MOLECULAR CARCINOGENESIS

Common Genetic Variants in TERT Contribute toRisk of Cervical Cancer in a Chinese Population

Sumin Wang,1 Jiangping Wu,1 Lingmin Hu,2,3,4 Chenyue Ding,2,3,4 Yanjing Kan,1 Yan Shen,1

Xiaojun Chen,5 Hongbing Shen,2,3,4 Xirong Guo,1 and Zhibin Hu2,3,4*1Nanjing Maternity and Child Health Hospital of Nanjing Medical University, Nanjing, China2Key Lab of Reproductive Medicine, Nanjing Medical University, Nanjing, China3Department of Epidemiology and Biostatistics, Ministry of Education Key Lab for Modern Toxicology, School of PublicHealth, Nanjing Medical University, Nanjing, China4Section of Clinical Epidemiology, Jiangsu Key Lab of Cancer Biomarkers, Prevention and Treatment, Cancer Center,Nanjing Medical University, Nanjing, China5Tumor Hospital of Nantong City, Nantong, China

Single-nucleotide polymorphisms (SNPs) of TERT rs2736098, rs2736100, and CLPTM1L rs402710 at 5p15.33 are

significantly associated with risk of a spectrum of cancers. However, cervical cancer has been rarely evaluated. In thisstudy, we genotyped the three SNPs in a case–control study with 1,033 cervical cancer cases and 1,053 cancer-freecontrols in a Chinese population. Logistic regression analyses showed that the two TERT SNPs both significantly associ-ated with cervical cancer risk in the recessive model (rs2736098, AA vs. AG/GG: adjusted OR ¼ 1.35, 95%

CI ¼ 1.06–1.72; rs2736100, CC vs. AC/AA: adjusted OR ¼ 1.38, 95% CI ¼ 1.11–1.73). However, no associationwas found between CLPTM1L rs402710 and cervical cancer. These results suggest that genetic variants in 5p15.33,especially in TERT, may be markers for susceptibility to cervical cancer. � 2012 Wiley Periodicals, Inc.

Key words: TERT; CLPTM1L; polymorphism; cervical cancer

INTRODUCTION

Cervical cancer is one of the common cancers inwomen throughout the world. It is responsible for250,000 deaths per year and approximately 80% ofcervical cancer cases emerged in developing coun-tries [1,2]. It is well established that human papil-lomavirus (HPV) infection is the primary andnecessary cause of cervical cancer [1,3,4]. However,only a small proportion of HPV infected womenprogress to cervical cancer, which suggests thathost genetic factors may also play a role in cervicalcarcinogenesis [4].The human 5p15.33 locus contains two known

susceptible genes: TERT (human telomerase reversetranscriptase) and CLPTM1L (alias CRR9; cleft lipand palate transmembrane 1 like), which havebeen implicated in carcinogenesis. Normally, TERTis only expressed in embryonic stem cells andgerm cells, and abnormal expression of TERTmRNA and protein has been observed in manytumors, including cervical cancer [5–11]. Telome-rase activation is an early-stage event in cervicalcarcinogenesis and is associated with both initia-tion and progression of cervical lesions [12].CLPTM1L is a predicted transmembrane proteinthat is expressed in a range of normal and malig-nant tissues including the cervix, lung, breast, ova-ry, and skin. Expression of CLPTM1L has been

shown to sensitize ovarian cancer cells to cisplatin-induced apoptosis [13].Recently, several genome-wide association studies

(GWAS) have reported variations in the TERT-CLPTM1L (5p15.33) locus contribute to predisposi-tion of cancers, they found that TERT rs2736098,rs2736100, and CLPTM1L rs401681 were associatedwith risk of many types of cancers [14–19]. In addi-tion, CLPTM1L rs402710 also found to be associat-ed with multiple cancers risk though not allcancers studied [20,21]. Furthermore, rs401681 inCLPTM1L existed high linkage disequilibrium (LD)with rs402710 (D0 ¼ 1 and r2 ¼ 0.89 for CHB), wasreported to be associated with cervical cancer risk

Additional Supporting Information may be found in the onlineversion of this article.

