Vascular Endothelial Growth Factor (VEGF)Gene Polymorphisms and Gastric Cancer Riskin a Chinese Han Population

Yong Zhou,1 Ni Li,2 Wen Zhuang,1 and Xiaoting Wu1*1Department of Gastrointestinal Surgery, West China Hospital, Sichuan University, Chengdu, China2Department of Ophthalmology, West China Hospital, Sichuan University, Chengdu, China

The association between vascular endothelial growth factor (VEGF) gene polymorphisms and gastric cancer risk isstill controversial and ambiguous. The objective of this study was to investigate the association between VEGF gene

polymorphisms and gastric cancer risk in Chinese Han patients. We extracted the peripheral blood samples in 150patients with gastric cancer and 150 controls. Polymerase chain reaction restriction fragment length polymorphism(PCR-RFLP) analysis was performed to detect three VEGF gene polymorphisms (�634G/C, þ936C/T, and þ1612 G/A)

in these patients. Patients with gastric cancer had a significantly higher frequency of 1612 AA genotype (OR¼ 6.26,95% CI¼ 1.80, 21.85; P¼ 0.004) than controls. Patients with cardia gastric cancer had a significantly lower frequencyof AA (OR¼ 0.11, 95% CI¼ 0.01, 0.89; P¼ 0.04) than those with noncardia gastric cancer. Patients with Lauren’sdiffuse-type gastric cancer had a significantly higher frequency of AA (OR¼ 3.41, 95% CI¼ 1.22, 9.55; P¼ 0.02)

than those with Lauren’s intestinal-type gastric cancer. The �634G/C and þ936C/T gene polymorphisms were notassociated with a risk of GC and its progression. This study suggests that the VEGF þ1612G/A gene polymorphismsmay be associated with gastric cancer in Chinese Han patients, and that difference in genotype distribution may be

associated with the location and Lauren’s classification of gastric cancer. � 2010 Wiley-Liss, Inc.

Key words: gastric cancer; VEGF; gene polymorphism

INTRODUCTION

Gastric cancer, the second leading cause of deathfrom cancer throughout the world, is an importanthealth problem. A 2005 analysis of the worldwideincidence of and mortality from cancer showed that934,000 cases of gastric cancer occurred in 2002 andthat 700,000 patients die annually of this disease [1].Despite the overall decline in gastric cancer rates inmost of theWesternWorld, gastric cancer remains aserious fatal disease throughout much of the rest ofthe world. Thirty-eight percent of worldwide casesoccur in China, where it remains the most commoncancer in both sexes as it is elsewhere in EasternAsia [2]. Conversely, the incidence rates of adeno-carcinomas of the proximal stomach and distalesophagus have been increasing, particularly inthe Western World [3]. Gastric carcinogenesis is acomplex, multistep, and multifactorial process, inwhich many factors are implicated.Angiogenesis is an essential step for tumor growth,

playing a critical role in invasion and metastasis [4].The vascular endothelial growth factor (VEGF) is oneof the most potent endothelial cell mitogens andplays and important role in angiogenesis [5–6].Different clinical studies have shown that high levelsof VEGF expression and increased microvesseldensity in tumors were associated with advanced

stage disease andworse prognosis for various types oftumors, including gastric cancer [7]. The VEGF geneis located on chromosome 6p21.3 and consists ofeight exons that exhibit alternative splicing to forma family of proteins [8]. Several single-nucleotidepolymorphisms (SNPs) in the VEGF gene have beenshown to affect the expression of the gene [9–10].Three potentially functional SNPs (�634 G/C,þ936C/T, and þ1612 G/A) in the 3-untranslatedregion (UTR) of the VEGF gene, were shown to affectVEGF plasma levels. VEGF gene polymorphismshave been reported to be associated with cancersof the lung [11], colorectum [12–13], breast, andbladder [14–17].The association between VEGF gene polymor-

phisms and gastric cancer risk is still controversialandambiguous [9,18–27]. Theobjectiveof this study

MOLECULAR CARCINOGENESIS 50:184–188 (2011)

� 2010 WILEY-LISS, INC.

Abbreviations: VEGF, vascular endothelial growth factor; SNPs,single nucleotide polymorphisms; OR, odds ratio; CI, confidenceinterval.

