Examination of Whole Blood DNA Methylation as a Potential Risk … · cohorts(training set:n¼72;validationset:n¼91)and67 cancer-free subjects (controls) who were attending the endoscopy

Research Article

Examination of Whole Blood DNA Methylation as a PotentialRisk Marker for Gastric Cancer

Tomomitsu Tahara1,4, Shinji Maegawa1, Woonbok Chung1, Judith Garriga1, Jaroslav Jelinek1,Marcos R.H. Est�ecio2,3, Tomoyuki Shibata4, Ichiro Hirata4, Tomiyasu Arisawa5, and Jean-Pierre J. Issa1

AbstractWhole bloodDNAmethylation analysis has been proposed to be a riskmarker for cancer that can be used

to target patients for preventive interventions. To test this,we examinedwholebloodDNAmethylationof 16

CpG island promoters and LINE1 repetitive element in patients with gastric cancer and control subjects.

Bisulfite pyrosequencing was used to quantify the methylation of 14 CpG island promoters (MINT25,

RORA,GDNF,CDH1,RARAB2,ER,CDH13,MYOD1, SFRP1,P2RX7, SLC16A12, IGF2,DPYS, andN33) and

LINE1 from 72 patients with gastric cancer, 67 control, and 52 healthy young individuals. Quantitative

methylation-specific real-time PCRwas also conducted for 3 CpG island promoters (MINT25,MYO3A, and

SOX11). Among all sites tested, only a marginal increase in the methylation of the SFRP1 promoter was

observed in the blood of patients with gastric cancer when compared with the control group (11.3 % vs

10.5%; age-adjusted P value: P¼ 0.009), and this association was also seen in a validation set of 91 patients

with gastric cancer (11.5% vs 10.5%; age-adjusted P value: P ¼ 0.001). The methylation of 9 sites (GDNF,

CDH1, RARAB2, CDH13,MYOD1, SFRP1, SLC16A12,DPYS,N33, and LINE1) and their mean Z score was

correlated with higher age (R¼ 0.41, P < 0.0001) andmarginally with telomere shortening (R¼�0.18, P¼0.01) but not with gastric cancer risk (other than SFRP1 methylation). Variability in whole blood DNA

methylation of cancer markers is primarily associated with aging, reflecting turnover of white blood cells,

and has no direct link to gastric cancer predisposition. SFRP1methylation inwhole bloodmay be associated

with gastric cancer risk. Cancer Prev Res; 6(10); 1093–100. �2013 AACR.

IntroductionGastric cancer is a major cause of cancer-related death,

worldwide.Helicobacter pylori (H. pylori) plays an importantrole in gastric carcinogenesis, although the majority ofindividuals with H. pylori infection do not develop gastriccancer (1). Surveillance of these high-risk patients usingreliable and accurate predictive markers is important forreducing the incidence of cancer and its mortality. It is nowwidely accepted that changes in DNAmethylation patterns,particularly promoter hypermethylation and global(genome-wide) hypomethylation, contribute to cancerdevelopment and tumor growth (2, 3). Neoplastic growthis frequently preceded by aberrant promoter methylation,

which leads to a loss-of-function for the genes that promotecell proliferation (4). Hypomethylation is thought to con-tribute to carcinogenesis by inducing genomic instability (5,6), leading to the formation of abnormal chromosomalstructures (7, 8). Gene promoter hypermethylation andglobal hypomethylation in tumor tissues are commonevents in the development of many types of cancer (4),whereas gene promoter hypermethylation and global hypo-methylation are also observed in aged or inflamed tissuesand are associated with cancer occurrence in their targetedtissues (9–11).

The DNA methylation status of various tissues has beenshown to be associated with aging and perhaps also expo-sures encountered throughout life (12, 13), and therefore isnow increasingly seen as a mechanism of cancer predispo-sition (14–17). Evaluating whole blood DNA methylationas a risk marker for cancer is of particular interest becauseperipheral blood DNA is a convenient tissue to assay forconstitutional methylation, as its collection is noninvasive.It is possible that the methylation status of cancer targettissues (i.e., neoplastic cells and the surrounding tissue/field defect) might reflect acquired or inherited somaticevents that are detectable in nontargeted tissues (methyla-tion memory of exposures/inheritance) and correlate withcancer susceptibility. Thus, epigenetic signatures in wholeblood DNA could reflect the interaction of host genetic andenvironmental factors associated with cancer susceptibility.

Authors' Affiliations: 1Fels Institute for Cancer Research & MolecularBiology, Temple University School of Medicine, Philadelphia, Pennsylva-nia; 2Department ofMolecular Carcinogenesis, The University of TexasMDAnderson Cancer Center, Smithville; 3Center for Cancer Epigenetics, TheUniversity of TexasMDAndersonCancerCenter, Houston, Texas; 4Depart-ment of Gastroenterology, Fujita Health University School of Medicine,Toyoake; and 5Department of Gastroenterology, Kanazawa Medical Uni-versity, Ishikawa, Japan

Corresponding Author: Tomomitsu Tahara, Temple University School ofMedicine, 3307 N. Broad Street, Room 154 PAHB, Philadelphia, PA 19140.Phone: 215-707-4300; Fax: 215-707-1454; E-mail:[email protected]

doi: 10.1158/1940-6207.CAPR-13-0034

�2013 American Association for Cancer Research.

