DNA repair genes play a major role in maintaining genomic stability through different repair pathways that are mediated by cell cycle control genes such as p53. We found previously that glioma patients were susceptible to γ-ray-induced chromosomal breaks, which may be influenced by genetic variation in genes involved in DNA strand breaks, such as XRCC1 in single-strand break repair, XRCC3 and RAD51 in homologous recombination repair, and XRCC7 in nonhomologous end joining double-strand break repair. Therefore, we tested the hypothesis that genetic polymorphisms in XRCC1, XRCC3, RAD51, XRCC7, and p53 were associated with risk of glioma in 309 patients with newly diagnosed glioma and 342 cancer-free control participants frequency matched on age (±5 years), sex, and self-reported ethnicity. We did not find any statistically significant differences in the distributions of XRCC1 Arg399Gln, XRCC3 Thr241Met, RAD51 G135C, and P53 Arg72Pro polymorphisms between the cases and the controls. However, the XRCC7 G6721T variant T allele and TT genotype were more common in the cases (0.668 and 43.4%, respectively) than in the controls (0.613 and 38.9%, respectively), and the differences were statistically significant (P = 0.045 and 0.040, respectively). The adjusted odds ratios were 1.78 (95% confidence interval, 1.08–2.94) and 1.86 (95% confidence interval, 1.12–3.09) for the GT heterozygotes and TT homozygotes, respectively. The combined T variant genotype (GT+TT) was associated with a 1.82-fold increased risk of glioma (95% confidence interval, 1.13–2.93). These results suggest that the T allele may be a risk allele, and this XRCC7 polymorphism may be a marker for the susceptibility to glioma. Larger studies are needed to confirm our findings and unravel the underlying mechanisms.

In 2004, ∼10,540 men and 7,860 women will develop a primary brain tumor, and an estimated 7,200 men and 5,490 women will die from the disease in the United States (1). Of these deaths, ∼90% will result from gliomas. The etiology of gliomas remains unclear, but epidemiological studies have shown that ionizing radiation and genetic alterations are established risk factors in subsets of brain tumor patients (2, 3, 4, 5, 6, 7).

Ionizing radiation induces various types of DNA damage, including single- and double-strand breaks. Dozens of genes are involved in DNA strand break repair to maintain genomic stability through different pathways mediated by cell cycle control genes (8, 9). Each DNA repair gene plays a unique role. For example, XRCC1 is involved in single-strand break repair, XRCC3 and RAD51 in homologous recombination repair, and XRCC7 in nonhomologous end joining double-strand break repair (8, 9, 10, 11). However, genetic alterations of these genes may affect the function of their proteins and lead to diseases or cancers (12, 13). The roles of common polymorphisms, XRCC1 Arg399Gln, XRCC3 Thr241Met, RAD51 G135C, and P53 Arg72Pro have been investigated in various cancers, and the results are mixed (14, 15, 16, 17, 18, 19, 20, 21). There is one newly identified variant (G6721T) in XRCC7 involved in the nonhomologous end joining, but the functional relevance is not yet known (10).

The X-ray repair cross-complementing group 7 (XRCC7) gene encodes the catalytic subunit of a DNA-activated protein kinase, which is involved in the nonhomologous end joining repair pathway in murine cells (22) and humans (10). Defects in the XRCC7 gene make the DNA-activated protein kinase activity undetectable in murine mutants and these cells sensitive to ionizing radiation (23). Furthermore, the XRCC7 gene is a strong candidate gene involved in severe combined immunodeficiency (23, 24).

Because we reported recently that glioma patients were susceptible to γ-ray-induced chromosomal breaks (6, 7), we hypothesized that the polymorphisms of these genes are associated with glioma risk. To test this hypothesis, we conducted a hospital-based case-control study using a restriction fragment length polymorphism assay to genotype for the variants of XRCC1, XRCC3, RAD51, P53, and XRCC7 genes in 309 patients with newly diagnosed glioma and 342 cancer-free controls frequency matched on age (±5 years), sex, and self-reported ethnicity.

Study Population.

