CpG Island Methylator Phenotype Involving Chromosome 3p … · 2017. 2. 15. · 3p Confers an Increased Risk of Non-small Cell Lung Cancer Zeyi Liu, ... In this study, we tested hypothesis

ORIGINAL ARTICLE

CpG Island Methylator Phenotype Involving Chromosome3p Confers an Increased Risk of Non-small Cell

Lung Cancer

Zeyi Liu, MS,* Wenwen Li, MS,* Zhe Lei, MS,* Jun Zhao, MD,† Xiao-Feng Chen, MD,‡Rengyun Liu, MS,* Xiaobei Peng, MS,* Zhi-hao Wu, PhD,§ Jun Chen, MD,§ Hongyu Liu, MD,§

Qing-Hua Zhou, MD,§ and Hong-Tao Zhang, PhD*

Purpose: This study aims to explore the association of CpG islandmethylator phenotype (CIMP) involving tumor suppressor genes onshort arm of chromosome 3 (3p) with increased risk of non-smallcell lung cancer (NSCLC).Methods and Materials: In this study, four NSCLC cell lines werecultured, and peripheral blood mononuclear cell (PBMC) specimensfrom 80 patients with NSCLC and 80 matched controls werecollected for 3p-involved CIMP (3pCIMP) analysis. 3pCIMP wasreferred to as having at least three synchronously methylated genesof 3p per sample. Methylation-specific polymerase chain reactionwas performed to examine the methylation status of each gene. DNAdemethylation of NSCLC cell lines was achieved through thetreatment with 5-aza-deoxycytidine. Logistic regression was used toassess odds ratios and 95% confidence intervals, which were ad-justed for gender, age, and smoking status.Results: Demethylation experiment showed that 3pCIMP statuscould play an important role in NSCLC cell proliferation. A total of97.5% of PBMC specimens from NSCLC patients presented pro-moter methylation of any one of six genes (hOGG1, RAR-B,SEMA3B, RASSF1A, BLU, or FHIT) on 3p. Interestingly, 3pCIMP�was found in 43.8% of NSCLC PBMC specimens and only in 6.3%of normal PBMC samples. The data suggest that 3pCIMP status issignificantly associated with NSCLC and normal PBMC samples

(p � 0.001). More importantly, the results show that 3pCIMPpositive carriers have a 12.8-fold increased risk of NSCLC (adjustedodds ratio, 12.8; 95% confidence interval, 4.38–37.4, p � 0.001) inChinese population.Conclusions: This is the first evidence of an association betweenPBMC 3pCIMP and risk for NSCLC.

Key Words: NSCLC, 3pCIMP, PBMC, Methylation, Risk.

(J Thorac Oncol. 2010;5: 790–797)

Epidemiological evidence has documented that lung cancerhas markedly increased incidence and mortality over thepast decade in China.1 Because at least 75% of patients withlung cancer have evidence of regional or distant metastasis atthe time of diagnosis, the overall 5-year survival rate of lungcancer is very poor.2 Non-small cell lung cancer (NSCLC) isthe most common type of lung cancer, accounting for ap-proximately 85% of all lung cancer cases and consistingmainly of adenocarcinoma, squamous cell carcinoma, andlarge cell carcinoma.3 It has been suggested that NSCLCcould result from accumulation of multiple genetic or epige-netic changes, including DNA methylation and histone acet-ylation4 and microRNA.5,6 Our recent studies provide supportfor the notion that DNA methylation could underlie epige-netically inactivation of tumor suppressor genes (TSGs) onshort arm of chromosome 3 (3p) in NSCLC,7 and CpG islandmethylator phenotype (CIMP) involving TSGs on 3p couldbe one of the frequent epigenetic events in carcinogenesis ofNSCLC.8

Multiple investigations have evaluated DNA methyl-ation changes present in NSCLC, elucidating a wide range ofmethylation frequencies for a variety of genes.9–13 A majorityof studies have compared the methylation aberrations oc-curred in NSCLC with those presented in the paired paracan-cerous lung tissues from the same patient.14 More recently,Brock et al.15 found aberrant patterns of promoter methyl-ation of APC, RASSF1A, p16, and CDH13 associated withearly recurrence in stage I NSCLC. Taken together, theseexperimental data are beneficial in identifying a panel ofcancer-specific methylation markers for early detection andprevention and targeted treatment of NSCLC. Importantly,

*Soochow University Laboratory of Cancer Molecular Genetics, School ofBasic Medicine and Biological Sciences, Medical College of SoochowUniversity, Suzhou; †The First Affiliated Hospital, Soochow University,Suzhou; ‡Department of Surgery, Shanghai Hospital for PulmonaryDiseases, Shanghai; and §Tianjin Key Laboratory of Lung CancerMetastasis and Tumor Microenviroment, Tianjin Lung Cancer Institute,Tianjin Medical University General Hospital, Tianjin, People’s Republicof China.

