ArticlePDF Available

Abstract and Figures

Objective: A noninvasive tool that allows individuals to be monitored who are at risk of developing a malignancy is an unmet need. Such a test would need to consist of a molecular signature that allows for gradual judgment to assess the efficacy of preventive strategies. Here we performed a proof-of-principle study to test whether a DNA methylation (DNAme) signature in fluid collected from the vagina is able to identify women with cervical or endometrial cancer. Materials and methods: DNA from vaginal fluid samples from 111 women (30, 8, 73 with endometrial cancer, cervical cancer, and benign gynecological conditions, respectively) were analyzed for DNAme using the Illumina 450k DNA methylation bead array assay, which allows the assessment of DNAme at more than 480.000 CpG sites. We developed a cervical and an endometrial cancer DNAme signature by comparing normal and cancerous cervical and endometrial samples from the publicly available The Cancer Genome Atlas data and developed deviation scores to assess the potential of discriminating cancer from a control sample using a vaginal fluid DNAme signature. Results: More than 60% of variations in DNAme in our vaginal fluid cannot be explained by those clinical or technical factors that we were aware of. Both the cervical and the endometrial cancer DNAme signature resulted in receiver operating characteristic area under the curve between 0.75 and 0.83 to discriminate controls and the cancers for which the signature has been designed for. Conclusions: Whole DNAme signatures based on array technologies in body fluids are able to discriminate cancer cases from controls.
Content may be subject to copyright.
DNA Methylation Signatures in Vaginal Fluid Samples for
Detection of Cervical and Endometrial Cancer
Konstantinos Doufekas, MBChB,* Shijie Charles Zheng, BSc,Þþ Shohreh Ghazali, MSc,*
Michael Wong, MBChB,* Yasmin Mohamed, MBBS,* Allison Jones, BSc,* Daniel Reisel, MBPhD,*
Tim Mould, MD,* Adeola Olaitan, MD,* Nicola Macdonald, MD,* Andrew Erich Teschendorff, PhD,*Þ§
and Martin Widschwendter, MD*
Objective: A noninvasive tool that allows individuals to be monitored who are at risk of
developing a malignancy is an unmet need. Such a test would need to consist of a molecular
signature that allows for gradual judgment to assess the efficacy of preventive strategies.
Here we performed a proof-of-principle study to test whether a DNA methylation (DNAme)
signature in fluid collected from the vagina is able to identify women with cervical or
endometrial cancer.
Materials and Methods: DNA from vaginal fluid samples from 111 women (30, 8, 73
with endometrial cancer, cervical cancer, and benign gynecological conditions, respectively)
were analyzed for DNAme using the Illumina 450k DNA methylation bead array assay,
which allows the assessment of DNAme at more than 480.000 CpG sites. We developed a
cervical and an endometrial cancer DNAme signature by comparing normal and cancerous
cervical and endometrial samples from the publicly available The Cancer Genome Atlas data
and developed deviation scores to assess the potential of discriminating cancer from a
control sample using a vaginal fluid DNAme signature.
Results: More than 60% of variations in DNAme in our vaginal fluid cannot be explained
by those clinical or technical factors that we were aware of. Both the cervical and the en-
dometrial cancer DNAme signature resulted in receiver operating characteristic area under
the curve between 0.75 and 0.83 to discriminate controls and the cancers for which the
signature has been designed for.
Conclusions: Whole DNAme signatures based on array technologies in body fluids are
able to discriminate cancer cases from controls.
ORIGINAL STUDY
International Journal of Gynecological Cancer &Volume 00, Number 00, Month 2016 1
*Department of Women’s Cancer, Elizabeth Garrett Anderson In-
stitute for Women’s Health, University College London, London,
United Kingdom; CAS Key Laboratory of Computational Biology,
CAS-MPG Partner Institute for Computational Biology, Shanghai; and
University of Chinese Academy of Sciences, Beijing, China; and §UCL
Cancer Institute, University College London, London, United Kingdom.
Address correspondence and reprint requests to Martin
Widschwendter, MD, FRCOG, UCL Department of Women’s
Cancer, UCL EGA Institute for Women’s Health, University
College London, Medical School Bldg, Room 340,
74 Huntley St, London WC1E 6AU, United Kingdom.
E-mail: M.Widschwendter@ucl.ac.uk.
This work was funded by the Eve Appeal
(http://www.eveappeal.org.uk/) and a grant from the
UCLH/UCL Comprehensive Biomedical Research Centre
project, and the work has been undertaken at UCLH/UCL,
which received a proportion of its funding from the
Department of Health NIHR Biomedical Research Centres
funding scheme. This project has received funding
from the European Union’s Horizon 2020 research
and innovation program under grant agreement no. 634570.
The authors declare no conflicts of interest. Konstantinos Doufekas,
Shijie Charles Zheng, and Shohreh Ghazali contributed equally.
Supplemental digital content is available for this article. Direct URL
citation appears in the printed text and is provided in the HTML and
PDF versions of this article on the journal’sWeb site (www.ijgc.net).
Copyright *2016 by IGCS and ESGO
ISSN: 1048-891X
DOI: 10.1097/IGC.0000000000000739
Copyright © 2016 by IGCS and ESGO. Unauthorized reproduction of this article is prohibited.
