Association of Sequence Variants on Chromosomes 20, 11, and 5 (20q13.33, 11q23.3, and 5p15.33) With Glioma Susceptibility in a Chinese Population

American Journal of Epidemiology ª The Author 2011. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. Vol. 173, No. 8 DOI: 10.1093/aje/kwq457 Advance Access publication: February 24, 2011 Original Contribution Association of Sequence Variants on Chromosomes 20, 11, and 5 (20q13.33, 11q23.3, and 5p15.33) With Glioma Susceptibility in a Chinese Population Hongyan Chen, Yuanyuan Chen, Yao Zhao, Weiwei Fan, Keke Zhou, Yanhong Liu, Liangfu Zhou, Ying Mao, Qingyi Wei, Jianfeng Xu, and Daru Lu* * Correspondence to Dr. Daru Lu, State Key Laboratory of Genetic Engineering, Fudan-VARI Genetic Epidemiology Center and MOE Key Laboratory of Contemporary Anthropology, School of Life Sciences, Fudan University, 220 Handan Road, Shanghai 200433, China (e-mail: drlu@fudan.edu.cn). Initially submitted May 25, 2010; accepted for publication November 29, 2010. Two genome-wide association studies of glioma in European populations identified 14 genetic variants strongly associated with risk of glioma, but it is unknown whether these variants are associated with glioma risk in Asian populations. The authors genotyped these 14 variants in 976 glioma patients and 1,057 control subjects to evaluate their associations with risk of glioma, particularly high-grade glioma (glioblastoma; n ¼ 312), in a Chinese population (2004 2009). Overall, the authors identified 3 susceptibility loci for glioma risk at 20q13.33 (RTEL1 rs6010620 (P ¼ 2.79 3 10 6 )), 11q23.3 (PHLDB1 rs498872 (P ¼ 3.8 3 10 6 )), and 5p15.33 (TERT rs2736100 (P ¼ 3.69 3 10 4 )) in this study population; these loci were also associated with glioblastoma risk (20q13.33: RTEL1 rs6010620 (P ¼ 3.57 3 10 7 ); 11q23.3: PHLDB1 rs498872 (P ¼ 7.24 3 10 3 ); 5p15.33: TERT rs2736100 and TERT rs2736098 (P ¼ 1.21 3 10 4 and P ¼ 2.84 3 10 4, respectively)). This study provides further evidence for 3 glioma susceptibility regions at 20q13.33, 11q23.3, and 5p15.33 in Chinese populations. Asian continental ancestry group; association; genetics; genetic variation; genome-wide association study; glioma; polymorphism, single nucleotide Abbreviations: CI, confidence interval; GWAS, genome-wide association study(ies); OR, odds ratio; SNP, single nucleotide polymorphism. Although primary brain tumors are relatively rare (5% 9% of all cancers), approximately 22,070 persons are diagnosed each year in the United States (1). Asian populations generally show lower incidence rates than those in Europe and North America. In China, in 2000, the annual incidence rate of brain tumors was less than 3.9 per 100,000 in men and 2.8 per 100,000 in women (2). However, incidence rates tend to be increased in large cities in China. According to the China s Health Statistics Yearbook 2009, the annual mortality rate in 2008 in China was approximately 3.13 per 100,000 population (3). Glioma derived from glial cells surrounding and supporting neurons (4) is the most common type of primary intracerebral neoplasm in China, as well as in the West, accounting for more than 40% of primary brain tumors in humans (5). Furthermore, fewer than 50% of glioma patients live longer than 5 years after diagnosis (6). The only contributing factor identified currently is ionizing radiation (4, 7), although an increased familial risk has also been recognized. At least 8 susceptibility loci have been found to be associated with glioma risk from a series of case-control studies in at least 2 populations via candidate-gene association studies. These include the DNA repair genes PRKDC (also known as XRCC7) G6721T (8, 9), XRCC1 W399R (9 11), PARP1 A762V (9, 11), MGMT F84L (11, 12), ERCC1 A8092C (13, 14), and ERCC2 Q751K (9, 14); the cell cycle gene EGFþ61 A/G (15, 16); and the inflammation gene IL13 R110G (17, 18). Although these candidate-gene association-study approaches are promising, serious limitations due to small sample sizes (very few studies have had 915