Disclosure: The authors declare no conflicts of interest.Address for correspondence: Hong-Tao Zhang, PhD, Soochow University Lab-

oratory of Cancer Molecular Genetics, Medical College of Soochow Uni-versity, 199 Ren’ai Road, Sino-Singapore Industrial Park, Suzhou 215123,People’s Republic of China. E-mail: [email protected]; and Qing-HuaZhou, MD, Tianjin Key Laboratory of Lung Cancer Metastasis and TumorMicroenvironment, Tianjin Lung Cancer Institute, Tianjin Medical Univer-sity General Hospital, 154 Anshan Road, Heping District, Tianjin 300052,People’s Republic of China. E-mail: [email protected]

Zeyi Liu and Wenwen Li contributed equally to this work.Copyright © 2010 by the International Association for the Study of LungCancerISSN: 1556-0864/10/0506-0790

Journal of Thoracic Oncology • Volume 5, Number 6, June 2010790

CORE Metadata, citation and similar papers at core.ac.uk

Provided by Elsevier - Publisher Connector

https://core.ac.uk/display/82300691?utm_source=pdf&utm_medium=banner&utm_campaign=pdf-decoration-v1

more particular attention was paid to assessing synchronousmethylation of multiple cancer-related genes (termed CIMP)and its implications to patients with NSCLC.16,17

Following the recently published reports on the pres-ence of CIMP in NSCLC,16,17 for the first time, we shed lighton the presence of chromosome 3p-involved CIMP(3pCIMP) that might play an important role in tumorigenesisof NSCLC.8 Although there is accumulating evidence sug-gesting that somatic epigenetic changes of putative TSGscontribute to carcinogenesis of NSCLC14 and DNA hyper-methylation frequently occurs in peripheral blood mononu-clear cell (PBMC) from colorectal and lung cancer pa-tients,18,19 little is known about an association of PBMC3pCIMP with the predisposition to developing NSCLC.

In this study, we tested hypothesis that some individu-als carrying PBMC 3pCIMP are more susceptible to devel-oping NSCLC. First, to investigate whether 3pCIMP statuscould play an important role in NSCLC cell proliferation, weexamined 3pCIMP status in four NSCLC cell lines before andafter the demethylating treatment. Then, we subjected bisul-fite-modified DNA of PBMC from 80 NSCLC patients and80 controls to analysis of 3pCIMP status. The results suggestthat 3pCIMP could occur frequently in PBMC from Chinesepopulation, indicating that PBMC 3pCIMP plays an impor-tant role in predisposition to NSCLC.

METHODSNSCLC Cell Lines and Specimens

Four NSCLC cell lines (SPC-A-1, A549, 95-D, andLTEP-a-2) purchased from Shanghai Institute of Biochemis-try and Cell Biology were cultured in Roswell Park MemorialInstitute (RPMI) 1640 medium (Gibco-Invitrogen Inc.) with10% fetal bovine serum (Gibco-Invitrogen Inc., Grand Island,NY) and incubated at 37°C in a humid atmosphere of 5% CO2.These NSCLC cell lines are from lung adenocarcinoma cells.

A total of 80 blood specimens were obtained afterinformed consent from patients who were diagnosed as pri-mary NSCLC in the First Affiliated Hospital of SoochowUniversity between January 2005 and March 2008. Patho-logic stages for NSCLC patients were determined accordingto the Revised International System for Staging Lung Can-cer.20 None of NSCLC patients had received either radiother-apy or chemotherapy before blood sampling. In addition toNSCLC blood samples, 80 blood samples were also collectedfrom individuals with no history of cancer as controls, whichwere randomly selected from the same geographic region,and their ages were in the similar range as the cancer patients.This research was approved to perform by the AcademicAdvisory Board of Soochow University.

PBMC was isolated by centrifugation at 3500 rpmfor 20 minutes after blood sampling.21 Genomic DNA wasisolated from NSCLC cell lines and the samples accordingto standard proteinase K digestion and phenol-chloroformextraction.

Bisulfite ModificationGenomic DNA was treated with sodium bisulfite before

methylation analysis according to the CpGenome Fast DNA

Modification Kit (Chemicon International, Inc.). The proce-dure for bisulfite modification was described in our previ-ously published article.8 Briefly, 1.0 �g of DNA in a volumeof 100 �l water were denatured for 10 minutes at 37°C byaddition of 7 �l of freshly prepared 3 M NaOH. Then 550 �lof freshly prepared DNA Modification Reagent (pH 5.0) wasadded and mixed well. The mixture was incubated at 55°C for20 hours. After adding 750 �l of binding buffer, we purifiedthe modified DNA with spin columns. Desulphonation wascompleted on the column by addition of 50 �l of freshlyprepared 20 mM NaOH/90% EtOH. Finally, DNA was elutedwith 30 to 45 �l of elution buffer and stored at �20°C untilanalyzing.

Methylation-Specific PCRDNA methylation was determined using methyla-

tion-specific polymerase chain reaction (PCR) (MSP) de-scribed elsewhere.22 MSP can differentiate methylatedalleles from unmethylated ones of an interested gene aftertreatment with sodium bisulfite. Primer sequences for MSPanalysis of RAR-B, SEMA3B, RASSF1A, BLU, and FHITwere reported previously23–27 except for that of hOGG1was designed by ourselves using Methyl Primer ExpressSoftware version 1.0 (Applied Biosystem Inc, Forster City,CA).8 All the cycling conditions of PCR consisted of aninitial denaturation at 95°C for 5 minutes, followed by 35cycles of 94°C for 30 seconds, various annealing temper-ature for 45 seconds, and 72°C for 1 minute 15 seconds.The expected PCR products were revealed by electro-phoresis and visualized with ethidium bromide staining.SPC-A-1 and A549 NSCLC cell lines were used as posi-tive controls for hOGG1, RAR-B, SEMA3B, BLU, andFHIT; and for RASSF1A, respectively. But distilled waterand unbisulfited DNA were served as negative controls.7