Abbreviations: HPV, human papillomavirus; GWAS, genome-wide association studies; LD, linkage disequilibrium; OR, odds ratio;CI, confidence interval.

Sumin Wang, Jiangping Wu, and Lingmin Hu contributed equallyto this work.

*Correspondence to: Department of Epidemiology and Biostatis-tics, School of Public Health, Nanjing Medical University, 140 Hanz-hong Rd, Nanjing 210029, China.

Received 25 August 2011; Revised 1 December 2011; Accepted14 December 2011

DOI 10.1002/mc.21872

Published online in Wiley Online Library(wileyonlinelibrary.com).

� 2012 WILEY PERIODICALS, INC.

in an Icelandic GWAS study with 249 cervicalcancer cases and 3,667 controls [14]. However, LDbetween TERT rs2736098 and rs2736100 (D0 ¼ 0.53and r2 ¼ 0.28 for CHB) is relatively low. Therefore,in this study, we hypothesized that three reportedgenetic variants (TERT rs2736098, rs2736100 andCLPTM1L rs402710) at 5p15.33 contribute to cervi-cal cancer risk in Chinese populations. To test thishypothesis, we genotyped these three SNPs in anongoing case–control study of 1,033 cervical can-cer cases and 1,053 frequency-matched cancer-freecontrols in a Chinese population.

MATERIALS AND METHODS

Participants

The study was approved by the Institutional Re-view Board of Nanjing Medical University. The1,033 newly diagnosed, histologically confirmedincident cervical cancer patients were consecutive-ly recruited between March 2006 and June 2010from the First Affiliated Hospital of Nanjing Medi-cal University and the Nantong Tumor Hospital,Jiangsu, China. A total of 1,053 controls were ran-domly selected from a pool of more than 40,000individuals, who participated in a community-based screening program for non-infectious dis-eases conducted in Jiangsu Province during thesame time period as the cases were recruited. All ofthe controls had no self-reported cancer historyand were frequency-matched to the cases on age(� 5 years). All of the cases and control subjectswere unrelated ethnic Han Chinese women. Afterinformed consent was obtained, each subject wasface-to-face interviewed by trained interviewersusing a structured questionnaire and a 5-ml venousblood sample was collected for genetic testing.

Genotyping

Genomic DNA was extracted from a leukocytepellet by traditional proteinase K digestion and fol-lowed by phenol–chloroform extraction and ethanolprecipitation. All the three SNPs were genotyped bythe TaqMan allelic discrimination Assay on an ABI7900 system (Applied Biosystems, Foster City, CA).The information on primers and probes were shownin Table 1. All the genotyping assays was performedwithout knowing the subjects’ case and control sta-tus, two blank (water) controls in each 384-well for-mat were used for quality control and more than10% samples were randomly selected to repeat,yielding a 100% concordant. The success rates ofgenotyping for rs2736098, rs2736100, and rs402710were 96%, 97%, and 99%, respectively.

Statistical Analysis

We used Student’s t-test or x2 test to evaluatedifferences in the distributions of demographiccharacteristics, selected variables (age, age at

menarche, age at natural menopausal, menopausalstatus, and parity), and genotypes of the threeSNPs between the cases and controls. The associa-tions between the three SNPs and cancer risks wereestimated by computing the odds ratios (ORs) andtheir 95% confidence intervals (CIs) from bothunivariate and multivariate logistic regressionanalyses with the adjustment on age, menopausalstatus, and parity. Homogeneity among differentstrata by selected variables (dichotomized groupsof age, menopausal status, parity, and dichoto-mized groups age at menarche) was assessed withthe x2 -based Q-test. Gene–environment multipli-cative interactions were tested by a general logisticregressionmodel: Y ¼ b0 þ b1 � A þ b2 � B þ b3 �(A � B) þ e. In the formula, Y is the logit of case–control status, b1 and b2 are the main effects offactor A and B, respectively, and b3 is the interac-tion term. All the statistical analyses wereperformed with SAS 9.1.3 software (SAS Institute,Cary, NC), and P < 0.05 in a two sided test wasconsidered statistically significant.