CancerPreventionResearch

www.aacrjournals.org 1093

Research. on March 24, 2020. © 2013 American Association for Cancercancerpreventionresearch.aacrjournals.org Downloaded from

Published OnlineFirst August 13, 2013; DOI: 10.1158/1940-6207.CAPR-13-0034

http://cancerpreventionresearch.aacrjournals.org/

In addition, rare cases of constitutionalDNAmethylationoftumor suppressor genes have been reported and proposedto be predisposing to cancer. To evaluate this concept ingastric cancer, we investigated the methylation status of 16CpG island promoters selected on the basis of either cancer-associated or age-related methylation, and also studied theLINE1 repetitive element (representative of global methyl-ation) in whole blood DNA in patients with gastric cancerand control subjects. We also investigated the associationbetween DNA methylation status and telomere lengthshortening, which is an indicator for cell turnover and aging(18, 19).

Materials and MethodsSamples analyzed

For screening,we usedDNA from8primary gastric cancertissues and 6 non-neoplastic gastric mucosae from healthysubjects. For testing, we use two different gastric cancercohorts (training set: n¼ 72; validation set: n¼ 91) and 67cancer-free subjects (controls) who were attending theendoscopy center of Fujita Health University (Toyoake,Aichi, Japan) from January 2005 to March 2010. Fivemilliliters of whole blood DNA was collected from eachparticipant in an EDTA tube and stored frozen until DNAextraction. Whole blood DNA extraction was carried outusing a commercial kit (FlexiGene DNA Kit, QIAGEN) andstored until processing for analysis. All patients with gastriccancer were admitted to Fujita Health University hospitalfor the treatment of gastric cancer. Noncancer patientsunderwent upper gastroscopy for various reasons, includingyearly screening for gastric cancer, a secondary completecheck-up after barium radiographic examination due to asuspicion of gastric cancer or peptic ulcer disease, andcomplaints of abdominal discomfort. They were finallydiagnosed as not having gastric cancer. To avoid confoun-ders, we excluded patients with chronic illness from thestudy. We also collected whole bloodDNA from 52 healthyyoung individuals recruited from Japanesemedical studentsand staff of the Fujita Health University School of Medicinefrom April 2006 to October 2007. The Ethics Committee ofthe Fujita Health University School of Medicine approvedthe protocol and written informed consent was obtainedfrom all subjects.

Selection of candidate panels and CpG methylationanalysis by bisulfite pyrosequencing and qMSP

The selection of genes is based on the hypothesis thatthe methylation status of targeted tissues (i.e., cancer andsurrounding tissue) might reflect acquired or inheritedsomatic events that are detectable in nontargeted tissues(methylation memory of exposures/inheritance) and cor-relate with cancer susceptibility. We reasoned that theremight be two approaches to selecting genes (all selectedfrom the literature), one based on frequency of methyl-ation in cancer (MINT25, RORA, GDNF, RARAB2,SLC16A12, SOX11, and MYO3A) and a separate onebased on methylation in normal inflamed/aged tissues(CDH1, ER, CDH13, MYOD1, SFRP1, P2RX7, IGF2, and

N33; refs. 9, 10, 20–26). Bisulfite-treated genomic DNAwas used to evaluate the methylation status of these CpGisland promoters by bisulfite pyrosequencing. We alsoevaluated the methylation status of the LINE1 repetitiveelement using bisulfite pyrosequencing. Bisulfite treat-ment of DNA was carried out with an EpiTect BisulfiteKit (Qiagen) according to the manufacturer’s protocol.Pyrosequencing was carried out using a PSQ96 systemwith a Pyro-Gold Reagent Kit (QIAGEN) and the resultswere analyzed using PyroMark Q96 ID software version1.0 (QIAGEN). The primers used for pyrosequencing arelisted in Supplementary Table S1. All bisulfite pyrose-quencing was conducted at least twice and averaged. Toincrease sensitivity, for 3 genes (MINT25, SOX11, andMYO3A), we conducted quantitative methylation-specificreal-time PCR (qMSP). The cycle threshold (Ct) values fortargeted genes were normalized in each reaction using aprimer/probe set for the reference gene, mC-LESS, aunique sequence that does not contain cytosines (27).Supplementary Table S2 lists primers and TaqMan probes(Applied Biosystems) for mC-LESS and the 3 genes exam-ined. Each qMSP reaction batch was controlled with onepositive (M. SssI methylase–treated DNA) and multipleblanks with no DNA. For each plate, mC-LESS (theinternal control) and the tested genes were determinedtogether to avoid inter-assay variation. All qMSP reactionswere conducted in triplicate and averaged.

Relative average telomere length measurementRelative telomere length was measured as a comparative

quantification, in particular as abundance of telomerictemplate versusa single-copy gene (T/S) by quantitativereal-time PCR as described previously (28). For the quan-titative real-time PCR, the iTaq SYBR Green Supermix (Bio-Rad) and StepOnePlus Real-Time PCR System (AppliedBiosystems) were used. The primers for telomeres andsingle-copy genes (H-globin) are listed in SupplementaryTable S3. All measurements were carried out in duplicateand averaged.