In this case-control analysis, we included 309 patients with newly diagnosed and histologically confirmed glioma recruited at The University of Texas M. D. Anderson Cancer Center from 1994 to 2000. These cases included 151 glioblastoma multiforme, categorized as high-grade glioma; 70 anaplastic astrocytoma, medium-grade glioma; and 88 others in the low-grade glioma group (oligodendroglioma, not-otherwise-specified astrocytoma, or mixed glioma). The 342 controls were cancer-free participants recruited at the M. D. Anderson Blood Bank (24.9%), visitors of other patients at M. D. Anderson who were biologically unrelated to the study patients (52.0%), and others from the Houston community residing in the vicinity of M. D. Anderson (23.1%). Included among these participants were 200 patients and 220 controls who participated in a study reported previously (7) and for whom DNA was available. Because genetic susceptibility is more identifiable in young individuals, we included only study participants between the ages of 20 and 60 years. The cases and controls were frequency matched on age (by ±5 years), sex, and ethnicity (all self-reported non-Hispanic whites). Each eligible participant was interviewed to obtain data regarding age, sex, and ethnicity. Only non-Hispanic whites were included because of the small numbers of other ethnic groups. After giving informed consent, each participant donated 20 ml of blood collected in heparinized tubes. The research protocol was approved by the M. D. Anderson Institutional Review Board.

Genotyping.

We used the commercially available Qiagen kit (Qiagen Inc., Valencia, CA) to extract DNA from peripheral blood leukocytes. The purified DNA was used to determine the genotypes: for the XRCC1 Arg399Gln polymorphism at nucleotide −28152 of exon 10, the XRCC3Thr241Met polymorphism at nucleotide −18607 of exon 7, the RAD51G135C polymorphism of untranslated region, and the p53 Arg72Pro polymorphism of exon 4, we used the published primers and the PCR-restriction fragment length polymorphism method to amplify PCR fragments, then digested them with the corresponding restriction enzymes to identify the three genotypes of each gene (25, 26, 27, 28); and for the XRCC7G6721T polymorphism, we designed the primers and developed a simple restriction fragment length polymorphism method. The sense primer was 5′-CGGCTGCCAACGTTCTTTCC-3′ (nucleotides 6626–6645), and the antisense primer was 5′-TGCCCTTAGTGGTTCCCTGG-3′ (complementary to nucleotides 6974–6993; GenBank accession no. L27425; Ref. 10). The primers amplified 368-bp fragment containing the G/T variant in intron 8, which was then subjected to digestion with PvuII (New England BioLabs, Inc., Beverly, MA) at 37°C overnight. The homologous GG allele had only one band of 368-bp; the heterozygous allele (GT) had three bands of 368, 274, and 94 bp; and the homozygous TT allele had two bands of 274 and 94 bp because of the gain of the restriction site (Fig. 1). The 20-μl PCR mixture contained ∼50 ng of genomic DNA, 5.0 pmol of each primer, 0.1 mm each deoxynucleoside triphosphate, 1 × PCR buffer [50 mm KCl, 10 mm Tris HCl (pH 9.0 at 25°C), and 0.1% Triton X-100], 1.5 mm MgCl2, and 1.0 units of Taq polymerase (Sigma-Aldrich Biotechnology, Saint Louis, MO). The PCRs were performed with a PTC-200 DNA Engine (MJ Research, Inc., Watertown, MA). The PCR profile consisted of an initial melting step of 95°C for 5 min; 30 cycles of 95°C for 30 s, 61°C for 30 s, and 72°C for 45 s; and a final elongation step of 72°C for 10 min with a minor modification for each gene. The PCR products were checked on a 1% agarose gel, and the digested products were visualized on a 2% agarose gel stained with ethidium bromide and photographed using the Digital Imaging System (Model IS-1000; Alpha Innotech Co., San Leandro, CA).

Statistical Analysis.

We first conducted a univariate analysis to evaluate the distribution of selected variables in the cases and controls. The χ2 test was then computed to evaluate the differences in the frequency distributions of age; sex; and the XRCC1, XRCC3, RAD51, P53, and XRCC7 genotypes between the cases and controls. We also evaluated the observed genotype frequencies with those calculated from the Hardy-Weinberg equilibrium theory (p2 + 2pq + q2 = 1, where p is the frequency of the variant allele and q = 1 − p). We applied univariate and multivariate logistic regression to calculate crude and adjusted odds ratios and 95% confidence intervals, respectively, for the association between the genotypes and risk of glioma. We stratified the data into subgroups of demographic variables and glioma histology. We used Statistical Analysis System software (Version 8; SAS Institute Inc., Cary, NC) to perform all of the statistical analyses.

