Genome-wide DNA Methylation
Profiling of Cell-Free Serum DNA
in Esophageal Adenocarcinoma
and Barrett Esophagus1,2
Rihong Zhai*, Yang Zhao*, Li Su*, Lauren Cassidy*,
Geoffrey Liu†and David C. Christiani*,‡
*Environmental and Occupational Medicine and
Epidemiology Program, Department of Environmental
Health, Harvard School of Public Health, Boston, MA, USA;
†Medical Oncology and Haematology, Department of
Medicine, Princess Margaret Hospital/Ontario Cancer
Institute, University of Toronto, Ontario, Canada;
‡Department of Medicine, Massachusetts General Hospital,
Boston, MA, USA
Aberrant DNA methylation (DNAm) is a feature of most types of cancers. Genome-wide DNAm profiling has been
performed successfully on tumor tissue DNA samples. However, the invasive procedure limits the utility of tumor
tissue for epidemiological studies. While recent data indicate that cell-free circulating DNAm (cfDNAm) profiles
reflect DNAm status in corresponding tumor tissues, no studies have examined the association of cfDNAm with
cancer or precursors on a genome-wide scale. The objective of this pilot study was to evaluate the putative signif-
icance of genome-wide cfDNAm profiles in esophageal adenocarcinoma (EA) and Barrett esophagus (BE, EA pre-
cursor). We performed genome-wide DNAm profiling in EA tissue DNA (n = 8) and matched serum DNA (n = 8), in
serum DNA of BE (n = 10), and in healthy controls (n = 10) using the Infinium HumanMethylation27 BeadChip that
covers 27,578 CpG loci in 14,495 genes. We found that cfDNAm profiles were highly correlated to DNAm profiles in
matched tumor tissue DNA (r = 0.92) in patients with EA. We selected the most differentially methylated loci to
perform hierarchical clustering analysis. We found that 911 loci can discriminate perfectly between EA and control
samples, 554 loci can separate EA from BE samples, and 46 loci can distinguish BE from control samples. These
results suggest that genome-wide cfDNAm profiles are highly consistent with DNAm profiles detected in
corresponding tumor tissues. Differential cfDNAm profiling may be a useful approach for the noninvasive screening
of EA and EA premalignant lesions.
Neoplasia (2012) 14, 29–33
DNA methylation (DNAm) is a fundamental epigenetic modifica-
tion in which a methyl group is added to the carbon-5 position of
the cytosine ring within the CpG dinucleotide . Extensive research
has revealed that DNAm is widely implicated in all crucial changes in
cancer cells, such as tumor-suppressor gene silencing, oncogene acti-
vation, and defective DNA repair [2,3]. Indeed, aberrant DNAm has
been detected in a variety of cancers, including esophagus, colon,
breast, liver, kidney, and lung [4–10]. To date, however, most stud-
ies of cancer DNAm have investigated DNAm in tissue-extracted
DNA. The invasive procedure and the likely existence of tissue het-
erogeneity limit the utility of tissue DNA for epidemiologic studies.
Therefore, it is desirable to develop less invasive and more accessible
Abbreviations: EA, esophageal adenocarcinoma; DNAm, DNA methylation; cfDNA,
cell-free circulating DNA
Address all correspondence to: David C. Christiani, MD, MPH, MS, or Rihong Zhai,
MD, PhD, Department of Environmental Health, Harvard School of Public Health,
665 Huntington Ave, FXB109, Boston, MA 02115. E-mail: firstname.lastname@example.org,
1This study was funded by National Institutes of Health grant RO1CA109193,
RO3CA110822, RO1CA074386, P50CA090578, and ES00002 and by the Flight
Attendant Medical Research Institute Award YCSA-062459. All authors disclosed
no conflict of interest; the institutes that provided the grant supports did not partic-
ipate in the study design, the collection, analysis, or interpretation of data.
2This article refers to supplementary materials, which is designated by Figure W1 and
is available online at www.neoplasia.com.