DNA Demethylation of NSCLC Cell LinesDNA demethylation was performed on NSCLC cells by

treatment with 5-aza-deoxycytidine (5-aza-CdR) (Sigma-Al-drich Inc, St. Louis, MO). Briefly, NSCLC cells were seededat a density of 5 � 104 cells in RPMI 1640 containing 10%fetal bovine serum on day 1 and allowed to grow understandard tissue culture conditions. On days 2, 4, and 6, themedium was changed to serum-free RPMI 1640 containing 5�M 5-aza-CdR. On days 3, 5, and 7, the medium waschanged to drug-free and serum-free RPMI 1640. On days 1,5, and 7, the cells were harvested for analysis of DNAmethylation.

MTT AssayProliferation of NSCLC cells before and after treatment

of demethylation agent 5-aza-CdR was assessed with theMTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazoliumbromide] (Sigma- Aldrich, Inc) method. Cells were seeded in96-well plates at a density of 5 � 104 cells/well and culturedunder standard tissue culture conditions for 24 hours. On days2, 4, and 6, the medium was changed to serum-free RPMI1640 containing 5 �M 5-aza-CdR. On days 1, 5, and 7, 20 �lof 5 mg/ml MTT were added, and the cells were incubated for5 hours at 37°C. One hundred microliters of dimethyl sulfox-

Journal of Thoracic Oncology • Volume 5, Number 6, June 2010 Association of 3p CpG Island Methylator Phenotype and NSCLC

Copyright © 2010 by the International Association for the Study of Lung Cancer 791

ide were quickly added to all wells and mixed thoroughly, sothat the dark blue crystals dissolved effectively. Absorbancewas measured at the wavelength of 550 nm. This experimentwas repeated three times. All data represent the mean of ninewells, which were subjected to statistical analysis.

Definition of CIMPAccording to the previously reported CIMP, which was

defined by average of number of methylated genes per tu-mor,28,29 we classified it into two categories, including CIMPpositive with at least three methylated genes and CIMPnegative with less than three methylated genes.

Statistical AnalysisChi-square test and Fisher exact test were used to

evaluate the correlation between two different categories.Logistic regression was performed to assess odds ratios(ORs) and 95% confidence intervals, which were adjusted forgender, age, and smoking status, using STATA version 10.0software package (StataCorp LP, College Station, TX). Thesensitivity, specificity, positive predictive value, and accu-racy for hOGG1, SEMA3B, RASSF1A, and 3pCIMP werecalculated using the 2 � 2 contingency table. Statisticaldifferences were considered to be significant at p � 0.05.Data analysis was performed using SigmaStat software (SystatSoftware, San Jose, CA) and SPSS version 11.5 software pack-age (SPSS Inc, Chicago, IL).

RESULTS

Relationship Between 3pCIMP and NSCLC CellProliferation

To determine whether 3pCIMP plays an importantrole in the proliferation of NSCLC cells, we performedDNA demethylation and MTT analyses and 3pCIMP eval-uation on four NSCLC cell lines. As illustrated in Figure 1,SPC-A-1 and A549 presented 3pCIMP positive before

(day 1) and after (days 5 and 7) treatment of demethylationagent 5-aza-CdR, respectively. However, 95-D and LTEP-a-2 showed 3pCIMP negative after (day 7) 5-aza-CdRtreatment, albeit that they presented 3pCIMP positive after(day 5) demethylation treatment.

Meanwhile, we examined the effect of demethylationtreatment on cell proliferation in these NSCLC cell lines. Asshown in Figure 2, cell proliferation was significantly de-creased in 95-D and LTEP-a-2 on day 7, but not in SPC-A-1and A549. These findings were compatible with 3pCIMPevaluation (Figure 1), indicating that 3pCIMP status couldplay an important role in NSCLC cell proliferation. 5-aza-CdR treatment is a currently accepted system to investigateeffects of hypomethylation, although this compound is rec-ognized to be capable of inducing other cellular effects.30

Methylation Profile of Multiple Genes on 3p inPBMC From NSCLC Patients

Methylation profile for PBMC specimens from 80NSCLC patients and 80 controls was determined using MSPmethod for six TSGs on 3p, including hOGG1, RAR-B,SEMA3B, RASSF1A, BLU, and FHIT. Representative bandsfor MSP analysis were exemplified in Figure 3, and thedetailed frequency distribution of aberrant methylation foreach gene was seen in Figure 4. With respect to NSCLC, thefrequency of promoter methylation of the gene was 22.5% forhOGG1, 20.0% for RAR-B, 91.3% for SEMA3B, 20.0% forRASSF1A, 48.8% for BLU, and 23.8% for FHIT, respectively.By contrast, lower frequencies of methylation for hOGG1(7.5%), RAR-B (8.8%), SEMA3B (47.5%), and RASSF1A

FIGURE 2. Cell proliferation analysis for non-small cell lungcancer (NSCLC) cell lines without and with treatment of de-methylation agent 5-aza-CdR. The cells were grown to 80%confluency when they were harvested on day 7. The y-axisrepresents the absorbance of 550 nm wavelength of MTTassay. Symbols � and � below 5-aza-CdR represent withoutand with treatment, respectively. On day 7, significant differ-ence in cell proliferation was observed between NSCLC celllines (95-D and LTEP-a-2) without and with 5-aza-CdR treat-ment. *p � 0.05.

