Alzheimer’s & Dementia: Diagnosis, Assessment & Disease Monitoring 8 (2017) 131-138Diagnostic Assessment & Prognosis
Early diagnosis of mild cognitive impairment and Alzheimer’s disease
based on salivary lactoferrinEva Carroa,b,*, Fernando Bartolom
ea,b, F
elix Bermejo-Parejaa,c,d, Alberto Villarejo-Galendea,c,d,
Jos
e Antonio Molinaa,c,d,e, Pablo Ortizf,g, Miguel Caleroa,h,i, Alberto Rabanoh,
Jos
e Luis Canteroa,j, Gorka Orivek,l,**
aNetworked Biomedical Research Center in Neurodegenerative Diseases (CIBERNED), Spain
bGroup of Neurodegenerative Diseases, Hospital 12 de Octubre Research Institute (imas12), Madrid, Spain
cNeurology Service Hospital Universitario 12 de Octubre, Madrid, Spain
dNeurodegenerative Disorders Group, Hospital 12 de Octubre Research Institute (imas12), Madrid, Spain
eDepartment of Medicine, Faculty of Medicine, Universidad Complutense de Madrid, Madrid, Spain
fDermatology Service Hospital Universitario 12 de Octubre, Madrid, Spain
gHospital 12 de Octubre Research Institute (imas12), Madrid, Spain
hAlzheimer Disease Research Unit, CIEN Foundation, Queen Sofia Foundation Alzheimer Center, Madrid, Spain
iChronic Disease Programme, Carlos III Institute of Health, Majadahonda, Madrid, Spain
jLaboratory of Functional Neuroscience, Pablo de Olavide University, Seville, Spain
kLaboratory of Pharmacy and Pharmaceutical Technology, Faculty of Pharmacy, University of the Basque Country, Vitoria, Spain
lNetworked Biomedical Research Center in Bioengineering, Biomaterials and Nanomedicine (CIBER-BBN), SpainAbstract Introduction: The Alzheimer’s disease (AD) process is likely initiated many years before clinicalThe present public
tained in the biomedi
nostics S.L. entity and
under the title “search
and other neurodege
The research was car
specifications provide
belong to Geroa Diagn
http://dx.doi.org/10.10
2352-8729/
 2017 T
license (http://creativeonset. Biomarkers of preclinical disease are critical for the development of disease-modifying or
even preventative therapies. Current biomarkers for early disease, including cerebrospinal fluid tau
and amyloid b (Ab) levels, structural and functional magnetic resonance imaging, and the use of brain
amyloid imaging, are limited because they are very invasive or expensive. Noninvasive biomarkers
may be a more accessible alternative, but none can currently detect preclinical AD with the required
sensitivity and specificity.
Methods: Here, we show a novel, straight-forward, and noninvasive approach for assessment of
early stages of cognitive decline. Salivary samples from cases of amnestic mild cognitive impairment
(aMCI) and AD, and neurology controls were analyzed.
Results: We have discovered and validated a new single saliva biomarker, lactoferrin, which in our
cross-sectional investigation perfectly discriminates clinically diagnosed aMCI and AD patients
from a cognitively healthy control group. The accuracy for AD diagnosis shown by salivary lac-
toferrin was greater than that obtained from core cerebrospinal fluid (CSF) biomarkers, including
total tau and CSF Ab42. Furthermore, salivary lactoferrin can be used for population screening and
for identifying those underdiagnosed subjects with very early stages of mild cognitive impairment
and AD.ation has been prepared on the basis of the results ob-
cal research project signed between the Geroa Diag-
the Hospital 12 de Octubre Biomedical Foundation
of new salivary biomarkers in Alzheimer’s disease
nerative diseases: possible diagnostic application.”
ried out in accordance with the scientific-technical
d by Geroa Diagnostics S.L. The results of this project
ostics S.L., who holds the exclusive ownership of the
intellectual property rights that protect them, with the exception of the ones
inherent to the author/inventor. This publication has been executed with the
prior authorization of the rights titleholder Geroa Diagnostics S.L.
*Corresponding author. Tel.: 134-913908765; Fax: 134-913908544.
**Corresponding author. Tel.: 134-663027696; Fax: 134-945013040.
E-mail address: carroeva@h12o.es (E.C.), gorka.orive@ehu.eus (G.O.)
16/j.dadm.2017.04.002
he Authors. Published by Elsevier Inc. on behalf of the Alzheimer’s Association. This is an open access article under the CC BY-NC-ND
commons.org/licenses/by-nc-nd/4.0/).
Table 1
Demographic and clin
Variable
n (F/M)
Age (years)
MMSE score
CDR score
APOE ε4 carriers
Education
Can read and write
Primary studies
Secondary studies
Abbreviations: aMC
MMSE, Mini–Mental
NOTE. Data are ex
E. Carro et al. / Alzheimer’s & Dementia: Diagnosis, Assessment & Disease Monitoring 8 (2017) 131-138132Conclusion: This biomarker may offer new insights in the early diagnostics for AD.

