|Home | About | Journals | Submit | Contact Us | Français|
ERCC2 and ERCC1 are important in DNA nucleotide excision repair and lie on chromosome 19q13.3 near a putative glioma suppressor region. We genotyped constitutive variants ERCC1 C8092A and ERCC2 K751Q and R156R in approximately 450 adults with glioma and 500 controls from two independent population-based series, uniformly reviewed patients’ tumors to determine histopathologic category, and determined a variety of tumor markers among astrocytic tumors. Odds ratios (ORs) for glioblastoma for those carrying two ERCC1 A alleles versus none or one were 1.67 in series 1 and 1.64 in series 2, which yielded a combined OR of 1.67 (95% CI, 0.93–3.02; P = 0.09), adjusted for age, gender, ethnicity, and series. Odds ratios for the ERCC2 variants were not consistently elevated or reduced for the two series in all cases versus controls. However, among whites, for those with ERCC2 K751Q genotype QQ versus QK/KK, the OR for nonglioblastoma histologies versus controls was 1.82 (95% CI, 0.97–3.44; P = 0.06). Also, among whites, glioma patients were significantly more likely than controls to be homozygous for variants in both ERCC1 C8092A and ERCC2 K751Q (OR, 3.2; 95% CI, 1.1–9.3). Given the numbers of comparisons made, these findings could be due to chance. However, the results might warrant clarification in additional series in conjunction with the nearby putative glioma suppressor genes (GLTSCR1 and GLTSCR2)
Numerous studies are addressing associations of polymorphisms in DNA repair genes and cancer risk (Goode et al., 2002) because accurate and efficient DNA repair is crucial to genomic integrity and fidelity. The DNA repair system is complex, governed by more than 125 genes, many of which are polymorphic (Ng and Henikoff, 2002; Ronen and Glickman, 2001; Zhu et al., 2004). Two DNA repair genes, ERCC2 and ERCC1, whose products are important in nucleotide excision repair lie near each other on chromosome 19q13.3 in a region near a putative glioma suppressor region (Smith et al., 2000).
We previously reported a significant association of the ERCC2 polymorphism R156R (C to A) with adult glioma; the age-gender-adjusted case-control odds ratio (OR)3 comparing those with CA/CC versus those with AA genotypes was 2.3, with a 95% CI of 1.3–4.2 (Caggana et al., 2001), and the ORs were significantly elevated for the histologic types glioblastoma multiforme, astrocytoma, and oligoastrocytoma. The K751Q (A to C) polymorphism (leading to a change from lysine [K] to glutamine [Q]) was significantly inversely associated with the R156R variant (Kendall correlation = −0.32; P = 0.0001); this variant also was inversely associated with adult glioma, though not significantly so (OR for one or two C alleles [KQ or KK] versus no C alleles [QQ] was 0.7; 95% CI, 0.4–1.1). In an overlapping group of adult glioma cases and controls, we also found that people with oligoastrocytoma were significantly more likely to be homozygous for the more common variant of ERCC1, C8092A, than controls (OR for CC versus CA/AA, 4.6; 95% CI, 1.6–13.2), but the C variant was not significantly more common in other histologic types of glioma (Chen et al., 2000). The C8092A polymorphism represents a silent variant in the 3′ untranslated region of ERCC1 but is a nonsynonymous variant in the gene ASE-1 (also called CAST) that overlaps with the complementary strand of ERCC1 (Hoeijmakers et al., 1989).
In this report, we examine these ERCC1 and ERCC2 polymorphisms in a new, independent series of adults with glioma and controls and combine information from the new and previous series to permit more robust comparisons.
The University of California San Francisco Committee on Human Research approved methods for this study, and subjects provided signed consent. Details of case-control ascertainment for these two series of subjects have been presented in detail elsewhere (Krishnan et al., 2003; Wiemels et al., 2002; Wrensch et al., 1997, 2004). We ascertained all adults newly diagnosed with glioma (International Classification of Disease for Oncology, morphology codes 9380–9481) in six San Francisco Bay Area counties (Alameda, Contra Costa, Marin, San Mateo, San Francisco, and Santa Clara) from August 1991 to April 1994 (series 1) and from May 1997 to August 1999 (series 2). Cases were ascertained within a median of seven weeks of diagnosis by using Northern California Cancer Center’s Rapid Case Ascertainment program as previously described (Wiemels et al., 2002; Wrensch et al., 1997, 2004). Controls ascertained through random-digit dialing with methods previously described (Wrensch et al., 1997) were frequency matched to cases by age, race, and gender. We began collecting blood specimens from willing subjects partway through the first series and asked all participants in series 2 to donate either a blood and/or buccal specimen. Our previous reports on ERCC1 and ERCC2 variants (Caggana et al., 2001; Chen et al., 2000) included about 150 white cases and an equal number of controls from series 1.