Key Words: Epigenetics, DNA methylation, Endometrial cancer, Cervical cancer,
Vaginal swab
Received February 21, 2016, and in revised form March 14, 2016.
Accepted for publication March 21, 2016.
(Int J Gynecol Cancer 2016;00: 00Y00)
Whereas a well-established protocol for screening for
cervical cancer exists, no such strategy has been de-
veloped for endometrial cancer. A simple self-collection test
that identifies women with either cervical or endometrial
cancer would be required to specifically identify women at
risk of developing endometrial cancer and allow continuous
monitoring of preventive strategies.
During the past 1 to 2 decades, the involvement of
epigenetic changes in cancer detection and monitoring has
been described. One of the best characterized epigenetic
modifications occurring during carcinogenesis is de novo
methylation of CG dinucleotide (CpG) islands.
1
Whereas
each epithelial tumor contains about 10 to 15 tumor sup-
pressor genes silenced by mutations, several hundred genes
are affected and silenced by DNA methylation (DNAme).
2
The discovery of DNAme has opened up an exciting new
approach to the development of tools for the early detection
and personalized treatment of cancer.
3
Numerous reports have shown that methylation signa-
tures can be detected in virtually any body fluid (serum/plasma,
vaginal fluid, smears, nipple fluid aspirate, etc).
4
Testing for
cancer-specific DNAme in body fluids has a number of ad-
vantages compared with competing strategies
5
: (1) DNAme is
biologically and chemically stable and it is relatively impervi-
ous to fluctuations in physiological states and sample collection
conditions, and the DNA analyte is resistant to degradation both
in vivo and after obtaining the sample, which is particularly
important given transport delays to processing laboratories. (2)
Once a specific region has acquired methylation, this methyl-
ation pattern is conserved throughout disease progression; this
largely reduces the likelihood of false-positive tests. (3) DNA
methylation constitutes a positively detectable signal, as op-
posed to a loss of signal, as in the case of chromosomal deletions.
(4) Selection of gene promoter CpG island hypermethylation
offers advantages for assay design compared with genetic al-
terations that tend to be interspersed throughout a given gene.
Hypermethylation typically occurs over a focal area of a gene
early in carcinogenesis and greatly simplifies both the design
and interpretation of analytical tests. (5) Finally, current tech-
nologies allow the testing of several thousands of CpGs in a
given sample, which offers the opportunity to not only detect
but also obtain a comprehensive picture of the biology and
nature of the cancer.
Whereas we and others have demonstratedVusing
various collection modalities
6
including vaginal fluid,
7Y10
cervical
smear samples,
11
uterine lavages,
12
endometrial brushings,
13
or
endometrial biopsies
14
Vthat it is feasible to detect tumor-
specific alterations at specific well-defined genetic loci in
endometrial cancer (eg, in genes like HAND2, which are crucial
for disease development
15
), no evidence exists that demon-
strates that one or more cancers can be detected based on body
fluid analysis on platforms that allow simultaneous testing of
many thousands of features. More importantly, risk identifi-
cation and monitoring the efficacy of preventive strategies for
endometrial cancer (ie, Mirena coil or oral progesterone in
women with complex atypical hyperplasia or Lynch syndrome)
require a multifeature signature as opposed to a single-gene-
mutation ‘‘yes/no’’ test.
In a proof-of-principle study, we analyzed vaginal fluid
samples and demonstrate the feasibility and challenges of
methylation array technologies to separate cervical and en-
dometrial cancer cases from controls.
MATERIALS AND METHODS
Study Population
The Cancer Genome Atlas Set
We analyzed the available 3 normal and 306 cancerous
cervical and 34 normal and 374 cancerous endometrial
samples for which DNAme data based on the Illumina 450k
methylation array were available at The Cancer Genome Atlas
(TCGA) repository.
Vaginal Fluid Set
After written informed consent, vaginal swabs from
women who presented with cervical (n = 8) or endometrial
(n = 30) cancer (cases) or women who presented with benign
gynecological conditions (n = 73) (controls) at the University
College London Hospital were taken at presentation in the
outpatient clinic or before surgery. There was no significant
difference in mean age between the cases (mean age, 63.4 years)
and the controls (mean age, 57 years) (P= 0.064). The majority
of endometrial cancer cases were endometrioid endometrial
cancer cases (n = 23), and the remaining were carcinosarcoma
(n = 2) and serous cancers (n = 5), with 18 and 12 stage 1 and
stage 2 (or higher), respectively. In the cervical cancer group,
5 and 3 were squamous and adenocarcinoma, respectively, with
2 and 6 stage 1 and stage 2 (or higher), respectively. Controls
were referred to our hospital because of abnormal bleeding
(n = 21), urinary incontinence and/or prolapse (n = 20),
benign adnexal masses (n = 12), uterine fibroids (n = 11),
endometriosis (n = 3), or other benign gynecological con-
ditions (n = 6).
The study was approved by the Research Ethics
Committee No. REC 14/LO/.
Doufekas et al International Journal of Gynecological Cancer &Volume 00, Number 00, Month 2016
2*2016 IGCS and ESGO
Copyright © 2016 by IGCS and ESGO. Unauthorized reproduction of this article is prohibited.