Statistical analysisMethylation levels and telomere length in whole blood

DNA between gastric cancer and control subjects werecompared using the Student t test. When the P values wereless than 0.05, the logistic regression model was used toadjust for age. The correlation between methylation levels,telomere length, and age was assessed using the Spearmancorrelation analysis. Two-sided P < 0.05 was consideredstatistically significant.

ResultsStudy populations

Table 1 describes the age and gender of the gastric cancer,control, and healthy young individuals groups. In thecomparison of gastric cancer and control subjects, the ageof the gastric cancer subjects was significantly higher thanthat of the controls (control vs gastric cancer training set:P¼0.01; control vs gastric cancer validation set: P < 0.0001).

Tahara et al.

Cancer Prev Res; 6(10) October 2013 Cancer Prevention Research1094




Prevalence of male gender was also significantly higher inthe gastric cancer validation set when compared with thehealthy young individuals (P ¼ 0.006).

Methylation status of candidate genes in primarygastric cancer tissues and non-neoplastic gastricmucosaeInitially, we evaluated the methylation status of 16 CpG

island promoters and LINE1 in 8 primary gastric cancertissues and 6 non-neoplastic gastric mucosae from healthysubjects by bisulfite pyrosequencing. This comparison wascarried out because we wished to include in our study genesthat are methylated in the normal mucosa of patients withcancer but not in healthy mucosa (potential field defect).We reasoned that there might be two approaches to select-ing genes (all selected from the literature), one based onfrequency of methylation in cancer (MINT25, RORA,GDNF, RARAB2, SLC16A12, SOX11, and MYO3A) and aseparate one based on methylation in normal inflamed/aged tissues (CDH1, ER, CDH13, MYOD1, SFRP1, P2RX7,IGF2, and N33). Considering the heterogeneity in methyl-ation status in individual tumors, most of the genes (exceptfor LINE1), showed higher methylation levels in gastriccancer tissues than in non-neoplastic gastric mucosa, asexpected on the basis of the selection criteria. (Supplemen-tary Fig. S1).

Methylation status of candidate genes in whole bloodDNA in patients with gastric cancer and controlpatientsFigure 1 shows the methylation status of 14 CpG island

promoters (MINT25, RORA, GDNF, CDH1, RARAB2, ER,CDH13, MYOD1, SFRP1, P2RX7, SLC16A12, IGF2, DPYS,and N33) and LINE1 in the gastric cancer training set,control, and healthy young individuals groups examinedby bisulfite pyrosequencing. We found a marginal increasein the methylation of the SFRP1 promoter in gastric cancercompared with the control group (11.3% vs 10.5%; age-adjusted P value: P ¼ 0.009), whereas the methylation ofSFRP1 was not significantly different between controls andhealthy young individuals (10.5 � % vs 10.0%, P ¼ 0.25).We did not find any association for the remaining 13 CpGisland promoters and LINE1 among gastric cancer andcontrol subjects (Fig. 1).We also examined themethylationstatus of whole blood DNA for 3 genes (MINT25, SOX11,

and MYO3A) using qMSP. However, we did not observe asignificant difference in the methylation status of these 3genes between the gastric cancer and control groups (Fig. 2).Because H. pylori infection is strong inducer of gastricmucosa methylation (10), we determined whetherH. pyloriinfection status would influence the methylation status ofwhole blood DNA. H. pylori status was available in 69cancer-free subjects (4 young individuals and 65 controls)and 71 patients with gastric cancer. As expected, H. pyloriinfectionwasmore prevalent in gastric cancer (61/70, 86%)than in cancer-free subjects (29/43, 62.3%, P ¼ 0.0006).However, we did not find any significant associationbetween DNA methylation and H. pylori infection status(all P values >0.1, data not shown).

Methylation status of candidate genes in whole bloodDNA in relation to aging

We investigated the association between whole bloodDNA methylation and aging using the Spearman correla-tion analysis. We combined the training set, control, andhealthy young individuals groups. One hundred and ninetyone subjects were included for this analysis. We foundsignificant positive correlations between 8 CpG island pro-moters (GDNF, CDH1, RARAB2, CDH13,MYOD1, SFRP1,SLC16A12, DPYS, and N33) and LINE1 with aging. Themethylation status of 2 genes (DPYS and N33) showed arelatively strong correlation with higher age (DPYS: R ¼0.37, P < 0.0001; N33: R ¼ 0.37, P < 0.0001; Fig. 3A). Also,the Z score of mean methylation for the nine genes wasclosely correlatedwith higher age (R¼ 0.41, P < 0.0001; Fig.3B).