Methods used for genotyping many of the series 1 subjects have been previously described (Caggana et al., 2001; Chen et al., 2000). For the remaining series 1 subjects and series 2 subjects, DNA was isolated from heparinized whole blood by using Qiagen column (Valencia, Calif.) purification. Buccal specimens were used to obtain results for ERCC2 R156R for 147 cases, ERCC2 K751Q for 89 subjects, and ERCC1 C8092A for 92 subjects. Buccal swabs were inserted into a 1.5-ml tube with 300 to 600 μl of 50mM NaOH and vortexed. We then removed the brush from the tube, making sure all liquid was reserved. The tube was boiled in a water bath at 9°C for 5 min. The tube was next centrifuged at 14,000 rpm for 1 min, and the liquid was then transferred into a freezing vial and the amount of liquid measured. The sample was neutralized by adding a 1:10 volume (10% final concentration) of 1 M Tris-EDTA, pH 8.0. The DNA concentration was measured by using Hoescht-33258 fluorimetry. Up to 10 μl was used in a 50-μl polymerase chain reaction (PCR). DNA was stored at −80°C.
Restriction enzymes for ERCC2 genotyping were purchased from NEN (NEN Life Sciences, Boston, Mass.), and PCR was carried out on an ABI 9600 thermocycler (Applied Biosystems, Foster City, Calif.). Each reaction included 0.2–0.6 μM of primers mixed with 50 ng of genomic DNA, 1.0–2.0 mM of MgCl2, 0.8 units of Taq DNA polymerase, and 200 μM of deoxyribonucleotide triphosphates in 10 mM Tris, 50 mM KCl at pH 8.3 in a final reaction volume of 50 μl. The amplification primers for each polymorphism are as follows.
ERCC1 intron 1 (rs3212981) (Rieder et al., 2002)
Quality control measures include blinded analyses, replicates of 10% of samples and positive controls (blood-derived DNA from all known genotypes), and negative controls for contamination (no DNA) were run routinely with patient samples.
For ERCC2 R156R, we compared genotypes from 340 of 364 of the series 1 subjects originally reported by Caggana et al. (2001) with those found by using the technique described above (the current method); 44 of 340 (13%) had discrepant genotypes, and we reanalyzed those a third time using the current method. The agreement for genotyping by the current method was 41/44 (93%); for these 44 subjects, the final genotype used was that found in two out of three assays. For the remaining 24 subjects from the first series that were not genotyped with the current method, we would expect about three subjects (13% times 24) to be potentially misclassified. For the remaining 574 subjects whose genotypes were assessed only by the current method, there is the potential for 7% misclassification. Therefore, for this genotype, we estimate about 43 [3 + (0.07*574)] of 938 total subjects genotyped (4.5%) to be potentially misclassified. For ERCC2 K751Q, 87 of 416 subjects from the Caggana et al. (2001) paper were genotyped by the original and current method, and eight (10%) were discrepant; of these, all eight gave the same result when assessed by the current method a second time and were assigned those genotypes. Thus, for the remaining 329 subjects for whom we used the Caggana et al.–reported genotypes, there is the possibility that 33 (approximately 10% of 329) are misclassified; this corresponds to about 3.5% (33/936) of those reported, assuming complete accuracy of the current method for those genotyped by that method alone.