*Correspondence to: Department of Gastrointestinal Surgery,West China Hospital, Sichuan University, 37 Guo Xue Rd., Chengdu610041, Sichuan Province, China.

Received 26 July 2010; Revised 12 October 2010; Accepted 12October 2010

DOI 10.1002/mc.20703

Published online 10 December 2010 in Wiley Online Library(wileyonlinelibrary.com).

was to investigate the association between VEGFgene polymorphisms and gastric cancer risk inChinese Han patients.

MATERIALS AND METHODS

Study Subjects

From 2008 to 2010, 150 healthy volunteers wereenrolled in this study as the control group and150patientswithgastric cancer fromthedepartmentof gastrointestinal surgery of the West China Hospi-tal in Sichuan University (Chengdu, China) wereenrolled as the study group. The Chinese Hanpopulation was collected from the same geographicregion. Informed consents were obtained accordingto theDeclaration ofHelsinki. Informed consentwasobtained from both groups that participated in thisstudy. The clinicopathologic findings of the cancergroup were collected. All the pathological cell typesin the cancer group were adenocarcinoma. Tumorlocation was defined as being either cardia or non-cardia (antrum and corpus). When a tumor waslocated on the border, the tumor sitewas determinedaccording to its dominant location. The tumor celltypes were classified as either differentiated or un-differentiated type. The tumor stages were dividedinto early stage (tumor limited to mucosal orsubmucosal layer) and advanced stage. The surgicalspecimens of gastric cancer were classified as intes-tinal type and diffuse types according to the Lauren’sclassification. Smoking habit was defined as non-smoker (smoked <100 cigarettes in lifetime) andsmoker. Alcohol consumption was defined as non-drinker and drinker (consumed more than one cup,200mL, per day).

Genotyping

The genomic DNA was prepared from peripheralblood by use of QIAamp DNA blood mini kit(QIAGEN Inc., Valencia, CA). The VEGF �634G/C,þ936C/T, and þ1612G/A gene polymorphismswere then determined using a polymerase chainreaction restriction fragment length polymorphism(PCR-RFLP) assay. Based on the GenBank referencesequence, the PCR primers designed for �634G/C,þ936C/T, and þ1612G/A were 5’-GTA GCA AGAGCT CCA GAG AGA AGT-3’ (forward) and 5’-TGGACG AAA AGT TTC AGT GCG ACG-3’ (reverse);5’-CTC GGT GAT TTA GCA GCA AG-3’ (forward)and 5’-CTCGGTGATTTAGCAGCAAG-3’ (reverse);5’-CACATGCTGCACGCGCATCTCA-3’ (forward)and 5’-ACC CCA GGA AGG GGA GCA GGA-3’(reverse), respectively. The PCR reactions werecarried out in a 20 mL reaction volume containing100ng genomic DNA, 25pmol/L each primer,0.2mmol/L deoxyuceotide triphosphates,10mmol/L Tris–HCl (pH8.3), 50mmol/L KCl,1.5mmol/L MgCl2, and 1 unit of Taq polymerase(Takara Shuzo Co., Otsu, Shiga, Japan). The PCR

cycle conditions consisted of an initial denaturationstep at 948C for 5min, followed by 35 cycles of 30 s at948C, 30 s at 628C, 30 s at 728C, andafinal elongationat 728C for 10min. The PCR products were digestedovernight with the appropriate restriction enzymes(New England Biolabs, Beverly, MA), which wereBsmFI, NlaIII, andMnlI for the �634G/C, þ936C/T,and þ1612G/A polymorphisms, respectively. Thedigested PCRproductswere resolved on a 3%agarosegel and stained with ethidium bromide for visual-ization under UV light. For quality control, thegenotyping analysis was done blind as regards thesubjects. The selected PCR-amplified DNA sampleswere also examined by DNA sequencing to confirmthe genotyping results.

Statistical Analysis

STATAprogramversion 8.0was used to perform allthe statistical analyses (StataCorp LP, TX). Compar-ison of genotypes of the VEGF variants between thecases and controls were evaluated by using the chi-square test. Crude odds ratios (OR) and adjusted ORsfor sex and age with 95% confidence interval (CI)were as calculated by logistic regression analysis.The P-value <0.05 was considered statisticallysignificant. All statistical tests were two-sided. Evi-dence for deviation from Hardy–Weinberg equili-brium of alleles at individual loci was assessed byexact tests.