 2017 The Authors. Published by Elsevier Inc. on behalf of the Alzheimer’s Association. This is an
open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/
4.0/).Keywords: Noninvasive biomarkers; Saliva; Lactoferrin; Alzheimer’s disease; Mild cognitive impairment; Dementia;Diagnosis1. Background
Alzheimer’s disease (AD) is the most common neurode-
generative disorder of the aging population, and because of
the increase in longevity, the prevalence of AD is expected to
raise dramatically. Because AD process is probably initiated
many years before the clinical onset [1], biomarkers of pre-
clinical disease are critical for the development of disease-
modifying or even preventative therapies [2]. Unfortunately,
current biomarkers for early disease, including cerebrospinal
fluid tau and amyloid b (Ab) levels [3], structural and func-
tional magnetic resonance imaging [4], and the use of brain
amyloid imaging or inflammaging [5,6], are limited because
they are either invasive, time consuming, or expensive.
Therefore, detecting AD at the earliest possible stage is
vital to enable trials of disease modification agents and
considerable efforts are being invested in the identification.
Saliva testing is currently used in areas of toxicology,
endocrinology, infectious diseases, and forensics, with es-
tablished diagnostic tests available for alcohol detection,
HIV infections, hormonal analyses, and drug testing.
Because saliva collection is noninvasive and relatively stress
free, saliva can serve as a potential alternative and universal
diagnostic fluid. Identification of Ab and tau [7,8], or a-Syn
and DJ-1 [9] in human saliva, proteins that are critically
involved in AD and Parkinson’s disease (PD), respectively,
support the potential diagnostic value of saliva for neurode-
generative diseases.
A history of systemic infection is a known risk factor for
AD [10–12]. Brain infections with bacteria or viruses are
implicated in AD pathogenesis [13], but the impact of
antimicrobial peptides on disease outcomes has not beenical characteristics of subjects from first training study
Control aMCI
91 (59/32) 44 (25/19)
73.7 6 6.88 75.16 6 5.13
29 6 0.8 26.8 6 1.16***
0 0.5
12.9% 42.1%**
22.2% 36%
33.3% 36%
44.4% 28%*
I, amnestic mild cognitive impairment; AD, Alzheimer’s
State Examination; NA, not applicable; CDR, Clinical Dem
pressed as mean 6 SD. *P , .05 versus control group; **Psufficiently explored. Saliva is one of the body’s first lines
of defense due to its composition of antimicrobial proteins.
Lactoferrin, one of the major antimicrobial peptides in
saliva, represents an important defensive element by
inducing a broad spectrum of antimicrobial effects against
bacteria, fungi, protozoa, viruses, and yeasts [14–17],
through its ability to decrease bacterial growth, biofilm
development, iron overload, reactive oxygen formation,
and regulating the inflammatory response [18,19].
The primary aim of our study was to investigate the po-
tential of an AD diagnostic biomarker in saliva. We first car-
ried out an AD diagnostic cross-sectional study and enrolled
274 participants at the Neurology Service at the Hospital
Universitario 12 de Octubre (Madrid, Spain). We defined
four groups of subjects according to their cognitive status:
amnestic mild cognitive impairment (aMCI), AD, PD, and
cognitively healthy control group. We discovered in this first
diagnostic training study that saliva lactoferrin, an iron- but
also Ab-binding [20,21] glycoprotein, was strongly
correlated with AD. We secondly validated the saliva
lactoferrin as AD biomarker in two new blinded and
independent cohorts. Finally, salivary levels of lactoferrin
were examined in two independent longitudinal cohorts
composed of healthy nondemented individuals.2. Methods
2.1. Subjects and clinical classification
For the cross-sectional study, we included four groups of
donors in the training study: (n5 80) AD patients; (n5 44)
aMCI patients; (n 5 59) PD patients; and (n 5 91) elderlyAD PD P value
80 (49/31) 59 (32/27) ns
76.2 6 5.33** 69.5 6 8.6** ,.01
19.25 6 1.76*** NA ,.001

1 NA
45.9%** NA ,.01
38.5% NA
33.3%
28.2%* ,.05
disease; PD, Parkinson’s disease; F, female; M, male; ns, not significant;
entia Rating.
, .01 versus control group; ***P , .001 versus control group.
E. Carro et al. / Alzheimer’s & Dementia: Diagnosis, Assessment & Disease Monitoring 8 (2017) 131-138 133nondemented controls, recruited from Neurology service of
the University Hospital 12 de Octubre (Madrid, Spain)
(Table 1). In the validation study, subjects from two indepen-
dent entities, Pablo de Olavide University from Sevilla,
Spain, and Alzheimer Disease Research Unit, CIEN Foun-
dation, Queen Sofia Foundation Alzheimer Center (Madrid,
Spain), were included and were divided into AD patients
(n 5 36), mild cognitive impairment (MCI) patients
(n 5 15), and elderly nondemented controls (n 5 40)
(Supplementary Table 1).
The AD diagnosis was established according to the Na-
tional Institute on Neurological Disorders and Stroke, and
the Alzheimer’s Disease and Related Disorders Association
guidelines [22]. MCI was diagnosed in patients with cogni-
tive impairment according to the Petersen criteria [23]. Dis-
ease severity was evaluated using Mini–Mental State
Examination (MMSE) scores. MCI patients had an MMSE
of 26.8 6 0.8; and AD patients had a MMSE of
19.25 6 0.7. PD patients were diagnosed under the criteria
of probable PD [24,25]. Inclusion criteria for cognitively
normal older individual subjects were MMSE scores of
29 6 0.8, no history or clinical signs of neurological or
psychiatric disease or cognitive symptoms, as shown in
Supplementary Table 2. Subjects’ consent was obtained ac-
cording to the Declaration of Helsinki, and approval was ob-
tained from the Research Ethic Committee of each entity.
