Content uploaded by Johannes Kornhuber
Author content
All content in this area was uploaded by Johannes Kornhuber on Nov 28, 2014
Content may be subject to copyright.
DEMENTIAS - ORIGINAL ARTICLE
Neurochemical dementia diagnostics: a simple algorithm
for interpretation of the CSF biomarkers
Piotr Lewczuk ÆRu
¨diger Zimmermann Æ
Jens Wiltfang ÆJohannes Kornhuber
Received: 29 April 2009 / Accepted: 17 July 2009 / Published online: 4 August 2009
ÓSpringer-Verlag 2009
Abstract Cerebrospinal fluid (CSF)-based neurochemical
dementia diagnostics (NDD) is a well-established diag-
nostic tool for neurodegenerative disorders, including
Alzheimer’s disease (AD). However, the direct comparison
of the concentrations of the biomarkers between laborato-
ries is often very misleading, due to relatively high inter-
laboratory discrepancies of normal/abnormal ranges (cutoff
values). Therefore, an interpretation tool might be useful
for centers performing NDD to facilitate a standardized,
diagnostic-oriented reporting of the data on biomarkers. In
this study, we present a simple, easy-to-implement algo-
rithm allowing diagnostic-relevant categorization of
patients according to the outcome of the CSF NDD results
and, correspondingly, enabling reporting of the data to
clinicians in a clear and easy-to-follow form. The algo-
rithm is flexible and cutoff values independent, meaning
each laboratory can easily supplement it with the cutoff
values and normal/abnormal ranges according to the needs;
the only prerequisite is to perform the standard CSF NDD
assays (amyloid bpeptides and Tau/pTau).
Keywords Alzheimer’s disease Neurodegeneration
Cerebrospinal fluid Interpretation algorithm Biomarkers
Introduction
Since neurodegeneration, including Alzheimer’s disease
(AD), strongly affects populations of all industrialized
countries, it is obvious that more and more laboratories
worldwide establish methods and protocols to improve
AD diagnosis. Among accepted biomarkers, cerebrospinal
fluid (CSF) concentrations of amyloid bpeptides (Ab
peptides) and Tau protein(s) along with hyperphospho-
rylated forms of the latter have been proven to fulfill the
criteria for a valid diagnostic test (The Working Group
on: ‘‘Molecular and Biochemical Markers of Alzheimer’s
Disease’’ 1998). This is not surprising because these
molecules are directly involved in the pathologic events
of the disease, namely deposition of senile plaques and
formation of neurofibrillary tangles, respectively. CSF
concentration of Abpeptides ending at the amino acid
position 42 (Ab42) is consequently found to be decreased,
and Tau/phospho-Tau proteins are increased in AD
(Welge et al. 2009 and reviewed in Lewczuk and Wilt-
fang 2008).
Due to many reasons, among them relatively high inter-
laboratory imprecision of the measurements of biomarkers
(Lewczuk et al. 2006), laboratory-specific cutoff values
must be elaborated and used in each diagnostic center, even
when the same assays are performed. Discrepancies in the
protocols of collection and storage of the CSF samples
make the direct comparison of the results even more
complex. Therefore, we postulate that physicians ordering
laboratory analysis obtain not only ‘raw’ concentrations,
but also an integrated laboratory report including diag-
nostic-relevant interpretation of the biomarkers’ constel-
lation. To facilitate this task, we have elaborated an
interpretation algorithm that might be easily implemented
by other interested groups.
P. Lewczuk R. Zimmermann J. Kornhuber (&)
Department of Psychiatry and Psychotherapy,
Universita
¨tsklinikum Erlangen, Schwabachanlage 6,
91054 Erlangen, Germany
e-mail: Johannes.Kornhuber@uk-erlangen.de
P. Lewczuk
e-mail: Piotr.Lewczuk@uk-erlangen.de
J. Wiltfang
Department of Psychiatry and Psychotherapy,
University of Duisburg-Essen, Essen, Germany
123
J Neural Transm (2009) 116:1163–1167
DOI 10.1007/s00702-009-0277-y
Assays and the interpretation algorithm
Assays
Biomarkers in the CSF are analyzed according to the
protocols described elsewhere (Lewczuk et al. 2004a,b).