RESULTS

Characteristics and risk factors in patients withgastric cancer and controls are presented in Table I.Cases and controls did not differ regarding gender(P¼ 0.64) or age (P¼0.14). In this study, gastriccancer cases were lighter than controls (P¼0.004).Smoking is a risk factor of gastric cancer (OR¼1.72,95% CI¼1.07, 2.76; P¼0.02). Drinking is also a riskfactor of gastric cancer (OR¼1.70, 95% CI¼1.07,2.71; P¼0.02). The allele distribution for the inves-tigated VEGF gene polymorphisms did not deviatefrom Hardy–Weinberg equilibrium for the controlsand gastric cancer patients.Patients with gastric cancer had a significantly

higher frequency of 1612 AA genotype (OR¼6.26,95%CI¼1.80, 21.85; P¼0.004) than controls (TableII). Patients with cardia gastric cancer had a signi-ficantly lower frequency of AA (OR¼ 0.11, 95%CI¼0.01, 0.89; P¼0.04) than those with noncardiagastric cancer (Table III). Patients with Lauren’sdiffuse-type gastric cancer had a significantly higherfrequency of AA (OR¼3.41, 95% CI¼1.22, 9.55;P¼ 0.02) than those with Lauren’s intestinal-typegastric cancer (Table III). When stratified by thestage and histological differentiation of gastriccancer, no statistically significant result wasobserved (Table III). The �634G/C and þ936C/Tgene polymorphisms were not associated with a riskof GC and its progression (Tables II and III).

VEGF POLYMORPHISMS AND GASTRIC CANCER RISK 185

Molecular Carcinogenesis

DISCUSSION

In the present study, we investigated the associa-tion between VEGF gene polymorphisms and gastriccancer risk in Chinese Han patients. Our studysuggests that the VEGF þ1612G/A gene polymor-phisms may be associated with gastric cancer inChinese Han patients, and that difference in geno-typedistributionmaybeassociatedwith the locationand Lauren’s classification of gastric cancer.

A genetic predisposition to gastric cancer has beensuggested by both epidemiological studies and casereports of gastric cancer families [28]. Recent studiessuggest that SNPs may be related to the tumori-genesis of gastric cancer [29]. Individual geneticsusceptibilitymay be critical in a variety of processesrelevant to gastric cancer tumorigenesis, such as(i)mucosalprotection in the faceofHelicobacter pyloriinfection and other carcinogens, (ii) the inflamma-tory response, which conditions the maintenance,

Table I. Characteristics and Risk Factors in Patients With Gastric Cancer and Controls

Variable

Cases Controls

P(n¼ 150) (n¼ 150)

Gender (%)Male 84(56.0) 80(53.3) 0.64Female 66(44.0) 70(46.7) 0.64

Age mean (SD) year 57.1(10.2) 55.4(9.7) 0.14Height mean (SD) cm 167.2(9.7) 166.7(9.5) 0.65Weight Mean(SD), kg 60.6(11.5) 64.3(10.7) 0.004Smoking status (%)Yes 66(44.0) 47(31.3) 0.02No 84(56.0) 103(68.7) 0.02

Drinking (%)Yes 70(46.7) 62(34.0) 0.02No 80(53.3) 88(66.0) 0.02

Location (%)Cardia 48(32.0) NANoncardia 102(68.0) NA

Stage (%)Advanced 91(60.7) NAEarly 59(39.3) NA

Lauren’s classification (%)Diffuse 42(28.0) NAIntestinal 108(72.0) NA

Histological differentiation (%)Differentiated 78(52.0) NAUndifferentiated 72(48.0) NA

NA, not applicable.

