M O L E C U L A R O N C O L O G Y 1 0 ( 2 0 1 6 ) 1 6 2 1 e1 6 2 6

available at www.sciencedirect.com

ScienceDirect

www.elsevier.com/locate/molonc

Prospective validation of a blood-based 9-miRNA profile for

early detection of breast cancer in a cohort of women
examined by clinical mammography

Maria B. Lynga,*, Annette R. Kodahlb, Harald Binderc, Henrik J. Ditzela,b,**

a
Institute of Molecular Medicine, University of Southern Denmark, Odense, Denmark
b
Department of Oncology, Odense University Hospital, Odense, Denmark
c
Institute of Medical Biostatistics, Epidemiology and Informatics (IMBEI), University Medical Center, Johannes
Gutenberg University, Mainz, Germany

A R T I C L E I N F O A B S T R A C T

Article history: Mammography is the predominant screening method for early detection of breast cancer,
Received 3 June 2016 but has limitations and could be rendered more accurate by combination with a blood-
Received in revised form based biomarker profile. Circulating microRNAs (miRNAs) are increasingly recognized as
23 September 2016 strong biomarkers, and we previously developed a 9-miRNA profile using serum and
Accepted 17 October 2016 LNA-based qPCR that effectively stratified patients with early stage breast cancer vs.
Available online 1 November 2016 healthy women. To further develop the test into routine clinical practice, we collected
serum of women examined by clinical mammography (N ¼ 197) according to standard
Keywords: operational procedures (SOPs) of the Danish Cancer Biobank. The performance of the circu-
Breast cancer lating 9-miRNA profile was analyzed in 116 of these women, including 36 with breast can-
Serum cer (aged 50e74), following a standardized protocol that mimicked a routine clinical set-up.
Circulating miRNA profile We confirmed that the profile is significantly different between women with breast cancer
Biomarker and controls (p-value <0.0001), with an AUC of 0.61. Significantly, one woman whose 9-
Early detection miRNA profile predicted a 73% probability of having breast cancer indeed developed the
disease within one year despite being categorized as clinically healthy at the time of blood
sample collection and mammography. We propose that this miRNA profile combined with
mammography will increase the overall accuracy of early detection of breast cancer.
ª 2016 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights
reserved.

Abbreviations: ER, estrogen receptor; HER2, human epidermal growth factor 2; miRNA, microRNA; SOP, standard operational proced-
ures; LNA, locked nucleic acid; AUC, area under the curve; ROC, receiver operating characteristic.
* Corresponding author. Institute of Molecular Medicine, University of Southern Denmark, J.B. Winsløws Vej 25, 3, 5000 Odense C,
Denmark.
** Corresponding author. Institute of Molecular Medicine, University of Southern Denmark, J.B. Winsløws Vej 25, 3, 5000 Odense C,
Denmark.
E-mail addresses: maria.lyng@rsyd.dk (M.B. Lyng), hditzel@health.sdu.dk (H.J. Ditzel).
http://dx.doi.org/10.1016/j.molonc.2016.10.004
1574-7891/ª 2016 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.
1622 M O L E C U L A R O N C O L O G Y 1 0 ( 2 0 1 6 ) 1 6 2 1 e1 6 2 6