FIGURE 1. 3p-involved CpG island methylator phenotype(3pCIMP) in four non-small cell lung cancer (NSCLC) celllines without and with treatment of demethylation agent5-aza-CdR. The variable day represents cell harvest schedule.On day 7, two NSCLC cell lines (SPC-A-1 and A549) with3pCIMP� have methylation of at least three genes, and twoNSCLC cell lines (95-D and LTEP-a-2) with 3pCIMP� presentmethylation of less than three genes. Symbols � and � be-low 5-aza-CdR represent without and with treatment, re-spectively. Black boxes, presence of unmethylation and meth-ylation; white boxes, presence of unmethylation. Symbols �and � below CIMP represent CIMP negative and positive,respectively.

Liu et al. Journal of Thoracic Oncology • Volume 5, Number 6, June 2010

Copyright © 2010 by the International Association for the Study of Lung Cancer792

FIGURE 3. The representative patterns of methyl-ated (M) and unmethylated (U) alleles in methyla-tion-specific polymerase chain reaction (MSP) analy-sis for six genes in peripheral blood mononuclearcell from patients with non-small cell lung cancer(NSCLC). Two samples (2CB and 48CB) with CpGisland methylator phenotype (CIMP)� have methyl-ation of at least three genes. Two samples (4CB and28CB) with CIMP� present methylation of less thanthree genes. MSP analysis of SPC-A-1 and A549NSCLC cell lines served as a positive control (Pos)for hOGG1, RAR-B, SEMA3B, BLU, and FHIT; and forRASSF1A, respectively. Distilled water and unbisul-fited DNA were used as negative controls (Neg 1and Neg 2, respectively).

FIGURE 4. Summary of methyl-ation for hOGG1, RAR-B, SEMA3B,RASSF1A, BLU, and FHIT in PBMCfrom 80 patients with NSCLC (A)and 80 controls (B). Green boxes,presence of unmethylation; yellowboxes, presence of unmethylationand methylation. CB, cancer bloodsample; NB, normal blood sample;�, CIMP positive; �, CIMP nega-tive; PBMC, peripheral blood mono-nuclear cell; NSCLC, non-small celllung cancer.



(2.5%) were found in controls. A significant difference inmethylation status for hOGG1, SEMA3B, and RASSF1A wasfound between PBMC from NSCLC patients versus PBMCfrom controls (p � 0.02, p � 0.001, and p � 0.001, respec-tively) (Figure 5).

Frequency of 3pCIMP in PBMC From NSCLCPatients and Normal Blood Samples

A total of 97.5% (78 of 80) of PBMC specimens fromNSCLC patients presented promoter methylation of any oneof six genes (hOGG1, RAR-B, SEMA3B, RASSF1A, BLU, orFHIT) on 3p. Correspondingly, a total of 77.5% (62 of 80) ofnormal blood samples had at least one gene methylated(Figure 4). A total of 26.3% (21 of 80) of NSCLC carried onegene, 27.5% (22 of 80) carried two genes, 32.5% (26 of 80)carried three genes, 10% (8 of 80) carried four genes, and

1.25% (1 of 80) carried five genes methylated (Figure 6).Correspondingly, a total of 33.8% (27 of 80) of normal bloodsamples had one gene, 37.5% (30 of 80) had two genes, 3.8%(3 of 80) had three genes, and 2.5% (2 of 80) had four genes,and none had five genes methylated (Figure 6). More impor-tantly, 3pCIMP� was observed in 43.8% (35 of 80) ofNSCLC PBMC and 6.3% (5 of 80) of normal blood samples;3pCIMP- was found in 56.2% (45 of 80) of NSCLC PBMCand 93.7% (75 of 80) of normal blood samples (Table 1).Using a chi-square test statistical analysis, we found that3pCIMP status was significantly associated with NSCLC andnormal blood samples (�2df � 1�28.0, p � 0.001).

3pCIMP of PBMC in Association with IncreasedRisk for NSCLC

In this study, we examined the relation of 3pCIMPstatus to clinical characteristics of NSCLC (Table 2). As aresult, no significant association between 3pCIMP status andmultiple clinical factors, including age, sex, smoking history,differentiation, tumor node metastasis stage, and histologicstratification was found (p � 0.05). To determine whether thepresence of 3pCIMP has an effect on increased risk ofNSCLC, we performed logistic regression to assess the ORsin the paired cohort. Consequently, methylation of hOGG1,SEMA3B, and RASSF1A, the gene pair hOGG1 and SEMA3B,hOGG1 and RASSF1A, and SEMA3B and RASSF1A wasobserved to increase significantly risk of NSCLC when com-pared with methylation-free ones (3.72-fold, p � 0.02; 7.97-fold, p � 0.001; 4.49-fold, p � 0.001; 9.95-fold, p � 0.001;infinite folds, p � 0.02; 10.00-fold, p � 0.001, respectively).More intriguingly, we found that 3pCIMP positive carriershad a 12.79-fold increased risk of developing NSCLC (ad-justed OR, 12.8; 95% confidence interval, 4.38–37.4, p �0.001) in Chinese population (Table 3).