Briefly, we measure CSF concentrations of two forms of
Abpeptides ending at the C-terminus of 42: Ab1-42 (the
assay of Innogenetics, Ghent, Belgium, specific for both
the C- and N-termini) and Abx-42 (the assay of The
Genetics, Zu
¨rich, Switzerland, unspecific for the N-termi-
nus, but specific for the C-terminus), as well as Abx-40
(The Genetics), total Tau (Innogenetics) and Tau phos-
phorylated at the position 181 (pTau181, Innogenetics).
Apolipoprotein E (APOE) genotyping is performed with
the assay of Innogenetics.
Algorithm
The algorithm for interpreting the outcome of the analysis
of the biomarkers is presented in Fig. 1. Each analysis
starts with the score equal to 0. Depending on the con-
centration of the biomarkers, the score can increase and the
final sum (the lowest, 0 points; the highest, 4 points)
defines the categorization of a given patient into one of the
four diagnostic groups, which is eventually presented to the
physician on the CSF integrated report.
Interpretation of the amyloid bresults
CSF concentration of Abpeptides ending at the amino acid
position of 42 is decreased in AD and, as recently
observed, Ab42/40 concentration ratio (AbR) seems to
even better reflect AD pathology than Ab42 alone, espe-
cially in cases with extremely low or extremely high total
Abconcentrations (Wiltfang et al. 2007). In such cases, a
discrepancy can be observed between Ab1-42 and/or
Abx-42 ‘raw’ concentration and the Ab42/40 ratio, and this
requires a corresponding comment on the CSF report.
Nevertheless, our current opinion is that the decreased
Ab1-42 or Abx-42 cannot exclude Abpathology even if
the Ab42/40 ratio is normal. Certainly more experimental
work is required to properly weigh the role of Ab42/40 in
the interpretation of the Abpathology in AD.
END: SCORING
+0 +1a+2
Tau/pTau
Rapidly progressing
neurodegeneration (e.g. CJD)
+1
pTau normal or in border zone
+0
SCORE:
0: no evidence of organic CNS disease
1: improbably AD
2-3: possibly AD
4:
p
robabl
y
AD
all biomarkers of the Tau/pTau group are normal
Tau>1200pg/mLc
at least one exceeds border
zone
+0 +1a+2 +2
b
Aβ42 fits AβR
+1
at least one exceeds
border zone
Amyloid βPeptides
all biomarkers of the Aβ group are normal
Fig. 1 Interpretation of the concentrations of CSF biomarkers starts
with the score =0 and, depending on the results, can increase up to 4
points. With positive answers for the questions in the algorithm’s
nodes, the interpreter follows green arrows, with negative answers,
red ones. Finally, the score obtained at the end of the interpretation
determines the categorization of the patient into one of the groups
listed in the insert. In special cases, an additional comment is given: a
the results are in a border zone (marginally increased concentration of
Tau/pTau or decreased Ab42 concentration/Ab42/40 ratio) and
patients should be treated cautiously. This comment is usually given
when the results are altered by not more than the assay’s imprecision;
bdiscrepancy between Ab42 concentration and Ab42/40 concentra-
tion ratio (usually in case of very low or very high total Abpeptides
CSF load); cin case of very high Tau concentration, rapidly
progressing neurodegeneration must be taken into account (for
example Creutzfeldt-Jakob disease should be considered as the
differential diagnosis); the value presented here is the cutoff of our
laboratory and can be easily adjusted
c
1164 P. Lewczuk et al.
123
Interpretation of the Tau/pTau181 results
Increased Tau concentration in the CSF is quite a sensitive
biomarker of a neurodegenerative process, however,
unspecific for any given disorder, especially AD (Itoh et al.