1. Introduction based biomarker assays have great clinical potential and miR-
NAs are excellent candidates.
Mammography is currently the standard screening method We previously reported the identification and retrospective
for breast cancer in many countries, fulfilling the WHO criteria validation of a circulating 9-miRNA profile for detection of
(Chestnov et al., 2014). However, the method has limitations early stage breast cancer using miRNAs isolated from serum
due to relatively high false-positive rates, and limitations in samples obtained in the late 90’s from 48 women with early
detecting smaller tumors, including those in dense breast tis- breast cancer and 24 healthy controls. Serum samples from
sue (Nelson et al., 2016a,b). In 2014, WHO published an an additional cohort of 111 women (60 with early stage breast
updated position on mammography screening (Chestnov cancer and 51 controls) were used for validation and
et al., 2014) to provide independent guidance on the balance confirmed the ability of the 9-miRNA profile to distinguish be-
between benefits and harm in women of different age groups. tween women with breast cancer and healthy individuals
In well-resourced settings, such as many Western Countries, (Kodahl et al., 2014).
biennial screening is suggested for women aged 50e69 years To further develop this assay for clinical use, we evaluated
(Mittmann et al., 2015), which is the current screening inter- the blood-based 9-miRNA profile of a prospective cohort of
val, and age-range, in Denmark. women undergoing clinical mammography (N ¼ 197). This
Mammograms are increasingly becoming digitalized cohort was comprised of 18% invasive cancer patients
(computerized) and being developed further as a new tomo- (N ¼ 36), and 82% clinically healthy controls (N ¼ 161), with
graphic technique, termed digital breast tomosynthesis an age-range of 50e74 years. Every second control was chro-
(DBT) (Gilbert et al., 2016). Despite these technological devel- nologically selected to obtain a 2:1 ratio (controls to cases),
opments, however, there are still limitations resulting in resulting in miRNA profiling of 116 women. Our finding sup-
false-positive and -negative results, and implementation of ports the utility of the 9-miRNA profile for early detection of
these advanced digitalized techniques is costly and conse- breast cancer, which would allow initiation of treatment at
quently difficult in many countries. Nuclear breast imaging an earlier time point.
in which radiotracers are used for functional screening rather
than X-ray-based anatomic screening (Berg, 2016) has been
tested in women with a high-risk of developing breast cancer, 2. Materials and methods
and has potential for population-based screening, although
likely not in the near future. 2.1. Breast cancer patients and healthy controls
Thus, there is great interest in identifying circulating bio-
markers to screen for early stage cancer using blood of Women undergoing clinical mammography due to recall from
seemingly healthy individuals. These circulating biomarkers screening mammography were asked to participate in the
include nucleic acid fragments shed into the blood stream study, and upon signing the informed consent, blood was
from cancerous cells, either apoptotic or necrotic, or as an drawn from 197 women (aged 50e74 years) from October
active secretory process (Schwarzenbach et al., 2014). The 2013 to July 2015. For miRNA profiling, all invasive breast can-
notion of a “liquid biopsy” has several advantages since cer cases were included (N ¼ 36), and every second chronolog-
circulating nucleotides (miRNA/DNA) are highly stable in ical database entry of a woman with no mammography-
the blood-stream (Kosaka et al., 2010; Mitchell et al., 2008), detected breast cancer was selected for the control group
likely represent the entire tumor vs. an isolated tumor nee- (N ¼ 80), resulting in a 2:1 ratio of healthy controls vs. breast
dle biopsy, permit automation of most of the analysis, and cancer cases. Breast cancer diagnosis was confirmed by biopsy
support compliance due to minimal discomfort. The pres- and surgical specimen. Follow-up of healthy controls consist-
ence of miRNA in the blood of cancer patients is believed ing of review of medical files and/or new mammograms was
to have a half-life ranging from 15 min to several hours, performed in April 2016; follow-up period: 0.7e2.4 years. Blood
similar to cell-free DNA (Schwarzenbach et al., 2014; samples were collected prior to clinical mammography and
Minchin et al., 2001; Botezatu et al., 2000), providing contin- serum was isolated according to SOPs of the Danish Cancer
uous markers of the cancer applicable to screening as well Biobank (DCB Herlev Hospital, 2016). Briefly, serum was pre-
as monitoring potential recurrence, and perhaps also pared within 1 h of blood collection by coagulation between
serving as early markers of the effect of certain cancer 30 and 120 min, followed by centrifugation at 4  C, 2000 g for
treatments. 10 min, and immediately stored after fractionation at
The first report of miRNAs in the blood stream in 2008 80  C. Clinical characteristics of those included in the statis-
(Chim et al., 2008; Lawrie et al., 2008) initiated an intense tical analysis are listed in Table 1. The study is approved by the
search for these markers for various diseases, but enthusiasm Regional Ethical Committee (ID: S-20100132), and all partici-
for using miRNAs has been challenged in recent years pants signed informed consent. The REMARK guidelines
(Witwer, 2015; Schwarzenbach et al., 2014). Issues raised were followed where possible (McShane et al., 2005).
include the risk that miRNAs may represent the body’s
“danger” response to a cancer more than the tumor itself. 2.2. Isolation of RNA from serum
There are also limitations with reproducibility of miRNA pro-
files attributed to variability of pre-analytical handling, popu- Total RNA was extracted from serum using the miRCURY
lation diversity and varying technologies used. Despite these RNA isolation kit (Exiqon, Denmark), including addition of
concerns, development in this field continues, since blood- 1 mg carrier-RNA per 60 mL lysis and finally eluted in 50 mL
 M O L E C U L A R O N C O L O G Y 1 0 ( 2 0 1 6 ) 1 6 2 1 e1 6 2 6 1623