Telomere length in gastric cancer and control subjectsand its relationship to aging and gene methylation

We examined the relative telomere length of whole bloodDNA with quantitative real-time PCR. Because of severalgastric cancer subjectswith short telomere length, the gastriccancer group showed shorter mean telomere length thanthe control group. However, this association was notsignificantly different (Fig. 4A). We investigated the clini-copathologic features of gastric cancer in 6 patients withgastric cancer with the lowest telomere length (histology,staging, location, etc.), but we did not find any significantassociations. We also investigated the association of ageand methylation status with telomere length. Shortened

Table 1. Study populations

VariablesHealthy youngindividuals

Control withoutgastric cancer

Gastric cancertraining set

Gastric cancervalidation set

N 52 67 72 91Age (mean � SEM)a 23.2 � 0.3 57.4 � 1.5 62.3 � 1.1 71.3 � 0.93Maleb 26 (50%) 41 (61.2%) 46 (63.9%) 67 (73.6%)

aControl without gastric cancer versus gastric cancer training set, P ¼ 0.01, Control without gastric cancer versus gastric cancervalidation set, P < 0.0001.bHealthy young individuals versus gastric cancer validation set, P ¼ 0.006.

Blood Methylation in Gastric Cancer

www.aacrjournals.org Cancer Prev Res; 6(10) October 2013 1095




telomere length was significantly correlated with higher age(R¼�0.26, P¼ 0.0003; Fig. 4B). We also found significantinverse correlation between shortened telomere length andmethylation of 4 CpG island promoters (DPYS, CDH13,MYOD1, and SLC16A12; Supplementary Fig. S2). A mar-ginal significant correlation was also found between short-ened telomere length and Z score of the mean methylationof nine age-related sites (GDNF, CDH1, RARAB2, CDH13,

MYOD1, SFRP1, SLC16A12, DPYS, N33, and LINE1; R ¼�0.18, P ¼ 0.01; Fig. 4C).

Methylation status of the SFRP1 gene in the gastriccancer validation set

The SFRP1 promoter displayed a significant increase inmethylation in the gastric cancer training set when com-pared with the control group. To confirm these data, we

MINT25 qMSP SOX11 qMSP MYO3A qMSP

MIN

T25 M

eth

yla

tion(4

0-d

Ct)

SO

X11 M

eth

yla

tion(4

0-d

Ct)

MY

O3A

Meth

yla

tion(4

0-d

Ct)45

40

35

30

25

20

50

40

30

20

10

0

50

40

30

20

10

0

Youn

g

Con

trol

GC

Youn

g

Con

trol

GC

Youn

g

Con

trol

GC

Figure 2. Methylation status of 3 promoter CpG islands in healthy young individuals (Young), cancer-free subjects (Control), and gastric cancer training set(gastric cancer) groups examined by qMSP. Horizontal bars represent mean methylation values. The relative level of methylated DNA is depicted as40-dCt [Ct of specific gene—Ct of mC-LESS (internal control)]. A higher 40-dCt represents more methylation of the target biomarker. Methylation levels inwhole blood DNA between gastric cancer and CONTROL were compared using the Student t test.

CDH1

Meth

yla

tion(%

)

Meth

yla

tion(%

)

Meth

yla

tion(%

)

Meth

yla

tion(%

)

Meth

yla

tion(%

)

Meth

yla

tion(%

)

Meth

yla

tion(%

)

Meth

yla

tion(%

)

Meth

yla

tion(%

)

Meth

yla

tion(%

)

Meth

yla

tion(%

)

Meth

yla

tion(%

)

Meth

yla

tion(%

)

Meth

yla

tion(%

)

Meth

yla

tion(%

)

25

20

15

10

5

0

20

15

10

5

0

20

15

10

5

0

15

10

5

0

80

70

60

50

20

15

10

5

0

60

40

20

0

60

40

20

0

15

10

5

0

15

10

5

0

40

30

20

10

0

40

30

20

10

0

30

20

10

0

40

30

20

10

0

20

10

0

Youn

g

Con

trol

GC

Youn

g

Con

trol

GC

Youn

g

Con

trol

GC

Youn

g

Con

trol

GC

Youn

g

Con

trol

GC

Youn

g

Con

trol

GC

Youn

g

Con

trol

GC

Youn

g

Con

trol

GC

Youn

g

Con

trol

GC

Youn

g

Con

trol

GC

Youn

g

Con

trol

GC

Youn

g

Con

trol

GC

Youn

g

Con

trol

GC

Youn

g

Con

trol

GC

Youn

g

Con

trol

GC

MINT25 RORA GDNF CDH13

P2RX7 SLC16A12 SFRP1P = 0.009

N33 ER

MYOD1 DPYS IGF2 RARB2 LINE1

Figure 1. Methylation status of 14 promoter CpG islands and LINE1 in healthy young individuals (Young), cancer-free subjects (Control), and gastric cancertraining set (gastric cancer) groups examined by bisulfite pyrosequencing. Horizontal bars represent mean methylation values. Methylation levels inwhole bloodDNAbetween gastric cancer andCONTROLwere compared using the Student t test and the logistic regressionmodel was used to adjust for agewhen the P values were less than 0.05.