The mean ages were 44.1 years (SD ±11.0) and 43.8 years (SD ±10.6) for the 309 cases and 342 controls, respectively (P = 0.686). The frequency matching on age and sex between the cases and controls appeared to be adequate (Table 1). The genotype distributions of the XRCC1 Arg399Gln, XRCC3 Thr241Met, RAD51 G135C, and P53 Arg72Pro between the cases and controls were not statistically significant, and their variant allele frequencies were not statistically different either between the cases and controls (Table 1). The distributions of these genotype frequencies were in agreement with those expected from the Hardy-Weinberg equilibrium model for controls (P = 0.997, 0.708, 0.545, and 0.986, respectively).

However, we found that the variant XRCC7 T allele was more common in the cases (0.668) than in the controls (0.613), and the difference was statistically significant (P = 0.045). The XRCC7 genotype frequencies between the cases and controls were also statistically significant (P = 0.040). Likewise, the difference in the distribution of combined T variant genotype (GT+TT) between the cases and controls was also statistically significant P = 0.012), suggesting that the T allele may be a risk allele for glioma. The distributions of the XRCC7 genotype frequencies were in agreement with those expected from the Hardy-Weinberg equilibrium model for both the cases (P = 0.765) and controls (P = 0.759).

To evaluate the difference in the controls obtained from different sources (blood bank, hospital, and community), we compared these genotype distributions among the controls. There was no statistically significant difference in these five genotype distributions among these three control groups (data not shown).

As shown in Table 2, we analyzed the distribution of the XRCC7 genotypes stratified by age, sex, and the histological type of the glioma and found that the differences in the distributions of the XRCC7 genotypes between the cases and controls were statistically significant in the older (P = 0.022), males (P = 0.034), high-grade glioma (P = 0.014), and low-grade glioma subgroups (P = 0.020).

The odds ratios for the GT, TT, and combined GT+TT genotypes by using the GG genotype as the reference are summarized in Table 3. After adjustment for age and sex, GT, TT, and combined (GT+TT) genotypes were all associated with a nearly 2-fold increased risk of glioma (odds ratios, 1.78; 95% confidence interval, 1.08–2.94 for GT; odds ratios, 1.86; 95% confidence interval, 1.12–3.09 for TT; and odds ratios, 1.82; 95% confidence interval, 1.13–2.93 for GT+TT). Stratifying these data by age, sex, and type of tumor histology showed that the association between the T variant genotype (GT, TT or GT+TT) and the increased risk of glioma was more pronounced in older (46–60 years; odds ratios, 2.58; 95% confidence interval, 1.23–5.43 for the GT+TT genotype) and male (odds ratios, 2.33; 95% confidence interval, 1.18–4.59 for the GT+TT genotype) participants. Additional stratification by age and sex revealed that the subgroup of males aged 46–60 years (cases/controls = 97/87) had the highest risk (odds ratios, 2.95; 95% confidence interval, 1.05–8.26; data not shown). The homozygous variant TT genotype was associated with a >2-fold increased risk of glioblastoma multiforme gliomas (odds ratios, 2.16; 95% confidence interval, 1.11–4.22), and the heterozygous GT variant genotype was associated with a nearly 3-fold risk for low-grade gliomas (odds ratios, 2.93; 95% confidence interval, 1.24–6.93; Table 3). However, significant odds ratios (2.39; 95% confidence interval, 1.04–5.53) for the GT/TT genotype was only found for the low-grade tumors.

We also investigated the effects of combined genotypes of these five polymorphisms on risk of glioma, but we did not find any significant association beyond the XRCC7 genotype (data not shown).

In this study, we selected five major genes involved in the single-strand break repair (XRCC1), homologous recombination (XRCC3 and RAD51), nonhomologous end joining (XRCC7) repair pathways and cell-cycle control (P53) to investigate the associations of their common polymorphisms with risk of glioma. The data revealed that only the XRCC7 G6721T polymorphism was associated with glioma risk. To our knowledge, this study is the first case-control analysis of the XRCC7 intron 8 G6721T polymorphism in risk of human cancers. It provides estimates of the prevalence of the G6721T genotypes from 342 cancer-free controls. The T allele frequency of 0.613 in this non-Hispanic white population was slightly lower than that (0.667) in healthy participants from the only previously published report that included 12 cancer patients and 27 family members (10). However, we found a significantly higher frequency of the XRCC7 T variant genotypes in the glioma patients than in the control participants, and this high frequency of the T variant genotype was associated with 1.82-fold increase in risk of glioma. The increased risk of glioma associated with the XRCC7 variant genotypes is more pronounced in older (46–60 years), male cases and patients with high-grade and low-grade gliomas. Because the variability is always observed by strata, the results need to be verified in other hypothesis-driven studies.