occurred in the blood samples was evaluated by determining

Table 1 e Clinical characteristics of the study patient cohort.
the ratio of miRNA-451 (expressed in red blood cells) and
Parameter Patients, Patients, Controls miRNA-23a (stable in blood, and not affected by hemolysis)
ERþ ER (Blondal et al., 2013) and samples exhibiting a ratio of >7
Total 31 5 80 were excluded.
Biopsies 31 5 4
Age, years 63 (50e70) 61 (56e63) 54 (50e74) 2.5. Normalization and data analysis
(median, range)
Surgery In our previous publication, miRNA-10b-5p was identified to
Lumpectomy 24 3
be the most stable miRNA, based on 65 miRNA assays, and
Mastectomy 7 1
Inoperablea 0 1
therefore used as reference miRNA. The risk score was calcu-
Diagnosis lated using the normalized (DCp,target ¼ log2 (Cp target
IDC 25 4 miRNA)log2 (Cp miRNA-10b-5p)) miRNA values, centered
ILC 2 0 and standardized to a standard deviation of 1, of the 9 miRNAs
Mixed IDC/ILC 3 0 (miR-15a, miR-18a, miR-107, miR-133a, miR-139-5p, miR-143,
Tubular 1 0
miR-145, miR-365, miR-425) and the following equation:
NAa 0 1
score ¼ 0.934 þ (0.152* DCp [miR-143-3p]) þ (0.435* DCp
Tumor size mm 12 (4e34) 18 (12e58)
(median, range) [miR-139-5p]) þ (0.341* DCp [miR-107]) þ (1.124* DCp [miR-
Nodal status 145-5p]) þ (0.702* DCp [miR-15a-5p]) þ (0.31* DCp [miR-425-
0 22 3 5p]) þ (0.18* DCp [miR-18a-5p]) þ (0.275* DCp [miR-133a-3p])
1e3 9 1 þ (0.134* DCp [miR-365a-3p]). This score can be converted
Selected LNs biopsieda 1 into a predicted probability of having breast cancer by the
HER2 status
equation ¼ 1/(1 þ exp(score)). Comparison between groups
Normal 30 5
was performed by means of t-tests; discrimination perfor-
Amplified 1 0
mance was assessed using ROC curves and the corresponding
a The inoperable patient. area under the curve (AUC).

