab42 and

tional studies, a high concordance between CSF

1990 1995 2000 2005 2010 2015

Low CSF CSF biomarkers CSF High CSF

High T-tau/low Aβ42

CSF biomarkers included in the

Total tau (T-tau)

Aβ42 and Aβ42/40 ratio

recently published [37] and have performed two

synaptic unit (dendrites – neurogranin, presynap- epitope masking by hydrophobic Ab peptides

You might also like