DNA Methylation Analyses
Vaginal swabs (cells and fluid dissolved in phosphate
buffered saline) were extracted using Qiagen DNeasy blood
and tissue kit (69506). All DNA samples were then bisulfite
modified using the EZ DNA Methylation Kit D5008 (Zymo
Research, Orange, Calif ) according to the manufacturer’s
instructions. DNA was then processed following standard
procedures for the Illumina 450k DNA methylation bead
arrays.
16,17
Statistical Analyses
Development of an Endometrial Cancer
DNAme Signature
To derive an endometrial cancer signature, we used the
pamr package
18
on the TCGA UCEC data set. One half of the
normal samples and one half of the cancer samples were randomly
selected as a training data set and the rest as a test data set. We only
used the top 5% variable features to train the pamr model. The
selection process and the cross-validation results are shown in
Supplementary Figure 1, http://links.lww.com/IGC/A384.
Development of a Cervical Cancer DNAme
Signature
Because the TCGA set only contained 3 normal cervical
samples, we used limma
19
model to derive a cervical cancer
signature by comparing normal with cancer samples. The Pvalue
histogram is shown in Supplementary Figure 2, http://links.lww.
com/IGC/A384. Features, which passed a false discovery rate
threshold of 0.05, were ranked by absolute beta value difference
between cancer and normal in decreasing order. The top 500
features were selected as cervical cancer signature.
Processing of Illumina 450k Methylation Data
The vaginal fluid data set was read in with functions in
R minfi package. Beta values with detection Pvalues more
than 0.05 were set to NA (not available). The NA percentages
across samples and probes are shown in Supplementary
Figure 3, http://links.lww.com/IGC/A384. Probes with more
than 10% NA were removed for the subsequent analysis.
impute.knn function from impute package with k=10was
applied, and BMIQ was applied to adjust type 2 probe bias.
Singular Value Decomposition
After BMIQ processing, we applied singular value
decomposition (SVD) on row-centered beta value matrix. The
number of component K to be used in diagnostic SVD heat
map was determined with Random Matrix Theory.
20
The as-
sociation Pvalues were calculated by linear regression model
for continuous or ordinal variables and by Kruskal-Wallis rank
sum test for categorical variables.
Deviation Scores
To compute deviation scores of each signature, we first
define reference samples. We then compute, for each CpG
consisting of the signature, the mean A-value, K
c
, and standard
deviation, R
c
, across the reference samples. For any given
sample, the deviation score is computed as:
DeviationScores sðÞ¼1
n~
n
cwcAcsjKc
Rc
;
where w
c
is +1(j1) if the corresponding CpG, c, is hyper-
methylated (hypomethylated) in cancer samples. In this for-
mula, nis the total number of CpGs in the corresponding
cancer signature.
For both of EC and CC signatures, we randomly se-
lected 25 normal vaginal samples as reference normal to
compute deviation scores in 100 times. According to the re-
sults of 100 runs (not shown here), we found that deviation
scores of 2 normal groups (reference and nonreference) were
similar. Hence, in the following analysis, we used all normal
samples in the vaginal fluid data set as reference.
We generated a box plot of Pearson correlation co-
efficients based on all beta values of normal samples with
each other and identified 1 outlier (median, G0.9), which was
removed. Then deviation scores based on EC and CC sig-
natures were computed. Receiver operating characteristic
(ROC) and box plots were plotted.
RESULTS
First, we analyzed all the 111 samples together to assess
whether any of the known clinical (eg, normal-tumor status,
age, menopausal status, etc) or technical (eg, DNA concen-
tration, DNA purity, etc) factors could explain the variation in
DNAme that exists across all the samples (Supplementary Fig. 4,
http://links.lww.com/IGC/A384). Although age, normal-tumor
status, grade, and stage of disease demonstrated a trend toward
an association with the top component of this data set, more
than 60% of the variation in our set (Supplementary Fig. 4,
http://links.lww.com/IGC/A384) cannot be explained by those
clinical or technical factors that we were aware of.
To assess the accuracy of our DNAme data, we tested
whether a known and validated algorithm developed by
Horvath and that is based on a specific DNAme signature and
correlates extremely well with chronologic age
21
is also
strongly associated with age. We observed an extremely strong
correlation between the chronologic age of our volunteers and
the age that is predicted purely based on a DNAme signature in
their vaginal fluid sample (r
2
=0.77,PG10
j36
; Fig. 1), pro-
viding support for the high quality of our data.
Next, we wanted to confirm that at least some of the
variations observed in the principal component (PC) analysis
are caused by normal-cancer status. Because PC1 and PC2 did
not correlate with any phenotypes, we assessed PC3 based on
disease status. Both cervical and endometrial cancer cases
showed a substantial deviation from the controls (Fig. 2).