Tahara et al.





evaluated its methylation status in a gastric cancer valida-tion set (n ¼ 91) and found a significant though marginalincrease of SFRP1 methylation in the gastric cancer valida-

tion set in comparison with the control group (11.5% vs10.5%; age adjusted P value: P ¼ 0.001; Fig. 5). We alsoinvestigated whether higher methylation of the SFRP1 gene

Figure 3. Age-related methylation. A, methylation status of 8 promoter CpG islands and LINE1 in relation to aging. The healthy young individuals, control, andgastric cancer training set groups were all included in this analysis. Statistical analyses were conducted using the Spearman correlation analysis. B,mean Z score of the methylation levels of 9 age-related loci in relation and aging. The healthy young individuals (young subjects), control, and gastric cancertraining set groups (control patients and patients with gastric cancer) were all included in this analysis. Statistical analyses were conducted using theSpearman correlation analysis. Left is a scatter plot of all the data by age, whereas right is a bar graph of all the patients divided by study group.






is associatedwith clinicopathologic characteristics of gastriccancer using the combined training and validation sets (n¼158); however, no significant association was foundbetween SFRP1 methylation and any clinicopathologicfeatures of gastric cancer (Table 2).

DiscussionIn this study, we tested whether methylation analysis of

selected loci in whole blood DNA could be useful as abiomarker of risk in gastric cancer. Among all the sites wetested, only a marginal increase in the methylation of theSFRP1 promoter was observed in both training and valida-tion sets. The SFRP1 gene is located at 8p11.21 and itsfrequent methylation has been documented in colorectalcancer tissues as well as in aged non-neoplastic colonmucosae (25). Our results also suggest that the SFRP1promoter is one of the regions where methylation in bloodcould reflect gastric cancer predisposition, but the differ-

ences between cases and controls are small and the datashould be verified in other cohorts. Overall, our negativeresults stand in contrast to other studies. In recent studies ongastric cancer, the methylation levels of the IGF2 differen-tially methylated region 0 (DMR0) and the Alu and LINE1repetitive elements tended to be lower in blood frompatients with gastric cancer than in control groups, but theassociation in overall subjects was not significant (P > 0.05for all; refs. 29, 30). Another study explored the associationbetween methylation in prediagnostic blood leukocyteDNA and gastric cancer risk in the prospective cohort(31). The result showed Alu methylation was inverselyassociated with gastric cancer risk, mainly among casesdiagnosed one or more years after. However, in overallsubjects, methylation was not significantly different amongthe cases and controls (31). This issue has also been lookedat in other tumors such as bladder cancer and others.Although several studies reported on promising results, thedifferences tended to be small and few of these have beenvalidated by others. For example, white blood cells (WBC)DNAmethylationof the LRE1 sequence andATM intragenicloci (ATMmvp2a) were significantly associated with the riskof head and neck and breast cancers (15, 17). However, thedifference in methylation levels for cases and controls inboth studies was not large (0.753 vs 0.747 for LRE1 in thehead and neck cancers; 76.8% vs 76.4% and 76.9% vs81.8% for 2 independent breast cancer cohorts; refs. 15, 17).There have also been several studies showing the possiblerole of constitutional methylation in blood DNA in cancerpredisposition (14–17).Wong and colleagues reported thatmethylation of the BRCA1 promoter in blood DNA wasmore frequent in early-onset breast cancer patients andcorrelated with higher BRCA1 methylation in tumors andBRCA1 mutation associated with pathologic features (16).It seems unlikely that any of the markers analyzed here fitthis constitutional methylation paradigm. Taken together,these results suggest thatmethylation inwhole bloodmightreflect cancer predisposition. However, these changes arerelatively small, with a large overlap between cases and

R = -0.26, P = 0.0003

A B CR = -0.18, P = 0.01

Re

lative

te

lom

ere

le

ng

th

Re

lative

te

lom

ere

le

ng

th4

3

2

1

4

3.75

3.5

3.25

3

2.75

2.5

2.25

2

1.75

1.5

Re

lative

te

lom

ere

le

ng

th

4

3.75

3.5

3.25

3

2.75

2.5

2.25

2

1.75

1.510 −1.5 −1 −0.5 0.5 1 1.5 2.5 32020 30 40 50

Age (y) Mean Z score of methylation of age

related 9 genes

60 70 80 90

Youn

g

Con

trol

GC

Figure 4. Relative telomere length examined by quantitative real-time PCR. A, telomere length in healthy young individuals (Young), cancer-free subjects(Control), and gastric cancer training set (gastric cancer) groups. Horizontal bars represent mean telomere length. B, telomere length, in relation toage. C, telomere length in relation to the mean Z score of the methylation levels of 9 age-related loci. Statistical analyses were conducted using the Studentt test (A) and Spearman correlation analysis (B, C).

20

15

10

5

0Control GC Validation set

P = 0.001

SF

RP

1 M

eth

yla

tion

(%

)

Figure5. Methylation statusofSFRP1 in cancer-free subjects (CONTROL)and the gastric cancer validation set. Horizontal bars represent meanmethylation values. Methylation levels in whole blood DNA betweengastric cancer and CONTROL were compared using the Student t testand the logistic regression model was used to adjust for age.