RESULTS

Selected characteristics of the 1,033 cervicalcancer cases and the 1,053 cancer-free controls aredescribed in Table 2. As expected, there was similardistribution of age in cases and controls (P ¼ 0.652).However, compared with the control subjects,the cervical cancer cases had significantly lowerage at menarche (P < 0.001) and higher parity(P ¼ 0.007).The genotype distributions of TERT rs2736098,

rs2736100, and CLPTM1L rs402710 in the casesand controls are described in Table 3. The observedgenotype frequencies for the three SNPs in thecontrols are all in Hardy–Weinberg equilibrium

Table 1. Information on Primers and Probes for Detect-ing the TERT–CLPTM1L Loci for TaqMan AllelicDiscrimination

Polymorphism Sequence (50–30)

rs2736098Primer F: CCGTGGTTTCTGTGTGGTGT

R: CGCCTGAGGAGTAGAGGAAGTProbe FAM-AGCACCACGCGGG-MGB

HEX-CACCACGCAGGCC-MGBrs2736100Primer F: GACGGGGAACAAAGGAGGA

R: GTTCTATCTCAGGCATCTTGACACCProbe FAM-CAAAGCTAAAGAAACAC-MGB

HEX-CAAAGCTACAGAACA-MGBrs402710Primer F: CGCTGAGACGGAGCAACG

R: CACCATGCCCACGTCTCACProbe FAM-CATACGCAGCCGCA-MGB

HEX-CATACGCAGCTGCA-MGB

2 WANG ET AL.

Molecular Carcinogenesis

(P ¼ 0.710 for rs2736098, P ¼ 0.637 for rs2736100,and P ¼ 0.741 for rs402710). In the multivariatelogistic regression models, TERT rs2736098 variantgenotype AA was associated with 1.35-fold in-creased risk for cervical cancer (95% CI ¼ 1.06–1.72), compared with the AG/GG genotype. ForTERT rs2736100, the CC genotype was also associ-ated with 1.38-fold increased cervical cancer risk(95% CI ¼ 1.11–1.73), compared with the AC/AAgenotype (Table 3). However, there was no associa-tion between CLPTM1L rs402710 and cervical can-cer risk (Table 3).

The associations of TERT rs2736098 and rs2736100with cervical cancer risk were further examined insubgroups by selected variables (SupplementaryTable 1). Test for interaction between the two SNPsand selected variables showed statistical evidencefor an interaction between parity and TERTrs2736100 in cervical cancer risk (P for inter-action ¼ 0.047) (Supplementary Table 1).

DISCUSSION

To the best of our knowledge, this is the firststudy that investigated the three reported genetic

Table 2. Demographic and Selected Variables in Cervical Cancer Cases and Controls

VariablesCases (n ¼ 1,033),

N (%)Controls (n ¼ 1053),

N (%) P-value

Age, year (mean � SD) 53.93 � 12.47 53.68 � 12.26 0.652Age at menarche, year (mean � SD) 15.68 � 1.97 16.06 � 1.92 <0.001Age at natural menopausal, year (mean � SD)a 49.02 � 3.86 49.18 � 4.06 0.496Menopausal status <0.001Premenopausal 418 (41.1) 425 (40.4)Natural menopause 545 (53.6) 620 (58.9)Unnatural menopause 54 (5.3) 8 (0.8)Parity 0.0070–1 428 (42.0) 496 (47.9)�2 591 (58.0) 540 (52.1)

aAge at natural menopause information was available in 539 breast cancer cases and 619 controls.