To assess whether a DNAme signature that is specific
for cervical cancer and specific for endometrial cancer would
pick up only cervical or only endometrial cancer based on
vaginal fluid analysis, we used the publicly available TCGA
data. Of note, there were DNAme data from only 3 normal
cervical samples available in this public repository but 306
cancerous samples, and hence we took the top 500 CpGs that
demonstrated the largest difference between the 3 normal and
International Journal of Gynecological Cancer &Volume 00, Number 00, Month 2016 Vaginal Fluid DNA Methylation Signatures
*2016 IGCS and ESGO 3
Copyright © 2016 by IGCS and ESGO. Unauthorized reproduction of this article is prohibited.
all cancer cases (Methods). As expected, the difference be-
tween the median deviation score of the controls and cervical
cancer vaginal fluid samples was highly significant (P=
0.001; Fig. 3A). Surprisingly, the magnitude of deviation
score difference between the controls and endometrial cancer
cases was the same (PG10
j6
) despite the fact that the sig-
nature was gained by comparing normal and cancerous cer-
vical samples. The resulting ROC area under the curves
(AUCs) were almost identical (ie, 0.83 and 0.80 for cervical
and endometrial cancers, respectively) (Fig. 3A). Finally, ap-
plying the TCGA-based endometrial cancer signature in the
vaginal fluid sample was almost equally effective compared
with the cervical cancer signature in discriminating both en-
dometrial (ROC AUC, 0.80) and cervical cancer cases (ROC
AUC, 0.75) from control volunteers with benign gynecological
conditions (Fig. 3B).
DISCUSSION
Here we provide evidence for the first time that DNAme
signatures in body fluids, based on a large number of inde-
pendent CpGs, are able to identify individuals with a specific
cancer. It was rather surprising that a 500-CpG signature that
was developed based on a comparison of 3 normal controls
with several hundred cancers was able to identify not only
those cancer entities that it was deemed to recognize (ie,
cervical cancer) but also a different cancer entity with an
entirely different histologic differentiation.
There is no immediate clinical relevance for our find-
ings because the current sensitivities and specificites for de-
tection of cervical cancerVcompared with alternative means of
detecting this cancerVare far too low. However, based on the
evidence we gathered within this cohort, we have acquired
robust indications as to how to improve algorithms that in fact
might be able to dramatically increase the accuracy of DNAme
signatureYbased cancer diagnostic tests: (1) The sample sets
that are used to develop a cancer signature have to contain a
sufficient number of normal and cancer tissues. (2) More than
60% of the variations in our data set could not be explained by
any of the data that we have collected or wereavailable from the
volunteers. We speculate that a variety of ‘‘contaminating’
DNA from a large variety of inflammatory cells, sperm cells,
and normal epithelial and stromal cells, which are present in
FIGURE 1. Horvath’s DNAme age of all vaginal fluid DNAme data. The chronological ages of volunteers are strongly
correlated with predicted Horvath’s predictor based on DNAme profiles (r
2
= 0.77, PG10
j36
).
FIGURE 2. Singular value decomposition PC3 of 3 disease statuses. The SVD PC3 values of cervical cancer and
endometrial cancer samples are significantly different from values of controls (PG1e-3).
Doufekas et al International Journal of Gynecological Cancer &Volume 00, Number 00, Month 2016
4*2016 IGCS and ESGO
Copyright © 2016 by IGCS and ESGO. Unauthorized reproduction of this article is prohibited.
vaginal fluid samples across both control and case volunteers,
are core factors for the relative low accuracy of our tests.
Identifying the source for these contaminating DNAme sig-
natures and extending some of the existing algorithms to ap-
plication in vaginal swab samples will be a priority for future
research. Specifically, we will develop DNAme signatures that
are specific for each subtype of potential contaminating cells
to calculate the proportion of these cells contributing to the
overall DNAme signal. Such tests based on vaginal fluid could
then form a solid basis to regularly monitor women who are at
high risk of developing endometrial cancer (ie, women with a
Lynch syndrome or morbidly obese women or very young
women with a complex atypical hyperplasia who do not want
to undergo a hysterectomy). The same strategy could also be
applied to assess and monitor the efficacy of preventive mea-
sures (ie, Mirena coil or oral progesterone).
Overall, we have provided the first evidence that entire
molecular signatures assessed by array technologies are
simple to obtain and vaginal fluid samples are a promising
tool to diagnose cancer. Using further adaptations to the
statistical algorithms and adjusting for background DNA
noise should facilitate the development of tools that allow risk
prediction and monitoring of preventive strategies.
REFERENCES
1. Jones PA, Baylin SB. The epigenomics of cancer. Cell.
2007;128:683Y692.
2. Widschwendter M, Jones PA. DNA methylation and breast
carcinogenesis. Oncogene. 2002;21:5462Y5482.
3. Laird PW. Principles and challenges of genome-wide DNA
methylation analysis. Nat Rev Genet. 2010;11:191Y203.
4. Wittenberger T, Sleigh S, Reisel D, et al. DNA methylation
markers for early detection of women’s cancer: promise and
challenges. Epigenomics. 2014;6:311Y327.
5. Jones A, Lechner M, Fourkala EO, et al. Emerging promise of
epigenetics and DNA methylation for the diagnosis and
management of women’s cancers. Epigenomics. 2010;2:9Y38.
6. Jiang SW, Li J, Podratz K, et al. Application of DNA methylation
biomarkers for endometrial cancer management. Expert Rev
Mol Diagn. 2008;8:607Y616.
7. Doufekas K, Hadwin R, Kandimalla R, et al. GALR1
methylation in vaginal swabs is highly accurate in identifying
women with endometrial cancer. Int J Gynecol Cancer.
2013;23:1050Y1055.
8. Fiegl H, Gattringer C, Widschwendter A, et al. Methylated DNA
collected by tamponsVa new tool to detect endometrial cancer.