Received 21 November 2011; Revised 10 January 2012; Accepted 11 January 2012
Copyright © 2012 Neoplasia Press, Inc. All rights reserved 1522-8002/12/$25.00
Volume 14 Number 1January 2012pp. 29–33
approaches that can substitute or complement tissue DNA for
The presence of cell-free DNA (cfDNA) in the plasma/serum in
healthy individuals and, in higher amounts, in cancer patients was
demonstrated three decades ago [11,12]. However, it is only recently
possible to use cfDNA as a marker for cancer diagnosis or progression
[13,14]. In cancer patients, cfDNA carries the same mutations (K-ras,
N-ras, p53) as those found in corresponding tumor tissues [15,16].
CfDNA also carries other features of the primary tumor, including
microsatellite instabilities, loss of heterozygosity, and epigenetic
[13,17,18]. Interestingly, DNAm patterns detected in cfDNA are
in high concordance with patterns observed in corresponding primary
tumor tissues [19,20]. Indeed, individual cfDNA methylation
(cfDNAm) markers have been linked to different types of cancer, in-
cluding esophageal adenocarcinoma (EA) [21–27]. However, there
has been relatively little attention on the association of cfDNAm with
cancer precursor; no studies have examined cfDNAm profiles in rela-
tion to cancer precursor and cancers on a genome-wide scale. The
objective of this study was to compare the concordance of genome-
wide DNAm profiles between tumor tissues and sera and to assess the
performance of genome-wide cfDNAm profiles in differentiating EA,
EA precursor (Barrett esophagus, BE), and control.
Materials and Methods
Genomic DNA Extraction from Tissue and Serum
Matched serum and tissue samples were obtained from eight ran-
domly selected EA patients. Additional serum samples were collected
from 10 BE patients and 10 healthy controls [28,29]. EA and BE were
incident cases of histologically confirmed patients, and all subjects were
recruited from Massachusetts General Hospital (Boston, MA). EA is
defined as a tumor center located at or above the gastroesophageal
junction and had at least two-thirds of the bulk tumor located in the
esophagus; and BE is defined as pathologically confirmed intestinal
metaplasia . Controls were among healthy friends and non–
blood-related family members of hospitalized patients. Controls were
recruited at the same institutions in the same period as EA cases
. Pathologist-identified tissue regions that have more than 70%
tumor cells without definite evidence of necrosis are considered as
tumor tissues and used for DNA extractions. Peripheral venous blood
sample was drawn for eachsubject, andtheserum sample wasseparated
within 2 hours. The serum was isolated by centrifugation at 2000 rpm
for 10 minutes at 40°C and stored at −80°C until analysis. CfDNA
was extracted from 800-μl aliquots of serum using the Maxwell 16
blood DNA kit (Promega, Madison, WI). Tissue DNA was isolated
from 100 mg of EA tumor tissue by Maxwell 16 tissue DNA purifica-
Massachusetts General Hospital and Harvard School of Public Health
Genome-wide DNA Methylation Analysis
We used the Illumina Infinium HumanMethylation27 BeadChip
(Illumina, San Diego, CA) to analyze DNAm profiles. The BeadChip
contains 27,578 highly informative CpG loci covering more than
14,495 genes . DNA samples were bisulfite converted, then
whole-genome amplified (WGA), enzymatically fragmented, and
hybridized to the array. The assay was performed according to the
manufacturer’s instructions and was done at the BioMedical Genomics
Center, University of Minnesota.
After scanning of the BeadChip, data files were managed using the
Illumina BeadStudio software Methylation module. Each CpG site
on the BeadChip is represented by two bead types representing the
methylated (M) and unmethylated (U) state at that site. The methyla-
tion value for each CpG locus is expressed as a β value, representing
a continuous measurement from 0 (unmethylated) to 1 (completely
methylated) according to the following calculation: β value = (signal
intensity of M probe) / [(signal intensity of M + U probes) + 100];
the average β values is based on the average intensity of all U and
M CpG probes for a given locus.