Written informed consent was obtained from all participants
or representatives.
For a 5-year longitudinal study, two independent and
blinded cohorts were included in this study. In the first inde-
pendent cohort, 116 healthy volunteers were enrolled from
Neurology Service at the University Hospital 12 de Octubre.
In a second independent cohort, 190 neurologically healthy
individuals selected from sample donors at the Biobanco
imas12 of the University Hospital 12 de Octubre were
included as a confirmatory group. At the end of the year 5
of the study, some of these individuals met criteria for
aMCI/AD.
Whole saliva was collected from patients with aMCI, AD,
as well as healthy control subjects including samples from
dementias other than AD (Table 1). Unstimulated whole
saliva was collected into sterile plastic containers precoated
with 2% sodium azide solution, as described previously [7].
Collected samples were immediately placed on ice and pre-
cleared by a low spin at 600! g for 10 minutes at 4C. Ali-
quoted 0.5-mL samples were stored at 280C after
treatment with Protease Inhibitor Cocktail (Roche). Protein
estimation was analyzed using a BCA protein assay kit
(Pierce, Rockford, IL, USA) according to the manufacturer’s
instructions.
Oral mucosa was collected into sterile plastic containers
according to the study by Aagaard et al. [26]. Briefly, 190
participants (M 5 110/F 5 80; average age 5 62 6 1.23)
drooled into a 50-mL collection tube after allowing saliva
to collect in the mouth for 
1 minute, centrifuged at
6000! g for 10 minutes at 4C, and pellets were stored at280C. Oral mucosa samples and united data from patients
included in this study were provided by the Biobanco imas12
in the Hospital 12 de Octubre integrated in the Spanish Hos-
pital Biobanks Network (RetBioH; www.redbiobancos.es)
following standard operation procedures with appropriated
approval of the Ethical and Scientific Committees.
Cerebrospinal fluid (CSF) samples (10 mL per subject)
were obtained by lumbar puncture on informed patients
consent. All samples were spun at 3000 rpm at 4C for 10mi-
nutes to remove any cells and debris, aliquoted in small vol-
umes, and stored in low bind polypropylene tubes at280C.
2.2. Genotyping
Genomic DNAwas extracted from peripheral blood using
Illustra blood genomic Prep Mini Spin Kit (GE Healthcare).
Apolipoprotein E (APOE) genotyping (2 3/3 3/4 3isoforms)
was performed using FRET probes.
2.3. Mass spectrometry
Saliva samples from four male subjects from each group
(MCI, AD, and elderly nondemented controls) were pooled
by mixing equal amounts. Fifty micrograms of each pool
were loaded on a SDS-PAGE gel. ImageQuant software
(GE Healthcare) was used for quantity determination.
Protein bands matching the corresponding molecular
weight were excised from the gel and distained. In-gel diges-
tion was carried out sequentially with trypsin and endopep-
tidase Asp-N for 16 h each. To validate the presence of
lactoferrin in human saliva, this protein was identified by
MALDI-TOF/TOF mass spectrometer 4800 Proteomics
Analyzer (Applied Biosystems, Framingham, MA) and
4000 Series ExplorerTM software (Applied Biosystems).
2.4. Biochemical analyses
Levels of lactoferrin in saliva samples were determined
using the lactoferrin human ELISA kit (ab108882, Abcam)
according to manufacturer’s instructions. Levels of endoge-
nous Ab42 and tau in CSF samples were determined using
the Ab42 human ELISA Innotest kit (Innogenetics), and
tau human ELISA Innotest kit (Innogenetics), respectively.
2.5. Statistical analysis
Qualitative variables are expressed with the percentage
and the 95% confidence interval. Quantitative variables are
expressed with the mean 6 standard deviation (SD) in
case they follow a normal distribution and with the median
and interquartile range in case they follow nonnormal distri-
bution. To determine the relation between salivary lactofer-
rin levels and the presence or not of AD, a multiple
regression analysis was performed. To assess the association
between levels of lactoferrin and the presence or absence of
disease, crude linear regression model is constructed and
adjusted for potentially confounding variables. To select
Fig. 1. Salivary lactoferrin levels in patients with aMCI, AD, and healthy controls. (A) Lactoferrin levels decrease in aMCI and AD compared with control
group. Boxplot graph shows median, interquartile range, and extreme values of each group. ***P , .001; Kruskal-Wallis test. For lactoferrin expression in
pooled saliva samples, see Supplementary Fig. 1. For additional data on lactoferrin levels in PD, see Supplementary Fig. 2. (B) Correlation between saliva levels
of lactoferrin and cognitive decline in aMCI and AD groups. Lactoferrin levels appeared to be negatively correlated with severity of the disease (r 520.742;
P, .001; Kendall’s tau correlation analysis). (C) Saliva levels of lactoferrin correlated with MMSE score, (r5 0.731; P, .001; Spearman’s correlation anal-
ysis). (D) Receiver operating characteristic (ROC) curve obtained for the test of saliva lactoferrin levels from the full control group and aMCI/AD group. The
area under the ROC curve (AUC), a measure of how well a parameter can distinguish between diagnostic groups, was 1 (95% CI 1–1). The ROC plots represent
sensitivity (true positive rate) versus 12 specificity (false positive rate). Salivary lactoferrin significantly correlates with Ab42 (E) and total tau (F) in CSF, based
on Spearman’s correlation analysis. Abbreviations: aMCI, amnestic mild cognitive impairment; AD, Alzheimer’s disease; CSF, cerebrospinal fluid; MMSE,
Mini–Mental State Examination; PD, Parkinson’s disease; T-tau, total tau.