Although the functional relevance of the XRCC7 polymorphism is unknown, several lines of evidence suggest that these findings are biologically plausible. The XRCC7 gene is located on chromosome 8q11 and encodes the catalytic polypeptide of DNA-activated protein kinase, which plays a key role in DNA nonhomologous end joining double-strand breaks (22). Studies have suggested that human chromosome 8q11 functionally corrects the hyper-radiosensitivity and variable (diversity) joining region recombination in severe combined immunodeficiency cells and complements the DNA double-strand break repair deficiency of severe combined immunodeficiency cells that are phenotypically sensitive to radiation-induced chromosome aberration (29, 30, 31). Although the functional significance of the XRCC7 intron G6721T polymorphism is unknown, this intronic single-nucleotide polymorphism might regulate splicing and cause mRNA instability (10) or may be a haplotype with other genetic changes in other disease-related genes through a linkage disequilibrium mechanism (12). However, these possibilities should be investigated in future studies.

We realize that population admixture is a known confounding factor for population-based association analysis, and may result in inflated type I error. In this study we only included Caucasians because >85% of our dataset were Caucasian. The ethnic distribution of our samples is similar to the distribution of gliomas reported by Surveillance, Epidemiology, and End Results and other registries. We followed the guidelines recommended by the Office of Management and Budget to respect the individual participant self-reporting their race and ethnicity (32) and also examined the consistency of the prevalence of the alleles we genotyped. The frequencies were in the ranges of frequencies published by ethnicity (12, 16, 17, 21, 33), suggesting the self-reported ethnicity was appropriate in this association study.

In conclusion, the XRCC7 6721T variant may contribute to risk of glioma, especially in older males, and might be associated with some glioma histologies. Although this hospital-based case-control study may have limitations resulting from bias of the case and control selection, it is unlikely that their genotypes would be influenced by this bias (34). Because this is the first report of an association between this XRCC7 polymorphism and glioma risk, our findings need to be validated in larger studies with more rigorous study designs. It is also important to investigate the role of known polymorphisms of genes involved in the nonhomologous end joining repair pathway in the development of glioma and their gene-gene or gene-environment interactions.

Grant support: National Cancer Institute Grants CA70917, CA55261, ES11740, ES07784, and CA16672.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Requests for reprints: Melissa L. Bondy, Department of Epidemiology, Unit 189, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, TX 77030. Phone: (713) 792-3020; Fax: (713) 792-0807; E-mail: mbondy@mdanderson.org

Fig. 1.

PCR-based restriction fragment length polymorphism genotyping for XRCC7 intron 8 (G6721T) polymorphism (PvuII). Lanes 5, 9, and 10, GG homozygotes; Lanes 1, 4, 11, 12, and 13: GT heterozygotes; and Lanes 2, 3, 6, 7, and 8, TT homozygotes.

Fig. 1.

PCR-based restriction fragment length polymorphism genotyping for XRCC7 intron 8 (G6721T) polymorphism (PvuII). Lanes 5, 9, and 10, GG homozygotes; Lanes 1, 4, 11, 12, and 13: GT heterozygotes; and Lanes 2, 3, 6, 7, and 8, TT homozygotes.

Close modal
Table 1

Distribution of selected variables and five genotypes in glioma patients and cancer-free controls