Given that these DNA methylation marker candidates(listed in Table 4) seem to associate preferentially withincreased risk of NSCLC in Chinese population, we com-pared the differences in their sensitivity and specificity fordiagnosis of NSCLC. As summarized in Table 4, the speci-

FIGURE 5. Frequency of methylation for six genes in pe-ripheral blood mononuclear cell from 80 patients with non-small cell lung cancer (NSCLC) and 80 controls. Significantlymethylated difference of hOGG1, SEMA3B, and RASSF1A wasobserved between NSCLCs and controls (*p � 0.02, p �0.001, and p � 0.001, respectively).

FIGURE 6. Methylated gene numbers per sample in 80non-small cell lung cancers and 80 normal blood samples.

TABLE 1. Characteristics of NSCLC Cases and Controls byAge, Sex, Smoking History, and 3pCIMP Status

VariableCases

(n � 80)Controls(n � 80) pa

Age (yr), mean � SD) 61.0 � 10.3 60.7 � 10.7 0.83

Sex

Male 59 (73.8) 50 (62.5) 0.18

Female 21 (26.2) 30 (37.5)

Smoking history

Yes 47 (58.7) 19 (23.8) �0.001

No 33 (41.3) 61 (76.2)

CIMP status

3pCIMP� 35 (43.8) 5 (6.3) �0.001

3pCIMP� 45 (56.2) 75 (93.7)

Values are expressed as n (%).NSCLC, non-small cell lung cancer; 3pCIMP, 3p-involved CpG island methylator

phenotype.a �2 test.



ficity for SEMA3B (52.5%) and its combinations (47.5% and52.5%) was much lower than hOGG1 (92.5%), RASSF1A(97.5%), the gene pair hOGG1 and RASSF1A (90.0%), and3pCIMP (93.8%), although the sensitivity for SEMA3B(91.3%) and its combinations (92.5% and 91.3%) was muchhigher than hOGG1 (22.5%), RASSF1A (20.0%), the genepair hOGG1 and RASSF1A (32.5%), and 3pCIMP (43.8%).Moreover, the positive predictive value and overall accuracyfor 3pCIMP (87.5% and 68.8%) were relatively higher com-pared with a majority of DNA methylation marker candi-dates. Taken together, 3pCIMP ranks first among these po-tential DNA methylation markers for NSCLC carcinogenesis.

DISCUSSIONReliable markers for NSCLC are still lacking, even

though more efforts on discovering them have been made.6Improved understanding of molecular events, including epi-genetic silencing of genes, may help to identify a panel ofearly tumor markers.31,32 Epigenetic silencing of TSGs viapromoter hypermethylation has been found to correlate withthe initiation and progression of various types of cancers,including lung cancer.33,34 3p may harbor multiple TSGs,which are involved in the early stages of bronchial carcino-genesis.35 However, no evidence for association of PBMCepigenetic changes regarding TSGs on 3p with increased riskof NSCLC has been reported. Therefore, in this study, wefocused on identifying the presence of 3pCIMP in NSCLC

PBMC and assessing an association between PBMC 3pCIMPand the predisposition to NSCLC.

In an effort to validate first the biologic function of3pCIMP, we performed MTT, 3pCIMP and demethylationanalyses in four NSCLC cell lines and found different3pCIMP status (either positive or negative) compatible withdifferentiately proliferated changes of NSCLC cells (eitherpreserved or decreased), suggesting that 3pCIMP could func-tion as an epigenetic silencing of TSGs in pathogenesis ofNSCLC. These findings are consistent with those from ourrecently available study8 of NSCLC tissues that noted asignificant association between somatic 3pCIMP and devel-opment of NSCLC. Given the fact that methylated status of aCpG site is more stably inherited than unmethylated status36

and that the best-known epigenetic marker is DNA methyl-

TABLE 2. Association Between 3pCIMP Status of PBMCand Clinical Characteristics of NSCLC

Variable Cases3pCIMP�(n � 35)

3pCIMP�(n � 45) pa

Age (yr), mean � SD 61 � 11.3 61.0 � 9.6 1.00

Sex

Male 59 22 37 0.09

Female 21 13 8

Smoking history

Yes 47 17 30 0.16

No 33 18 15

Differentiation

Well 3 1 2 0.58

Moderate 41 16 25

Low 36 18 18

TNM stage

Stage I 26 10 16 0.74

Stage II 34 15 19

Stage III � IV 20 10 10

Histology

AC 22 12 10 0.13

SCC 37 14 23

LCC 10 2 8

ACC 8 6 2

Carc 3 1 2

a �2 test.AC, adenocarcinoma; SCC, squamous cell carcinoma; LCC, large cell carcinoma;

ACC, alveolar cell carcinoma; Carc, carcinoid carcinoma; NSCLC, non-small cell lungcancer; 3pCIMP, 3p-involved CpG island methylator phenotype.