Cancer Epidemiol Biomarkers Prev. 2004;13:882Y888.
9. Widschwendter A, Gattringer C, Ivarsson L, et al. Analysis of
aberrant DNA methylation and human papillomavirus DNA in
cervicovaginal specimens to detect invasive cervical cancer and
its precursors. Clin Cancer Res. 2004;10:3396Y3400.
10. Bakkum-Gamez JN, Wentzensen N, Maurer MJ, et al.
Detection of endometrial cancer via molecular analysis of
DNA collected with vaginal tampons. Gynecol Oncol. 2015;
137:14Y22.
11. Kinde I, Bettegowda C, Wang Y, et al. Evaluation of DNA from
the Papanicolaou test to detect ovarian and endometrial cancers.
Sci Transl Med. 2013;5:167ra4.
FIGURE 3. Cancer signature deviation scores. A, Cervical cancer (CC) signature deviation scores. Left panel:
Box plot of CC signature deviation scores across different disease statuses. One-tailed Wilcoxon rank sum test shows
that both of the median deviation scores in CC and endometrial cancer (EC) samples were significantly different,
with Pvalues of approximately 1e-3 and 1.37e-6, respectively, for CC and EC. Right panel: ROC plot of deviation
scores of CC and EC compared with control samples. The 2 ROC plots are almost identical, with resulting AUC values
of 0.825 and 0.796, respectively, for CC and EC. B, Endometrial cancer signature deviation scores. Left panel:
Box plot of EC signature deviation scores across different disease statuses. One-tailed Wilcoxon rank sum test shows
that both of the median deviation scores in CC and EC samples were significantly different, with Pvalues of ab 9.81e-3
and 7.96e-7, respectively, for CC and EC. Right panel: ROC plot of deviation scores of CC and EC compared with
control samples. The ROC plots indicate that EC signature is almost equally powerful as CC signature, with similar
resulting AUC values of 0.753 and 0.803 for CC and EC, respectively.
International Journal of Gynecological Cancer &Volume 00, Number 00, Month 2016 Vaginal Fluid DNA Methylation Signatures
*2016 IGCS and ESGO 5
Copyright © 2016 by IGCS and ESGO. Unauthorized reproduction of this article is prohibited.
12. Maritschnegg E, Wang Y, Pecha N, et al. Lavage of the uterine
cavity for molecular detection of mu¨llerian duct carcinomas: a
proof-of-concept study. J Clin Oncol. 2015;33:4293Y4300.
13. Wentzensen N, Bakkum-Gamez JN, Killian JK, et al. Discovery
and validation of methylation markers for endometrial cancer.
Int J Cancer. 2014;135:1860Y1868.
14. Lai HC, Wang YC, Yu MH, et al. DNA methylation as a
biomarker for the detection of hidden carcinoma in endometrial
atypical hyperplasia. Gynecol Oncol. 2014;135:552Y559.
15. Jones A, Teschendorff AE, Li Q, et al. Role of DNA methylation
and epigenetic silencing of HAND2 in endometrial cancer
development. PLoS Med. 2013;10:e1001551.
16. Teschendorff AE, Gao Y, Jones A, et al. DNA methylation
outliers in normal breast tissue identify field defects that are
enriched in cancer. Nat Commun. 2016;7:10478.
17. Sandoval J, Gao Y, Jones A, et al. Validation of a DNA
methylation microarray for 450,000 CpG sites in the human
genome. Epigenetics. 2011;6:692Y702.
18. Tibshirani R, Hastie T, Narasimhan B, et al. Diagnosis of
multiple cancer types by shrunken centroids of gene expression.
Proc Natl Acad Sci USA. 2002;99:6567Y6572.
19. Smyth GK. Linear models and empirical bayes methods for
assessing differential expression in microarray experiments.
Stat Appl Genet Mol Biol. 2004;3:Article3.
20. Teschendorff AE, Zhuang J, Widschwendter M. Independent
surrogate variable analysis to deconvolve confounding factors in
large-scale microarray profiling studies. Bioinformatics.
2011;27:1496Y1505.
21. Horvath S. DNA methylation age of human tissues and cell
types. Genome Biol. 2013;14:R115.
Doufekas et al International Journal of Gynecological Cancer &Volume 00, Number 00, Month 2016
6*2016 IGCS and ESGO
Copyright © 2016 by IGCS and ESGO. Unauthorized reproduction of this article is prohibited.
... 107 Other promising sampling methods include vaginal tampons: epigenomic analyses of DNA obtained with tampons or vaginal swabs have shown feasibility to detect endometrial cancer in few proof-of-concepts. 46,48,108,109 Tampons are noninvasive and well-accepted absorbent hygiene products; however, the use of tampons could be inadequate in screening settings among women without bleeding symptoms. The studies summarized in Table 2 provide evidence of the potential detection of disease using new biomarkers in minimally invasive sampling methods. ...
