CSF soluble amyloid precursor proteins in
the diagnosis of incipient Alzheimer disease
R. Perneczky, MD
A. Tsolakidou, PhD
J. Diehl-Schmid, MD
T. Grimmer, MD
H. Fo ¨rstl, MD
A. Kurz, MD
P. Alexopoulos, MD
Objective: To explore if soluble amyloid precursor proteins (sAPP) in CSF improve the identifica-
tion of patients with incipient Alzheimer disease (AD) in a group of patients with mild cognitive
Methods: A cohort study with follow-up assessments of 58 patients with MCI with baseline CSF
sampling was conducted: 21 patients had progressed to probable AD (MCI-AD), 27 still had MCI,
8 had reverted to normal (MCI-NAD), and 2 patients with frontotemporal dementia (FTD) were
excluded. Sixteen additional patients with FTD were included to explore the specificity of the CSF
markers. CSF concentrations of sAPP?, sAPP?, tau, and A?1-42were measured with sensitive
and specific ELISAs. Associations between diagnostic status, CSF protein concentrations, and
other patient characteristics were explored using multiple logistic regression analyses with step-
wise variable selection. The optimal sensitivity and specificity of the best models were derived
from receiver operating characteristic curves.
Results: The MCI-AD group had significantly higher sAPP? concentrations than the MCI-NAD and
the FTD groups. A combination of sAPP?, tau, and age differentiated the MCI-AD and the MCI-
NAD groups with a sensitivity of 80.00% and a specificity of 81.00%. The best model for the
differentiation of the MCI-AD and the FTD groups included sAPP? and tau, and showed a sensitiv-
ity of 95.20% and a specificity of 81.20%. A?1-42and sAPP? did not significantly contribute to
Conclusion: These findings suggest that sAPP? may be clinically useful, and superior to A?1-42, in
the early and differential diagnosis of incipient AD. Neurology®2011;77:35–38
AD ? Alzheimer disease; AUC ? area under the curve; CDR ? Clinical Dementia Rating; FTD ? frontotemporal dementia; MCI ?
mild cognitive impairment; MCI-AD ? mild cognitive impairment progressed to probable Alzheimer disease; MMSE ? Mini-
Mental State Examination; NAD ? no Alzheimer disease; NPV ? negative predictive value; PPV ? positive predictive value;
sAPP ? soluble amyloid precursor protein; VIF ? variance-inflation factor.
Patients with mild cognitive impairment (MCI)1progress to clinically diagnosable Alzheimer
disease (AD) at a rate of up to 15% per year. Some patients, however, may never progress or
may even revert to normal.2Given this variable prognosis and in view of the development of
disease-modifying strategies, which will probably show the strongest impact if applied early,
biomarkers capable of identifying predementia AD are of great interest. The soluble amyloid
precursor proteins (sAPP) ? and ? mirror fundamental early events of AD pathogenesis3and
CSF concentration changes have been reported in AD.4We therefore explored whether sAPP?
and sAPP? improved the sensitivity and specificity of the detection of incipient AD in MCI
compared with the established biomarkers tau and A?1-42.5
METHODS Study sample. Seventy-five consecutive patients with MCI1(Clinical Dementia Rating,6CDR score of 0.5) with
CSF sampling for diagnostic purposes were contacted. Follow-up data were acquired for 58 patients. The remaining 17 were lost to
follow-up; the lost patients did not significantly differ from the analyzed group regarding CSF biomarker concentrations, age at
lumbar puncture, APOE ?4 allele carrier status, gender distribution, and Mini-Mental State Examination (MMSE)7scores. The
From the Department of Psychiatry and Psychotherapy, Technische Universita ¨t Mu ¨nchen, Klinikum rechts der Isar, Mu ¨nchen, Germany.
Study funding: Supported by the Bund der Freunde der Technischen Universita ¨t Mu ¨nchen e.V. (22592) and the Kommission fu ¨r Klinische Forschung
of Klinikum rechts der Isar Mu ¨nchen (B06-09).
Disclosure: Author disclosures are provided at the end of the article.