Average β values were analyzed without normalization as recom-
mended by Illumina. The differences of DNAm levels (β value) be-
tween groups were analyzed using Student’s t test. Correlations of
mean DNAm levels of 27,578 CpG loci between tissue and serum
DNA samples were analyzed using the Pearson test. We selected the
most differentially methylated loci (a β value difference of >0.23 
and a P ≤ .0000006 between groups) to run hierarchical clustering
analysis. The choice of 0.23 as a criterion for a difference in β was
based largely on replicate experiments by Illumina that showed that
the HumanMethylation27 BeadChip could reliably detect a differ-
ence in β < 0.20 with a less than 1% false-positive rate . Un-
supervised hierarchical clustering analysis was carried out using the
R software program.
AverageDNAyields (range) were446.6ng(350.0-627.7ng),426.5 ng
(268.1-784.8 ng), 154.8 ng (145.1-172.2 ng), and 156.4 ng (128.2-
182.1 ng) for EA sera, matched EA tissue, BE sera, and control sera
samples, respectively. The OD 260/280 ratio, a measure of DNA
purity with respect to protein contamination, ranged from 1.32 to
1.80. Comparison of the number of CpG sites that could be success-
fully interrogated (as determined by BeadStudio software) revealed that
the call rates in both tissue DNA samples and sera DNA samples were
Figure 1. Correlation of DNAm levels (β values) between serum
DNA samples and matched tissue DNA samples (r = 0.92).
Genome-wide DNAm Profiling of cfDNAZhai et al.Neoplasia Vol. 14, No. 1, 2012
allgreater than 99.9%.Comparison ofβvaluedistribution between EA
sera DNA and matched tissue DNA showed similar patterns with high
peaks of hypomethylated loci and low peaks of the hypermethylated
loci (Figure W1). To investigate whether the methylation levels of sera
DNA were consistent with tissue DNA samples in EA, we carried out a
direct comparison of DNAm levels (β values) of 27,578 loci between
tissue DNA and matched serum DNA. Figure 1 shows that highly
correlated (r = 0.92) results can be achieved for matched tissue and
serum DNA, suggesting that genome-wide DNAm profiles in cfDNA
reflect DNAm profiles in tumor tissue DNA.
Clustering analyses showed that 911 loci perfectly discriminated
between EA and control samples, 544loci separated EAfrom BE sam-
ples, and 46 loci distinguished BE from control samples (Figure 2).
The large number of differentiated DNAm markers is consistent with
gene expression studies [33,34].
Genome-wide DNAm profiling has been performed successfully on
DNA recovered from several different types of tumor tissues includ-
ing ovarian , breast [31,36], prostate , and parathyroid 
cancers on the HumanMethylation27 BeadChip. Despite the robust
methylation profiling results from tumor tissues, little information
exists regarding the methylation analysis of cfDNA samples using
high-density methylation arrays. To our knowledge, this study is the
first to evaluate cfDNAm profiling on a genome-wide scale using the
Humanmethylation27 platform. Our data, although exploratory, sug-
gest that cfDNA obtained from serum can produce excellent DNAm
profiling on a genome-wide scale and may serve as a useful tool to
develop DNAm-based biomarkers for clinical application.