3. Results
RNase-free water. The samples were immediately stored at
80  C until further analysis.
3.1. Patient characteristics

2.3. miRNA real-time PCR A consecutive cohort of women (N ¼ 197) who had undergone
clinical mammography due to recall from screening
Reverse transcription was conducted with 2 ml RNA in 10 ml re- mammography was evaluated, 36 of which had invasive
actions using the miRCURY LNA Universal RT microRNA breast cancer. The controls were then selected to obtain a
PCR, Polyadenylation and cDNA synthesis kit (Exiqon) in tech- 2:1 ratio of controls to cases, choosing every second woman
nical triplicates. cDNA was diluted 50 and assayed in 10 mL included chronologically in the study, leading to a total of
PCR reactions according to the protocol for miRCURY LNA 116 samples for miRNA profiling. The clinical characteristics
Universal RT microRNA PCR; each microRNA was assayed of these participants are summarized in Table 1. Since our
once by qPCR on the microRNA Ready-to-Use PCR, custom previous study identified no difference in miRNA expression
Pick and Mix Panel using ExiLENT SYBR Green master mix according to lymph node involvement, age, menopausal sta-
(Exiqon). Negative controls excluding template from the tus or tumor size between breast cancer samples, these pa-
reverse transcription reaction was performed and profiled in rameters were not used to further stratify the patients. The
the same manner as the samples. Amplification was per- qPCR data generated was analyzed as in our previous study
formed in a LightCycler 480 Real-Time PCR System (Roche, (Kodahl et al., 2014), including stringent quality control mea-
Germany) in 384 well plates, and the amplification curves sures. Two serum samples, both from individuals in whom
were analyzed using the Roche LC software both for determi- no breast cancer was detected by clinical mammography,
nation of Cq (by the 2nd derivative method) and for melting were excluded due to the risk of hemolysis to eliminate any
curve analysis. potential influence on our results.

2.4. Sample quality 3.2. Clinical utility of the 9-miRNA profile

Raw data and all assays were inspected for distinct melting The 9-miRNA profile was significantly different between pa-
curves and the Tm was verified to be within known specifica- tients with ERþ breast cancer and healthy controls
tions for the assay. Only assays detected with 3 Cqs less than (p < 0.0001, (95% CI [0.41; 0.86])) with an AUC of 0.58
the negative control, and with Cq <37, were included in the (Figure 1A) following a completely standardized flow of anal-
data analysis. DNA spike-ins’ were used to assess the quality ysis, confirming our previously published results. We also
of the RNA extraction, RT and qPCR steps. These assays investigated the performance of the profile for the complete
were not used for normalization. Whether hemolysis had dataset, including the ER breast cancer samples (Figure 1B).
1624 M O L E C U L A R O N C O L O G Y 1 0 ( 2 0 1 6 ) 1 6 2 1 e1 6 2 6

1
A
1
0.8

Predicted probability
0.8
0.6

0.6
SensiƟvity

0.4

0.4 0.2

0
0.2 -2 -1 0 1 2 3 4 5 6
Score

0
0 0.2 0.4 0.6 0.8 1 Figure 2 e Visualization of the relation of the risk score to the
predicted probability of having breast cancer. The score was
1 - Specificity
calculated from the normalized qPCR expression data of the 9
B miRNAs using the formula provided in Material and Methods.
1