E. Carro et al. / Alzheimer’s & Dementia: Diagnosis, Assessment & Disease Monitoring 8 (2017) 131-138134
E. Carro et al. / Alzheimer’s & Dementia: Diagnosis, Assessment & Disease Monitoring 8 (2017) 131-138 135potential confounding variables to be included in the model,
a bivariate analysis with each of them and the dependent var-
iable (lactoferrin levels) was performed. We only selected
statistically significant (P , .05) variables. The bivariate
analysis included a Student’s t-test (two-tailed) or Mann-
Whitney U-test for single pairwise comparisons depending
on variable distribution, and the Kruskal-Wallis test for mul-
tiple comparisons. Statistical significance was set at P, .05.
The Spearman rank and Kendall’s tau correlation was used
for correlation analyses. In addition, receiver operating char-
acteristic (ROC) curve analysis was used to determine cutoff
points of salivary lactoferrin. The cutoff level for dichoto-
mizing values was selected as the situation optimizing sensi-
tivity and specificity. The classification performance of
salivary lactoferrin was assessed using area under the ROC
curve (AUC). The ROC can be understood as a plot of the
probability of classifying correctly the positive samples
against the rate of incorrectly classifying true negative sam-
ples. Thus, the AUC measure of an ROC plot is a measure of
predictive accuracy. Youden index was calculated to select
the optimum cutoff point. Data analysis was performed us-
ing the SPSS v20.0 software. A nominal P value of less
than .05 was considered to indicate statistical significance.
The statistical power of the study was 80%.Table 2
Demographic, clinical, and biochemical characteristics of subjects from a
subcohort study
Variable Control AD P value
n (F/M) 68 (43/25) 59 (44/15) ns
Age (years) 69.53 6 8 80.07 6 7.6*** ,.001
APOE ε4 carriers 17.64% 49.15%** ,.01
CSF total tau (pg/mL) 250.71 6 195.87 650.56 6 469.71** ,.01
CSF Ab42 (pg/mL) 983.05 6 461.83 366.97 6 163*** ,.001
Saliva lactoferrin
(mg/mL)
10.24 6 1.96 4.78 6 1.11*** ,.001
Abbreviations: AD, Alzheimer’s disease; F, female; M, male; ns, not sig-
nificant; CSF, cerebrospinal fluid.
NOTE. Age data are expressed as mean 6 SD. Biomarkers data are ex-
pressed as median (interquartile range). **P, .01, ***P, .001 versus con-
trol group.3. Results
Lactoferrin expression was first analyzed in pooled saliva
samples fromADand aMCI patients, and control subjects, af-
ter SDS-PAGE fractionation and identification by mass spec-
trometry analysis (31% sequence coverage; Supplementary
Fig. 1A). The band intensity analysis showed reduced lacto-
ferrin levels in aMCIandADcomparedwith the control group
(Supplementary Fig. 1B). Further confirmation by ELISA
analysis in each individual donor sample revealed that sali-
vary lactoferrin levels were significantly reduced in aMCI
and AD patients compared with the healthy control group
(Fig. 1A). A significant negative association was found be-
tween stages of disease (aMCI and AD) and salivary lactofer-
rin levels (Kendall’s tau 5 20.742; P , .001; Fig. 1B).
Salivary lactoferrin concentration was also correlated with
MMSE score in patients with aMCI and AD (r 5 0.731;
P , .001; Fig. 1C). Furthermore, we found that APOE ε4
allele status correlated with decreased lactoferrin levels in
saliva (Pearson’s5 20.204; P, .001).
Using linear regression analysis, we discovered that pa-
tients suffering from AD and aMCI had 6.432 mg (95% con-
fidence interval [CI]: 6.850–6.014; P , .001) and 5.310 mg
(95% CI: 5.810–4.810; P , .001) of salivary lactoferrin
per milliliter less than cognitively healthy participants,
respectively. We used these results to build linear classifier
models that would distinguish the aMCI/AD groups
from the control group, using ROC analysis. An AUC of 1
(95% CI 1–1) was obtained being the sensitivity 100%
(95% CI 96.90%–100%) and specificity 100% (95% CI
95.95%–100%) for aMCI/AD and healthy control groupclassification (Fig. 1D), and the cutoff value was 7.43 mg/
mL (Youden index: 1).