This study revealed several interesting observations. First, we ob-
served a high degree of concordance in DNAm profiles between
cfDNA and EA tumor tissue DNA on a genome-wide scale. This
proof-of-principle study suggests that a comprehensive analysis of
cfDNAm profiles has the potential to reflect genome-wide DNAm
alterations in primary tumor tissues. Second, we found that differen-
tial cfDNAm profiles can distinguish EA and BE from controls, as
well as EA from BE. The present results suggest that cfDNAm pro-
files may be a valuable biomarker for early detection of EA. Third, we
also observed a trend of increasing numbers of aberrant cfDNAm
markers with controls, BE, and EA (Figure 2). Our results are in
agreement with previous reports in which it was found that aberrant
tissues’ DNAm levels accumulated gradually with the histologic
changes from normal to BE, then from BE to esophageal carcinoma
[39–42]. These data suggest that cfDNAm profiles may reflect the in-
creasing involvement of aberrant DNAm in the process of normal-BE-
carcinoma sequence. However, because this is not a longitudinal study,
we cannot distinguish whether the abnormal cfDNAm markers seen
in the preneoplastic sera develop concurrently with adenocarcinoma,
or whether theaberrant cfDNAmmarkers inBE represent a predispos-
ing event that give rise to adenocarcinoma. Nevertheless, cfDNAm
profiles that might identify patients with cancer or at elevated risk
for developing cancer would have the potential to provide an opportu-
nity for early intervention. Furthermore, a benefit of using cfDNAm
marker is that serum can be easily obtained when biopsies are not avail-
able or when the exact position of the primary lesion is not clear. A
prospective longitudinal study should help reveal whether these aber-
rant cfDNAm profiles in normal or BE subjects are predictive of EA.
Our results should be viewed with caution. First, for the Human-
Methylation27 BeadChip, Illumina recommends a starting DNA
material of 500 ng or higher to achieve better performance. In this
study, the average DNA quantity was less than 500 ng and the overall
methylation marker completion rate was less than 100%, suggesting
that the starting DNA quantity of more than 500 ng is thus necessary
to obtain more robust methylation profiling results in future studies.
Moreover, the exploratory nature of this study is such that we are
unable to conduct sensitivity analysis and identify DNAm markers
number of tissue and serum samples from patients with EA. More works
in larger sample sizes are required to assess the roles of genome-wide
cfDNAm profiles in cancer diagnosis. Furthermore, cfDNAm profiling
MethyLight and MethDet platforms [43,44].
In conclusion, our data suggest that whole-genome amplified DNA
derived from stored serum have high potential to be used for genome-
wide methylation profiling. Genome-wide cfDNAm profiles may be a
useful noninvasive biomarker of EA and EA precursor lesions. Further
Figure 2. Hierarchical clustering of methylation values (β) from samples indicates that samples can be separated into distinct groups by
cfDNAm profiles. Columns represent samples. Rows represent CpG loci. Color represents methylation level (β) from 0 to 1 as per color
bar (red indicates low methylation level; yellow, high methylation level).
Neoplasia Vol. 14, No. 1, 2012Genome-wide DNAm Profiling of cfDNAZhai et al.
studies on cfDNAm are needed to compare the reproducibility of the
DNAm results obtained with different DNAm platforms and larger
numbers of samples. In addition, comparison of DNAm performance
between cfDNA obtained from fresh serum and DNA obtained from
archived serum samples stored in different environmental conditions
The authors thank the patients for their cooperation and participa-
tion in this study. We also thank Andrea Shafer and Salvatore Mucci
for data collection, entry, and management.
 Jones PA and Baylin SB (2007). The epigenomics of cancer. Cell 128, 683–692.
 Selaru FM, David S, Meltzer SJ, and Hamilton JP (2009). Epigenetic events in
gastrointestinal cancer. Am J Gastroenterol 104, 1910–1912.
 Esteller M (2008). Epigenetics in cancer. N Engl J Med 358, 1148–1159.
 Jin Z, Cheng Y, Gu W, Zheng Y, Sato F, Mori Y, Olaru AV, Paun BC, Yang J,
Kan T, et al. (2009). A multicenter, double-blinded validation study of methyl-
ation biomarkers for progression prediction in Barrett’s esophagus. Cancer Res
 Schulmann K, Sterian A, Berki A, Yin J, Sato F, Xu Y, Olaru A, Wang S, Mori Y,
Deacu E, et al. (2005). Inactivation of p16, RUNX3, and HPP1 occurs early
in Barrett’s-associated neoplastic progression and predicts progression risk.