Table 3. Logistic Regression Analyses on Associations Between TERT–CLPTM1L Variations and Risk of Cervical Cancer

Variable Cases (n ¼ 1033), N (%) Controls (n ¼ 1053), N (%) OR (95% CI) OR (95% CI)a

rs2736098 (G>A)GG 375 (37.8) 397 (39.1) 1.00AG 444 (44.7) 480 (47.3) 0.98 (0.81–1.19) 0.98 (0.80–1.18)AA 174 (17.5) 138 (13.6) 1.34 (1.03–1.74) 1.33 (1.02–1.74)DOM 618/375 618/397 1.06 (0.88–1.27) 1.06 (0.88–1.27)REC 174/819 138/877 1.35 (1.06–1.72) 1.35 (1.06–1.72)ADD 1.11 (0.98–1.26) 1.11 (0.98–1.26)

rs2736100 (A>C)AA 322 (31.9) 352 (35.0) 1.00AC 462 (45.7) 480 (47.7) 1.05 (0.86–1.28) 1.02 (0.84–1.25)CC 226 (22.4) 174 (17.3) 1.42 (1.11–1.82) 1.40 (1.09–1.81)DOM 688/322 654/352 1.15 (0.96–1.38) 1.12 (0.93–1.36)REC 226/784 174/832 1.38 (1.11–1.72) 1.38 (1.11–1.73)ADD 1.17 (1.04–1.33) 1.16 (1.03–1.32)

rs402710 (C>T)CC 479 (46.9) 494 (47.2)CT 445 (43.5) 453 (43.3) 1.01 (0.85–1.22) 1.03 (0.86–1.24)TT 98 (9.6) 99 (9.5) 1.02 (0.75–1.39) 1.01 (0.74–1.38)DOM 543/479 552/494 1.02 (0.85–1.21) 1.02 (0.86–1.22)REC 98/924 99/947 1.02 (0.76–1.36) 1.00 (0.74–1.34)ADD 1.01 (0.89–1.15) 1.01 (0.89–1.16)

OR, odds ratio; CI, confidence interval; DOM, heterozygote/mutational homozygote vs. wild homozygote; REC, mutational homozy-gote vs. heterozygote/wild homozygote; ADD, mutational homozygote vs. heterozygote vs. wild homozygote.aAdjusted by age, menopausal status, and parity.

GENETIC VARIANTS IN TERT AND CERVICAL CANCER 3

Molecular Carcinogenesis

variants (TERT rs2736098, rs2736100, and CLPTM1Lrs402710) at 5p15.33 and cervical cancer risk inChinese populations. Overall, we found that theTERT rs2736098 and rs2736100, but not CLPTM1Lrs402710, may contribute to cervical cancer risk.Several lines of evidence support our findings.

Studies have found that telomerase is stronglyrepressed in normal tissues but expressed in mostcancer cells, suggesting that the activation of telo-merase is a critical step in human carcinogenesis[22,23]. TERT, as the reverse transcriptase compo-nent of telomerase, tightly regulate the telomeraseactivity at both transcriptional and posttransla-tional levels. Introduction of the HPV-16 E6/E7gene, and the TERT gene into normal endometrialglandular cells was sufficient to generate immortal-ized cells [24]. These findings underscore the im-portance of TERT in the etiology of cancer, likelycervical cancer as well.