To evaluate whether this reduction of lactoferrin was spe-
cific to AD, we measured lactoferrin concentration in saliva
samples from 59 PD patients (Table 1). Salivary lactoferrin
levels in PD patients were higher (P , .001) than those
observed in the control group (Supplementary Fig. 2). These
findings are in agreementwith neuronal upregulation of lacto-
ferrin in PD patients, as previously reported [27,28]. The
linear regression model excluded patient’s age as a potential
confounding variable previously selected in the bivariate
analysis. Additional analysis ruled out the potential
influence of comorbidities, including diabetes, obesity,
cardiovascular disease, and hypertension on salivary
lactoferrin levels.
We then validated the cutoff value of saliva lactoferrin in
two new blinded and independent cohorts enrolling 91 addi-
tional participants with the same standardized clinical assess-
ments used in the previous study.Demographic characteristics
of participants recruited in the Alzheimer Disease Research
Unit, CIEN Foundation, Queen Sofia Foundation Alzheimer
Center (Madrid, Spain), and Pablo de Olavide University
(Sevilla, Spain) are shown in Supplementary Table 1. Results
showed that the previously determined cutoff value of saliva
lactoferrin (7.43 mg/mL) perfectly classified all patients
(MCI/AD; n 5 51) and all cognitively healthy subjects
(n5 40).
The relationship between saliva lactoferrin and CSF total
tau and CSFAb42 levels was also addressed in a 127-subject
subcohort (Table 2). Table 3 summarizes the correlation be-
tween the analytes in the control group and AD patients,
based on Spearman’s r correlation analysis. As shown in
Table 3, and Fig. 1E and F, saliva lactoferrin significantly
correlates with CSF Ab42 (r 5 0.688, P , .0001; Fig. 1E),
and CSF total tau (r 5 20.601, P , .0001; Fig. 1F).
In our case-control studies, nondemented control subjects
from clinical cohorts were subjected to strict inclusion and
exclusion criteria. These clinical cohorts have been used in
the training and validation studies as well as in the CSF
Table 3
Correlation matrix of saliva and CSF biomarkers in the control group versus
AD patients
Correlations
Biomarkers Saliva lactoferrin CSF Ab42 CSF total tau
Saliva lactoferrin 1 0.688*** 20.601***
CSF Ab42 0.688*** 1 20.529***
CSF total tau 20.601*** 20.529*** 1
Abbreviations: CSF, cerebrospinal fluid; AD, Alzheimer’s disease.
NOTE. ***P , .0001.
E. Carro et al. / Alzheimer’s & Dementia: Diagnosis, Assessment & Disease Monitoring 8 (2017) 131-138136biomarker correlation study. However, one should be
cautious when generalizing from findings based on control
samples recruited from clinic. Thus, we also decided to
analyze salivary samples from “nonclinical” subject groups
(with less detail available about the participants, but being
more representative of the general population). We used
two independent longitudinal cohorts composed of healthy
nondemented individuals. A first cohort with 116 subjects,
including caregivers, family, and other volunteers
(Supplementary Table 3), was recruited between 2009 to
2014 at the Neurology Service, and the Group of Neurode-
generative Diseases at the Hospital Universitario 12 de Oc-
tubre, and one saliva sample [7] was collected. A second
cohort of 190 apparently neurologically healthy subjects
(Supplementary Table 4) was recruited at the Biobanco
imas12 of the Hospital Universitario 12 de Octubre (Madrid,
Spain), and oral mucosa [26] samples were collected at one
time point from 2011 to 2012. The clinical status of all pa-
tients was evaluated in a follow-up examination (July
2015). Levels of lactoferrin in oral mucosa were equal to
those measured in saliva samples. We found 18 subjects
with abnormally reduced lactoferrin levels (,7.43 mg/mL)
and 288 with normal/high lactoferrin levels (
7.43 mg/
mL). After checking their clinical diagnostic status, we de-Table 4
Demographic characteristics of subjects from both longitudinal cohorts
Subjects Gender Age Onset
Lt levels
(mg/mL)
Neurological
diagnose
1 M 82 2 3.01 aMCI
2 F 70 4 3.17 aMCI
3 F 71 5 3.69 aMCI
4 F 68 5 5.10 aMCI/AD
5 F 81 1 1.65 aMCI/AD
6 F 77 2 1.89 aMCI
7 M 83 3 6.18 aMCI/AD
8 M 88 4 4.45 aMCI
9 F 96 4 5.02 AD
10 M 66 4 3.66 aMCI
11 M 82 4 2.51 AD
12 M 67 4 5.92 aMCI
13 M 84 4 4.17 AD
14 M 85 3 4.13 aMCI/AD
Abbreviations: Lt, lactoferrin; M, male; aMCI, amnestic mild cognitive
impairment; F, female; AD, Alzheimer’s disease.tected that over the course of the study (range 1–5 years,
Table 4), 14 of 18 subjects had converted to a clinical diag-
nosis of aMCI or AD, whereas none of the subjects with a
negative test value had converted to aMCI or AD. The
ROC for the collectively analyzed data from both indepen-
dent cohorts classified underdiagnosed patients and healthy
control groups with an AUC of 0.984 (95% CI 0.932–1),
with a sensitivity of 100%, and a specificity of 98.6%
(Supplementary Fig. 3).4. Discussion
Herein, we present the discovery and validation of sali-
vary lactoferrin as a novel aMCI/AD diagnostic biomarker.