Blocks from tumors classified as astrocytic (glioblastoma multiforme, anaplastic astrocytoma, and astrocytoma) were collected and processed for immunohistochemical and genetic markers as previously described (Chen et al., 2001). Details for determining extent of TP53 expression by immunohistochemistry and for presence or absence of p53 mutation in exons 5–8 also have been previously described (Chen et al., 2001). In addition, we assessed expression of epidermal growth factor receptor (EGFR) and MDM2 and amplification of these genes. Immunohistochemical staining for EGFR and MDM2 was performed as previously described (Simmons et al., 2001). Briefly, 5-μm sections were deparaffinized in histological grade xylene for 10 min, rehydrated through sequential 95%–70% ethanol, and placed in phosphate-buffered saline (PBS). Microwave antigen retrieval was performed by placing the slides in 10 mM of citrate buffer, pH 6.0, and microwaving for 12 min, followed by two to three 5-min washes in PBS. Endogenous peroxidase activity was blocked with 3% hydrogen peroxide in PBS/0.05% Tween-20 or 3% hydrogen peroxide in methanol for 10 to 20 min. Sections were then washed and blocked for 20 min in the appropriate serum from the same species as the secondary antibody diluted to 10% in PBS. The primary antibody was applied to the sections in a humid chamber overnight at 4°C. After washing, the secondary antibody was applied according to directions in the Envision kit from Dako (Ely, U.K.). Detection of the antibody was performed with diaminobenzadine for 1 to 5 min. Sections were then counterstained with light hematoxylin and mounted. Primary antibodies used were anti-EGFR (Ab-1, clone 528) and anti-MDM2 (Ab-1), both from Oncogene Research Products (San Diego, Calif.). Scoring for EGFR was for membrane/cytoplasmic staining on a three-point scale, where 0 = no staining, 1 = weak/focal staining, and 2 = strong/diffuse staining. Scoring for MDM2 was for nuclear staining on a four-point scale from 0 to 3: 0, no staining; 1, <5% of nuclei with positive staining; 2, 5% to 30% of nuclei stained; and 3, >30% of positive nuclei.
The following methods were used to assess EGFR and MDM2 amplification. EGFR and MDM2 amplification were measured by a quantitative PCR method (ABI 7900) using the generic DNA binding dye SYBR Green I (Stratagene, La Jolla, Calif.), which has been shown to be equivalent to TAQman (ABI) to assess gene copy number (De Preter et al., 2002). Quality control measures for the real-time SYBR Green assay included running triplicate determinations for both the targets (EGFR or MDM2) and control gene (GAPDH). The patient DNA was extracted from paraffin-embedded tumor tissue in parallel with presumed normal DNA extracted from paraffin-embedded, histologically confirmed normal tissue obtained from a glioma patient who consented to this study; all of the paraffin tissue obtained for this case was histologically normal. Cell line DNAs served as positive and negative controls for amplification that were run with each experiment. We used the cell lines A431 (amplified for EGFR), HT29 (negative EGFR amplification), Rh18 (MDM2 amplified) and Rh30 (negative MDM2 amplification). A431 and HT29 were obtained from the American Type Tissue Culture Association (Manassas, Va.), and Rh18 and Rh30 were provided by P. Houghton (St. Jude Children’s Research Hospital). Target gene cycle threshold values for patients were calculated relative to the change in the reference gene (GAPDH) when compared to normal DNA. The copy number (CN) provided the relative quantification ratio according to the following formula.
where ΔCT EGFR = cycle threshold (CT) patient minus CT normal, ΔCT GAPDH = CT patient minus CT normal, CN = ratio of delta CT of the target gene to the delta CT of the reference gene, and PCR efficiency (E) =0.95 (De Preter et al., 2002).
First, we examined genotype frequencies for cases (all, glioblastoma, and nonglioblastoma) and controls by series, by gender within series, and for whites within series. We also examined genotype frequencies by age group 40, 41–60, and >60 for controls, all cases, and cases who had either glioblastoma or other histologies. Among glioblastoma cases versus controls, we also compared ERCC1 and ERCC2 genotype frequencies by presence or absence in the tumor of p53 mutation, EGFR or MDM2 amplification, and immunohistochemical positivity for p53, EGFR or MDM2.
Odds ratios and 95% CIs were calculated with logistic regression in SAS PROC LOGISTIC (SAS Institute, 1990). Models were run separately for each polymorphism variant as either dominant (var/var and var/wt vs. wt/wt) or recessive (var/var vs. var/wt and wt/wt) and adjusted for the matching factors age, gender, and ethnicity and for series in models that combined series. Also, a model was fit that included the number of variant alleles in either ERCC1 C8092A and ERCC2 R156R or ERCC1 C8092A and ERCC2 K751Q. We also compared various combinations of ERCC1 and ERCC2 variants by ethnicity separately in cases and controls and by histologic subtypes of cases using chi-square tests or their exact equivalent, if warranted, using SAS PROC FREQ. Haplotype frequencies for ERCC2 K751Q and R156R and case-control ORs for haplotypes among whites controlling for age and gender were estimated by the methods of Shaid et al. (2002) and R software (R Development Core Team, 2005).