TABLE 3. Methylation Status of Six Genes Located onChromosome 3p Among NSCLC Patients and Controls andTheir Contributions to the Risk for NSCLC

Cases(N � 80),

n (%)

Controls(N � 80),

n (%) p OR (95% CI)a

hOGG1

U 62 (77.5) 74 (92.5) 1.00

M 18 (22.5) 6 (7.5) 0.02 3.72 (1.07–12.9)

RAR-B

U 64 (80.0) 73 (91.3) 1.00

M 16 (20.0) 7 (8.7) 0.07 2.24 (0.67–7.51)

SEMA3B

U 7 (8.7) 42 (52.5) 1.00

M 73 (91.3) 38 (47.5) �0.001 7.97 (3.01–21.1)

RASSF1A

U 64 (80.0) 78 (97.5) 1.00

M 16 (20.0) 2 (2.5) 0.001 4.49 (0.90–22.3)

BLU

U 41 (51.2) 47 (58.8) 1.00

M 39 (48.8) 33 (41.2) 0.43 1.31 (0.59–2.90)

FHIT

U 61 (76.2) 61 (76.2) 1.00

M 19 (23.8) 19 (23.8) 1.00 1.37 (0.52–3.56)

hOGG1 and SEMA3B

USM 63 (78.8) 78 (97.5) 1.00

SM 17 (21.2) 2 (2.5) �0.001 9.95 (2.11–46.9)

hOGG1 and RASSF1A

USM 75 (93.8) 80 (100) 1.00

SM 5 (6.2) 0 (0.0) 0.02 Infinite (1.36-�)

SEMA3B and RASSF1A

USM 64 (80.0) 78 (97.5) 1.00

SM 16 (20.0) 2 (2.5) 0.001 10.0 (2.11–47.5)

3pCIMP status

3pCIMP� 45 (56.2) 75 (93.7) 1.00

3pCIMP� 35 (43.8) 5 (6.3) �0.001 12.8 (4.38–37.4)

a ORs and 95% CIs were calculated in a logistic regression model and adjusted forage, sex, and smoking status.

NSCLC, non-small cell lung cancer; OR, odds ratio; CI, confidence interval; U,unmethylated; M, methylated; USM, unsynchronously methylated; SM, synchronouslymethylated.



ation,37 we hypothesize that 3pCIMP might be emerging as apromising risk factor of NSCLC.

On the basis of what we observed earlier, additionalstudies are needed to address the question of whether3pCIMP could contribute to increased risk of developingNSCLC. By using a population-based investigation, we foundthat the frequency of PBMC methylation for six TSGs pro-moter ranged from 20.0% for RAR-B (or RASSF1A) to 91.3%for SEMA3B. Although 3p TSG methylation is frequent inNSCLC PBMC, we cannot exclude the possibility that thedetected TSG methylation is derived from circulating tumorcells in the peripheral blood. As summarized in our recentlypublished article in Lung Cancer,8 the frequencies of meth-ylation for hOGG1, RAR-B, SEMA3B, RASSF1A, BLU, andFHIT in 60 NSCLC tumors are 65%, 50%, 57%, 67%, 65%,and 47%, respectively, showing that the methylation fre-quency for hOGG1, RAR-B, RASSF1A, BLU, and FHIT wasfar lower in NSCLC PBMC than paired NSCLC tissues.Especially, the frequency for VHL and hMLH1 that were notconsidered as contributing to PBMC 3pCIMP was zero in allsubjects, including NSCLC patients and controls (data notshown). Despite of discrepant frequencies between our reportand Seng et al. (hMLH1 in 12.0% and 35.7% of NSCLCtumors, respectively),8,38 the results suggested that hMLH1methylation could be a late event in NSCLC progression.More importantly, 3pCIMP� was found in 43.8% of NSCLCPBMC specimens but only in 6.3% of normal PBMC sam-ples. More excitingly, 3pCIMP-positive carriers have a 12.8-fold increased risk of developing NSCLC in Chinese popu-lation, and combined analysis with sensitivity and specificityhas shown that 3pCIMP might become preferentially a sur-rogate for other DNA methylation marker candidates ofNSCLC, including hOGG1, SEMA3B, and RASSF1A.

Although CIMP etiology remains elusive, it exists inseveral malignant solid tumors.16 Meanwhile, the correlationof CIMP with clinicopathologic factors in various humancancers was well established.39–41 However, we did not findan association between 3pCIMP and multiple clinical factors,including age, sex, smoking history, differentiation, tumornode metastasis stage, and histologic stratification of NSCLC,which was in accordance with the data from Suzuki et al.17The results demonstrate that 3pCIMP could be in relation toNSCLC initiation rather than progression. Considering thattreatment with demethylating agents can restore gene expres-

sion of inactivating TSGs,42 administration of demethylatingagents may become an ideal candidate for individuals with3pCIMP positive to decrease risk of developing NSCLC.

In conclusion, we provide strong evidence of an asso-ciation between PBMC 3pCIMP and risk for NSCLC, whichsupports the suggestion from Russo et al.19 that peripherallymphocytes could potentially be used as a surrogate forbronchial epithelial cells to detect altered DNA methylationin development and progression of lung cancer. Better un-derstanding of 3pCIMP may provide new insight into cancerprevention and therapeutic interventions.

ACKNOWLEDGMENTSSupported by grants from National Natural Science

Foundation of China (30400533, 30672400, and 30973425 toH.-T.Z.), Science and Technology Committee of JiangsuProvince (BK2008162 to H.-T.Z.), SRF for ROCS, StateEducation Ministry (2008890 to H.-T.Z.), Qing-Lan Projectof Education Bureau of Jiangsu Province (to H.-T.Z.), “333”Project of Jiangsu Province (to H.-T.Z.), Key Project fromNational Natural Science Foundation of China (30430300 toQ.-H.Z.), and Key Projects of Tianjin Sci-Tech Support Pro-gram (07SYSYSF05000 and 06YFSZSF05300 to Q.-H.Z.).