2001). A very high CSF concentration of Tau points to
‘rapidly progressing neurodegeneration’, which means that
Creutzfeldt-Jakob disease (CJD) must be taken into con-
sideration as differential diagnosis (Otto et al. 2002). On
the other hand, even very high CSF Tau concentrations
cannot exclude AD, although it seems to be less probable
in such cases compared to those when Tau is moderately
increased. Therefore, although in cases with CSF Tau
concentration higher than 1,200 pg/mL (or any other lab-
oratory-specific cutoff value), interpretation is directed into
the block of ‘rapidly progressing neurodegeneration’, AD
cannot be completely excluded. pTau181 is more specific
for AD; however, a mild-to-moderate increase of the CSF
concentration of phosphorylated Tau protein can be
observed in rapidly progressing neurodegeneration, too
(Buerger et al. 2006), and hence AD cannot be definitely
confirmed even if the CSF pTau concentration is very high.
Border zones
Each laboratory analysis is characterized by a certain
margin of imprecision and, as a matter of fact, restrictive
application of mathematical rules on how to interpret a
given value in relation to a defined cutoff can bring about
serious misinterpretations. The presented algorithm deals
with this problem by knots ‘border zones’: their role is to
correctly interpret for example the results within the
assay’s imprecision. Neglecting such rules in the inter-
pretation process easily leads to overinterpretation of only
slightly increased Tau/pTau or only slightly decreased Ab
peptides/ratio: Tau increase by 1% or so is indeed
‘increased’, but such a result must not be reported without
a comment. The definition of border zones certainly
depends on the laboratory’s experience, assay performance
and so on; in our case, it is usually 5–10% of the cutoff.
Diagnosis-relevant categorization of patients:
AD and ‘rapidly progressing neurodegeneration’
Results of the CSF analysis with both groups of biomarkers
(Tau/pTau and Ab) in pathologic ranges are interpreted as
‘probable AD’ (4 points in the interpretation diagram).
Results of the CSF analysis with all biomarkers in normal
ranges are interpreted as ‘no evidence of organic CNS
disease’ (0 points). Results in-between, either with normal
Tau/pTau and abnormal Ab, or vice versa, with pathologic
Tau/pTau and normal Ab, are interpreted as ‘possible AD’
(2–3 points). The isolated very high concentration of Tau is
interpreted as ‘suspected rapidly progressing neurodegen-
eration, improbable AD’ (1 point), but this same concen-
tration of Tau accompanied by pathologic Ab
concentrations/ratio would shift the interpretation to ‘pos-
sible’ (2–3 points) or even ‘probable’ AD (4 points)
depending on whether pTau was normal or not,
respectively.
Validation of the algorithm
To validate the algorithm, we prepared a Microsoft Excel-
based macro executing the algorithm’s rules and compared
the interpretations of 100 reports performed by an expert
clinical neurochemist with the interpretations delivered by
the software. We found 85% of the diagnostic group cat-
egorizations identical, whereby the remaining 15% differed
slightly by one diagnostic category (for example ‘possibly
AD’ vs. ‘probably’ or ‘improbably AD’ vs. ‘no evidence of
organic CNS disorder’). In all five cases with very high Tau
concentration, both the software and the human expert
delivered interpretation of ‘suspected rapidly progressing
neurodegeneration’.
Discussion
Relatively high inter-laboratory imprecision of the con-
centrations of the NDD biomarkers (Lewczuk et al. 2006)
excludes simple copying and pasting of the cutoff values
from one center to another, and lack of any systematic
quality control program for NDD biomarkers makes the
situation even more difficult (Verwey et al. 2009). More-
over, similar to neurochemical diagnosis of virtually all
diseases of the nervous system, central and peripheral,
CSF-based AD diagnosis is more precise when more bio-
markers are taken into account. Currently, no single CSF
biomarker is pathognomonic for AD. This, in turn, calls for
interpretation algorithms that could be easily implemented
in clinical neurochemistry laboratories to enable compari-
son of the results among centers.