According to our first study, salivary lactoferrin perfectly
classified all aMCI/AD patients and all cognitively healthy
subjects and it showed a very high correlation with validated
CSF biomarkers. Furthermore, in our second study
(“nonclinical” cohorts), we found apparently healthy indi-
viduals with low levels of saliva lactoferrin who were at
high risk of converting to aMCI/AD dementia (more than
77%). As a consequence, and although more clinical studies
are needed, we suggest that salivary lactoferrin is a precise
and robust biomarker for aMCI/AD diagnosis and may
help to identify, after a general population screening, those
“apparently healthy” subjects that suffer from underdiag-
nosed later stage preclinical AD or even MCI.
It is fairly well recognized that clinical diagnosis of
aMCI/AD patients is challenging [29,30]. However,
accuracy of AD clinical diagnosis by the specialized
physicians in this study is around 90%, similar to that
reported elsewhere [31]. Interestingly, after a more in-
depth evaluation of the results obtained in the biomarker cor-
relation study (salivary lactoferrin vs. CSF total tau and CSF
Ab42), we found that although all cognitively normal partic-
ipants had normal/high lactoferrin levels (
7.3 mg/mL), 7 of
68 (w10%) may have preclinical AD pathology, based on
their CSF Ab42 levels. After monitoring the clinical evolu-
tion of these 7 subjects, we found that one of them, with lac-
toferrin levels of 8 mg/mL (close to the cutoff value),
converted to MCI 6 years later. This may suggest that the
decline of salivary lactoferrin happens in a later stage of
the preclinical AD process, mainly when subtle cognitive
deficit appears, taking into account the hypothetical model
of the chronology [32].
In addition, we found that 7 of 59 (11.9%) patients clin-
ically diagnosed with AD showed normal levels of CSF
Ab42. Four of these 7 patients had normal CSF Ab42 levels
but high CSF total tau levels. The latter may suggest that
salivary lactoferrin may also work as a biomarker of
cortical/cognitive dysfunction associated to other types of
dementia in addition to AD. In fact, these 4 patients were
clinically diagnosed with mixed AD dementia, including
vascular component and dementia with Lewy bodies. Our re-
sults also provide evidence that salivary lactoferrin may
work for identifying “apparently healthy” subjects that
E. Carro et al. / Alzheimer’s & Dementia: Diagnosis, Assessment & Disease Monitoring 8 (2017) 131-138 137suffer from later stage preclinical AD or aMCI, a high num-
ber of which are currently underdiagnosed.
Lactoferrin, an important modulator of immune response
and inflammation [33], represents an important defensive
element by inducing a broad spectrum of antimicrobial ef-
fects [14,16,17]. A novel role for antimicrobial peptides
has been proposed in AD pathology as pathogen-targeting
agents and markers of brain infections that are involved in
the aggregation of amyloid [34], reinforcing the relationship
between AD and brain infections [35].
To our knowledge, this is the first published report of a
saliva-based single biomarker with very high accuracy for
early diagnosis of AD. Saliva lactoferrin robustly classifies
aMCI and AD patients from healthy control subjects. The
accuracy for detection is equal to or greater than that ob-
tained from other published blood and CSF studies
[36,37]. However, saliva is by far more convenient and
easier to obtain and costs less to acquire compared with
blood and CSF. In addition, our biomarker consists of a
single protein, lactoferrin, in contrast to others based on a
set of proteins, lipids, or arrays of RNAs, making it more
useful for screening in large-scale clinical trials and for
future clinical use.
Further longitudinal cohort analyses are needed to
address how salivary lactoferrin marker may help to differ-
entiate between AD and other neurodegenerative diseases,
including dementia with Lewy bodies or frontotemporal de-
mentia. Finally, we plan to study the correlation of salivary
lactoferrin levels with core CSF biomarkers, and PET neuro-
imaging, and potential confounding variables, including co-
morbid disorders, physiological status, or diet. Although
these new studies are highly recommended, our initial study
provides robust evidence for the capability of salivary lacto-
ferrin to identify patients suffering from aMCI/AD. More-
over, it may also work for identifying “apparently healthy”
subjects that suffer from later stage preclinical AD or
aMCI, a high number of which are currently underdiag-
nosed. We believe our results may represent a significant
advance in the National Institute on Aging and Alzheimer’s
Association consensus statement mandate for biomarkers of
preclinical AD.Acknowledgments
The authors are grateful to Juan Carlos del Castillo who
participated in the promotion of this study. The authors
specially thank Raquel Cobos, Consuelo Pascual, Desiree
Antequera, Jose Javier Aguirre, Bel
en Gonz
alez-Lahera,
Alicia Maroto, and Ana B. Pastor for their helpful support
and technical assistance. The authors want to particularly
acknowledge the patients enrolled in this study for their
participation and the Biobanco imas12 in Hospital 12 de Oc-
tubre integrated in the in Spanish Hospital Biobanks
Network (RetBioH; www.redbiobancos.es). Biobank is sup-
ported by Instituto de Salud Carlos III. This work was sup-
ported by a grant from GEROADiagnostics (grant 2015/82).Authors’ contributions: E.C. and G.O. conceived of the
study. Experiments were coordinated and scientifically
directed by E.C., and the article written by E.C. and G.O.