TERT rs2736098 and rs2736100 have beenreported to be associated with risks of lung cancer[14,20,25], specifically lung adenocarcinoma, glio-ma, and testicular germ cell tumor (rs2736100)[16,17,25], as well as breast cancer, basal cell carci-noma, carcinomas of the urinary bladder, andprostate cancer (rs2736098) [14,26]. However, inan Icelandic GWAS with 249 cervical cancer casesand 3667 controls, Rafnar et al. [14] did not find asignificant association between rs2736098 andcervical cancer risk. In the contrary, this studyshowed an association between rs401681 inCLPTM1L and cervical cancer risk [14]. In the cur-rent study, we found both TERT rs2736098 andrs2736100, but not CLPTM1L rs402710 (in highlyLD with rs401681: D0 ¼ 1 and r2 ¼ 0.89), were sig-nificantly associated with cervical cancer risk inthe recessive model. We also calculate the LDvalue based on our own data for the two SNPs inTERT (D0 ¼ 0.53 and r2 ¼ 0.24), which is similar toHapMap data (D0 ¼ 0.53 and r2 ¼ 0.28 for CHB).We then concluded that both rs2736098 andrs2736100 in TERT are important markers of cervi-cal cancer susceptibility in Chinese populations.The discrepancies between our current study andthe published studies might be due to the differ-ence in LD structure and marker SNPs betweendifferent ethnic populations, but also the study de-sign and sample size. To date, no study conductedassociation analysis between rs2736100 or rs402710and cervical cancer. Only one study researched onrs2736098 and cervical cancer [14], we then per-formed a mini meta-analysis on rs2736098 andcervical cancer susceptibility with allele geneticmodel by pooling our current data and the pub-lished data. As shown in Supplementary Figure 1,rs2736098 was associated with cervical cancer risk(OR ¼ 1.12, P ¼ 0.042). In our study, TERTrs2736100 was significantly interacted with parity,a known risk factor for cervical cancer, but how

parity is related to telomerase activity, HPV infec-tion and cervical cancer remains largely unknown.Although TERT rs2736098 was correlated withshorter telomeres, no published data are availableabout the effect of TERT rs2736098 and rs2736100on TERT expression or telomerase activity, andwhy the genetic effects of the two SNPs was in therecessive model also warrants further investiga-tions. However, in the current study, we did nothave the tissue samples for HPV typing. AlthoughHPV infection is a necessary pathogenic factorfor cervical cancer, it is possible that associationsmay be stronger with one particular subtype overanother.To further define the potential causal SNP

around TERT, we searched the chromosome regionabout 300 kb around TERT in the HapMap data-base (HapMapData Rel 24/phaseII Mar08, on NCBIB36 assembly, dbSNP b126) for both European andChinese Han populations. However, there was noSNP linked to the two SNPs with r2 above 0.40 ineither European or Chinese Han populations.Therefore, further studies are recommend to finemapping the region, based on comprehensive rese-quencing analysis, to narrow the set of genetic var-iants worthy of functional studies.In conclusion, we found that TERT rs2736098

and rs2736100 at 5p15.33 are important markersof cervical cancer susceptibility in Chinese popula-tions. Further validation studies with different eth-nic background, together with the resequencing ofthe marked region, HPV typing, and functionalevaluations are warranted.

ACKNOWLEDGMENTS

This work was supported in part by the Programfor Changjiang Scholars and Innovative ResearchTeam in University (IRT0631).

REFERENCES

1. Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statis-tics, 2002. CA Cancer J Clin 2005;55:74–108.

2. Parkin DM, Bray F. Chapter 2: The burden of HPV-relatedcancers. Vaccine 2006;24:11–25.

3. Walboomers JM, Jacobs MV, Manos MM, et al. Humanpapillomavirus is a necessary cause of invasive cervicalcancer worldwide. J Pathol 1999;189:12–19.

4. Bosch FX, Lorincz A, Munoz N, Meijer CJ, Shah KV. Thecausal relation between human papillomavirus and cervi-cal cancer. J Clin Pathol 2002;55:244–265.

5. Hiyama E, Yokoyama T, Tatsumoto N, et al. Telomeraseactivity in gastric cancer. Cancer Res 1995;55:3258–3262.

6. Kim HR, Christensen R, Park NH, Sapp P, Kang MK. Ele-vated expression of hTERT is associated with dysplastic celltransformation during human oral carcinogenesis in situ.Clin Cancer Res 2001;7:3079–3086.

7. Clark GM, Osborne CK, Levitt D, Wu F, Kim NW. Telome-rase activity and survival of patients with node-positivebreast cancer. J Natl Cancer Inst 1997;89:1874–1881.