The reported algorithm is cutoff value-independent,
which means that each laboratory performing analyses
presented here (Tau/pTau and Ab) can, or actually even
should, establish its own cutoff values. For this reason, we
purposely do not present any concrete numbers, apart of the
Tau cutoff for rapid neurodegeneration, which, of course,
can be easily modified to fit a laboratory’s experience. We
feel that this is the most important difference compared to
the ‘dementia markers calculator’ distributed by Innoge-
netics, where the cutoff values for the assays of the com-
pany are fixed according to the experience we have in our
laboratory in Erlangen (Dr. M von Darl, personal infor-
mation). Such an approach has certain advantages too, for
Neurochemical dementia diagnostics 1165
123
example it is easy to use; however, we feel that the algo-
rithm with unfixed normal/abnormal ranges is easier to
implement if the cutoff values in a given center differ from
those suggested by the manufacturer. Similarly, it must be
stressed that the algorithm presented in this report is assay-
unspecific, which means that the assays of biomarkers of
the Aband Tau groups, as used currently in our laboratory,
can be replaced by other assays or even other analytical
platforms. The only prerequisite is that Abpeptide ending
at the amino acid position of 42 is measured along with its
ratio to Abending at the position of 40, Tau, and phos-
phorylated form of the latter.
Since no data currently exists to show whether one of
the biomarkers, or a group of biomarkers, has stronger
impact on the AD diagnosis than the others, we treat them
equally (equal number of points for altered Tau/pTau and
Ab). Nevertheless, if convincing evidences are published
showing that one of the groups of biomarkers is more
important that the other, in terms of specificity and sensi-
tivity, this algorithm can be easily modified by adjusting
the number of points given for this biomarker. Similarly,
for the current version of the interpretation algorithm, only
established and generally accepted routine NDD biomark-
ers were considered (CSF amyloid bpeptides, Tau and
pTau proteins). Currently, we do not consider these ‘can-
didate biomarkers’ that, although promising, still need
more studies to be validated in terms of diagnosing
patients. Good examples for these might be CSF soluble
amyloid precursor proteins (Lewczuk et al. 2008; Portelius
et al. 2009), decreased phospholipase A2 activity in the
CSF of AD patients (Smesny et al. 2008) or alterations in
the concentration of amyloid bpeptides in the blood
(Graff-Radford et al. 2007; Hansson et al. 2008).
Currently, discrepant results are presented in literature
regarding the question whether APOE genotype affects the
metabolism of AD biomarkers (Strittmatter et al. 1993),
which would mean that normal/abnormal ranges should be
defined considering the APOE status of a given patient
(Arai et al. 1995; Blomberg et al. 1996; Golombowski et al.
1997). Since it is currently unclear if APOE genotype
should influence the CSF biomarkers cutoffs, we currently
use one set of cutoffs for all possible APOE genotypes.
Nevertheless, the algorithm presented here can be easily
adjusted to meet the needs of different cutoffs for different
APOE genotypes, if it turns out to be necessary.
Validation of the algorithm with an MS Excel-based
macro showed not only high compatibility with the inter-
pretations done by a clinical neurochemist, dealing with
‘border zone’ cases perhaps even more objectively, but also
proved that this algorithm can easily be implemented as a
computer software to further simplify interpretation of
NDD results. In comparison with other diagnostic tools, the
sensitivity of the CSF NDD methods, including their
interpretation according to the currently suggested algo-
rithm, seems to be higher than, for example, neuroimaging
techniques (Weih et al. 2009), which further validates it as
a useful diagnostic support tool. Further verification pro-
cedures, especially comparison with postmortem analysis,
would certainly further validate the algorithm. However, it
must be treated carefully for two reasons: (a) neither Ab
plaques nor Tau/pTau tangles are pathognomonic for
Alzheimer’s disease, and (b) there is usually substantial
time difference between the diagnostic lumbar puncture
and the autopsy, and hence neuropathology does not nec-
essarily reflect the stage of the disease when the CSF
biomarkers were analyzed, usually years previously. Cur-
rently we are working on the integration of the software
into a laboratory database.
In conclusion, we present a flexible, easy-to-implement
algorithm to interpret Aband Tau/pTau CSF concentra-
tions in the context of neurochemical dementia diagnostics.
We hope that this algorithm would help to standardize
diagnosis-oriented interpretation of the CSF NDD
biomarkers.
Acknowledgments We thank all the technical coworkers of the
Laboratory for Clinical Neurochemistry and Neurochemical Demen-
tia Diagnostics at the Department of Psychiatry and Psychotherapy,
Erlangen, for years of performing NDD analysis with excellent
quality. A part of this work was presented at the 9th International
Conference on Alzheimer’s and Parkinson’s Diseases, Prague, 2009.