0.8

0.6
4. Discussion
SensiƟvity

In this study, we verified our circulating 9-miRNA profile

0.4
(Kodahl et al., 2014) in a prospectively collected cohort of
post-menopausal women undergoing clinical mammography
0.2
due to recall after screening mammography. A ratio of 2:1 of
clinically healthy controls vs. breast cancer patients, respec-
0
0 0.2 0.4 0.6 0.8 1 tively, was chosen to minimize the potential risk of bias, as
1 - Specificity
some of the healthy controls may have clinically undetectable
cancer or inflammation that could potentially affect the re-
Figure 1 e ROC curve analysis using the 9-miRNA profile for sults. The set-up of this study mimicked a diagnostic test in
discriminating breast cancer cases from controls (N [ 78), i.e. a standard clinical setting, following clear SOPs and core facil-
without a suspicious finding on the mammogram. A) Graph only ity techniques.
including the ERD breast (N [ 31) cancer cases, with an Although mammography is a widely used and approved
AUC [ 0.58 (p < 0.0001, (95% CI [0.41; 0.86])). B) Graph of the method for detecting breast cancer, it is not optimal since
ERD and ERL breast cancer samples combined (N [ 36), having an the test can be affected by breast density, is operator-
AUC [ 0.61 (p < 0.0001, (95% CI [0.35; 0.77])). dependent and is quite uncomfortable. Twenty seven percent
of the women included in our cohort had a biopsy taken at
time of the clinical mammography (N ¼ 53/197), and 68%
(N ¼ 36) were found to have cancer. The women selected for
This profile also reached significance (p < 0.0001, 95% CI biopsy were not only based on X-ray mammography, but
[0.35; 0.77]), and there was a slight increase in the AUC also clinical exams and ultrasound. Screening mammography
(0.61) for early detection of combined ERþ and ER breast has a high percentage of recall, and most turn out not to have
cancer. cancer. In a recent American meta-analysis (2016), a 42% cu-
The qPCR data for the 9-miRNAs can be transformed into a mulative rate of false-positives over 10 years were found
risk score that can be converted mathematically to a predicted upon biennial screening (Nelson et al., 2016a,b). Furthermore,
probability of breast cancer (Figure 2). This predicted probabil- the invasive lobular carcinomas (ILCs) have long been recog-
ity functions as a continuous value, and its performance was nized as difficult to image by X-ray due to the growth pattern
evaluated by reviewing whether any of the women in whom of diffuse infiltration of single rows of malignant cells, which
no breast cancer was detected by clinical mammography sub- does not destroy the underlying anatomic structures (Johnson
sequently developed breast cancer within a follow-up period et al., 2015). A normal distribution (Lakhani et al., 2016) of
of 0.7e2.4 years. One woman was diagnosed with bilateral invasive ductal carcinomas (IDCs) (80%) and ILCs (11%) was
breast cancer 348 days after blood sampling, and did not observed in this study.
have a biopsy taken at clinical mammography. Her 9-miRNA Cell-free (cf) miRNAs have all the characteristics of an
profile score was 0.978, resulting in a predicted probability of optimal blood-based cancer biomarker, such as stability in
having breast cancer of 0.73 (i.e. 73%). both the blood stream and during handling and storage, and
To assess the sensitivity and specificity of the 9-miRNA likely reflect the tumor in its entirety compared to a single bi-
profile, a cut-off of 0.56 (risk score value) was chosen, yielding opsy. Cell-free miRNAs have been determined to be RNase-
0.70 and 0.47, or 0.73 and 0.44, respectively, for the ERþ group resistant in plasma as they were found to be bound with com-
only or the combined ERþ and ER group. plexes of the protein AGO2, but a minority were also found to
 M O L E C U L A R O N C O L O G Y 1 0 ( 2 0 1 6 ) 1 6 2 1 e1 6 2 6 1625

be associated with vesicles or exosomes (Mitchell et al., 2008;