Address correspondence and
reprint requests to Dr. Robert
Perneczky, Department of
Psychiatry and Psychotherapy,
Technische Universita ¨t Mu ¨nchen,
Klinikum rechts der Isar, 81675
Mu ¨nchen, Germany
Copyright © 2011 by AAN Enterprises, Inc.
diagnoses were established by consensus of 2 experienced clini-
cians (R.P., P.A.) blind to the CSF data. A total of 21 patients
had progressed to probable AD according to National Institute
of Neurological and Communicative Disorders and Stroke–
Alzheimer’s Disease and Related Disorders Association criteria8
(MCI-AD), 27 still had MCI, and 8 had reverted to normal (no
AD group; MCI-NAD). Within the MCI-NAD group, patients
with MCI at follow-up did not significantly differ from patients
who had reverted to normal regarding CSF biomarker concen-
trations, age at lumbar puncture, APOE ?4 allele carrier status,
gender distribution, and baseline MMSE scores. Two patients
were diagnosed with frontotemporal dementia (FTD)9and were
excluded from the analyses. Sixteen additional patients with
FTD9were included in order to explore the specificity of the
markers for AD.
Standard protocol approvals, registrations, and patient
consent. The study was conducted according to the 1975 Dec-
laration of Helsinki; patients gave written informed consent.
Laboratory methods and statistical analysis. APOE
genotyping was performed using a standard PCR.10The APOE
genotype was dichotomized for the present study into individu-
als with one or 2 copies of the ?4 allele and those without any
copies of the ?4 allele. CSF concentrations of total tau, A?1-42,
sAPP?, and sAPP? were measured in duplicate with commercial
ELISAs according to the manufacturers’ instructions (tau/A?1-42:
Innogenetics, Gent, Belgium; sAPP?/sAPP?: IBL, Gunma,
The characteristics of the 3 patient subsamples were com-
pared using parametric tests for normally distributed data (one-
way analysis of variance with post hoc Bonferroni test, ?2test).
Associations between diagnostic status and other patient charac-
teristics were explored using multiple logistic regression analyses
with stepwise forward variable selection. The variance-inflation
factor (VIF) and tolerance were calculated in order to check for
multicollinearity in the dataset. Only sAPP? and sAPP? showed
relevant multicollinearity (VIF ?4 and tolerance ?0.3) as re-
ported before4; these 2 predictors were therefore tested in sepa-
rate statistical models. The optimal sensitivity and specificity of
the best statistical models were derived from receiver operating
characteristic curves. Positive and negative predictive values
(PPV, NPV) were also calculated.
RESULTS The CSF concentrations of tau and
A?1-42were in the expected range for this type of
sample.5The patient groups did not differ regarding
baseline MMSE scores, APOE ?4 allele carrier status,
educational background, gender distribution, and
length of the follow-up period (also true within the
MCI-NAD group for differences between patients
with stable MCI and subjects who had reverted
to normal). The MCI-AD group was older than
the MCI-NAD group (p ? 0.02, n ? 56) and the
sAPP? and tau concentrations were higher in the
MCI-AD compared with both the MCI-NAD
(sAPP?: p ? 0.04; tau: p ? 0.01, n ? 56) and the
FTD groups (sAPP?: p ? 0.001; tau: p ? 0.001;
n ? 37). The sAPP? concentration was higher in the
MCI-AD than in the FTD group (p ? 0.001, n ?
37; table, figures 1 and 2).
In the logistic regression analysis with diagnostic
status (MCI-AD vs MCI-NAD) as the dependent
variable and age, tau, A?1-42, and sAPP? concentra-
tion as the independent variables, age (? ? 0.12,
SE ? 0.05, p ? 0.01, n ? 56), tau (? ? 0.003,
SE ? 0.002, p ? 0.05, n ? 56), and sAPP? (? ?
0.002, SE ? 0.001, p ? 0.05, n ? 56) were signifi-
cant, whereas A?1-42(p ? 0.44, n ? 56) was not
significant. The results remained significant after re-
moving the subjects who had reverted to normal
from the MCI-NAD group (not shown). The sensi-
TableCharacteristics of the study sample
MCI-AD (n ? 21)MCI-NAD (n ? 35)FTD (n ? 16)
Age at lumbar puncture, y,
67.95 (8.81)61.91 (7.79) 63.63 (6.08)
Follow-up period, mo,
33.10 (21.87) 33.57 (19.55) NA
APOE ?4 carrier, yes: no
Tau, ng/L, mean (SD)a,b
542.10 (276.66)340.20 (203.77)200.44 (100.59)
A?1-42, ng/L, mean (SD)
622.95 (275.61) 789.91 (383.12)790.56 (186.30)
sAPP?, ng/mL, mean (SD)b
373.73 (141.27)298.26 (155.73) 187.05 (89.74)
sAPP?, ng/mL, mean (SD)a,b
1200.29 (452.40)931.88 (399.46) 630.32 (258.93)
Abbreviations: FTD ? frontotemporal dementia; MCI ? mild cognitive impairment;
MCI-AD ? mild cognitive impairment progressed to probable Alzheimer disease; MMSE ?