Article
Full-text available
Due to the anatomical continuity of the uterine cavity with the cervix, genomic exploitation of material from routine Pap smears and other non‐invasive sampling methods represent a unique opportunity to detect signs of disease using biological material shed from the upper genital tract. Recent research findings offer a promising perspective in the detection of endometrial cancer, but certain questions need to be addressed in order to accelerate the implementation of novel technologies in a routine screening or clinical setting. We discuss here new perspectives on detection of endometrial cancer using genomic and other biomarkers in minimally invasive sampling methods with a special focus on public health classic screening criteria, highlighting current gaps in knowledge. This article is protected by copyright. All rights reserved.
... Urine could be a source of tumor DNA for detecting prostate and bladder cancer, though hypermethylation of glutathione S-transferase P1 (GSTP1) or other gene panels (e.g., RASSF1A, APC, p14) do not have sufficient sensitivity in detecting early stage disease [117,118]. Vaginal fluid could be used to distinguish endometrial or cervical cancer patients from healthy controls using a DNA methylation signature composed of 500 CpGs, which seemed to reflect methylation patterns seen in tumor biopsies [119]. Finally, sputum or bronchial lavage fluid can be used to aid in the diagnosis of lung cancer. ...
Article
Full-text available
Aberrant epigenetic modifications are an early event in carcinogenesis, with the epigenetic landscape continuing to change during tumor progression and metastasis-these observations suggest that specific epigenetic modifications could be used as diagnostic and prognostic biomarkers for many cancer types. DNA methylation, post-translational histone modifications, and non-coding RNAs are all dysregulated in cancer and are detectable to various degrees in liquid biopsies such as sputum, urine, stool, and blood. Here, we will focus on the application of liquid biopsies, as opposed to tissue biopsies, because of their potential as non-invasive diagnostic tools and possible use in monitoring therapy response and progression to metastatic disease. This includes a discussion of septin-9 (SEPT9) DNA hypermethylation for detecting colorectal cancer, which is by far the most developed epigenetic biomarker assay. Despite their potential as prognostic and diagnostic biomarkers, technical issues such as inconsistent methodology between studies, overall low yield of epigenetic material in samples, and the need for improved histone and non-coding RNA purification methods are limiting the use of epigenetic biomarkers. Once these technical limitations are overcome, epigenetic biomarkers could be used to monitor cancer development, disease progression, therapeutic response, and recurrence across the entire cancer care continuum.
... Along the same line, Sun and coworkers [86] analyzed in cervical vaginal lavages methylation at 14 CpG sites within the HPV16 L1 region and noticed a significant increase in methylation in samples from women with CIN3+ compared to the HPV16 genomes from women without CIN3+, indicating that hyper/hypomethylation of viral CpG sites may constitute a potential biomarker for precancerous and cancerous cervix disease [86]. In a high-throughput experiment using the Illumina 450 k DNA methylation array, Doufekis et al. [87] investigated the DNA methylation in vaginal fluid samples at more than 480,000 CpG sites and found a DNA methylation signature for cervical and endometrial cancer which resulted in a ROC area under the curve between 0.75 and 0.83. ...
Article
Full-text available
Purpose: Despite improvement in vaccines against human papilloma virus (HPV), the causative agent of cervical cancer, screening women for cervical precancer will remain indispensable in the coming 30-40 years. A simple test that could be performed at home or at a doctor's practice and that informs the woman whether she is at risk would significantly help make a broader group of patients who aware that they need medical treatment. Cervical vaginal fluid (CVF) is a body fluid that is very well suited for such a test. Methods: Narrative review of cervical (pre)cancer candidate biomarkers from cervicovaginal fluid, is based on a detailed review of the literature. We will also discuss the possibilities that these biomarkers create for the development of a self-test or point-of-care test for cervical (pre)cancer. Results: Several DNA, DNA methylation, miRNA, and protein biomarkers were identified in the cervical vaginal fluid; however, not all of these biomarkers are suited for development of a simple diagnostic assay. Conclusions: Proteins, especially alpha-actinin-4, are most suited for development of a simple assay for cervical (pre)cancer. Accuracy of the test could further be improved by combination of several proteins or by combination with a new type of biomarker, e.g., originating from the cervicovaginal microbiome or metabolome.
... For instance, whole blood is composed of at least 7-8 main cell subtypes (neutrophils, eosinophils, basophils, CD14 + monocytes, CD4 + T cells, CD8 + T cells, CD19 + B cells and CD56 + natural killer cells). A major component of cervical smears is immune cell infiltrates [22]. Because DNAm is highly cell-type specific [11], variations in cell-type composition between phenotypes can therefore confound analyses. ...
Article
A major challenge faced by epigenome-wide association studies (EWAS) is cell-type heterogeneity. As many EWAS have already demonstrated, adjusting for changes in cell-type composition can be critical when analyzing and interpreting findings from such studies. Because of their importance, a great number of different statistical algorithms, which adjust for cell-type composition, have been proposed. Some of the methods are 'reference based' in that they require a priori defined reference DNA methylation profiles of cell types that are present in the tissue of interest, while other algorithms are 'reference free.' At present, however, it is unclear how best to adjust for cell-type heterogeneity, as this may also largely depend on the type of tissue and phenotype being considered. Here, we provide a critical review of the major existing algorithms for correcting cell-type composition in the context of Illumina Infinium Methylation Beadarrays, with the aim of providing useful recommendations to the EWAS community.