Recruitment of patients and control individuals, and disease
assessment in centres, was done by F.B., F.B.-P., A.V.,
J.A.M., P.O., M.C., A.R., and J.L.C.
The authors declare competing financial interests: E.C. and
G.O. are founders of GEROA Diagnostics. Correspondence
and requests for materials should be addressed to carroeva@
h12o.es and gorka.orive@ehu.eus.
Supplementary data
Supplementary data related to this article can be found at
http://dx.doi.org/10.1016/j.dadm.2017.04.002.RESEARCH IN CONTEXT
1. Systematic review: The authors reviewed the litera-
ture using PubMed and reported key publications.
Current biomarkers for early disease, including cere-
brospinal fluid tau and amyloid b levels, structural
and functional neuroimaging, are limited because
they are very invasive or expensive. Noninvasive bio-
markers may be a more accessible alternative, but
none can currently detect preclinical Alzheimer’s
disease (AD) with the required sensitivity and speci-
ficity.
2. Interpretation: This study showed a novel, straight-
forward, and noninvasive approach for assessment
of early stages of cognitive decline. We have
discovered and validated a new single saliva
biomarker, lactoferrin, which in our cross-sectional
investigation perfectly discriminates clinically diag-
nosed amnestic mild cognitive impairment and AD
patients from control group.
3. Future directions: Although future prospective
studies will be highly recommended, this biomarker
may represent a major turning point in the early diag-
nostics and decision making for patient care in AD.
References
[1] Bateman RJ, Xiong C, Benzinger TL, Fagan AM, Goate A, Fox NC.
Clinical and biomarker changes in dominantly inherited Alzheimer’s
disease. N Engl J Med 2012;367:795–804.
[2] Blennow K, Zetterberg H. Understanding biomarkers of neurodegen-
eration: ultrasensitive detection techniques pave the way for mecha-
nistic understanding. Nat Med 2015;21:217–9.
[3] Hulstaert F, Blennow K, Ivanoiu A, Schoonderwaldt HC,
Riemenschneider M, De Deyn PP, et al. Improved discrimination of
E. Carro et al. / Alzheimer’s & Dementia: Diagnosis, Assessment & Disease Monitoring 8 (2017) 131-138138AD patients using beta-amyloid(1-42) and tau levels in CSF.
Neurology 1999;52:1555–62.
[4] Small SA, Perera GM, DeLaPaz R, Mayeux R, Stern Y. Differential
regional dysfunction of the hippocampal formation among elderly with
memory decline and Alzheimer’s disease. Ann Neurol 1999;45:466–72.
[5] Klunk WE, Engler H, Nordberg A, Wang Y, Blomqvist G, Holt DP,
et al. Imaging brain amyloid in Alzheimer’s disease with Pittsburgh
Compound-B. Ann Neurol 2004;55:306–19.
[6] Franceschi C, Bonafe M, Valensin S, Paolisso G, Yashin AI, Monti D,
et al. Inflamm-aging. An evolutionary perspective on immunosenes-
cence. Ann N YAcad Sci 2000;908:244–54.
[7] Bermejo-Pareja F, Antequera D, Vargas T, Molina JA, Carro E. Saliva
levels of Abeta1-42 as potential biomarker of Alzheimer’s disease: a
pilot study. BMC Neurol 2010;10:108.
[8] Shi M, Sui YT, Peskind ER, Li G, Hwang H, Devic I, et al. Salivary tau
species are potential biomarkers of Alzheimer’s disease. J Alzheimers
Dis 2011;27:299–305.
[9] Devic I, Hwang H, Edgar JS, Izutsu K, Presland R, Pan C, et al. Sali-
vary a-synuclein and DJ-1: potential biomarkers for Parkinson’s dis-
ease. Brain 2011;134:e178.
[10] Kamer AR, Craig RG, Dasanayake AP, Brys M, Glodzik-Sobanska L,
de Leon MJ. Inflammation and Alzheimer’s disease: possible role of
periodontal diseases. Alzheimers Dement 2008;4:242–50.
[11] Holmes C, Cunningham C, Zotova E, Woolford J, Dean C, Kerr S,
et al. Systemic inflammation and disease progression in Alzheimer dis-
ease. Neurology 2009;73:768–74.
[12] Iwashyna TJ, Ely EW, Smith DM, Langa KM. Long-term cognitive
impairment and functional disability among survivors of severe sepsis.
JAMA 2010;304:1787–94.
[13] Hill JM, Clement C, Pogue AI, Bhattacharjee S, Zhao Y, Lukiw WJ.
Pathogenic microbes, the microbiome, and Alzheimer’s disease
(AD). Front Aging Neurosci 2014;6:127.
[14] van der Strate BW, Beljaars L, Molema G, Harmsen MC, Meijer DK.
Antiviral activities of lactoferrin. Antiviral Res 2001;52:225–39.
[15] Beljaars L, van der Strate BW, Bakker HI, Reker-Smit C, van Loenen-
Weemaes AM, Wiegmans FC, et al. Inhibition of cytomegalovirus
infection by lactoferrin in vitro and in vivo. Antiviral Res 2004;
63:197–208.