Oncogene 24, 4138–4148.
 Wang JS, GuoM, Montgomery EA, ThompsonRE, Cosby H,HicksL, Wang S,
Herman JG, and Canto MI (2009). DNA promoter hypermethylation of p16
and APC predicts neoplastic progression in Barrett’s esophagus. Am J Gastroenterol
 Arai E, Ushijima S, Fujimoto H, Hosoda F, Shibata T, Kondo T, Yokoi S,
Imoto I, Inazawa J, Hirohashi S, et al. (2009). Genome-wide DNA methylation
profiles in both precancerous conditions and clear cell renal cell carcinomas are
correlated with malignant potential and patient outcome. Carcinogenesis 30,
 Arai E, Ushijima S, Gotoh M, Ojima H, Kosuge T, Hosoda F, Shibata T,
Kondo T, Yokoi S, Imoto I, et al. (2009). Genome-wide DNA methylation
profiles in liver tissue at the precancerous stage and in hepatocellular carcinoma.
Int J Cancer 125, 2854–2862.
 Belinsky SA, Klinge DM, Dekker JD, Smith MW, Bocklage TJ, Gilliland FD,
Crowell RE, Karp DD, Stidley CA, and Picchi MA (2005). Gene promoter
methylation in plasma and sputum increases with lung cancer risk. Clin Cancer
Res 11, 6505–6511.
 Hoque MO, Begum S, Topaloglu O, Chatterjee A, Rosenbaum E, van
Criekinge W, Westra WH, Schoenberg M, Zahurak M, Goodman SN, et al.
(2006). Quantitation of promoter methylation of multiple genes in urine DNA
and bladder cancer detection. J Natl Cancer Inst 98, 996–1004.
 Shapiro B, Chakrabarty M, Cohn EM, and Leon SA (1983). Determination of
circulating DNA levels in patients with benign or malignant gastrointestinal
disease. Cancer 51, 2116–2120.
 Leon SA, Shapiro B, Sklaroff DM, and Yaros MJ (1977). Free DNA in the
serum of cancer patients and the effect of therapy. Cancer Res 37, 646–650.
 Levenson VV (2010). DNA methylation as a universal biomarker. Expert Rev
Mol Diagn 10, 481–488.
 Pinzani P, Salvianti F, Pazzagli M, and Orlando C (2010). Circulating nucleic
acids in cancer and pregnancy. Methods 50, 302–307.
 Anker P, Lefort F, Vasioukhin V, Lyautey J, Lederrey C, Chen XO, Stroun M,
Mulcahy HE, and Farthing MJ (1997). K-ras mutations are found in DNA
extracted from the plasma of patients with colorectal cancer. Gastroenterology
 Mulcahy HE, Lyautey J, Lederrey C, Chen XQ, Lefort F, Vasioukhin V, Anker P,
Alstead EM, Farthing M, and Stroun M (2000). Plasma DNA K-ras mutations in
patients with gastrointestinal malignancies. Ann N Y Acad Sci 906, 25–28.
 Beau-Faller M, Gaub MP, Schneider A, Ducrocq X, Massard G, Gasser B,
Chenard MP, Kessler R, Anker P, Stroun M, et al. (2003). Plasma DNA
microsatellite panel as sensitive and tumor-specific marker in lung cancer
patients. Int J Cancer 105, 361–370.
 Sanchez-Cespedes M, Monzo M, Rosell R, Pifarre A, Calvo R, Lopez-Cabrerizo
MP, and Astudillo J (1998). Detection of chromosome 3p alterations in serum
DNA of non–small-cell lung cancer patients. Ann Oncol 9, 113–116.
 Board RE, Knight L, Greystoke A, Blackhall FH, Hughes A, Dive C, and
Ranson M (2008). DNA methylation in circulating tumour DNA as a bio-
marker for cancer. Biomark Insights 2, 307–319.