8. Wisman GB, De Jong S, Meersma GJ, et al. Telomerase in(pre)neoplastic cervical disease. Hum Pathol 2000;31:1304–1312.

4 WANG ET AL.

Molecular Carcinogenesis

9. Tatsumoto N, Hiyama E, Murakami Y, et al. High telome-rase activity is an independent prognostic indicator ofpoor outcome in colorectal cancer. Clin Cancer Res 2000;6:2696–2701.

10. Falchetti ML, Pallini R, D’Ambrosio E, et al. In situ detec-tion of telomerase catalytic subunit mRNA in glioblastomamultiforme. Int J Cancer 2000;88:895–901.

11. Wang L, Soria JC, Kemp BL, Liu DD, Mao L, Khuri FR.hTERT expression is a prognostic factor of survival inpatients with stage I non-small cell lung cancer. Clin CancerRes 2002;8:2883–2889.

12. Zheng PS, Iwasaka T, Yokoyama M, Nakao Y, Pater A,Sugimori H. Telomerase activation in in vitro and in vivocervical carcinogenesis. Gynecol Oncol 1997;66:222–226.

13. Yamamoto K, Okamoto A, Isonishi S, Ochiai K, Ohtake Y.A novel gene, CRR9, which was up-regulated in CDDP-resistant ovarian tumor cell line, was associated withapoptosis. Biochem Biophys Res Commun 2001;280:1148–1154.

14. Rafnar T, Sulem P, Stacey SN, et al. Sequence variants atthe TERT-CLPTM1L locus associate with many cancertypes. Nat Genet 2009;41:221–227.

15. Hu Z, Wu C, Shi Y, et al. A genome-wide associationstudy identifies two new lung cancer susceptibility loci at13q12.12 and 22q12.2 in Han Chinese. Nat Genet 2011;43:792–796.

16. Turnbull C, Rapley EA, Seal S, et al. Variants near DMRT1,TERT and ATF7IP are associated with testicular germ cellcancer. Nat Genet 2010;42:604–607.

17. Shete S, Hosking FJ, Robertson LB, et al. Genome-wideassociation study identifies five susceptibility loci forglioma. Nat Genet 2009;41:899–904.

18. Petersen GM, Amundadottir L, Fuchs CS, et al.A genome-wide association study identifies pancreaticcancer susceptibility loci on chromosomes 13q22.1,1q32.1 and 5p15.33. Nat Genet 2010;42:224–228.

19. Stacey SN, Sulem P, Masson G, et al. New common var-iants affecting susceptibility to basal cell carcinoma. NatGenet 2009;41:909–914.

20. McKay JD, Hung RJ, Gaborieau V, et al. Lung cancersusceptibility locus at 5p15.33. Nat Genet 2008;40:1404–1406.

21. Gago-Dominguez M, Jiang X, Conti DV, et al. Geneticvariations on chromosomes 5p15 and 15q25 and bladdercancer risk: findings from the Los Angeles–Shanghai blad-der case–control study. Carcinogenesis 2011;32:197–202.

22. Hahn WC, Counter CM, Lundberg AS, Beijersbergen RL,Brooks MW, Weinberg RA. Creation of human tumourcells with defined genetic elements. Nature 1999;400:464–468.

23. Kim NW, Piatyszek MA, Prowse KR, et al. Specific associa-tion of human telomerase activity with immortal cells andcancer. Science 1994;266:2011–2015.

24. Kyo S, Nakamura M, Kiyono T, et al. Successful immortali-zation of endometrial glandular cells with normal structur-al and functional characteristics. Am J Pathol 2003;163:2259–2269.

25. Landi MT, Chatterjee N, Yu K, et al. A genome-wideassociation study of lung cancer identifies a region ofchromosome 5p15 associated with risk for adenocarcino-ma. Am J Hum Genet 2009;85:679–691.

26. Savage SA, Chanock SJ, Lissowska J, et al. Genetic varia-tion in five genes important in telomere biology and riskfor breast cancer. Br J Cancer 2007;97:832–836.

GENETIC VARIANTS IN TERT AND CERVICAL CANCER 5

Molecular Carcinogenesis