Chapter
Gynecological cancer is a grave threat to women’s health. The human papillomavirus (HPV) infection is the main risk of cervical cancer. The incidence and mortality of cervical cancer have gradually decreased since the clinical use of pap smears. However, the sensitivity of pap smears still varies with different screening equipment, which seriously affects the effectiveness of screening for cervical cancer. On the other hand, the etiology of endometrial cancer and ovarian cancer is not clear. Although the treatment response is good in the early stage, the late diagnosis results in poor treatment outcomes and high mortality. Therefore, developing biomarkers for early detection and effective screening methods of these three diseases is necessary and urgent. The DNA methylomic studies of these diseases can have the opportunity to discover genes related to diseases and serve as indicators of disease diagnosis and prognosis biomarkers.
Article
Full-text available
Endometrial cancer (EC) is the most common gynaecological cancer in the UK. Ninety percent of women with EC present with postmenopausal bleeding (PMB), but less than 10% of women with PMB have a sinister underlying cause. National Institute for Health and Care Excellence guidance advises that symptomatic postmenopausal women undergo urgent investigation; however, guidance is unclear for premenopausal women. Current investigations for PMB, including transvaginal ultrasound sonography, endometrial biopsy and/or outpatient hysteroscopy, have advantages and disadvantages. Novel detection tools are in development, which combine minimally invasive sampling with genomic, proteomic and single cell technologies. To understand who is at risk of EC and who should be referred for urgent investigations. To understand the evidence underpinning the current diagnostic pathway for EC. To highlight unique and promising perspectives for EC detection and their potential to transform clinical care. Current diagnostics for EC are invasive and often painful. There is an urgent need for high‐quality randomised controlled trials to inform effective pain‐relief options. Premenopausal women with suspected EC do not fit criteria for urgent investigations. How can we identify those at highest risk to ensure they are fast‐tracked appropriately? Novel diagnostic tools hold promise, but they must be robustly validated before being introduced into clinical practice.
Article
Full-text available
Identifying molecular alterations in normal tissue adjacent to cancer is important for understanding cancer aetiology and designing preventive measures. Here we analyse the DNA methylome of 569 breast tissue samples, including 50 from cancer-free women and 84 from matched normal cancer pairs. We use statistical algorithms for dissecting intra- and inter-sample cellular heterogeneity and demonstrate that normal tissue adjacent to breast cancer is characterized by tens to thousands of epigenetic alterations. We show that their genomic distribution is non-random, being strongly enriched for binding sites of transcription factors specifying chromatin architecture. We validate the field defects in an independent cohort and demonstrate that over 30% of the alterations exhibit increased enrichment within matched cancer samples. Breast cancers highly enriched for epigenetic field defects, exhibit adverse clinical outcome. Our data support a model where clonal epigenetic reprogramming towards reduced differentiation in normal tissue is an important step in breast carcinogenesis.
Article
Full-text available
Purpose: Type II ovarian cancer (OC) and endometrial cancer (EC) are generally diagnosed at an advanced stage, translating into a poor survival rate. There is increasing evidence that Müllerian duct cancers may exfoliate cells. We have established an approach for lavage of the uterine cavity to detect shed cancer cells. Patients and methods: Lavage of the uterine cavity was used to obtain samples from 65 patients, including 30 with OC, five with EC, three with other malignancies, and 27 with benign lesions involving gynecologic organs. These samples, as well as corresponding tumor tissue, were examined for the presence of somatic mutations using massively parallel sequencing (next-generation sequencing) and, in a subset, singleplex analysis. Results: The lavage technique could be applied successfully, and sufficient amounts of DNA were obtained in all patients. Mutations, mainly in TP53, were identified in 18 (60%) of 30 lavage samples of patients with OC using next-generation sequencing. Singleplex analysis of mutations previously determined in corresponding tumor tissue led to further identification of six patients. Taken together, in 24 (80%) of 30 patients with OC, specific mutations could be identified. This also included one patient with occult OC. All five analyzed lavage specimens from patients with EC harbored mutations. Eight (29.6%) of 27 patients with benign lesions tested positive for mutations, six (75%) as a result of mutations in the KRAS gene. Conclusion: This study proved that tumor cells from ovarian neoplasms are shed and can be collected via lavage of the uterine cavity. Detection of OC and EC and even clinically occult OC was achieved, making it a potential tool of significant promise for early diagnosis.
Article
Full-text available
Breast, ovarian and endometrial cancers cause significant morbidity and mortality. Despite the presence of existing screening, diagnostic and treatment modalities, they continue to pose considerable unsolved challenges. Overdiagnosis is a growing problem in breast cancer screening and neither screening nor early diagnosis of ovarian or endometrial cancer is currently possible. Moreover, treatment of the diversity of these cancers presenting in the clinic is not sufficiently personalized at present. Recent technological advances, including reduced representation bisulfite sequencing, methylation arrays, digital PCR, next-generation sequencing and advanced statistical data analysis, enable the analysis of methylation patterns in cell-free tumor DNA in serum/plasma. Ongoing work is bringing these methods together for the analysis of samples from large clinical trials, which have been collected well in advance of cancer diagnosis. These efforts pave the way for the development of a noninvasive method that would enable us to overcome existing challenges to personalized medicine.