Mini-Mental State Examination; NA ? not applicable; NAD ? no Alzheimer disease; sAPP ?
soluble amyloid precursor protein.
aSignificant difference between the MCI-AD and MCI-NAD groups at p ? 0.05.
bSignificant difference between the MCI-AD and FTD groups at p ? 0.05.
Figure 1 Bar charts showing mean CSF soluble
amyloid precursor protein (sAPP) ?
and sAPP? concentrations in the 3
FTD ? frontotemporal dementia; MCI-AD ? mild cognitive
impairment progressed to probable Alzheimer disease;
MCI-NAD ? no Alzheimer disease group.
Neurology 77July 5, 2011
tivity and specificity of the best model were 81.00%
and 80.00%, respectively (area under the curve
[AUC] ? 0.79). The PPV was 70.08% and the NPV
In a second logistic regression analysis with diag-
nostic status (MCI-AD vs FTD) as the dependent
variable and independent variables as defined above,
tau (? ? 0.010, SE ? 0.005, p ? 0.03, n ? 37) and
sAPP? (? ? 0.004, SE ? 0.00, p ? 0.04, n ? 37)
were significant, whereas A?1-42(p ? 0.07, n ? 37)
and age (p ? 0.21, n ? 37) were not significant. The
sensitivity and specificity of the best model were
95.20% and 81.20%, respectively (AUC ? 0.92).
The PPV was 95.23% and the NPV 62.50%.
sAPP?, which was tested in independent regression
analyses due to its strong positive correlation with
sAPP?,4did not significantly contribute to the mod-
els (results not shown). Stepwise backward variable
selection did not significantly alter the results.
DISCUSSION Our study strongly supports sAPP?
as a biomarker of early AD and its differentiation
from other causes of dementia without A? pathol-
ogy. These findings suggest that sAPP? may be clin-
ically useful, and superior to A?1-42, in the
identification of patients with AD in the MCI stage
and in the differentiation of incipient AD from
FTD. sAPP? is a measure of the first critical step
leading to A? pathology3that may show its best diag-
nostic performance in combination with markers of
neuronal injury or neurodegeneration such as CSF
tau or MRI hippocampal atrophy. One possible ex-
planation for the inferior diagnostic accuracy of CSF
A?1-42is that it measures events far downstream
from the initial pathogenic steps of A? production.
Thus, it is influenced by several additional factors
such as proteolytic cleavage and clearance from brain.
Furthermore, decreased CSF levels of A?1-42are
probably just an indirect reflection of A? deposits in
cerebral plaques.11In addition, A?1-42in CSF only
partially mirrors the neurotoxic and synaptotoxic ef-
fects of A? production; a biomarker such as sAPP?
which is closer to the initiation of APP processing
may provide information more proximate to AD
pathogenesis, which may result in more accurate di-
Limitations of our study include patient recruit-
ment at a specialized memory clinic, which may re-
strict the generalization of the results to the general
population with incipient AD. Since no control
group was included, differences in CSF protein con-
centrations between the patient groups and cogni-
tively healthy subjects could not be determined.
However, a healthy control group was not needed to
explore the added value of sAPP for the identification
of incipient AD. An additional concern is the lack of
pathologic confirmation of AD and FTD. However,
the validity of present clinical diagnostic criteria
compared with autopsy has been reported to be very
good in study cohorts recruited at specialized centers,
but some of the variation in CSF protein concentra-
tions might still be due to comorbid other patholo-
gies. Finally, the recruitment of a modest number of
patients at a single academic center and the relatively
short follow-up period may have resulted in an un-
derestimation of the predictive value of sAPP?.
Therefore, a replication of our results in larger multi-
center studies is urgently needed. Further studies are
warranted to explore the clinical value of sAPP? for
the differentiation of AD from healthy aging and
from other neurodegenerative disorders, and to in-
vestigate its use as a marker for anti-amyloid treat-
ment response. A? imaging such as with [11C]
Pittsburgh compound B might prove particularly
useful in this regard since it provides an independent
method of assessing the A? burden in the brain,11
which may well be a correlate of sAPP turnover.