916 Chen et al. more than 500 cases and controls) and the characteristic heterogeneity of gliomas have led to failure to confirm these findings in subsequent independent replication studies. With the rapid advances in genotyping technology, it is now possible to characterize inherited variation in very large numbers of genes to comprehensively examine genetic variation in glioma etiology and prognosis. The number of genome-wide association studies (GWAS) has been growing rapidly, leading to the discovery and replication of many new susceptibility loci. Recently, GWAS of glioma in populations of European ancestry were completed in the United States and the United Kingdom (19, 20). One GWAS reported on by Shete et al. (19) implicated 14 single nucleotide polymorphisms (SNPs) in 5 genes meeting the criteria for genome-wide significance (P < 2.5 3 10 8 ). Among these 14 SNPs, odds ratios ranged from 1.18 to 1.36, with respective P values ranging between 1.07 3 10 8 and 2.34 3 10 18. The 5 genes were TERT, CCDC26, CDKN2A/B, PHLDB1, and RTEL1. The other GWAS identified 3 SNPs located in the CDKN2B and RTEL1 genes that were strongly associated with risk of high-grade glioma (glioblastoma) (P < 10 7 ) (20). Among these 3 SNPs, 2 (rs142829 and rs6010620) were also reported by Shete et al. (19). Because of potential differences in genetic background by ethnicity, it is extremely important to understand the consequences of inheriting these variants in other ethnic populations. Therefore, we comprehensively evaluated all glioma risk variants identified by these 2 GWAS in a Chinese Han population using a hospital-based case-control design. MATERIALS AND METHODS Study subjects Using the same recruitment method described elsewhere (21 23), we recruited 976 patients with histopathologically confirmed glioma and 1,057 healthy controls between October 2004 and July 2009 from the Department of Neurosurgery at Huashan Hospital, Fudan University (Shanghai, China). There were no restrictions on age, sex, or histologic type, but patients with a self-reported history of cancer and patients with previous radiotherapy or chemotherapy for unknown disease conditions were excluded. Additionally, diagnoses of potentially eligible cases were validated by trained abstractors who reviewed pathologic and medical records of all cases to confirm that there were no undiagnosed occult primary tumors at the time of recruitment. All controls, who were frequency-matched to the cases by age (65 years), sex, and residential area (urban area or countryside), were selected from visitors to the trauma outpatient clinic and persons undergoing annual check-ups at the same hospital. These controls had no known central nervous system-related diseases, self-reported history of cancer at any site, or history of radiotherapy/chemotherapy for unknown disease conditions. No evidence of demographic differences was found between the trauma outpatients and the annual check-up subjects. All cases and controls were from Shanghai and the surrounding provinces (Zhejiang, Jiangsu, and Anhui) in eastern China and had a Han Chinese ethnic background. We previously published information on questionnaire design, interview, and DNA sample collection (22). Briefly, a shortened structured questionnaire was adapted from a questionnaire originally created for a brain tumor study in the Department of Epidemiology at the M. D. Anderson Cancer Center (Houston, Texas). The questionnaire was used to elicit information on demographic factors, history of occupational exposure to ionizing radiation, and the health characteristics of study subjects. Prior to enrollment, subjects were informed of the objectives, procedures, and voluntary nature of this study in a cover letter attached to the questionnaire. Among persons who agreed and signed an informed consent form for participation in the study, each subject underwent a face-to-face questionnaire interview and provided a one-time blood sample (3 5 ml) for DNA extraction. A separate signed consent form was obtained for acquiring and reviewing medical records. Participation rates for cases and controls in this study were 85.7% and 79.2%, respectively. Selection of SNPs for genotyping The 14 previously identified SNPs, representing 5 distinct loci, were genotyped in the current study. Of these, 12 SNPs were from the study by Shete et al. (19) and the other 2 (rs16904140 and rs2297440) were excluded because of strong linkage disequilibrium with rs4295627 and rs6010620, respectively (D# ¼ 1 and r 2 ¼ 1 in the Chinese population in the HapMap Phase II data). We also included 1 SNP (rs4809324) implicated in the risk of high-grade glioma (20). In addition, we selected another SNP (rs2736098) located in the TERT gene that has been reported to be associated with risk of multiple types of cancer (24). We used the white blood cell fractions of the whole blood samples for extraction of genomic DNA using the Qiagen Blood Kit (Qiagen, Chatsworth, California). After extraction, genomic DNA was diluted to a final concentration