PL and RZ are supported by Fond fu
¨r Forschung und Lehre am
Klinikum Erlangen (ELAN; grant No.: 08.12.11.1). PL is a consultant
of Innogenetics.
References
Arai H, Terajima M, Miura M, Higuchi S, Muramatsu T, Machida N,
Seiki H, Takase S, Clark CM, Lee VM et al (1995) Tau in
cerebrospinal fluid: a potential diagnostic marker in Alzheimer’s
disease. Ann Neurol 38:649–652
Blomberg M, Jensen M, Basun H, Lannfelt L, Wahlund LO (1996)
Increasing cerebrospinal fluid tau levels in a subgroup of
Alzheimer patients with apolipoprotein E allele epsilon 4 during
14 months followup. Neurosci Lett 214:163–166
Buerger K, Otto M, Teipel SJ, Zinkowski R, Blennow K, DeBernardis
J, Kerkman D, Schroder J, Schonknecht P, Cepek L, McCulloch
C, Moller HJ, Wiltfang J, Kretzschmar H, Hampel H (2006)
Dissociation between CSF total tau and tau protein phosphor-
ylated at threonine 231 in Creutzfeldt-Jakob disease. Neurobiol
Aging 27:10–15
Golombowski S, Muller-Spahn F, Romig H, Mendla K, Hock C
(1997) Dependence of cerebrospinal fluid tau protein levels on
apolipoprotein E4 allele frequency in patients with Alzheimer’s
disease. Neurosci Lett 225:213–215
Graff-Radford NR, Crook JE, Lucas J, Boeve BF, Knopman DS,
Ivnik RJ, Smith GE, Younkin LH, Petersen RC, Younkin SG
(2007) Association of low plasma abeta42/abeta40 ratios with
increased imminent risk for mild cognitive impairment and
Alzheimer disease. Arch Neurol 64:354–362
Hansson O, Zetterberg H, Vanmechelen E, Vanderstichele H,
Andreasson U, Londos E, Wallin A, Minthon L, Blennow K
1166 P. Lewczuk et al.
123
(2008) Evaluation of plasma Abeta(40) and Abeta(42) as
predictors of conversion to Alzheimer’s disease in patients with
mild cognitive impairment. Neurobiol Aging
Itoh N, Arai H, Urakami K, Ishiguro K, Ohno H, Hampel H, Buerger
K, Wiltfang J, Otto M, Kretzschmar H, Moeller HJ, Imagawa M,
Kohno H, Nakashima K, Kuzuhara S, Sasaki H, Imahori K
(2001) Large-scale, multicenter study of cerebrospinal fluid tau
protein phosphorylated at serine 199 for the antemortem
diagnosis of Alzheimer’s disease. Ann Neurol 50:150–156
Lewczuk P, Esselmann H, Bibl M, Beck G, Maler JM, Otto M,
Kornhuber J, Wiltfang J (2004a) Tau protein phosphorylated at
threonine 181 in CSF as a neurochemical biomarker in
Alzheimer’s disease: original data and review of the literature.
J Mol Neurosci 23:115–122
Lewczuk P, Esselmann H, Otto M, Maler JM, Henkel AW, Henkel
MK, Eikenberg O, Antz C, Krause WR, Reulbach U, Kornhuber
J, Wiltfang J (2004b) Neurochemical diagnosis of Alzheimer’s
dementia by CSF Ab42, Ab42/Ab40 ratio and total tau.
Neurobiol Aging 25:273–281
Lewczuk P, Beck G, Ganslandt O, Esselmann H, Deisenhammer F,
Regeniter A, Petereit HF, Tumani H, Gerritzen A, Oschmann P,
Schroder J, Schonknecht P, Zimmermann K, Hampel H, Burger
K, Otto M, Haustein S, Herzog K, Dannenberg R, Wurster U,
Bibl M, Maler JM, Reubach U, Kornhuber J, Wiltfang J (2006)
International quality control survey of neurochemical dementia
diagnostics. Neurosci Lett 409:1–4
Lewczuk P, Kamrowski-Kruck H, Peters O, Heuser I, Jessen F, Popp
J, Burger K, Hampel H, Frolich L, Wolf S, Prinz B, Jahn H,
Luckhaus C, Perneczky R, Hull M, Schroder J, Kessler H, Pantel
J, Gertz H-J, Klafki H-W, Kolsch H, Reulbach U, Esselmann H,
Maler JM, Bibl M, Kornhuber J, Wiltfang J (2008) Soluble
amyloid precursor proteins in the cerebrospinal fluid as novel
potential biomarkers of Alzheimer’s disease: a multicenter study.