Mathivanan et al., 2010). Authors’ contribution
The published miRNA blood-based profiles for detection of
early breast cancer have minimal overlap in which miRNAs MBL, ARK and HJD participated in the study design. MBL coor-
are included (Ng et al., 2013; Zhu et al., 2009; Cuk et al., 2013; dinated the project, performed data interpretation and wrote
Heneghan et al., 2010; Zhang et al., 2015; Freres et al., 2016; the first draft of the manuscript. HB performed the statistical
Hagrass et al., 2015; Mishra et al., 2015). This lack of consis- analysis. ARK, HB and HJD assisted in writing the manuscript,
tency could be due to redundancy in miRNA function, and and all authors have read and approved the final manuscript.
several miRNAs could be deregulated in the same manner un-
der certain conditions, replacing one miRNA with another.
Another explanation for the lack of overlap could be the tech- Funding
nical approach, from the pre-analytical techniques of blood
sampling and isolation of serum/plasma, to the use of assays The public and private foundations that supported the study
(qPCR, array or sequencing) and reagents. had no role in the design and conduct of this study; in the
All women in this validation study were above the age of 50 collection, management, analysis and interpretations of the
and thus most likely post-menopausal, representing the com- data; or in the preparation, review or approval of the manu-
mon mammography screening population. Our profile was script. The authors had full access to all the data in the study
previously developed, including thorough investigation of and had the final responsibility for the decision to submit the
the normalization procedure, and we could significantly iden- manuscript for publication.
tify invasive breast cancer cases among healthy controls This work was supported by the Danish Cancer Society
(Kodahl et al., 2014). This was underscored by one woman [grant number: A6362, H.J. Ditzel], A Race Against Breast Can-
with a 73% predicted probability of breast cancer who had cer (H.J. Ditzel), Odense University Hospital Research Council
no evidence of breast cancer at the time of clinical mammog- (H.J. Ditzel), Dean’s Fund for Project Development at Univer-
raphy and blood sample, but in whom a palpable lump and a sity of Southern Denmark (H.J. Ditzel), AgeCare OUH EliteCen-
breast cancer diagnosis was confirmed nearly a year later. ter (H.J. Ditzel), and Danish Center for Translational Breast
Based on our blood test, this individual could potentially Cancer Research (H.J. Ditzel).
have undergone further medical exams and had treatment
initiated much earlier. As our population was restricted in
available ERebreast cancer samples, we will need to further Conflict of interest
investigate this in a larger ER breast cancer population to
confirm the all-round performance of this 9-miRNA profile. In- The authors declare no conflict of interest.
vestigations of non-malignant breast lesions to identify the
potential to discriminate between these and invasive breast
cancer will also be required. Furthermore, it would be very
Acknowledgements
interesting to examine the profile in younger women, as their
denser breast tissue poses a challenge for mammography.
The Danish CancerBiobank (DCB) is acknowledged for biolog-
ical material and for the data regarding handling and storage.
We would like to thank the Mammography Screening
5. Conclusion personnel, particularly Bente Magelund, and the Department
of Clinical Biochemistry and Pharmacology, Odense Univer-
Overall, while screening mammography has generally been
sity Hospital, for excellent assistance with the recruitment
shown to reduce breast cancer mortality, a recent study indi-
of women and blood drawing, and M. Kat Occhipinti for edito-
cates that the effect may only be borderline statistically signif-
rial assistance.
icant, particularly for younger women (Nelson et al., 2016a,b).
MicroRNAs are very potent blood-based biomarkers, but their
R E F E R E N C E S
use in the clinical setting is still limited due to lack of repro-
ducibility, limited sample sizes investigated and the lack of in-
dependent validation cohorts. Ideally, a blood-based
biomarker profile combined with mammogram findings could Berg, W.A., 2016. Nuclear breast imaging: clinical results and
future directions. J. Nucl. Med. 57 (Suppl. 1), 46Se52S.
increase both sensitivity and specificity and render an earlier,
Blondal, T., Jensby, N.S., Baker, A., Andreasen, D., Mouritzen, P.,
more precise, diagnosis of breast cancer. This potential of the Wrang, T.M., Dahlsveen, I.K., 2013. Assessing sample and
9-miRNA profile was underscored by the subsequent identifi- miRNA profile quality in serum and plasma or other biofluids.
cation of a interval breast cancer in one woman who, in our Methods 59, S1eS6.
initial analysis, had a 73% risk of breast cancer based on our Botezatu, I., Serdyuk, O., Potapova, G., Shelepov, V., Alechina, R.,
9-miRNA profile blood test, but no detectable disease at the Molyaka, Y., Ananev, V., Bazin, I., Garin, A., Narimanov, M.,
Knysh, V., Melkonyan, H., Umansky, S., Lichtenstein, A., 2000.
time of mammography and blood sampling. As the data ma-
Genetic analysis of DNA excreted in urine: a new approach for
tures, we will investigate whether more interval cancers are
detecting specific genomic DNA sequences from cells dying in
diagnosed, and further assess the miRNA profile to see an organism. Clin. Chem. 46, 1078e1084.
whether the 9-miRNA profile could have indicated the pres- Chestnov, O., Sepulveda, C., Murad, H., Heneghan, C.,
ence of a mammographically-undetectable cancer. Gartlehner, G., Bramley, D., Norris, S., 2014. WHO Position
1626 M O L E C U L A R O N C O L O G Y 1 0 ( 2 0 1 6 ) 1 6 2 1 e1 6 2 6