of 15 20 ng/ll for the genotyping assays. Polymorphism spanning fragments were amplified by polymerase chain reaction, and genotyping was performed with the Mass- ARRAY iplex platform (Sequenom, San Diego, California) through the use of an allele-specific matrix-assisted laser desorption/ionization time-of-flight mass spectrometry assay (25). Primers for amplification and extension reactions were designed using MassARRAY Assay Design software, version 3.1 (Sequenom), and SNP genotypes were obtained according to the iplex protocol provided by the manufacturer. We examined genotyping quality by means of a detailed quality control procedure that ensured a >95% successful call rate with duplicate calling of genotypes, internal positive control samples, and subsequent Hardy- Weinberg equilibrium testing. Statistical analysis We used Fisher s exact test to test for deviation from Hardy-Weinberg equilibrium among controls for each SNP and the v 2 test to compare the differences in demographic characteristics and frequency distributions of genotypes and alleles between cases and controls. The most common genotype in controls was used as the reference group. We

Glioma Susceptibility Genes in Chinese Populations 917 Table 1. Demographic Characteristics of Glioma Cases and Controls in a Chinese Study Population, 2004 2009 Cases (n 5 976) Controls (n 5 1,057) P Value From x 2 Test No. % No. % Sex 0.617 Male 581 59.5 633 59.9 Female 377 38.7 419 39.6 Missing data 18 1.7 5 0.5 Mean age, years 42.3 (16.3) a 42.1 (18.3) Age group 0.126 Children (18 years) 80 8.2 68 6.4 Adults (>18 years) 896 91.8 989 93.6 Cigarette smoking 0.34 Never smoker 555 57.1 625 59.1 Ever smoker 320 32.9 413 39.1 Missing data 101 10 19 1.8 Family history of cancer 0.003 (first-degree relatives) No 682 70.2 803 76.0 Yes 170 17.5 138 13.1 Missing data 124 12.4 116 11.0 Histologic type Astrocytic glioma 360 37.2 Glioblastoma 312 32.2 Other glioma 296 30.6 Missing data 8 <1 a Numbers in parentheses, standard deviation. performed unconditional logistic regression analysis, with adjustment for age and sex, to calculate odds ratios and 95% confidence intervals as estimates of the relative risk for each SNP and multiple SNPs. All statistical tests were 2-sided. Akaike s Information Criterion was employed to determine the best-fitting model for each SNP (26). For SNPs in TERT and RTEL1, pairwise linkage disequilibrium was estimated in control subjects using Haploview (27), in which haplotype blocks were inferred using the default option of the solid spine method. The software package Haplo.stats (http://www.mayo.edu/hsr/sfunc.html) was used to perform the haplotype analysis. The statistical significance of the global and haplotype-specific tests (haplo.score) was expressed as a permutation P value (P sim ; minimal simulation: 10,000 with a significance level less than 0.05). Haplotype frequencies and odds ratios (with 95% confidence intervals) were calculated for each haplotype, with adjustment for age and sex, using a generalized linear model (haplo.glm). In stratified analyses, case patients were classified into groups of high-grade tumors (glioblastomas) and low-grade tumors (including astrocytomas, oligodendrogliomas, mixed gliomas, and other low-grade gliomas). We also evaluated the cumulative effects of the risk alleles, which were independently associated with glioma risk, by counting the total number of risk alleles per person from the 3 independently confirmed loci (TERT rs2736100, PHLDB1 rs498872, and RTEL1 rs6010620) (categories were 0 1, 2, 3, 4, and 5 6). All statistical analyses were performed using SPSS software, version 17.0 (SPSS, Inc., Chicago, Illinois), unless indicated otherwise. Ethics Written informed consent was obtained from each participant, and the study protocol was approved by the School of Life Sciences of Fudan University Ethics Board. RESULTS Sample characteristics Characteristics of the 976 case patients and 1,057 control subjects are presented in Table 1. Because participants were frequency-matched for age, sex, and residential area under our study design, the distributions of age (age at diagnosis for cases and age at recruitment for controls) and sex were comparable between cases and controls. The mean age was 42.3 years for cases and 42.1 years for controls. Approximately 60% of both cases and controls were male. Similar to the previous study (22), cases were slightly more likely to report a family history of cancer (among first-degree relatives) than controls (17.5% vs. 13.1%; P ¼ 0.003).