Tahara et al.





controls and the potential usefulness of blood DNA meth-ylation as a screening/diagnostic biomarker for cancer maytherefore be limited.Our studyhad limitationsworth discussing. The selection

of genes was based on the hypothesis that the methylationstatus of targeted tissues (i.e., cancer and surroundingtissue) might be reflected in nontargeted tissues and corre-late to cancer susceptibility. Therefore, we selected genesthat have been reported to be hypermethylated in severalcancers, including gastric cancer as well as in inflamed oraged tissues (9, 10, 20–26). It is possible that the methyl-ationof other genesmight serve as bettermarkers of risk andthis can be addressed in future studies using genome widemethylation analysis technologies (32, 33). Our data sug-gest that true differences might be small however, and thesestudies will need careful attention to sample size, valida-tion, and quantitation to avoid the possibilities of false–positive findings. Another issue is technical, pyrosequen-cing is quantitative but has a background of approximately5% methylation and is therefore not adequate to detectmethylation below that range. For 3 selected genes(MINT25, SOX11, and MYO3A), we examined the methyl-ation status of whole blood DNA by qMSP. Because qMSPtargets only methylated molecules relative to a referencegene, it would potentially be more sensitive to detectingmethylated molecules, even those derived from circulatingtumor DNA. All 3 genes showed higher methylation ingastric cancer tissues relative to healthy gastric mucosa

(Supplementary Fig. S1). However, we did not find anydifference in themethylation status ofwholebloodbetweengastric cancer and control subjects. It is also worth notingthat whole blood DNA may include circulating DNAderived from tumor cells; it is likely to be in small amounts(<1%; refs. 34, 35), but itmaybe a confounder in evaluatingthe significance of marginally positive results (such as forSFRP1). Finally, it is important to note that a formal study ofa risk marker requires follow-up of individuals positive forthat marker rather than a case–control study (as conductedhere), and this is most relevant to positive results.

Our study confirms a firm association between aging andpromoter CpG island methylation; 9 sites (GDNF, CDH1,RARAB2, CDH13,MYOD1, SFRP1, SLC16A12,DPYS,N33,and LINE1) showed a significant increase in methylationthroughout the aging process. The Z score of mean meth-ylation for these 9 sites had a relatively good correlationwith aging (R¼0.41,P<0.0001).Moreover,methylationofthe 9 sites was inversely associated with telomere length.Telomere length shortening has been observed in agedblood DNA and inflamed tissues and is considered to beassociated with cell turnover (18, 19). Therefore, our resultsindicate that DNA methylation in the blood increases atmany sites throughout the lifespan and this methylationincrease is partly associated with the turnover of whiteblood cells. On the other hand, we did not find anydifference in the mean methylation levels of these 9 age-related sites between gastric cancer and control groups (datanot shown). Several other studies have examined age-relat-edmethylation inwhole blood and found a number of sitesthat are hypermethylated with age (13, 36), whereas veryfew were associated with age-related phenotypes (31). Thislack of association between the severity of age-relatedchanges in DNA methylation and disease occurrence maybe due to the tissue specificity of the link between aberrantmethylation and disease, which may therefore not bedetectable in whole blood. Alternatively, a link betweenmethylation and disease is more likely if aberrant methyl-ation is rate limiting. For cancer, it is possible that the rate-limiting step is acquisition of a mutation in a hypermethy-lated "field" (37, 38), a property that cannot bemeasured byDNA methylation alone. Although aging is the dominantfactor in accounting for white blood cell methylation CpGisland methylation, only a small fraction of the variationcan be explained by age (R ¼ 0.41). We have previouslyreported that chronic inflammation and folate intake areassociated with methylation change (11, 39), but thisinformation was not available for the current patients. H.pylori was previously associated with increased gastricmucosa methylation (11), but this does not seem to extendtoWBCs. Thus,many factors that contribute tomethylationvariation in blood remain unclear.

Grant SupportThis work was supported by NIH grants CA158112 and CA098006 to

(J.-P.J. Issa). J.-P.J. Issa is an American Cancer Society Clinical Researchprofessor supported by a generous gift from the F. M. Kirby Foundation.

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby marked

Table 2. SFRP1 methylation status andclinicopathologic subtypes of gastric cancer

Variables (n)SFRP1 methylation(mean � SEM%)

GenderMale (110) 11.4 � 0.26Female (48) 11.6 � 0.31

H. pylori statusNegative (38) 11.2 � 0.46Positive (120) 11.5 � 0.23

HistologyIntestinal (110) 11.5 � 0.26Diffuse (48) 11.3 � 0.31

LocationAntrum (49) 11.9 � 0.43Body (81) 11.2 � 0.25Cardia (28) 11.4 � 0.48

StagingEarly (82) 11.8 � 0.33Advanced (76) 11.2 � 0.23

MorphologyElevated or protruding (40) 11.8 � 0.44Depressed or with ulceration (105) 11.3 � 0.25Scirrhous (13) 11.1 � 0.55

NOTE: Methylation status was not determined for 5 cases.






advertisement in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.

Received January 31, 2013; revised July 10, 2013; accepted August 7, 2013;published OnlineFirst August 13, 2013.

References1. Uemura N, Okamoto S, Yamamoto S, Matsumura N, Yamaguchi S,

Yamakido M, et al. Helicobacter pylori infection and the developmentof gastric cancer. N Engl J Med 2001;345:784–9.