 Gormally E, Caboux E, Vineis P, and Hainaut P (2007). Circulating free DNA
in plasma or serum as biomarker of carcinogenesis: practical aspects and biolog-
ical significance. Mutat Res 635, 105–117.
 Liggett TE, Melnikov A, Yi Q, Replogle C, Hu W, Rotmensch J, Kamat A,
Sood AK, and Levenson V (2011). Distinctive DNA methylation patterns of
cell-free plasma DNA in women with malignant ovarian tumors. Gynecol Oncol
 Liggett TE, Melnikov AA, Marks JR, and Levenson VV (2011). Methylation
patterns in cell-free plasma DNA reflect removal of the primary tumor and drug
treatment of breast cancer patients. Int J Cancer 128, 492–499.
 Lee BB, Lee EJ, Jung EH, Chun HK, Chang DK, Song SY, Park J, and Kim
DH (2009). Aberrant methylation of APC, MGMT, RASSF2A, and Wif-1 genes
in plasma as a biomarker for early detection of colorectal cancer. Clin Cancer Res
 Melnikov AA, Scholtens D, Talamonti MS, Bentrem DJ, and Levenson VV
(2009). Methylation profile of circulating plasma DNA in patients with pancre-
atic cancer. J Surg Oncol 99, 119–122.
 Iyer P, Zekri AR, Hung CW, Schiefelbein E, Ismail K, Hablas A, Seifeldin IA, and
DNA of Egyptian hepatocellular carcinoma patients. Exp Mol Pathol 88, 107–111.
 Kawakami K, Brabender J, Lord RV, Groshen S, Greenwald BD, Krasna MJ,
Yin J, Fleisher AS, Abraham JM, Beer DG, et al. (2000). Hypermethylated APC
DNA in plasma and prognosis of patients with esophageal adenocarcinoma.
J Natl Cancer Inst 92, 1805–1811.
 Usadel H, Brabender J, Danenberg KD, Jeronimo C, Harden S, Engles J,
Danenberg PV, Yang S, and Sidransky D (2002). Quantitative adenomatous
polyposis coli promoter methylation analysis in tumor tissue, serum, and plasma
DNA of patients with lung cancer. Cancer Res 62, 371–375.
 Bradbury PA, Zhai R, Hopkins J, Kulke MH, Heist RS, Singh S, Zhou W, Ma C,
Xu W, Asomaning K, et al. (2009). Matrix metalloproteinase 1, 3 and 12 poly-
morphisms and esophageal adenocarcinoma risk and prognosis. Carcinogenesis 30,
 Zhai R, Chen F, Liu G, Su L, Kulke MH, Asomaning K, Lin X, Heist RS,
Nishioka NS, Sheu CC, et al. (2010). Interactions among genetic variants in
apoptosis pathway genes, reflux symptoms, body mass index, and smoking indi-
cate two distinct etiologic patterns of esophageal adenocarcinoma. J Clin Oncol
 Ye W, Held M, Lagergren J, Engstrand L, Blot WJ, Mclaughlin JK, and Nyren O
(2004). Helicobacter pylori infection and gastric atrophy: risk of adenocarcinoma
and squamous-cell carcinoma of the esophagus and adenocarcinoma of the gastric
cardia. J Natl Cancer Inst 96, 388–396.
 Thirlwell C, Eymard M, Feber A, Teschendorff A, Pearce K, Lechner M,
Widschwendter M, and Beck S (2010). Genome-wide DNA methylation analysis
of archival formalin-fixed paraffin-embedded tissue using the Illumina Infinium
HumanMethylation27 BeadChip. Methods 52, 248–254.
 Lleras RA, Adrien LR, Smith RV, Brown B, Jivraj N, Keller C, Sarta C,
Schlecht NF, Harris TM, Childs G, et al. (2011). Hypermethylation of a cluster
of kruppel-type zinc finger protein genes on chromosome 19q13 in oropharyngeal
squamous cell carcinoma. Am J Pathol 178, 1965–1974.