VariableCases (n = 309)Controls (n = 342)P              *
No.(%)No.(%)
Age (yr)      0.536 
 20–45 147 (47.6)  171 (50.0)  
 46–60 162 (52.4)  171 (50.0)  
Sex      0.184 
 Male 167 (54.0)  167 (48.8)  
 Female 142 (46.0)  175 (51.2)  
XRCC1 Arg399Gln      0.375 
 Arg/Arg 134 (43.4)  131 (38.3)  
 Arg/Gln 138 (44.7)  162 (47.4)  
 Gln/Gln 37 (12.0)  49 (14.3)  
 Arg/Gln + Gln/Gln 175 (56.6)  211 (61.7) 0.189 
 Gln allele frequency 0.343  0.380  0.184 
XRCC3 Thr241Met      0.727 
 Thr/Thr 134 (43.4)  147 (43.0)  
 Thr/Met 138 (44.7)  147 (43.0)  
 Met/Met 37 (12.0)  48 (14.0)  
 Thr/Met + Met/Met 175 (56.6)  195 (57.0) 0.922 
 Met allele frequency 0.343  0.355  0.693 
RAD51 G135C      0.098 
 GG 265 (85.8)  301 (88.0)  
 GC 40 (12.9)  41 (12.0)  
 CC (1.3)  (0.0)  
 GC+CC 44 (14.2)  41 (12.0) 0.395 
 C allele frequency 0.078  0.060  0.240 
P53 Arg72Pro      0.673 
 Arg/Arg 165 (53.4)  194 (56.7)  
 Arg/Pro 126 (40.8)  128 (37.4)  
 Pro/Pro 18 (5.8)  20 (5.9)  
 Arg/Pro + Pro/Pro 144 (46.6)  148 (43.3) 0.394 
 Pro allele frequency 0.262  0.246  0.549 
XRCC7 G6721T      0.040 
 GG 30 (9.7)  56 (16.4)  
 GT 145 (46.9)  153 (44.7)  
 TT 134 (43.4)  133 (38.9)  
 GT+TT 279 (90.3)  286 (83.6) 0.012 
 T allele frequency 0.668  0.613  0.045 
VariableCases (n = 309)Controls (n = 342)P              *
No.(%)No.(%)
Age (yr)      0.536 
 20–45 147 (47.6)  171 (50.0)  
 46–60 162 (52.4)  171 (50.0)  
Sex      0.184 
 Male 167 (54.0)  167 (48.8)  
 Female 142 (46.0)  175 (51.2)  
XRCC1 Arg399Gln      0.375 
 Arg/Arg 134 (43.4)  131 (38.3)  
 Arg/Gln 138 (44.7)  162 (47.4)  
 Gln/Gln 37 (12.0)  49 (14.3)  
 Arg/Gln + Gln/Gln 175 (56.6)  211 (61.7) 0.189 
 Gln allele frequency 0.343  0.380  0.184 
XRCC3 Thr241Met      0.727 
 Thr/Thr 134 (43.4)  147 (43.0)  
 Thr/Met 138 (44.7)  147 (43.0)  
 Met/Met 37 (12.0)  48 (14.0)  
 Thr/Met + Met/Met 175 (56.6)  195 (57.0) 0.922 
 Met allele frequency 0.343  0.355  0.693 
RAD51 G135C      0.098 
 GG 265 (85.8)  301 (88.0)  
 GC 40 (12.9)  41 (12.0)  
 CC (1.3)  (0.0)  
 GC+CC 44 (14.2)  41 (12.0) 0.395 
 C allele frequency 0.078  0.060  0.240 
P53 Arg72Pro      0.673 
 Arg/Arg 165 (53.4)  194 (56.7)  
 Arg/Pro 126 (40.8)  128 (37.4)  
 Pro/Pro 18 (5.8)  20 (5.9)  
 Arg/Pro + Pro/Pro 144 (46.6)  148 (43.3) 0.394 
 Pro allele frequency 0.262  0.246  0.549 
XRCC7 G6721T      0.040 
 GG 30 (9.7)  56 (16.4)  
 GT 145 (46.9)  153 (44.7)  
 TT 134 (43.4)  133 (38.9)  
 GT+TT 279 (90.3)  286 (83.6) 0.012 
 T allele frequency 0.668  0.613  0.045 
*

Two-sided χ2 test.