Table II. VEGF Gene Polymorphisms Among Gastric Cancer Patients and Controls

Genotype Cases n (%) Controls n (%) OR (95%CI) P

�634 GG 74(49.3) 76(50.7) 0.95(0.60, 1.49) 0.82�634 GC 47(31.3) 44(29.3) 1.10(0.67, 1.80) 0.71�634 CC 29(19.3) 30(20.0) 0.96(0.54, 1.69) 0.89�634 G allele frequency 195(65.0) 196(65.3) 0.99(0.70, 1.38) 0.93�634 C allele frequency 105(35.0) 104(34.7) 1.01(0.73, 1.42) 0.93þ936 CC 97(64.7) 94(62.7) 1.09(0.68, 1.75) 0.72þ936 CT 45(30.0) 49(32.7) 0.88(0.54, 1.44) 0.62þ936 TT 8(5.3) 7(4.7) 1.15(0.41, 3.26) 0.79þ936 C allele frequency 239(79.7) 237(79.0) 1.04(0.70, 1.55) 0.84þ936 T allele frequency 61(20.3) 63(21.0) 0.96(0.65, 1.43) 0.84þ1612 GG 104(69.3) 112(74.7) 0.77(0.46, 1.27) 0.30þ1612 GA 29(19.3) 35(23.3) 0.79(0.45, 1.37) 0.40þ1612 AA 17(11.3) 3(2.0) 6.26(1.80, 21.85) 0.004þ1612 G allele frequency 237(79.0) 259(86.3) 0.60(0.39, 0.92) 0.02þ1612 A allele frequency 63(21.0) 41(13.7) 1.68(1.09, 2.58) 0.02

186 ZHOU ET AL.

Molecular Carcinogenesis

severity and outcome of the Helicobacter pyloriinfection, (iii) the functioning of carcinogen detox-ification and antioxidant protection, (iv) cell pro-liferation ability, (v) the intrinsic variability of DNArepair processes, and (vi) the cell apoptotic pathway[30–31]. The mechanism of human gastric tumori-genesis is still relatively unknown, and singlenucleotide polymorphisms can be used as a tool insearching for genetic variations of the disease geneand susceptibility, and to increase understanding ofthe disease mechanism [32].Several studies have investigated the association

between VEGF gene polymorphisms and gastriccancer risk. However, the results remained incon-sistent. Tahara et al. [19] have investigated the effectof VEGF gene polymorphisms on the risk of gastricpremalignant condition in a Japanese population.They found that the þ1612G/A, but not þ936C/Tpolymorphisms of the VEGF gene was associatedwith gastric premalignant condition in older indi-viduals [19]. Their data also suggested that theþ1612G/A, but not þ936C/T polymorphisms inthe 3’-UTR of VEGF gene was associated with thesusceptibility to gastric cancer in the Japanesepopulation [18]. The study by Bae et al. [9] suggestedthat the VEGF þ936C/T polymorphism was asusceptibility factor for gastric cancer in a Koreanpopulation. The study by Chae et al. [27] evaluatedthe potential association of three VEGF gene poly-morphisms (�460T/C, þ405G/C, and 936C/T) withthe susceptibility to andclinicopathologic character-istics of gastric cancer in a Korean population.They found that the þ405C and þ936T allele wasassociated with a significantly decreased susceptibil-ity to gastric cancer [27]. The studybyGuanet al. [20]suggested that the VEGF-634G/C SNP might be amarker for susceptibility to gastric cancer in a USpopulation. The study by Tzanakis et al. suggestedthat the VEGF-634G/C SNP might be a marker forsusceptibility to gastric cancer in aGreekpopulation.But no significant differences in the remaining ofVEGF polymorphisms (�2578C/A, �1154G/A, andþ936C/T) between gastric cancer cases and controlswere detected. The study by Al-Moundhri et al. [22]suggested that VEGF polymorphisms (þ405 G/C,�460 T/C, and þ936C/T) had no role in gastriccancer risk predisposition in an Omani population.The data by Ke et al. [23] did not support a significantassociation between VEGF SNPs (�2578C/A,�1498T/C, �634G/C, and þ936C/T) and the risk ofgastric cancer in a high risk Chinese population. Thereasons for such a discrepancy include the frequencyof the polymorphism in the population studied orlinkage disequilibrium with other, perhaps undis-covered, functional SNPs in the VEGF gene [33].Some shortcomings of this study should be

discussed. Firstly, these results should be interpretedwith caution because the population only fromChina, which reduces the possibility of confounding