[16] Orsi N. The antimicrobial activity of lactoferrin: current status and per-
spectives. Biometals 2004;17:189–96.
[17] Gifford JL, Hunter HN, Vogel HJ. Lactoferricin: a lactoferrin-derived
peptide with antimicrobial, antiviral, antitumor and immunological
properties. Cell Mol Life Sci 2005;62:2588–98.
[18] Valenti P, Catizone A, Pantanella F, Frioni A, Natalizi T, Tendini M,
et al. Lactoferrin decreases inflammatory response by cystic fibrosis
bronchial cells invaded with Burkholderia cenocepacia iron-
modulated biofilm. Int J Immunopathol Pharmacol 2011;24:1057–68.
[19] Berlutti F, Schippa S, Morea C, Sarli S, Perfetto B, Donnarumma G,
et al. Lactoferrin downregulates pro-inflammatory cytokines upex-
pressed in intestinal epithelial cells infected with invasive or noninva-
sive Escherichia coli strains. Biochem Cell Biol 2006;84:351–7.
[20] Ito S, Ohtsuki S, Kamiie J, Nezu Y, Terasaki T. Cerebral clearance of
human amyloid-beta peptide (1-40) across the blood-brain barrier is
reduced by self-aggregation and formation of low-density lipoproteinreceptor-related protein-1 ligand complexes. J Neurochem 2007;
103:2482–90.
[21] Jaeger S, Pietrzik CU. Functional role of lipoprotein receptors in Alz-
heimer’s disease. Curr Alzheimer Res 2008;5:15–25.
[22] McKhann GM, Knopman DS, Chertkow H, Hyman BT, Jack CR Jr,
Kawas CH, et al. The diagnosis of dementia due to Alzheimer’s dis-
ease: recommendations from the National Institute on Aging-Alz-
heimer’s Association workgroups on diagnostic guidelines for
Alzheimer’s disease. Alzheimers Dement 2011;7:263–9.
[23] Petersen RC.Mild cognitive impairment as a diagnostic entity. J Intern
Med 2004;256:183–94.
[24] Gelb DJ, Oliver E, Gilman S. Diagnostic criteria for Parkinson disease.
Arch Neurol 1999;56:33–9.
[25] Calne DB, Snow BJ, Lee C. Criteria for diagnosing Parkinson’s dis-
ease. Ann Neurol 1992;32:S125–7.
[26] Aagaard K, Petrosino J, Keitel W, Watson M, Katancik J, Garcia N,
et al. The Human Microbiome Project strategy for comprehensive
sampling of the human microbiome and why it matters. FASEB J
2013;27:1012–22.
[27] Faucheux BA, Nillesse N, Damier P, Spik G, Mouatt-Prigent A,
Pierce A, et al. Expression of lactoferrin receptors is increased in the
mesencephalon of patients with Parkinson disease. Proc Natl Acad
Sci U S A 1995;92:9603–7.
[28] Leveugle B, Faucheux BA, Bouras C, Nillesse N, Spik G, Hirsch EC,
et al. Cellular distribution of the iron-binding protein lactotransferrin
in the mesencephalon of Parkinson’s disease cases. Acta Neuropathol
1996;91:566–72.
[29] Dubois B, Feldman HH, Jacova C, Hampel H, Molinuevo JL,
Blennow K, et al. Advancing research diagnostic criteria for Alz-
heimer’s disease: the IWG-2 criteria. Lancet Neurol 2014;13:614–29.
[30] Dubois B, Hampel H, Feldman HH, Scheltens P, Aisen P, Andrieu S,
et al. Preclinical Alzheimer’s disease: definition, natural history, and
diagnostic criteria. Alzheimers Dement 2016;12:292–323.
[31] Knopman DS, Rocca WA, Cha RH, Edland SD, Kokmen E. Incidence
of vascular dementia in Rochester, Minn, 1985-1989. Arch Neurol
2002;59:1605–10.
[32] Jack CR Jr, Knopman DS, Jagust WJ, Shaw LM, Aisen PS,
Weiner MW, et al. Hypothetical model of dynamic biomarkers of
the Alzheimer’s pathological cascade. Lancet Neurol 2010;9:119–28.
[33] Legrand D, Elass E, Carpentier M, Mazurier J. Lactoferrin: a modu-
lator of immune and inflammatory responses. Cell Mol Life Sci
2005;62:2549–59.
[34] Welling MM, Nabuurs RJ, van der Weerd L. Potential role of antimi-
crobial peptides in the early onset of Alzheimer’s disease. Alzheimers
Dement 2015;11:51–7.
[35] Alonso R, Pisa D,Marina AI,Morato E, R
abanoA, Carrasco L. Fungal
infection in patients with Alzheimer’s disease. J Alzheimers Dis 2014;
41:301–11.
[36] Mapstone M, Cheema AK, Fiandaca MS, Zhong X, Mhyre TR,
MacArthur LH, et al. Plasma phospholipids identify antecedent mem-
ory impairment in older adults. Nat Med 2014;20:415–8.
[37] Roe CM, Fagan AM, Williams MM, Ghoshal N, Aeschleman M,
Grant EA, et al. ImprovingCSF biomarker accuracy in predicting prev-
alent and incident Alzheimer disease. Neurology 2011;76:501–10.