2. Issa JP. Aging, DNA methylation and cancer. Crit Rev Oncol Hematol1999;32:31–43.

3. Feinberg AP, Tycko B. The history of cancer epigenetics. Nat RevCancer 2004;4:143–53.

4. Jones PA, Baylin SB. The epigenomics of cancer. Cell 2007;128:683–92.

5. Eden A, Gaudet F, Waghmare A, Jaenisch R. Chromosomal instabilityand tumors promoted by DNA hypomethylation. Science 2003;300:455.

6. Gaudet F, Hodgson JG, Eden A, Jackson-Grusby L, Dausman J, GrayJW, et al. Induction of tumors in mice by genomic hypomethylation.Science 2003;300:489–92.

7. Ehrlich M. DNA methylation in cancer: too much, but also too little.Oncogene 2002;21:5400–13.

8. Ehrlich M. DNA hypomethylation, cancer, the immunodeficiency, cen-tromeric region instability, facial anomalies syndrome and chromo-somal rearrangements. J Nutr 2002;132:2424S–9S.

9. Issa JP,OttavianoYL,CelanoP,HamiltonSR,DavidsonNE,BaylinSB.Methylation of the oestrogen receptor CpG island links ageing andneoplasia in human colon. Nat Genet 1994;7:536–40.

10. Issa JP, Ahuja N, Toyota M, Bronner MP, Brentnall TA. Acceleratedage-related CpG island methylation in ulcerative colitis. Cancer Res2001;61:3573–7.

11. Maekita T, Nakazawa K, Mihara M, Nakajima T, Yanaoka K, Iguchi M,et al. High levels of aberrant DNA methylation in Helicobacter pylori-infected gastric mucosae and its possible association with gastriccancer risk. Clin Cancer Res 2006;12:989–95.

12. Christensen BC, Houseman EA, Marsit CJ, Zheng S, Wrensch MR,Wiemels JL, et al. Aging and environmental exposures alter tissue-specific DNA methylation dependent upon CpG island context. PLoSGenet 2009;5:e1000602.

13. Teschendorff AE, Menon U, Gentry-Maharaj A, Ramus SJ, Weisen-berger DJ, Shen H, et al. Age-dependent DNA methylation of genesthat are suppressed in stem cells is a hallmark of cancer. Genome Res2010;20:440–6.

14. Marsit CJ, Koestler DC, Christensen BC, Karagas MR, Houseman EA,Kelsey KT. DNAmethylation array analysis identifies profiles of blood-derived DNAmethylation associated with bladder cancer. J Clin Oncol2011;29:1133–9.

15. Hsiung DT, Marsit CJ, Houseman EA, Eddy K, Furniss CS, McCleanMD, et al. Global DNA methylation level in whole blood as a biomarkerin head and neck squamous cell carcinoma. Cancer Epidemiol Bio-markers Prev 2007;16:108–14.

16. Wong EM, Southey MC, Fox SB, Brown MA, Dowty JG, Jenkins MA,et al. Constitutional methylation of the BRCA1 promoter is specificallyassociated with BRCA1mutation-associated pathology in early-onsetbreast cancer. Cancer Prev Res 2011;4:23–33.

17. Brennan K, Garcia-Closas M, Orr N, Fletcher O, Jones M, Ashworth A,et al. Intragenic ATM methylation in peripheral blood DNA as abiomarker of breast cancer risk. Cancer Res 2012;72:2304–13.

18. Takubo K, Izumiyama-Shimomura N, Honma N, Sawabe M, Arai T,Kato M, et al. Telomere lengths are characteristic in each humanindividual. Exp Gerontol 2002;37:523–31.

19. von Zglinicki T, Martin-Ruiz CM. Telomeres as biomarkers for ageingand age-related diseases. Curr Mol Med 2005;5:197–203.

20. KusanoM, ToyotaM, Suzuki H, Akino K, Aoki F, FujitaM, et al. Genetic,epigenetic, and clinicopathologic features of gastric carcinomas withtheCpG islandmethylator phenotype and an associationwith Epstein-Barr virus. Cancer 2006;106:1467–79.

21. Watanabe Y, Kim HS, Castoro RJ, Chung W, Estecio MR, Kondo K,et al. Sensitive and specific detection of early gastric cancer with DNA

methylation analysis of gastric washes. Gastroenterology 2009;136:2149–58.

22. Kwabi-AddoB,ChungW,ShenL, IttmannM,Wheeler T, Jelinek J, et al.Age-related DNA methylation changes in normal human prostatetissues. Clin Cancer Res 2007;13:3796–802.

23. Yamamoto E, Toyota M, Suzuki H, Kondo Y, Sanomura T, MurayamaY, et al. LINE-1 hypomethylation is associated with increased CpGisland methylation in Helicobacter pylori-related enlarged-fold gastri-tis. Cancer Epidemiol Biomarkers Prev 2008;17:2555–64.

24. Chung W, Kwabi-Addo B, Ittmann M, Jelinek J, Shen L, Yu Y, et al.Identification of novel tumor markers in prostate, colon and breastcancer by unbiased methylation profiling. PLoS ONE 2008;3:e2079.

25. Shen L, Toyota M, Kondo Y, Lin E, Zhang L, Guo Y, et al. Integratedgenetic and epigenetic analysis identifies three different subclasses ofcolon cancer. Proc Natl Acad Sci U S A 2007;104:18654–9.