Table

III.StratificationAnalysisofVEGFGenePolymorphismsin

GastricCancer

Variable

Cases

�634GG

�634GC

�634CC

þ936CC

þ936CT

þ936TT

þ1612GG

þ1612GA

þ1612AA

nP

nP

nP

nP

nP

nP

nP

nP

nP

Location

Cardia

48

24

0.91

15

0.99

90.90

31

0.99

14

0.88

30.73

36

0.30

11

0.45

10.04

Noncardia

102

50

32

20

66

31

568

18

16

Stage

Advanced

91

44

0.77

28

0.85

19

0.55

58

0.77

27

0.91

50.91

62

0.69

17

0.80

12

0.38

Early

59

30

19

10

39

18

342

12

5Lauren’sclassification

Diffuse

42

21

0.92

13

0.95

80.96

27

0.95

12

0.81

30.54

25

0.11

80.96

90.02

Intestinal

108

53

34

21

70

33

579

21

8Histologicaldifferentiation

Differentiated

78

38

0.88

25

0.84

15

0.97

50

0.88

23

0.88

40.91

54

0.98

15

0.97

90.93

Undifferentiated

72

36

22

14

47

22

450

14

8

VEGF POLYMORPHISMS AND GASTRIC CANCER RISK 187

Molecular Carcinogenesis

from ethnicity, but it does not permit extrapolationof the results to other ethnic groups. Secondly,further study is required to delineate the precisemechanisms involved and, potentially, to facilitatethe design of effective clinical trials. Thirdly,although this is a hospital based case control study,the selection bias may not be avoidable and thesubjects may not be representative of the generalpopulation. Finally, the study is based onunadjustedestimates.In conclusion, this study suggests that the VEGF

þ1612 G/A gene polymorphisms may be associatedwith gastric cancer inChineseHanpatients, and thatdifference in genotype distribution may be associ-ated with the location and Lauren’s classification ofgastric cancer.

ACKNOWLEDGMENTS

This study was supported by National NaturalScience Foundation of China (No.30901427).

REFERENCES

1. Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statistics2002. CA Cancer J Clin 2005;55:74–108.

2. Parkin DM, Bray F, Ferlay J, Pisani P. Estimating the worldcancer burden: Globocan 2000. Int J Cancer 2001;94:153–156.

3. Devesa SS, Fraumeni JF, Jr. The rising incidence of gastriccardia cancer. J Natl Cancer Inst 1999;91:747–749.

4. Ferrara N, Gerber HP, LeCouter J. The biology of VEGF and itsreceptors. Nat Med 2003;9:669–676.

5. Roy H, Bhardwaj S, Yla-Herttuala S. Biology of vascularendothelial growth factors. FEBS Lett 2006;580:2879–2887.

6. Belinsky GS, Claffey KP, Nambiar PR, Guda K, RosenbergDW. Vascular endothelial growth factor and enhancedangiogenesis do not promote metastatic conversion of anewly established azoxymethane-induced colon cancer cellline. Mol Carcinog 2005;43:65–74.

7. Iordache S, Saftoiu A, Georgescu CV, et al. Vascularendothelial growth factor expression and microvesseldensity-two useful tools for the assessment of prognosisand survival in gastric cancer patients. J Gastrointestin LiverDis 2010;19:135–139.

8. Vincenti V, Cassano C, Rocchi M, Persico G. Assignment ofthe vascular endothelial growth factor gene to humanchromosome 6p21.3. Circulation 1996;93:1493–1495.

9. Bae SJ, Ahn DH, Hong SP, et al. Gender-specific associationbetween polymorphism of vascular endothelial growthfactor (VEGF 936C>T) gene and patients with stomachcancer. Yonsei Med J 2008;49:783–791.

10. Watson CJ, Webb NJ, Bottomley MJ, Brenchley PE.Identification of polymorphisms within the vascular endo-thelial growth factor (VEGF) gene: Correlationwith variationin VEGF protein production. Cytokine 2000;12:1232–1235.

11. Lee SJ, Lee SY, Jeon HS, et al. Vascular endothelial growthfactor gene polymorphisms and risk of primary lung cancer.Cancer Epidemiol Biomarkers Prev 2005;14:571–575.

12. Bae SJ, Kim JW, Kang H, Hwang SG, Oh D, Kim NK. Gender-specific association between polymorphism of vascularendothelial growth factor (VEGF 936 C>T) gene and coloncancer in Korea. Anticancer Res 2008;28:1271–1276.

13. Dassoulas K, Gazouli M, Rizos S, et al. Common poly-morphisms in the vascular endothelial growth factor geneand colorectal cancer development, prognosis, and survival.Mol Carcinog 2009;48:563–569.