918 Chen et al. Table 2. Association of 14 Previously Identified Single Nucleotide Polymorphisms With the Risks of Glioma and Glioblastoma in a Chinese Population, 2004 2009 Single Nucleotide Polymorphism Chromosome Gene Location on Chromosome P Value for HWE Risk Allele Risk Allele Frequency (Proportion) Estimated Risk for All Histologic Types P Value b for Glioblastoma 95% Odds P Controls Cases Ratio a Confidence Value b Interval rs2736100 5 TERT 1339516 0.0235 G 0.416 0.475 1.26 1.11, 1.43 3.69e -04* 1.21e -04* rs2853676 5 TERT 1341547 0.0277 A 0.84 0.805 1.26 1.07, 1.48 6.70e -03* 0.065 rs2736098 5 TERT 1347086 0.2543 A 0.349 0.389 1.21 1.06, 1.38 4.58e -03* 2.84e -04* rs10464870 8 CCDC26 130547005 0.1725 C 0.174 0.178 1.02 0.86, 1.20 0.819 0.365 rs891835 8 CCDC26 130560934 0.7681 G 0.127 0.118 0.92 0.76, 1.11 0.37 0.94 rs1077236 8 CCDC26 130709683 0.0283 A 0.287 0.32 1.15 1.01, 132 0.039 0.081 rs4295627 8 CCDC26 130754639 0.3913 T c 0.746 0.757 0.94 0.82, 1.09 0.44 0.72 rs1063192 9 CCDKN2A/B 21993367 1 C 0.185 0.199 1.1 0.94, 1.30 0.25 0.375 rs2157719 9 CCDKN2A/B 22023366 0.6461 G c 0.116 0.141 1.26 1.05, 1.52 0.015 0.071 rs1412829 9 CCDKN2A/B 22033926 0.79 C 0.115 0.136 1.22 1.01, 1.48 0.038 0.069 rs4977756 9 CCDKN2A/B 22058652 0.7987 G 0.215 0.227 1.06 0.91, 1.23 0.44 0.575 rs498872 11 PHLDB1 111782577 0.5889 T 0.281 0.35 1.38 1.20, 1.57 3.80e -06* 7.24e -03* rs6010620 20 RTEL1 61780283 0.0112 G c 0.266 0.331 1.39 1.20, 1.59 2.79e -06* 3.57e -07* rs4809324 20 RTEL1 61788664 0.6173 C 0.117 0.136 1.2 1.00, 1.45 0.057 0.11 Abbreviation: HWE, Hardy-Weinberg equilibrium. * P < 0.01 a Odds ratio for carriers of the minor allele (adjusted for age and sex). b Adjusted for age and sex. c Ancestral allele. Among cases, 312 (32.2%) had glioblastoma and 656 (67.8%) had low-grade glioma (including 360 (37.2%) astrocytomas and 296 (30.6%) other low-grade gliomas). Analysis of associations with individual SNPs Table 2 shows allele frequencies for the 14 selected SNPs by case/control status. All markers were in Hardy-Weinberg equilibrium in control subjects (P > 0.01). Five of these SNPs were statistically significantly associated with glioma risk, with a significance level of 0.01 (P < 0.01); we assumed that 5 independent tests were performed, and the significance level was adjusted accordingly (rs6010620 on 20q13.33 (RTEL1, P ¼ 2.79 3 10 6 ), rs498872 on 11q23.3 (PHLDB1, P ¼ 3.8 3 10 6 ), and rs2736100, rs2853676, and rs2736098 on 5p15.33 (TERT; P ¼ 3.69 3 10 4, P ¼ 0.007, and P ¼ 0.005, respectively)). The remaining 9 SNPs did not reach this predefined statistically significant level. When we evaluated these 14 selected SNPs for associations with glioblastoma risk, rs6010620 was identified as having a strong association (P ¼ 3.57 3 10 7 ; Table 2), and another 2 markers (rs2736100 and rs2736098 (TERT; P ¼ 1.21 3 10 4 and P ¼ 2.84 3 10 4, respectively) and rs498872 (PHLDB1, P ¼ 7.24 3 10 3 )) were also significantly associated with giloblastoma risk. Table 3 shows the genotype frequency distributions of the 5 identified risk SNPs in cases and controls. Specifically, significant associations were observed for 3 SNPs (P ¼ 6.86 3 10 4 for rs2736100, P ¼ 1.95 3 10 6 for rs498872, and P ¼ 2.35 3 10 5 for rs6010620) in a recessive model, for 1 SNP (P ¼ 9.59 3 10 4 for rs2853676) in a dominant model, and for 1 SNP (P ¼ 0.0035 for rs2736098) in a log-additive model, based on the best fit of Akaike s Information Criterion. Analysis of the cumulative effects of multiple risk SNPs We further assessed the cumulative effects of the 3 most significant risk SNPs (TERT rs2736100, PHLDB1 rs498872, and RTEL1 rs6010620). Overall, glioma risk increased with increasing numbers of risk variant alleles. Persons carrying 5 or 6 risk alleles had a 3-fold increased risk of developing glioma compared with those who carried 0 or 1 risk allele (adjusted odds ratio (OR) ¼ 3.62, 95% confidence interval (CI): 1.79, 7.32; P ¼ 3.31 3 10 4 (Table 4)). Analysis of haplotype associations Two TERT SNPs (rs2736100 and rs2853676) and 2 RETL1 SNPs (rs6010620 and rs4809324) were in significant linkage disequilibrium (see Web Figure 1, which is posted on the Journal s Web site (http://aje.oxfordjournals. org/)). Haplotype-specific analysis revealed frequency differences between cases and controls for the GA haplotype in TERT (P sim ¼ 0.0068) and the GT haplotype in RTEL1 (P sim ¼ 0.00e þ00 ) (Table 5). Further logistic