Paper on Mammography Screening. http://apps.who.int/iris/ McShane, L.M., Altman, D.G., Sauerbrei, W., Taube, S.E., Gion, M.,
bitstream/10665/137339/1/9789241507936_eng.pdf? Clark, G.M., 2005. Reporting recommendations for tumor
ua¼1&ua¼1. marker prognostic studies (REMARK). J. Natl. Cancer Inst. 97,
Chim, S.S., Shing, T.K., Hung, E.C., Leung, T.Y., Lau, T.K., 1180e1184.
Chiu, R.W., Lo, Y.M., 2008. Detection and characterization of Minchin, R.F., Carpenter, D., Orr, R.J., 2001. Polyinosinic acid and
placental microRNAs in maternal plasma. Clin. Chem. 54, polycationic liposomes attenuate the hepatic clearance of
482e490. circulating plasmid DNA. J. Pharmacol. Exp. Ther. 296,
Cuk, K., Zucknick, M., Heil, J., Madhavan, D., Schott, S., 1006e1012.
Turchinovich, A., Arlt, D., Rath, M., Sohn, C., Benner, A., Mishra, S., Srivastava, A.K., Suman, S., Kumar, V., Shukla, Y.,
Junkermann, H., Schneeweiss, A., Burwinkel, B., 2013. 2015. Circulating miRNAs revealed as surrogate molecular
Circulating microRNAs in plasma as early detection markers signatures for the early detection of breast cancer. Cancer Lett.
for breast cancer. Int. J. Cancer 132, 1602e1612. 369, 67e75.
DCB Herlev Hospital, 2016. The Danish Cancer Biobank. http:// Mitchell, P.S., Parkin, R.K., Kroh, E.M., Fritz, B.R., Wyman, S.K.,
www.danskcancerbiobank.dk/. Pogosova-Agadjanyan, E.L., Peterson, A., Noteboom, J.,
Freres, P., Wenric, S., Boukerroucha, M., Fasquelle, C., Thiry, J., O’Briant, K.C., Allen, A., Lin, D.W., Urban, N., Drescher, C.W.,
Bovy, N., Struman, I., Geurts, P., Collignon, J., Schroeder, H., Knudsen, B.S., Stirewalt, D.L., Gentleman, R., Vessella, R.L.,
Kridelka, F., Lifrange, E., Jossa, V., Bours, V., Josse, C., Nelson, P.S., Martin, D.B., Tewari, M., 2008. Circulating
Jerusalem, G., 2016. Circulating microRNA-based screening microRNAs as stable blood-based markers for cancer
tool for breast cancer. Oncotarget 7, 5416e5428. detection. Proc. Natl. Acad. Sci. U.S.A 105, 10513e10518.
Gilbert, F.J., Tucker, L., Young, K.C., 2016. Digital breast Mittmann, N.M., Stout, N.K., Lee, P., Tosteson, A.N., Trentham-
tomosynthesis (DBT): a review of the evidence for use as a Dietz, A., Alagoz, O., Yaffe, M.J., 2015. Total cost-effectiveness
screening tool. Clin. Radiol. 71, 141e150. of mammography screening strategies. Health Rep. 26, 16e25.
Hagrass, H.A., Sharaf, S., Pasha, H.F., Tantawy, E.A., Nelson, H.D., Fu, R., Cantor, A., Pappas, M., Daeges, M.,
Mohamed, R.H., Kassem, R., 2015. Circulating microRNAs e a Humphrey, L., 2016a. Effectiveness of breast cancer screening:
new horizon in molecular diagnosis of breast cancer. Genes systematic review and meta-analysis to update the 2009 U.S.