26. Chung W, Bondaruk J, Jelinek J, Lotan Y, Liang S, Czerniak B, et al.Detection of bladder cancer using novel DNA methylation biomar-kers in urine sediments. Cancer Epidemiol Biomarkers Prev 2011;20:1483–91.

27. Weisenberger DJ, Trinh BN, Campan M, Sharma S, Long TI, Ana-nthnarayan S, et al. DNA methylation analysis by digital bisulfitegenomic sequencing and digital MethyLight. Nucleic Acids Res2008;36:4689–98.

28. McGrath M, Wong JY, Michaud D, Hunter DJ, De Vivo I. Telomerelength, cigarette smoking, and bladder cancer risk in men andwomen.Cancer Epidemiol Biomarkers Prev 2007;16:815–9.

29. Yuasa Y, Nagasaki H, Oze I, Akiyama Y, Yoshida S, Shitara K, et al.Insulin-like growth factor 2 hypomethylation of blood leukocyteDNA is associated with gastric cancer risk. Int J Cancer 2012;131:2596–603.

30. Hou L,WangH, Sartori S,GawronA, Lissowska J, Bollati V, et al. Bloodleukocyte DNA hypomethylation and gastric cancer risk in a high-riskPolish population. Int J Cancer 2010;127:1866–74.

31. Gao Y, Baccarelli A, Shu XO, Ji BT, Yu K, Tarantini L, et al. Bloodleukocyte Alu and LINE-1 methylation and gastric cancer risk in theShanghai Women's Health Study. Br J Cancer 2012;106:585–91.

32. Est�ecio MR, Issa JP. Tackling the methylome: recent methodologicaladvances in genome-wide methylation profiling. Genome Med 2009;1:106.

33. Jelinek J, Liang S, Lu Y, He R, Ramagli LS, Shpall EJ, et al. ConservedDNA methylation patterns in healthy blood cells and extensivechanges in leukemia measured by a new quantitative technique.Epigenetics 2012;7:1368–78.

34. Sturgeon SR, Balasubramanian R, Schairer C, Muss HB, Ziegler RG,Arcaro KF. Detection of promoter methylation of tumor suppressorgenes in serumDNAof breast cancer cases and benign breast diseasecontrols. Epigenetics 2012;7:1258–67.

35. Philipp AB, Stieber P, Nagel D, Neumann J, Spelsberg F, Jung A, et al.Prognostic role ofmethylated free circulatingDNA in colorectal cancer.Int J Cancer 2012;131:2308–19.

36. Bell JT, Tsai PC, Yang TP, Pidsley R, Nisbet J, Glass D, et al. Epigen-ome-wide scans identify differentially methylated regions for age andage-related phenotypes in a healthy ageing population. PLoS Genet2012;8:e1002629.

37. Shen L, KondoY,RosnerGL, Xiao L, HernandezNS, Vilaythong J, et al.MGMT promoter methylation and field defect in sporadic colorectalcancer. J Natl Cancer Inst 2005;97:1330–8.

38. Niwa T, Ushijima T. Induction of epigenetic alterations by chronicinflammation and its significance on carcinogenesis. Adv Genet2010;71:41–56.

39. Wallace K, Grau MV, Levine AJ, Shen L, Hamdan R, Chen X, et al.Association between folate levels and CpG Island hypermethyla-tion in normal colorectal mucosa. Cancer Prev Res 2010;3:1552–64.

Tahara et al.





2013;6:1093-1100. Published OnlineFirst August 13, 2013.Cancer Prev Res Tomomitsu Tahara, Shinji Maegawa, Woonbok Chung, et al. Marker for Gastric CancerExamination of Whole Blood DNA Methylation as a Potential Risk

Updated version

10.1158/1940-6207.CAPR-13-0034doi:

Access the most recent version of this article at:

Material

Supplementary

1

http://cancerpreventionresearch.aacrjournals.org/content/suppl/2013/08/19/1940-6207.CAPR-13-0034.DCAccess the most recent supplemental material at:

Cited articles

http://cancerpreventionresearch.aacrjournals.org/content/6/10/1093.full#ref-list-1

This article cites 39 articles, 16 of which you can access for free at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerpreventionresearch.aacrjournals.org/content/6/10/1093To request permission to re-use all or part of this article, use this link



http://cancerpreventionresearch.aacrjournals.org/lookup/doi/10.1158/1940-6207.CAPR-13-0034

http://cancerpreventionresearch.aacrjournals.org/content/suppl/2013/08/19/1940-6207.CAPR-13-0034.DC1

http://cancerpreventionresearch.aacrjournals.org/content/suppl/2013/08/19/1940-6207.CAPR-13-0034.DC1

http://cancerpreventionresearch.aacrjournals.org/content/6/10/1093.full#ref-list-1

http://cancerpreventionresearch.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://cancerpreventionresearch.aacrjournals.org/content/6/10/1093


Examination of Whole Blood DNA Methylation as a Potential Risk … · cohorts(training set:n¼72;validationset:n¼91)and67 cancer-free subjects (controls) who were attending the endoscopy

Documents