Table 2

Stratification analysis of XRCC7 genotype frequencies

Cases (n = 309)Controls (n = 342)P              *
GG No. (%)GT No. (%)TT No. (%)GG No. (%)GT No. (%)TT No. (%)
Age (yr)        
 20–45 19 (12.9) 73 (49.7) 55 (37.4) 29 (17.0) 76 (44.4) 66 (38.6) 0.512 
 46–60 11 (6.8) 72 (44.4) 79 (48.8) 27 (15.8) 77 (45.0) 67 (39.2) 0.022 
Sex        
 Male 14 (8.4) 76 (45.5) 77 (46.1) 30 (18.0) 66 (39.5) 71 (42.5) 0.034 
 Female 16 (11.3) 69 (48.6) 57 (40.1) 26 (14.9) 87 (49.7) 62 (35.4) 0.536 
Tumor histology        
 High-grade glioma 14 (9.3) 59 (39.1) 78 (51.7) 56 (16.4) 153 (44.7) 133 (38.9) 0.014 
 Medium-grade glioma 9 (12.9) 33 (47.1) 28 (40.0)    0.760 
 Low-grade glioma 7 (8.0) 53 (60.2) 28 (31.8)    0.020 
Cases (n = 309)Controls (n = 342)P              *
GG No. (%)GT No. (%)TT No. (%)GG No. (%)GT No. (%)TT No. (%)
Age (yr)        
 20–45 19 (12.9) 73 (49.7) 55 (37.4) 29 (17.0) 76 (44.4) 66 (38.6) 0.512 
 46–60 11 (6.8) 72 (44.4) 79 (48.8) 27 (15.8) 77 (45.0) 67 (39.2) 0.022 
Sex        
 Male 14 (8.4) 76 (45.5) 77 (46.1) 30 (18.0) 66 (39.5) 71 (42.5) 0.034 
 Female 16 (11.3) 69 (48.6) 57 (40.1) 26 (14.9) 87 (49.7) 62 (35.4) 0.536 
Tumor histology        
 High-grade glioma 14 (9.3) 59 (39.1) 78 (51.7) 56 (16.4) 153 (44.7) 133 (38.9) 0.014 
 Medium-grade glioma 9 (12.9) 33 (47.1) 28 (40.0)    0.760 
 Low-grade glioma 7 (8.0) 53 (60.2) 28 (31.8)    0.020 
*

Two-sided χ2 test.

High-grade glioma: glioblastoma multiforme; medium-grade glioma: anaplastic astrocytoma; low-grade glioma: including oligodendroglioma, NOS astrocytoma, or mixed glioma.

Table 3

Logistic regression analysis of XRCC7 genotypes stratified by selected variables and histological types

Adjusted odds ratio (95% confidence interval)*
NGGGTTTGT/TT
All participants 309/342 1.00 1.78 (1.08–2.94) 1.86 (1.12–3.09) 1.82 (1.13–2.93) 
Age (yr)      
 20–45 147/171 1.00 1.48 (0.76–2.88) 1.31 (0.66–2.60) 1.40 (0.75–2.64) 
 46–60 162/171 1.00 2.34 (1.08–5.08) 2.87 (1.32–6.26) 2.58 (1.23–5.43) 
Sex      
 Male 167/167 1.00 2.42 (1.18–4.96) 2.25 (1.10–4.60) 2.33 (1.18–4.59) 
 Female 142/175 1.00 1.29 (0.64–2.60) 1.52 (0.74–3.12) 1.39 (0.71–2.70) 
Tumor histology      
 High-Grade Glioma 151/342 1.00 1.49 (0.75–2.93) 2.16 (1.11–4.22) 1.81 (0.96–3.43) 
 Medium-grade glioma 70/342 1.00 1.44 (0.64–3.23) 1.43 (0.63–3.26) 1.44 (0.67–3.09) 
 Low-grade glioma 88/342 1.00 2.93 (1.24–6.93) 1.78 (0.72–4.37) 2.39 (1.04–5.53) 
Adjusted odds ratio (95% confidence interval)*
NGGGTTTGT/TT
All participants 309/342 1.00 1.78 (1.08–2.94) 1.86 (1.12–3.09) 1.82 (1.13–2.93) 
Age (yr)      
 20–45 147/171 1.00 1.48 (0.76–2.88) 1.31 (0.66–2.60) 1.40 (0.75–2.64) 
 46–60 162/171 1.00 2.34 (1.08–5.08) 2.87 (1.32–6.26) 2.58 (1.23–5.43) 
Sex      
 Male 167/167 1.00 2.42 (1.18–4.96) 2.25 (1.10–4.60) 2.33 (1.18–4.59) 
 Female 142/175 1.00 1.29 (0.64–2.60) 1.52 (0.74–3.12) 1.39 (0.71–2.70) 
Tumor histology      
 High-Grade Glioma 151/342 1.00 1.49 (0.75–2.93) 2.16 (1.11–4.22) 1.81 (0.96–3.43) 
 Medium-grade glioma 70/342 1.00 1.44 (0.64–3.23) 1.43 (0.63–3.26) 1.44 (0.67–3.09) 
 Low-grade glioma 88/342 1.00 2.93 (1.24–6.93) 1.78 (0.72–4.37) 2.39 (1.04–5.53) 
*

Adjusted for age and sex, accordingly.