As described in detail elsewhere (Krishnan et al., 2003, Wiemels et al., 2002, Wrensch et al., 1997, 2004), there were a total of 1129 eligible cases as reported by the Rapid Case Ascertainment program from both series 1 and 2. Of these, full interviews were obtained for 81% (896 patients), and 2% (23 patients) completed only a telephone interview. Only 11% (128 patients) refused to enter the study, 1% (22 patients) were too ill or had a language problem, 4% (47 patients) could not be located, and 1% (13 patients) did not get a doctor’s consent to contact. Of those cases who completed the interview, 3% (23 patients) had to be dropped from the study because a neuropathology review indicated that the subject did not have glioma or medulloblastoma (n = 6), the diagnosis date of the original tumor was before the catchment period (n = 3), residence was outside the six Bay Area counties (n = 1), permission for review was not obtained (n = 4), tumor specimens were unavailable (n = 4), or tumor specimens were insufficient for diagnosis (n = 5). Of the 873 case participants, 472 (54%) provided a blood or buccal specimen, and ERCC genotyping was conducted for 419 to 450 (Table 1).
Of the 15,894 phone numbers in both series contacted through random-digit dialing, 8% (n = 1316) yielded eligible controls. Of the total phone numbers contacted, 19% (n = 3086) were not in service, 14% (n = 2242) were businesses, 7% (n = 1047) were faxes or modems, 12% (n = 1964) had no response after 10 calls, 12% (n = 1824) were refusals, 5% (n = 827) had language or health problems, 20% (n = 3158) were eligible but the quota for their age/race/gender had been filled, and 3% (n = 430) were either too young, had multiple lines, lived out of the area, or were closed out. Of the potentially eligible controls, 66% (n = 864) completed a full interview, 13% (n = 169) completed an abbreviated telephone interview only, 14% (n = 184) refused to enter the study, 1% (n = 17) either had a language problem or were too ill to interview, 3% (n = 43) were not located, and upon further examination 3% (n = 39) were either related to cases (n = 3), lived out of the area (n = 9), or were good matches but the study closed (n = 27). Of the 864 controls from series 1 and 2, 558 (65%) provided blood or buccal specimens, and 494 to 519 were genotyped for ERCC1 or ERCC2 variants (Table 1).
For all participants and those subjects genotyped for ERCC1 and ERCC2 variants, Table 1 shows numbers of subjects, basic demographic characteristics (age, gender, ethnicity), and histologic distribution of cases. We have previously noted differences between the series (Wrensch et al., 2004). The most notable was that a higher proportion of subjects were genotyped from series 2 than from series 1 because funding for collecting specimens was first obtained partway through series 1. For series 1, we attempted blood draws on all cases in which the patient was thought to be alive at the time funding was available and on approximately equal numbers of controls (491 in all). Of these, 357 subjects (73%) provided a blood specimen. Also, note that although the ages of cases and controls were matched for the overall group, genotyped cases are on average younger than genotyped controls, primarily because case survival (and therefore ability to obtain constitutive material for genotyping) decreases with age. The histological distributions of cases in the two series were similar, except that series 2 had a higher percentage of cases with oligodendroglioma and lower percentage with oligoastrocytoma. Because the proportion of oligodendroglioma and oligoastrocytoma changed substantially between series 1 and 2, we combined these two histologies in subsequent analyses, as tumors with an oligodendroglial component. Table 1 describes the total study population as well as the subset of cases and controls having a result for at least one ERCC gene. In all, 1017 people were genotyped for at least one ERCC variant, with 958, 936, and 938 being genotyped for ERCC1 C8092A, ERCC2 K751Q, and R156R, respectively.
Table 2 shows genotype frequencies for ERCC1 and ERCC2 variants and case-control ORs by series and overall. For glioblastoma, ORs for those carrying two ERCC1 A alleles were 1.67 in series 1 and 1.64 in series 2, for a nearly statistically significant age-, gender-, ethnicity-, and series-adjusted OR of 1.67 (95% CI, 0.93–3.02; P = 0.09) in the combined series. None of the ORs for the ERCC2 variants was consistently elevated or reduced in either series in all cases versus controls or in cases with glioblastoma or other histologies.