Mol Psychiatry (in press)
Lewczuk P, Wiltfang J (2008) Neurochemical dementia diagnostics:
state of the art and research perspectives. Proteomics 8:1292–1301
Otto M, Wiltfang J, Cepek L, Neumann M, Mollenhauer B, Steinacker
P, Ciesielczyk B, Schulz-Schaeffer W, Kretzschmar HA, Poser S
(2002) Tau protein and 14–3-3 protein in the differential diagnosis
of Creutzfeldt-Jakob disease. Neurology 58:192–197
Portelius E, Brinkmalm G, Tran AJ, Zetterberg H, Westman-
Brinkmalm A, Blennow K (2009) Identification of novel APP/
Abeta isoforms in human cerebrospinal fluid. Neurodegener Dis
6:87–94
Smesny S, Stein S, Willhardt I, Lasch J, Sauer H (2008) Decreased
phospholipase A2 activity in cerebrospinal fluid of patients with
dementia. J Neural Transm 115:1173–1179
Strittmatter WJ, Weisgraber KH, Huang DY, Dong LM, Salvesen GS,
Pericak-Vance M, Schmechel D, Saunders AM, Goldgaber D,
Roses AD (1993) Binding of human apolipoprotein E to
synthetic amyloid beta peptide: isoform-specific effects and
implications for late-onset Alzheimer disease. Proc Natl Acad
Sci USA 90:8098–8102
The Working Group on: ‘‘Molecular and Biochemical Markers of
Alzheimer’s Disease’’ (1998) Consensus report of the working
group on: ‘‘Molecular and Biochemical Markers of Alzheimer’s
Disease’’. The Ronald and Nancy Reagan Research Institute of
the Alzheimer’s Association and the National Institute on Aging
Working Group. Neurobiol Aging 19:109–116
Verwey NA, van der Flier WM, Blennow K, Clark C, Sokolow S, De
Deyn PP, Galasko D, Hampel H, Hartmann T, Kapaki E,
Lannfelt L, Mehta PD, Parnetti L, Petzold A, Pirttila T, Saleh L,
Skinningsrud A, Swieten JC, Verbeek MM, Wiltfang J, Younkin
S, Scheltens P, Blankenstein MA (2009) A worldwide multi-
centre comparison of assays for cerebrospinal fluid biomarkers in
Alzheimer’s disease. Ann Clin Biochem 46:235–240
Weih M, Krinninger M, Zimmermann R, Lewczuk P, Svitek J,
Schaller G, Degirmenci U, Richter-Schmidinger T, Wiltfang J,
Kuwert T, Kornhuber J, Schmidt D (2009) Sensitivity of
neurochemical dementia diagnostics in CSF compared to
99mTc-SPECT in Alzheimer’s dementia. Fortschr Neurol Psy-
chiatr 77:407–411
Welge V, Fiege O, Lewczuk P, Mollenhauer B, Esselmann H, Klafki
HW, Wolf S, Trenkwalder C, Otto M, Kornhuber J, Wiltfang J,
Bibl M (2009) Combined CSF tau, p-tau181 and amyloid-beta
38/40/42 for diagnosing Alzheimer’s disease. J Neural Transm
116:203–212
Wiltfang J, Esselmann H, Bibl M, Hull M, Hampel H, Kessler H, Frolich
L, Schroder J, Peters O, Jessen F, Luckhaus C, Perneczky R, Jahn H,
Fiszer M, Maler JM, Zimmermann R, Bruckmoser R, Kornhuber J,
Lewczuk P (2007) Amyloid beta peptide ratio 42/40 but not A beta
42 correlates with phospho-Tau in patients with low- and high-CSF
A beta 40 load. J Neurochem 101:1053–1059
Neurochemical dementia diagnostics 1167
123