Dr. Perneczky participated in drafting/revising the manuscript, study con-
cept or design, analysis or interpretation of data, acquisition of data, sta-
tistical analysis, study supervision, and obtaining funding. Dr. Tsolakidou
participated in drafting/revising the manuscript, study concept or design,
Figure 2Scatterplot showing CSF soluble amyloid precursor protein (sAPP) ?
concentrations vs A?1-42concentrations in the 3 patient groups
FTD ? frontotemporal dementia; MCI-AD ? mild cognitive impairment progressed to prob-
able Alzheimer disease; NAD ? no Alzheimer disease.
Neurology 77July 5, 2011
analysis or interpretation of data, acquisition of data, and study super- Download full-text
vision. A. Arnold participated in analysis or interpretation of data,
contribution of vital reagents/tools/patients, and acquisition of data. Dr.
Diehl-Schmid participated in drafting/revising the manuscript and acqui-
sition of data. Dr. Grimmer participated in drafting/revising the manu-
script and acquisition of data. Dr. Fo ¨rstl participated in drafting/revising
the manuscript, study concept or design, and study supervision. Dr. Kurz
participated in drafting/revising the manuscript, study concept or design,
and analysis or interpretation of data. Dr. Alexopoulos participated in
drafting/revising the manuscript, study concept or design, analysis or in-
terpretation of data, acquisition of data, and statistical analysis.
The authors thank Dorottya Ruisz for proofreading and Tamara Eisele for
Dr. Perneczky serves on the editorial boards of the Journal of Alzheimer’s
Disease, Open Journal of Nuclear Medicine, and Open Longevity Science and
has received speaker honoraria from Janssen. Dr. Tsolakidou, A. Arnold,
and Dr. Diehl-Schmid report no disclosures. Dr. Grimmer serves on a
scientific advisory board for Bristol-Myers Squibb. Dr. Fo ¨rstl reports no
disclosures. Dr. Kurz has received speaker honoraria from Eisai Inc. and
receives research support from the German Federal Ministry of Health.
Dr. Alexopoulos receives research support from Komission fuer Klinische
Forschung des Klinikums rechts der Isar der TU Muenchen and Bund der
Freunde der TU.
1. Winblad B, Palmer K, Kivipelto M, et al. Mild cognitive
impairment: beyond controversies, towards a consensus:
report of the International Working Group on Mild Cog-
nitive Impairment. J Intern Med 2004;256:240–246.
2. DeCarli C. Mild cognitive impairment: prevalence, prog-
nosis, aetiology, and treatment. Lancet Neurol 2003;2:
Hardy J, Selkoe DJ. The amyloid hypothesis of Alzhei-
mer’s disease: progress and problems on the road to thera-
peutics. Science 2002;297:353–356.
Lewczuk P, Kamrowski-Kruck H, Peters O, et al. Soluble
amyloid precursor proteins in the cerebrospinal fluid as
novel potential biomarkers of Alzheimer’s disease: a multi-
center study. Mol Psychiatry 2010;15:138–145.
Hulstaert F, Blennow K, Ivanoiu A, et al. Improved dis-
crimination of AD patients using beta-amyloid(1-42) and
tau levels in CSF. Neurology 1999;52:1555–1562.
Morris JC. The Clinical Dementia Rating (CDR): current
version and scoring rules. Neurology 1993;43:2412–2414.
Folstein MF, Folstein SE, McHugh PR. “Mini-mental
state”: a practical method for grading the cognitive state of
patients for the clinician. J Psychiatr Res 1975;12:189–
McKhann G, Drachman D, Folstein M, Katzman R, Price
D, Stadlan EM. Clinical diagnosis of Alzheimer’s disease:
report of the NINCDS-ADRDA Work Group under the
auspices of Department of Health and Human Services
Task Force on Alzheimer’s Disease. Neurology 1984;34:
Neary D, Snowden JS, Gustafson L, et al. Frontotemporal
lobar degeneration: a consensus on clinical diagnostic cri-
teria. Neurology 1998;51:1546–1554.
Wenham PR, Price WH, Blandell G. Apolipoprotein E
genotyping by one-stage PCR. Lancet 1991;337:1158–
Grimmer T, Riemenschneider M, Forstl H, et al. Beta am-
yloid in Alzheimer’s disease: increased deposition in brain
is reflected in reduced concentration in cerebrospinal fluid.
Biol Psychiatry 2009;65:927–934.
Neurology 77July 5, 2011