Glioma Susceptibility Genes in Chinese Populations 919 Table 3. Genotype Frequencies for 5 Glioma Susceptibility Single Nucleotide Polymorphisms Among Cases and Controls and Their Association With Glioma Risk in a Chinese Population, 2004 2009 Gene, Single Nucleotide Polymorphism, and Haplotype Cases Controls No. % No. % P Value From x 2 Test Adjusted Odds Ratio a Logistic Regression Analysis 95% Confidence Interval P Value a TERT rs2736100 TT 244 25.6 334 32.2 0.001 1.00 Reference 8.93e -04 GT 515 54 542 52.3 1.34 1.09, 1.65 0.001 GG 194 20.4 160 15.4 1.63 1.25, 2.14 3.64e -04 TT vs. GT/GG 1.41 1.16, 1.71 6.86e -04 rs2853676 GG 620 65.4 723 69.5 3.94e -04 1.00 Reference 0.002 GA 286 30.2 302 29 1.12 0.92, 1.36 0.261 AA 42 4.4 16 1.5 2.80 1.55, 5.08 0.001 GG/GA vs. AA 2.71 1.50, 4.90 9.59e -04 rs2736098 GG 351 36.8 430 41.6 0.022 1.00 Reference 0.015 AG 461 48.4 486 47 1.18 0.98, 1.43 0.088 AA 141 14.8 117 11.3 1.50 1.13, 1.99 0.005 Log-additive 1.22 1.07, 1.39 0.0035 PHLDB1 rs498872 CC 387 40.5 533 51.3 5.05e -06 1.00 Reference 7.46e -06 CT 470 49.2 427 41.1 1.50 1.25, 1.81 1.90e -05 TT 99 10.4 78 7.5 1.78 1.28, 2.46 6.05e -04 CC vs. CT/TT 1.24 1.04, 1.49 1.95e -06 RTEL1 rs6010620 AA 411 42.9 547 52.6 2.23e -06 1.00 Reference 7.18e -06 AG 454 47.4 438 42.1 1.38 1.14, 1.66 6.37e -04 GG 93 9.7 55 5.3 2.17 1.51, 3.11 2.52e -05 AA vs. AG/GG 1.47 1.23, 1.75 2.35e -05 a Adjusted for age and sex. regression analyses revealed that haplotypes GG and GA in the TERT block were associated with increased risk of glioma (adjusted OR ¼ 1.22 (95% CI: 1.06, 1.41) and adjusted OR ¼ 1.34 (95% CI: 1.13, 1.60), respectively) and that haplotypes GT and GC in RTEL1 were significantly associated with increased risk of glioma (adjusted OR ¼ 1.46 (95% CI: 1.23, 1.73) and adjusted OR ¼ 1.29 (95% CI: 1.06, 1.56), respectively) compared with the most common haplotypes, TG and AT, respectively. DISCUSSION Performing targeted evaluation of high-risk variants for glioma identified from GWAS in European populations becomes more efficient in other populations, because smaller numbers of SNPs need to be typed and evaluated, thus incurring a lower cost. In this hospital-based case-control validation study of glioma in a Chinese population, we systematically evaluated 14 previously reported glioma risk loci identified through GWAS in European populations. Three glioma susceptibility loci from chromosomes 20, 11, and 5 (20q13.33, 11q23.3, and 5p15.33) were statistically significantly associated with both overall glioma risk (P values ranged from 0.0067 to 2.79 3 10 6 ) and highgrade glioma (glioblastoma) risk (P values ranged from 0.0072 to 3.57 3 10 7 ). Consistent with results of single- SNP analysis (allele and genotype), the haplotypes GG and GA in TERT and the haplotypes GT and GC in RTEL1 were significantly associated with an increased risk of glioma. These results suggest that some glioma risk variants identified in European populations are also associated with risk in Chinese Han populations. The strongest signal was located on chromosome 20 at 20q13.33 in RTEL1 (regulator of telomere elongation helicase 1), which is a DNA helicase crucial for regulation of telomere length in mice. The loss of RTEL1 has been associated with shortened telomere length, chromosome breaks, and translocations (28), because it maintains genomic stability directly by suppressing homologous recombination (29). Two SNPs (rs4809324 and rs6010620) in RTEL1 were previously reported to be associated with glioma risk (19, 20). However, only rs6010620 was associated with risk in