 Barrett MT, Yeung KY, Ruzzo WL, Hsu L, Blount PL, Sullivan R, Zarbl H,
Delrow J, Rabinvitch PS, and Reid BJ (2002). Transcriptional analyses of
Barrett’s metaplasia and normal upper GI mucosae. Neoplasia 4, 121–128.
 van Baal JW, Milano F, Rygiel AM, Bergman JJ, Rosmolen WD, van Deventer
SJ, Wang KK, Peppelenbosch MP, and Krishnadath KK (2005). A comparative
analysis by SAGE of gene expression profiles of Barrett’s esophagus, normal
squamous esophagus, and gastric cardia. Gastroenterology 129, 1274–1281.
 Bauerschlag DO, Ammerpohl O, Brautigam K, Schem C, Lin Q, Weigel MT,
Hilpert F, Arnold N, Maass N, Meinhold-Heerlein I, et al. (2011). Progression-
free survival in ovarian cancer is reflected in epigenetic DNA methylation pro-
files. Oncology 80, 12–20.
 Fackler MJ,UmbrichtCB,WilliamsD,ArganiP,CruzLA, MerinoVF,TeoWW,
Zhang Z, Huang P, Visvananthan K, et al. (2011). Genome-wide methylation
analysis identifies genes specific to breast cancer hormone receptor status and risk
of recurrence. Cancer Res 71, 6195–6207.
Genome-wide DNAm Profiling of cfDNAZhai et al.Neoplasia Vol. 14, No. 1, 2012
 Kobayashi Y, Absher DM, Gulzar ZG, Young SR, McKenney JK, Peehl DM,
Brooks JD, Myers RM, and Sherlock G (2011). DNA methylation profiling
reveals novel biomarkers and important roles for DNA methyltransferases in
prostate cancer. Genome Res 21, 1017–1027.
 StarkerLF,SvedlundJ,UdelsmanR,DralleH,Akerstrom G,WestinG,LiftonRP,
parathyroid tumors. Genes Chromosomes Cancer 50, 735–745.
 Zou H, Molina JR, Harrington JJ, Osborn NK, Klatt KK, Romero Y, Burgart LJ,
and Ahquist DA (2005). Aberrant methylation of secreted frizzled-related protein
genes in esophageal adenocarcinoma and Barrett’s esophagus. Int J Cancer 116,
 Zou H, Osborn NK, Harrington JJ, Klatt KK, Molina JR, Burgart LJ, and
Ahquist DA (2005). Frequent methylation of eyes absent 4 gene in Barrett’s
esophagus and esophageal adenocarcinoma. Cancer Epidemiol Biomarkers Prev
JH, DeMeester SR, DeMeester TR, Skinner KA, et al. (2001). Epigenetic patterns
in the progression of esophageal adenocarcinoma. Cancer Res 61, 3410–3418.
 Sato F, Jin Z, Schulmann K, Wang J, Greenwald BD, Ito T, Kan T, Hamilton
JP, Yang J, Paun B, et al. (2008). Three-tiered risk stratification model to pre-
dict progression in Barrett’s esophagus using epigenetic and clinical features.
PLoS One 3, e1890.
 Levenson VV and Melnikov AA (2011). The MethDet: a technology for bio-
marker development. Expert Rev Mol Diagn 11, 807–812.
 Campan M, Weisenberger DJ, Trinh B, and Laird PW (2009). MethyLight.
Methods Mol Biol 507, 325–337.
Neoplasia Vol. 14, No. 1, 2012 Genome-wide DNAm Profiling of cfDNAZhai et al.
_-FLA]_fig1",5,"place_anchor">_-FLA]_fig1",5,"pla- Download full-text
Figure W1. The β methylation value distributions in matched EA serum DNA (A) and tissue DNA (B).