We thank the patients with NSCLC for their participa-tion and cooperation.

REFERENCES1. Yang L, Parkin DM, Li LD, et al. Estimation and projection of the

national profile of cancer mortality in China: 1991–2005. Br J Cancer2004;90:2157–2166.

2. Mulshine JL, Sullivan DC. Clinical practice. Lung cancer screening.N Engl J Med 2005;352:2714–2720.

3. Brambilla E, Travis WD, Colby TV, et al. The new World HealthOrganization classification of lung tumours. Eur Respir J 2001;18:1059–1068.

4. Zhong S, Fields CR, Su N, et al. Pharmacologic inhibition of epigeneticmodifications, coupled with gene expression profiling, reveals noveltargets of aberrant DNA methylation and histone deacetylation in lungcancer. Oncogene 2007;26:2621–2634.

5. Takamizawa J, Konishi H, Yanagisawa K, et al. Reduced expression ofthe let-7 microRNAs in human lung cancers in association with short-ened postoperative survival. Cancer Res 2004;64:3753–3756.

6. Hu Z, Chen J, Tian T, et al. Genetic variants of miRNA sequences andnon-small cell lung cancer survival. J Clin Invest 2008;118:2600–2608.

7. Zhang HT, Chen XF, Wang MH, et al. Defective expression of trans-forming growth factor beta receptor type II is associated with CpG

TABLE 4. Sensitivity, Specificity, and PPV and Accuracy of hOGG1, SEMA3B, RASSF1A,and 3pCIMP for Diagnosis of NSCLC When Tested Independently or in Combinations

Methylated gene Sensitivity (%) Specificity (%) PPV (%) Accuracy (%)

hOGG1 18/80 (22.5) 74/80 (92.5) 18/24 (75.0) 92/160 (57.5)

SEMA3B 73/80 (91.3) 42/80 (52.5) 73/111 (65.8) 115/160 (71.9)

RASSF1A 16/80 (20.0) 78/80 (97.5) 16/18 (88.9) 94/160 (58.8)

hOGG1 and SEMA3B 74/80 (92.5) 38/80 (47.5) 74/116 (63.8) 112/160 (70.0)

hOGG1 and RASSF1A 26/80 (32.5) 72/80 (90.0) 26/34 (76.5) 98/160 (61.3)

SEMA3B and RASSF1A 73/80 (91.3) 42/80 (52.5) 73/111 (65.8) 115/160 (71.9)

3pCIMP 35/80 (43.8) 75/80 (93.8) 35/40 (87.5) 110/160 (68.8)

NSCLC, non-small cell lung cancer; 3pCIMP, 3p-involved CpG island methylator phenotype; PPV, positive predictive value.



methylated promoter in primary non-small cell lung cancer. Clin CancerRes 2004;10:2359–2367.

8. Liu Z, Zhao J, Chen XF, et al. CpG island methylator phenotypeinvolving tumor suppressor genes located on chromosome 3p in non-small cell lung cancer. Lung Cancer 2008;62:15–22.

9. Zochbauer-Muller S, Fong KM, Virmani AK, et al. Aberrant promotermethylation of multiple genes in non-small cell lung cancers. CancerRes 2001;61:249–255.

10. Toyooka S, Toyooka KO, Maruyama R, et al. DNA methylation profilesof lung tumors. Mol Cancer Ther 2001;1:61–67.

11. Toyooka S, Maruyama R, Toyooka KO, et al. Smoke exposure, histo-logic type and geography-related differences in the methylation profilesof non-small cell lung cancer. Int J Cancer 2003;103:153–160.

12. Zochbauer-Muller S, Minna JD, Gazdar AF. Aberrant DNA methylationin lung cancer: biological and clinical implications. Oncologist 2002;7:451–457.

13. Gu J, Berman D, Lu C, et al. Aberrant promoter methylation profile andassociation with survival in patients with non-small cell lung cancer.Clin Cancer Res 2006;12:7329–7338.

14. Feng Q, Hawes SE, Stern JE, et al. DNA methylation in tumor andmatched normal tissues from non-small cell lung cancer patients. CancerEpidemiol Biomarkers Prev 2008;17:645–654.

15. Brock MV, Hooker CM, Ota-Machida E, et al. DNA methylationmarkers and early recurrence in stage I lung cancer. N Engl J Med2008;358:1118–1128.

16. Marsit CJ, Houseman EA, Christensen BC, et al. Examination of a CpGisland methylator phenotype and implications of methylation profiles insolid tumors. Cancer Res 2006;66:10621–10629.

17. Suzuki M, Shigematsu H, Iizasa T, et al. Exclusive mutation in epider-mal growth factor receptor gene, HER-2, and KRAS, and synchronousmethylation of nonsmall cell lung cancer. Cancer 2006;106:2200–2207.

18. Suter CM, Martin DI, Ward RL. Germline epimutation of MLH1 inindividuals with multiple cancers. Nat Genet 2004;36:497–501.

19. Russo AL, Thiagalingam A, Pan H, et al. Differential DNA hypermeth-ylation of critical genes mediates the stage-specific tobacco smoke-induced neoplastic progression of lung cancer. Clin Cancer Res 2005;11:2466–2470.

20. Mountain CF. Revisions in the International System for Staging LungCancer. Chest 1997;111:1710–1717.

21. Roupret M, Hupertan V, Catto JW, et al. Promoter hypermethylation incirculating blood cells identifies prostate cancer progression. Int JCancer 2008;122:952–956.