920 Chen et al. Table 4. Association Between the Cumulative Effect of the 3 Most Statistically Significant Single Nucleotide Polymorphisms a and the Risk of Glioma in a Chinese Population, 2004 2009 No. of Risk Alleles Cases Controls No. % No. % 95% Odds Ratio b Confidence Interval P Value b 0 1 258 26.4 427 40.4 1.00 Reference 2 324 33.2 338 32.0 1.59 1.28, 1.96 3.33e -05 3 245 25.1 197 18.6 2.05 1.61, 2.62 7.97e -09 4 122 12.5 83 7.9 2.37 1.72, 3.27 1.32e -07 5 6 27 2.8 12 1.1 3.62 1.79, 7.32 3.31e -04 a Risk alleles included rs2736100 (TERT), rs498872 (PHLDB1), and rs6010620 (RTEL1). b Adjusted for age and sex in a logistic regression model. our Chinese study population (OR ¼ 1.39, 95% CI: 1.20, 1.59; P ¼ 2.79 3 10 6 ). Rs6010620 is located in intron 12 of RTEL1 and was found to be in strong linkage disequilibrium (D# ¼ 0.933, r 2 ¼ 0.572) with a missense polymorphism (rs3208008, Gln1042His) in exon 32 of RTEL1. This glutamine-to-histidine exchange may cause a conformational change in the RTEL1 protein, thus affecting its activity. Coincidentally, the third-strongest risk locus for glioma was at 5p15.33, where TERT (telomerase reverse transcriptase), also a telomerase-related gene, is located. TERT is the telomerase catalytic subunit that maintains telomeres and cell immortalization, and it has an established role in glioma grade and prognosis (30, 31). In this study population, carriers of the rs2736100 G allele and the rs2853676 A allele exhibited a statistically significant increased risk of glioma (OR ¼ 1.26 (95% CI: 1.12, 1.43; P ¼ 0.0002) and OR ¼ 1.26 (95% CI: 1.07, 1.48; P ¼ 0.0067), respectively). Carriers of the GA haplotype containing the G risk allele of rs2736100 and the A risk allele of rs2853676 had a 1.34- fold increased risk of developing glioma compared with those carrying the TG haplotype, which was consistent with individual SNP associations. The other SNP, rs2736098, located in exon 2 of TERT, was previously reported to be associated with multiple tumors, including basal cell carcinoma (32), lung cancer (33), bladder cancer (24), and prostate cancer (24). To the best of our knowledge, this is the first report of an association between rs2736098 and glioma risk (OR ¼ 1.2, 95% CI: 1.06, 1.38; P ¼ 0.0046). Rs498872, which is mapped to the 5#-untranslated region of PHLDB1 within a 101-kilobase linkage disequilibrium block on 11q23.3, was the second most strongly associated marker in our study population (OR ¼ 1.38, 95% CI: 1.20, 1.57; P ¼ 3.8 3 10 6 ). This SNP was also reported to be associated with rates of change in diastolic blood pressure (34). The 11q23.3 locus is commonly deleted in neuroblastoma (35); however, there is no direct functional evidence for a role of PHLDB1 in initiation of tumors. All of the confirmed variants showed the same direction of association as that seen in the European GWAS. In the previous study, the effect sizes were in the range of odds ratios of 1.2 1.4 (19, 20); in our study, 4 of the 5 risk SNPs had odds ratios of approximately 1.4, and 1 SNP in TERT was associated with an almost 2.5-fold increased risk. To understand the cumulative effects of these variants on glioma risk, we created a variable to combine the effects of risk alleles per person from the 3 independently confirmed loci (rs2736100, rs498872, and rs6010620). We found that these 3 independent SNPs had a strong cumulative effect on the risk of glioma. For example, persons carrying 5 or 6 risk alleles had a 3-fold increased risk of developing glioma compared with those who carried 0 or 1 risk allele, indicating the importance of the combined effects from independent risk loci in the etiology of glioma; however, this prediction strategy requires further confirmation in Table 5. Frequency Distributions of Haplotypes in the TERT and RTEL1 Genes Among Cases and Controls and Their Associations With Glioma Risk in a Chinese Population, 2004 2009 Gene and Haplotype a Total No. of Subjects Cases Controls Logistic Regression Analysis No. % No. % Adjusted Odds Ratio b 95% Confidence Interval P Value b TERT TG 2,217 1,005 51.5 1,212 57.3 1.00 Reference GG 1,144 576 29.5 568 26.9 1.22 1.06, 1.41 0.006 0.2413 GA 670 355 18.2 315 14.9 1.34 1.13, 1.60 0.001 0.0068 TA 35 16 0.8 19 0.9 1.03 0.53, 2.01 0.93 0.9999 RTEL1 AT 2,872 1,309 67.1 1,563 73.9 1.00 Reference GT 80 44 6.6 36 5.6 1.46 1.23, 1.73 1.15e -05 0.00e þ00 GC 890 452 67.9 438 68.7 1.29 1.06, 1.56 0.01 0.0696 AC 210 109 16.4 101 15.8 1.20 0.17, 8.54 0.855 NA Abbreviation: NA, not applicable. a The order of polymorphisms was rs2736100 and rs2853676 in the TERT block and rs6010620 and rs4809324 in the RETL1 block. b Adjusted for age and sex. c Generated by means of a permutation test with 10,000 simulations. P sim c