14. Eroglu A, Ozturk A, Cam R, Akar N. Vascular endothelialgrowth factor gene 936C/T polymorphism in breast cancerpatients. Med Oncol 2008;25:54–55.

15. Jacobs EJ, Feigelson HS, Bain EB, et al. Polymorphisms in thevascular endothelial growth factor gene and breast cancer inthe Cancer Prevention Study II cohort. Breast Cancer Res2006;8:R22.

16. Krippl P, LangsenlehnerU, RennerW, et al. A common936C/T genepolymorphismof vascular endothelial growth factor isassociated with decreased breast cancer risk. Int J Cancer2003;106:468–471.

17. Garcia-Closas M, Malats N, Real FX, et al. Large-scaleevaluation of candidate genes identifies associationsbetween VEGF polymorphisms and bladder cancer risk. PLoSGenet 2007;3:e29.

18. Tahara T, Shibata T, Nakamura M, et al. Effect of poly-morphisms in the 3’ untranslated region (3’-UTR) of vascularendothelial growth factor gene on gastric cancer and pepticulcer diseases in Japan. Mol Carcinog 2009;48:1030–1037.

19. Tahara T, Arisawa T, Shibata T, et al. Effect of polymorphismsin the 3’-untranslated region (3’-UTR) of VEGF gene ongastric pre-malignant condition. Anticancer Res 2009;29:485–489.

20. Guan X, Zhao H, Niu J, Tang D, Ajani JA, Wei Q. The VEGF-634G>C promoter polymorphism is associated with risk ofgastric cancer. BMC Gastroenterol 2009;9:77.

21. Guan X, Zhao H, Niu J, Tan D, Ajani JA, Wei Q. Poly-morphismsof TGFB1andVEGFgenes and survival of patientswith gastric cancer. J Exp Clin Cancer Res 2009;28:94.

22. Al-Moundhri MS, Al-Nabhani M, Burney IA, Al-Farsi AA, Al-Bahrani B. Gastric cancer risk predisposition and prognosticsignificance of vascular endothelial growth factor (VEGF)gene polymorphisms—A case-control study in an Omanipopulation. Mol Carcinog 2009;48:1170–1176.

23. Ke Q, Liang J, Wang LN, et al. Potentially functionalpolymorphisms of the vascular endothelial growth factorgene and risk of gastric cancer.Mol Carcinog 2008;47:647–651.

24. Kim JG, Sohn SK, ChaeYS, et al. Vascular endothelial growthfactor gene polymorphisms associated with prognosis forpatients with gastric cancer. Ann Oncol 2007;18:1030–1036.

25. Tzanakis N, Gazouli M, Rallis G, et al. Vascular endothelialgrowth factor polymorphisms in gastric cancer development,prognosis, and survival. J Surg Oncol 2006;94:624–630.

26. Demetrick DJ. Re: Vascular endothelial growth factor poly-morphisms in gastric cancer development, prognosis, andsurvival by Tzanakis N, Gazouli M, Rallis G. et al. J Surg Oncol2006;94:549–550.

27. Chae YS, Kim JG, Sohn SK, et al. Investigation of vascularendothelial growth factor gene polymorphisms and itsassociation with clinicopathologic characteristics in gastriccancer. Oncology 2006;71:266–272.

28. Bevan S, Houlston RS. Genetic predisposition to gastriccancer. Q J Med 1999;92:5–10.

29. Machado JC, Pharoah P, Sousa S, et al. Interleukin 1Band interleukin 1RN polymorphisms are associated withincreased risk of gastric carcinoma. Gastroenterology2001;121:823–829.

30. Gonzalez CA, Sala N, Capella G. Genetic susceptibility andgastric cancer risk. Int J Cancer 2002;100:249–260.

31. Lai KC, ChenWC, Tsai FJ, Li SY, Jeng LB. Arginine and prolinealleles of the p53gene are associatedwith different locationsof gastric cancer. Hepatogastroenterology 2005;52:944–948.

32. Loder N. Genetic variations can point the way to diseasegenes. Nature 1999;401:734.

33. Hofmann G, Langsenlehner U, Renner W, et al. Commonsingle nucleotide polymorphisms in the vascular endothelialgrowth factor gene and colorectal cancer risk. J Cancer ResClin Oncol 2008;134:591–595.

188 ZHOU ET AL.

Molecular Carcinogenesis