N = numbers of cases/controls.

High-grade glioma: glioblastoma multiforme; medium-grade glioma: anaplastic astrocytoma; low-grade glioma: including oligodendroglioma, NOS astrocytoma, or mixed glioma.

We thank Peter Inskip for his critical review and insightful comments and suggestions, Yawei Qiao and Zhigang Duan for their technical support, Phyllis Adatto for assistance in recruiting study participants and data management, Joyce Brown for assistance in preparing the manuscript, and David Galloway (Department of Scientific Publications, The University of Texas M. D. Anderson Cancer Center) for scientific editing.

1
Jemal A, Tiwari RC, Murray T, et al American Cancer Society. Cancer statistics, 2004.
CA Cancer J Clin
,
54
:
8
-29,  
2004
.
2
Ron E, Modan B, Boice JD, Jr, et al Tumors of the brain and nervous system after radiotherapy in childhood.
N Engl J Med
,
20
:
1033
-39,  
1988
.
3
Thomas TL, Stolley PD, Stemhagen A, et al Brain tumor mortality risk among men with electrical and electronics jobs: a case-control study.
J Natl Cancer Inst
,
79
:
233
-8,  
1987
.
4
Inskip PD, Linet MS, Heineman EF. Etiology of brain tumors in adults.
Epidemiol Rev
,
17
:
382
-414,  
1995
.
5
Preston-Martin S, Mack WJ. Neoplasms of the nervous system Schottenfeld D Fraumeni JF eds. .
Cancer Epidemiology and Prevention
,
p. 1231
-81, Oxford University Press Oxford  
1996
.
6
Bondy ML, Kyritsis AP, Gu J, et al Mutagen sensitivity and risk of gliomas: a case-control analysis.
Cancer Res
,
56
:
1484
-6,  
1996
.
7
Bondy ML, Wang LE, El-Zein R, et al Gamma-radiation sensitivity and risk of glioma.
J Natl Cancer Inst
,
93
:
1553
-7,  
2001
.
8
Ruttan CC, Glickman BW. Coding variants in human double-strand break DNA repair genes.
Mutat Res
,
509
:
175
-200,  
2002
.
9
Dasika GK, Lin SC, Zhao S, et al DNA damage-induced cell cycle checkpoints and DNA strand break repair in development and tumorigenesis.
Oncogene
,
18
:
7883
-99,  
1999
.
10
Sipley JD, Menninger JC, Hartley KO, et al Gene for the catalytic subunit of the human DNA-activated protein kinase maps to the site of the XRCC7 gene on chromosome 8.
Proc Natl Acad Sci USA
,
92
:
7515
-9,  
1995
.
11
Caldecott KW. XRCC1 and DNA strand break repair.
DNA Repair (Amst)
,
2
:
955
-69,  
2003
.
12
Goode EL, Ulrich CM, Potter JD. Polymorphisms in DNA repair genes and associations with cancer risk.
Cancer Epidemiol Biomark Prev
,
11
:
1513
-30,  
2002
.
13
Friedberg EC. DNA damage and repair.
Nature
,
421
:
436
-40,  
2003
.
14
Butkiewicz D, Rusin M, Enewold L, et al Genetic polymorphisms in DNA repair genes and risk of lung cancer.
Carcinogenesis
,
22
:
593
-7,  
2001
.
15
Hsieh LL, Chien HT, Chen IH, et al The XRCC1 399Gln polymorphism and the frequency of p53 mutations in Taiwanese oral squamous cell carcinomas.
Cancer Epidemiol Biomark Prev
,
12
:
439
-43,  
2003
.
16
Zhou W, Liu G, Miller DP, et al Polymorphisms in the DNA repair genes XRCC1 and ERCC2, smoking, and lung cancer risk.
Cancer Epidemiol Biomark Prev
,
12
:
359
-65,  
2003
.
17
Kuschel B, Auranen A, McBride S, et al Variants in DNA double-strand break repair genes and breast cancer susceptibility.
Hum Mol Genet
,
11
:
1399
-407,  
2002
.
18