Glioma Susceptibility Genes in Chinese Populations 921 additional studies before it can be used for future assessment of glioma risk in the general population. Overall, 9 of the previously identified risk SNPs were not associated with glioma in our Chinese Han study population. However, 8 of these 9 SNPs showed the same direction of association as in the previous reports, of which 3 SNPs (2 in CCDKN2A/B (rs2157719 and rs1412829) and 1 in CCDC26 (rs1077236)) reached a nominal significance level of 0.05. There are several possible reasons for our inability to replicate the previous results for these variants. First, there may have been a lack of statistical power to detect a small effect from very-low-penetrance SNPs; second, the risk allele may have an effect among some populations, but its magnitude may vary in different populations because of genetic heterogeneity, gene-environment interactions, and/or type 1 error; and third, some of the variants may confer risk in one population but not in others. The results from our study strongly suggest the presence of glioma risk variants at 3 genetic loci in Chinese Han populations. However, it is possible that other SNPs in these genes or other genes in or around these regions may be important in Chinese populations as well. Further fine mapping studies of these regions and subsequent functional studies are needed to identify causal variants to understand the etiology of glioma in Chinese populations. ACKNOWLEDGMENTS Author affiliations: State Key Laboratory of Genetic Engineering, Fudan-VARI Genetic Epidemiology Center and MOE Key Laboratory of Contemporary Anthropology, School of Life Sciences, Fudan University, Shanghai, China (Hongyan Chen, Yuanyuan Chen, Weiwei Fan, Daru Lu); Department of Neurosurgery, Huashan Hospital, Fudan University, Shanghai, China (Yao Zhao, KekeZhou, Liangfu Zhou, Ying Mao); Department of Epidemiology, Division of Cancer Prevention and Population Sciences, University of Texas M. D. Anderson Cancer Center, Houston, Texas (Yanhong Liu, Qingyi Wei); and Center for Cancer Genomics and Center for Human Genomics, Wake Forest University School of Medicine, Winston-Salem, North Carolina (Jianfeng Xu). Drs. Hongyan Chen, Yuanyuan Chen, and Yao Zhao contributed equally to this work. Drs. Ying Mao and Daru Lu jointly directed the work. This work was partially supported by the Shanghai Science and Technology Research Program (grants 09JC1402200 and 10410709100), the Natural Science Foundation of China (grants 30800622 and 81001114), the Scientific Research Foundation for the Returned Overseas Chinese Scholars (State Education Ministry), the Doctoral Fund of the Ministry of Education of China, and the Shanghai Key Subject Project for Public Health (grant 08GWZX0301). The authors thank Haishi Zhang and Fengping Huang for subject enrollment. They also thank all staff members of the Department of Neurosurgery of Huashan Hospital for their cooperation during data collection. Conflict of interest: none declared. REFERENCES 1. Jemal A, Siegel R, Ward E, et al. Cancer statistics, 2009. CA Cancer J Clin. 2009;59(4):225 249. 2. Ohgaki H, Kleihues P. Epidemiology and etiology of gliomas. Acta Neuropathol. 2005;109(1):93 108. 3. National Bureau of Statistics of China. China s Health Statistics Yearbook 2009 [in Chinese]. Peking, China: Peking Union Medical College Press; 2009:254. (http://www.moh.gov.cn/ publicfiles/business/htmlfiles/zwgkzt/ptjnj/year2009/t-9.htm). (Accessed May 24, 2010). 4. Little MP, de Vathaire F, Shamsaldin A, et al. Risks of brain tumour following treatment for cancer in childhood: modification by genetic factors, radiotherapy and chemotherapy. Int J Cancer. 1998;78(3):269 275. 5. Xue QC, Pu PY, Yang YS, et al. A survey of 790 cases of astrocytoma. Clin Neurol Neurosurg. 1990;92(1):27 33. 6. Bondy ML, Scheurer ME, Malmer B, et al. Brain tumor epidemiology: consensus from the Brain Tumor Epidemiology Consortium. Cancer. 2008;113(suppl 7):1953 1968. 7. Bondy ML, Wang LE, El-Zein R, et al. Gamma-radiation sensitivity and risk of glioma. J Natl Cancer Inst. 2001;93(20): 1553 1557. 8. Wang LE, Bondy ML, Shen H, et al. Polymorphisms of DNA repair genes and risk of glioma. Cancer Res. 2004;64(16): 5560 5563. 9. McKean-Cowdin R, Barnholtz-Sloan J, Inskip PD, et al. Associations between polymorphisms in DNA repair genes and glioblastoma. Cancer Epidemiol Biomarkers Prev. 2009; 18(4):1118 1126. 