Cancer 6, 281e287. Preventive Services Task Force Recommendation. Ann. Intern.
Heneghan, H.M., Miller, N., Lowery, A.J., Sweeney, K.J., Newell, J., Med. 164, 244e255.
Kerin, M.J., 2010. Circulating microRNAs as novel minimally Nelson, H.D., Pappas, M., Cantor, A., Griffin, J., Daeges, M.,
invasive biomarkers for breast cancer. Ann. Surg. 251, Humphrey, L., 2016b. Harms of breast cancer screening:
499e505. systematic review to update the 2009 U.S. Preventive
Johnson, K., Sarma, D., Hwang, E.S., 2015. Lobular breast cancer Services Task Force Recommendation. Ann. Intern. Med 164,
series: imaging. Breast Cancer Res. 17, 94. 256e267.
Kodahl, A.R., Lyng, M.B., Binder, H., Cold, S., Gravgaard, K., Ng, E.K., Li, R., Shin, V.Y., Jin, H.C., Leung, C.P., Ma, E.S., Pang, R.,
Knoop, A.S., Ditzel, H.J., 2014. Novel circulating microRNA Chua, D., Chu, K.M., Law, W.L., Law, S.Y., Poon, R.T.,
signature as a potential non-invasive multi-marker test in ER- Kwong, A., 2013. Circulating microRNAs as specific biomarkers
positive early-stage breast cancer: a case control study. Mol. for breast cancer detection. PLoS One 8, e53141.
Oncol. 8, 874e883. Schwarzenbach, H., Nishida, N., Calin, G.A., Pantel, K., 2014.
Kosaka, N., Iguchi, H., Ochiya, T., 2010. Circulating microRNA in Clinical relevance of circulating cell-free microRNAs in cancer.
body fluid: a new potential biomarker for cancer diagnosis and Nat. Rev. Clin. Oncol. 11, 145e156.
prognosis. Cancer Sci. 101, 2087e2092. Witwer, K.W., 2015. Circulating microRNA biomarker studies:
Lakhani, S.R., Ellis, I.O., Schnitt, S.J., Tan, P.H., van de Vijver, M.J., pitfalls and potential solutions. Clin. Chem. 61, 56e63.
2016. WHO Classification of Tumours of the Breast. IARC Press. Zhang, L., Xu, Y., Jin, X., Wang, Z., Wu, Y., Zhao, D., Chen, G.,
Lawrie, C.H., Gal, S., Dunlop, H.M., Pushkaran, B., Liggins, A.P., Li, D., Wang, X., Cao, H., Xie, Y., Liang, Z., 2015. A circulating
Pulford, K., Banham, A.H., Pezzella, F., Boultwood, J., miRNA signature as a diagnostic biomarker for non-invasive
Wainscoat, J.S., Hatton, C.S., Harris, A.L., 2008. Detection of early detection of breast cancer. Breast Cancer Res. Treat. 154,
elevated levels of tumour-associated microRNAs in serum of 423e434.
patients with diffuse large B-cell lymphoma. Br. J. Haematol. Zhu, W., Qin, W., Atasoy, U., Sauter, E.R., 2009. Circulating
141, 672e675. microRNAs in breast cancer and healthy subjects. BMC. Res.
Mathivanan, S., Ji, H., Simpson, R.J., 2010. Exosomes: extracellular Notes 2, 89.
organelles important in intercellular communication.
J. Proteomics 73, 1907e1920.