Jakubowska A, Narod SA, Goldgar DE, et al Breast cancer risk reduction associated with the RAD51 polymorphism among carriers of the BRCA1 5382insC mutation in Poland.
Cancer Epidemiol Biomark Prev
,
12
:
457
-9,  
2003
.
19
Wang WW, Spurdle AB, Kolachana P, et al A single nucleotide polymorphism in the 5′ untranslated region of RAD51 and risk of cancer among BRCA1/2 mutation carriers.
Cancer Epidemiol Biomark Prev
,
10
:
955
-60,  
2001
.
20
Kawajiri K, Nakachi K, Imai K, Watanabe J, Hayashi S. Germ line polymorphisms of p53 and CYP1A1 genes involved in human lung cancer.
Carcinogenesis
,
14
:
1085
-9,  
1993
.
21
Pierce LM, Sivaraman L, Chang W, et al Relationships of TP53 codon 72 and HRAS1 polymorphisms with lung cancer risk in an ethnically diverse population.
Cancer Epidemiol Biomark Prev
,
9
:
1199
-204,  
2000
.
22
Blunt T, Finnie NJ, Taccioli GE, et al Defective DNA-dependent protein kinase activity is linked to V(D)J recombination and DNA repair defects associated with the murine scid mutation.
Cell
,
80
:
813
-23,  
1995
.
23
Errami A, Overkamp WJ, He D, et al A new X-ray sensitive CHO cell mutant of ionizing radiation group 7, XR-C2, that is defective in DSB repair but has only a mild defect in V(D)J recombination.
Mutat Res
,
461
:
59
-69,  
2000
.
24
Woo RA, McLure KG, Lees-Miller SP, Rancourt DE, Lee PW. DNA-dependent protein kinase acts upstream of p53 in response to DNA damage.
Nature
,
394
:
700
-4,  
1998
.
25
Sturgis EM, Castillo EJ, Li L, et al Polymorphisms of DNA repair gene XRCC1 in squamous cell carcinoma of the head and neck.
Carcinogenesis
,
20
:
2125
-9,  
1999
.
26
Duan Z, Shen H, Lee JE, et al DNA repair gene XRCC3 241Met variant is not associated with risk of cutaneous malignant melanoma.
Cancer Epidemiol Biomark Prev
,
11
:
1142
-3,  
2002
.
27
Levy-Lahad E, Lahad A, Eisenberg S, et al A single nucleotide polymorphism in the RAD51 gene modifies cancer risk in BRCA2 but not BRCA1 carriers.
Proc Natl Acad Sci USA
,
98
:
3232
-6,  
2001
.
28
Shen H, Zheng Y, Sturgis EM, Spitz MR, Wei Q. P53 codon 72 polymorphism and risk of squamous cell carcinoma of the head and neck: a case-control study.
Cancer Lett
,
183
:
123
-30,  
2002
.
29
Komatsu K, Ohta T, Jinno Y, Niikawa N, Okumura Y. Functional complementation in mouse-human radiation hybrids assigns the putative murine scid gene to the pericentric region of human chromosome 8.
Hum Mol Genet
,
2
:
1031
-4,  
1993
.
30
Kirchgessner CU, Tosto LM, Biedermann KA, et al Complementation of the radiosensitive phenotype in severe combined immunodeficient mice by human chromosome 8.
Cancer Res
,
53
:
6011
-6,  
1993
.
31
Banga SS, Hall KT, Sandhu AK, Weaver DT, Athwal RS. Complementation of V(D)J recombination defect and X-ray sensitivity of scid mouse cells by human chromosome 8.
Mutat Res
,
315
:
239
-47,  
1994
.
32
Federal Register.
Part II: Office of Management and Budget
,
62(210)
:
58781
-90,  
1997
.
33
Wu X, Zhao H, Amos CI, et al p53 Genotypes and Haplotypes Associated With Lung Cancer Susceptibility and Ethnicity.
J Natl Cancer Inst
,
94
:
681
-90,  
2002
.
34
Garte S, Gaspari L, Alexandrie AK, et al Metabolic gene polymorphism frequencies in control populations.
Cancer Epidemiol Biomark Prev
,
10
:
1239
-48,  
2001
.