10. Kiuru A, Lindholm C, Heinävaara S, et al. XRCC1 and XRCC3 variants and risk of glioma and meningioma. J Neurooncol. 2008;88(2):135 142. 11. Liu Y, Scheurer ME, El-Zein R, et al. Association and interactions between DNA repair gene polymorphisms and adult glioma. Cancer Epidemiol Biomarkers Prev. 2009;18(1): 204 214. 12. Felini MJ, Olshan AF, Schroeder JC, et al. DNA repair polymorphisms XRCC1 and MGMT and risk of adult gliomas. Neuroepidemiology. 2007;29(1-2):55 58. 13. Chen P, Wiencke J, Aldape K, et al. Association of an ERCC1 polymorphism with adult-onset glioma. Cancer Epidemiol Biomarkers Prev. 2000;9(8):843 847. 14. Wrensch M, Kelsey KT, Liu M, et al. ERCC1 and ERCC2 polymorphisms and adult glioma. Neuro Oncol. 2005;7(4): 495 507. 15. Bhowmick DA, Zhuang Z, Wait SD, et al. A functional polymorphism in the EGF gene is found with increased frequency in glioblastoma multiforme patients and is associated with more aggressive disease. Cancer Res. 2004;64(4): 1220 1223. 16. Costa BM, Ferreira P, Costa S, et al. Association between functional EGFþ61 polymorphism and glioma risk. Clin Cancer Res. 2007;13(9):2621 2626. 17. Schwartzbaum J, Ahlbom A, Malmer B, et al. Polymorphisms associated with asthma are inversely related to glioblastoma multiforme. Cancer Res. 2005;65(14): 6459 6465. 18. Amirian E, Liu Y, Scheurer ME, et al. Genetic variants in inflammation pathway genes and asthma in glioma susceptibility. Neuro Oncol. 2010;12(5):444 452.

922 Chen et al. 19. Shete S, Hosking FJ, Robertson LB, et al. Genome-wide association study identifies five susceptibility loci for glioma. Nat Genet. 2009;41(8):899 904. 20. Wrensch M, Jenkins RB, Chang JS, et al. Variants in the CDKN2B and RTEL1 regions are associated with highgrade glioma susceptibility. Nat Genet. 2009;41(8): 905 908. 21. Liu Y, Zhou K, Zhang H, et al. Polymorphisms of LIG4 and XRCC4 involved in the NHEJ pathway interact to modify risk of glioma. Hum Mutat. 2008;29(3):381 389. 22. Liu Y, Zhang H, Zhou K, et al. Tagging SNPs in nonhomologous end-joining pathway genes and risk of glioma. Carcinogenesis. 2007;28(9):1906 1913. 23. Zhou K, Liu Y, Zhang H, et al. XRCC3 haplotypes and risk of gliomas in a Chinese population: a hospital-based case-control study. Int J Cancer. 2009;124(12):2948 2953. 24. Rafnar T, Sulem P, Stacey SN, et al. Sequence variants at the TERT-CLPTM1L locus associate with many cancer types. Nat Genet. 2009;41(2):221 227. 25. Jurinke C, van den Boom D, Cantor CR, et al. Automated genotyping using the DNA MassArray technology. Methods Mol Biol. 2002;187:179 192. 26. Akaike H. A new look at the statistical model identification. IEEE Trans Automat Contr. 1974;19:716 23. 27. Barrett JC, Fry B, Maller J, et al. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics. 2005;21(2):263 265. 28. Ding H, Schertzer M, Wu X, et al. Regulation of murine telomere length by Rtel: an essential gene encoding a helicaselike protein. Cell. 2004;117(7):873 886. 29. Barber LJ, Youds JL, Ward JD, et al. RTEL1 maintains genomic stability by suppressing homologous recombination. Cell. 2008;135(2):261 271. 30. Wager M, Menei P, Guilhot J, et al. Prognostic molecular markers with no impact on decision-making: the paradox of gliomas based on a prospective study. Br J Cancer. 2008;98(11):1830 1838. 31. Wang L, Wei Q, Wang LE, et al. Survival prediction in patients with glioblastoma multiforme by human telomerase genetic variation. J Clin Oncol. 2006;24(10):1627 1632. 32. Stacey SN, Sulem P, Masson G, et al. New common variants affecting susceptibility to basal cell carcinoma. Nat Genet. 2009;41(8):909 914. 33. Zienolddiny S, Skaug V, Landvik NE, et al. The TERT- CLPTM1L lung cancer susceptibility variant associates with higher DNA adduct formation in the lung. Carcinogenesis. 2009;30(8):1368 1371. 34. Rutherford S, Cai G, Lopez-Alvarenga JC, et al. A chromosome 11q quantitative-trait locus influences change of bloodpressure measurements over time in Mexican Americans of the San Antonio Family Heart Study. Am J Hum Genet. 2007;81(4):744 755. 35. Guo C, White PS, Weiss MJ, et al. Allelic deletion at 11q23 is common in MYCN single copy neuroblastomas. Oncogene. 1999;18(35):4948 4957.