22. Herman JG, Graff JR, Myohanen S, et al. Methylation-specific PCR: anovel PCR assay for methylation status of CpG islands. Proc Natl AcadSci USA 1996;93:9821–9826.

23. Youssef EM, Lotan D, Issa JP, et al. Hypermethylation of the retinoicacid receptor-beta(2) gene in head and neck carcinogenesis. Clin CancerRes 2004;10:1733–1742.

24. Kuroki T, Trapasso F, Yendamuri S, et al. Allelic loss on chromosome3p21.3 and promoter hypermethylation of semaphorin 3B in non-smallcell lung cancer. Cancer Res 2003;63:3352–3355.

25. Lo KW, Kwong J, Hui AB, et al. High frequency of promoter hyper-

methylation of RASSF1A in nasopharyngeal carcinoma. Cancer Res2001;61:3877–3881.

26. Liu XQ, Chen HK, Zhang XS, et al. Alterations of BLU, a candidatetumor suppressor gene on chromosome 3p21.3, in human nasopharyn-geal carcinoma. Int J Cancer 2003;106:60–65.

27. Zochbauer-Muller S, Fong KM, Maitra A, et al. 5 CpG island methyl-ation of the FHIT gene is correlated with loss of gene expression in lungand breast cancer. Cancer Res 2001;61:3581–3585.

28. Oue N, Mitani Y, Motoshita J, et al. Accumulation of DNA methylationis associated with tumor stage in gastric cancer. Cancer 2006;106:1250–1259.

29. Ottini L, Falchetti M, Lupi R, et al. Patterns of genomic instability ingastric cancer: clinical implications and perspectives. Ann Oncol 2006;17:vii97–vii102.

30. Grunwald C, Koslowski M, Arsiray T, et al. Expression of multipleepigenetically regulated cancer/germline genes in nonsmall cell lungcancer. Int J Cancer 2006;118:2522–2528.

31. An C, Choi IS, Yao JC, et al. Prognostic significance of CpG islandmethylator phenotype and microsatellite instability in gastric carcinoma.Clin Cancer Res 2005;11:656–663.

32. Tsou JA, Galler JS, Siegmund KD, et al. Identification of a panel ofsensitive and specific DNA methylation markers for lung adenocarci-noma. Mol Cancer 2007;6:70.

33. Herman JG, Baylin SB. Gene silencing in cancer in association withpromoter hypermethylation. N Engl J Med 2003;349:2042–2054.

34. Baylin SB, Ohm JE. Epigenetic gene silencing in cancer—a mechanismfor early oncogenic pathway addiction? Nat Rev Cancer 2006;6:107–116.

35. Breuer RH, Postmus PE, Smit EF. Molecular pathology of non-small-cell lung cancer. Respiration 2005;72:313–330.

36. Watanabe N, Okochi-Takada E, Yagi Y, et al. Decreased fidelity inreplicating DNA methylation patterns in cancer cells leads to densemethylation of a CpG island. Curr Top Microbiol Immunol 2006;310:199–210.

37. Esteller M. Epigenetics in cancer. N Engl J Med 2008;358:1148–1159.38. Seng TJ, Currey N, Cooper WA, et al. DLEC1 and MLH1 promoter

methylation are associated with poor prognosis in non-small cell lungcarcinoma. Br J Cancer 2008;99:375–382.

39. Brock MV, Gou M, Akiyama Y, et al. Prognostic importance ofpromoter hypermethylation of multiple genes in esophageal adenocar-cinoma. Clin Cancer Res 2003;9:2912–2919.

40. Ogino S, Odze RD, Kawasaki T, et al. Correlation of pathologic featureswith CpG island methylator phenotype (CIMP) by quantitative DNAmethylation analysis in colorectal carcinoma. Am J Surg Pathol 2006;30:1175–1183.

41. Zhang C, Li Z, Cheng Y, et al. CpG island methylator phenotypeassociation with elevated serum alpha-fetoprotein level in hepatocellularcarcinoma. Clin Cancer Res 2007;13:944–952.

42. Suzuki M, Sunaga N, Shames DS, et al. RNA interference-mediatedknockdown of DNA methyltransferase 1 leads to promoter demethyl-ation and gene re-expression in human lung and breast cancer cells.Cancer Res 2004;64:3137–3143.



CpG Island Methylator Phenotype Involving Chromosome 3p Confers an Increased Risk of Non-small Cell Lung CancerMETHODSNSCLC Cell Lines and SpecimensBisulfite ModificationMethylation-Specific PCRDNA Demethylation of NSCLC Cell LinesMTT AssayDefinition of CIMPStatistical Analysis

RESULTSRelationship Between 3pCIMP and NSCLC Cell ProliferationMethylation Profile of Multiple Genes on 3p in PBMC From NSCLC PatientsFrequency of 3pCIMP in PBMC From NSCLC Patients and Normal Blood Samples3pCIMP of PBMC in Association with Increased Risk for NSCLC

DISCUSSIONACKNOWLEDGMENTSREFERENCES

Documents

CpG Island Methylator Phenotype Involving Chromosome 3p … · 2017. 2. 15. · 3p Confers an Increased Risk of Non-small Cell Lung Cancer Zeyi Liu, ... In this study, we tested hypothesis