Article
Full-text available
Background: Endometrial cancer incidence is continuing to rise in the wake of the current ageing and obesity epidemics. Much of the risk for endometrial cancer development is influenced by the environment and lifestyle. Accumulating evidence suggests that the epigenome serves as the interface between the genome and the environment and that hypermethylation of stem cell polycomb group target genes is an epigenetic hallmark of cancer. The objective of this study was to determine the functional role of epigenetic factors in endometrial cancer development. Methods and findings: Epigenome-wide methylation analysis of >27,000 CpG sites in endometrial cancer tissue samples (n = 64) and control samples (n = 23) revealed that HAND2 (a gene encoding a transcription factor expressed in the endometrial stroma) is one of the most commonly hypermethylated and silenced genes in endometrial cancer. A novel integrative epigenome-transcriptome-interactome analysis further revealed that HAND2 is the hub of the most highly ranked differential methylation hotspot in endometrial cancer. These findings were validated using candidate gene methylation analysis in multiple clinical sample sets of tissue samples from a total of 272 additional women. Increased HAND2 methylation was a feature of premalignant endometrial lesions and was seen to parallel a decrease in RNA and protein levels. Furthermore, women with high endometrial HAND2 methylation in their premalignant lesions were less likely to respond to progesterone treatment. HAND2 methylation analysis of endometrial secretions collected using high vaginal swabs taken from women with postmenopausal bleeding specifically identified those patients with early stage endometrial cancer with both high sensitivity and high specificity (receiver operating characteristics area under the curve = 0.91 for stage 1A and 0.97 for higher than stage 1A). Finally, mice harbouring a Hand2 knock-out specifically in their endometrium were shown to develop precancerous endometrial lesions with increasing age, and these lesions also demonstrated a lack of PTEN expression. Conclusions: HAND2 methylation is a common and crucial molecular alteration in endometrial cancer that could potentially be employed as a biomarker for early detection of endometrial cancer and as a predictor of treatment response. The true clinical utility of HAND2 DNA methylation, however, requires further validation in prospective studies. Please see later in the article for the Editors' Summary.
Article
We demonstrate the feasibility of detecting EC by combining minimally-invasive specimen collection techniques with sensitive molecular testing. Prior to hysterectomy for EC or benign indications, women collected vaginal pool samples with intravaginal tampons and underwent endometrial brushing. Specimens underwent pyrosequencing for DNA methylation of genes reported to be hypermethylated in gynecologic cancers and recently identified markers discovered by profiling over 200 ECs. Methylation was evaluated individually across CpGs and averaged across genes. Differences between EC and benign endometrium (BE) were assessed using two-sample t-tests and area under the curve (AUC). Thirty-eight ECs and 28 BEs were included. We evaluated 97 CpGs within 12 genes, including previously reported markers (RASSF1, HSP2A, HOXA9, CDH13, HAAO, and GTF2A1) and those identified in discovery work (ASCL2, HTR1B, NPY, HS3ST2, MME, ADCYAP1, and additional CDH13 CpG sites). Mean methylation was higher in tampon specimens from EC v. BE for 9 of 12 genes (ADCYAP1, ASCL2, CDH13, HS3ST2, HTR1B, MME, HAAO, HOXA9, and RASSF1) (all p<0.05). Among these genes, relative hypermethylation was observed in EC v. BE across CpGs. Endometrial brush and tampon results were similar. Within tampon specimens, AUC was highest for HTR1B (0.82), RASSF1 (0.75), and HOXA9 (0.74). This is the first report of HOXA9 hypermethylation in EC. DNA hypermethylation in EC tissues can also be identified in vaginal pool DNA collected via intravaginal tampon. Identification of additional EC biomarkers and refined collection methods are needed to develop an early detection tool for EC. Copyright © 2015. Published by Elsevier Inc.
Article
The prognosis of endometrial cancer is strongly associated with stage at diagnosis, suggesting that early detection may reduce mortality. Women who are diagnosed with endometrial carcinoma often have a lengthy history of vaginal bleeding, which offers an opportunity for early diagnosis and curative treatment. We performed DNA methylation profiling on population-based endometrial cancers to identify early detection biomarkers and replicated top candidates in two independent studies. We compared DNA methylation values of 1500 probes representing 807 genes in 148 population-based endometrial carcinoma samples and 23 benign endometrial tissues. Markers were replicated in another set of 69 carcinomas and 40 benign tissues profiled on the same platform. Further replication was conducted in The Cancer Genome Atlas and in prospectively collected endometrial brushings from women with and without endometrial carcinomas. We identified 114 CpG sites showing methylation differences with p-values of ≤10(-7) between endometrial carcinoma and normal endometrium. Eight genes (ADCYAP1, ASCL2, HS3ST2, HTR1B, MME, NPY, and SOX1) were selected for further replication. Age-adjusted odds ratios for endometrial cancer ranged from 3.44 (95%-CI: 1.33-8.91) for ASCL2 to 18.61 (95%-CI: 5.50-62.97) for HTR1B. An area under the curve (AUC) of 0.93 was achieved for discriminating carcinoma from benign endometrium. Replication in The Cancer Genome Atlas and in endometrial brushings from an independent study confirmed the candidate markers. This study demonstrates that methylation markers may be used to evaluate women with abnormal vaginal bleeding to distinguish women with endometrial carcinoma from the majority of women without malignancy. © 2014 Wiley Periodicals, Inc.