Among whites, allele frequencies for the more common (wild-type) allele for ERCC1 C8092A, ERCC2 R156R, and ERCC2 K751Q were 0.74, 0.55, and 0.63 for controls and 0.74, 0.55, and 0.61 for glioma cases. Alleles at the three loci were in Hardy-Weinberg equilibrium (P values were 0.09, 0.51, and 0.27 among controls and 0.98, 0.82, and 0.64 among cases for ERCC1 C8092A, ERCC2 R156R, and K751Q, respectively).
Also, note controls’ genotype frequencies did not differ significantly in series 1 and 2 for ERCC1 C8092A (P = 0.88) or for ERCC2 R156R (P = 0.13). However, overall controls’ genotype frequencies for ERCC2 K751Q did differ significantly between series (P = 0.05), but the white controls’ genotype frequencies for this polymorphism did not differ significantly between series (P = 0.15).
Ethnic differences in genotype frequencies of the ERCC1 and ERCC2 variants were apparent in both cases and controls (Table 3). Among cases, whites and Latinos were significantly less likely than other ethnic groups to have ERCC1 AA genotype, and there were significant ethnic differences in percentages of cases that had ERCC2 K751Q CA or CC genotypes and R156R AC or AA genotypes. In controls, the percentages of people with ERCC2 variants differed significantly among ethnic groups. There was also substantial, although not statistically significant, ethnic variability for ERCC1 AA genotypes, with frequencies ranging from none (0%) among 19 Asians to 20% among five controls of other (not Asian, black, Latino, or white) ethnicity to 5% among 444 whites. Because we lack sufficient numbers of subjects among the nonwhite ethnicities to further stratify case-control comparisons, we limited subsequent analyses to whites.
For whites only, we estimated logistic models that included the three ERCC polymorphisms and adjusted for age, gender, and series. For all cases versus controls and glioblastoma cases versus controls, none of the ORs for the three polymorphisms achieved statistical significance (Table 4). The OR for white glioblastoma versus controls for ERCC1 AA versus AC or CC genotype (OR, 1.77; 95% CI, 0.87–3.58; P = 0.11) was similar to that observed in Table 2 for all glioblastoma cases versus controls (OR, 1.67; 95% CI, 0.93–3.02; P = 0.09); the elevated P value is due to there being fewer cases in the model having results for all three polymorphisms and restrictions to whites (162 versus 221). Also, white nonglioblastoma cases had a nearly significant excess of ERCC2 K751Q genotype CC versus AA/AC compared to controls (OR, 1.82; 95% CI, 0.97–3.45; P = 0.06). In whites, we also examined all the age- and gender-adjusted case-control models with univariate and pairwise inclusions of the ERCC1 and ERCC2 polymorphisms (results not shown). In addition to the results above, the nonglioblastoma cases versus control OR for the ERCC2 K751Q QQ versus KQ/KK was approximately significant (P = 0.07) for the model that included the two ERCC2 polymorphisms. We also estimated haplotype frequencies for ERCC2 R156R and K751Q among whites but found no significant associations with case-control status (results not shown).
We found no significant differences in the genotype frequencies of ERCC1 and ERCC2 variants among four histologic categories of white glioma cases, glioblastoma, lower grade astrocytoma (anaplastic astrocytoma and astrocytoma), oligodendroglial tumors (oligoastrocytoma and oligodendroglioma), and other histologies (results not shown).
Among whites with glioblastoma, we did not detect statistically significant or suggestive variation in any of the ERCC genotype frequencies with presence or absence of tumor p53 gene mutation or protein expression or EGFR gene amplification or protein expression (results not shown). However, as shown in Table 5, the ERCC1 genotype AA was overrepresented among subjects whose tumors were positive for expression of MDM2, although the difference was not statistically significant. The CC genotype was overrepresented among those whose tumors had MDM2 amplification (P = 0.09), though again, the results were not statistically significant.
Among whites, glioma cases were significantly more likely than controls to have two variant alleles at both ERCC1 C8092A and ERCC2 K751Q (OR, 3.2; 95% CI, 1.1–9.3; P = 0.03) (Table 6). There was no linear response with number of variant alleles up to 4; that is, the ORs for 0, 2, or 3 variant alleles versus 1 variant allele (chosen as referent because it was the most common category) were very close to 1. We found very similar results when analyses were restricted to white glioblastoma cases versus controls (results not shown).
To our knowledge, except for our two previous reports (Caggana et al., 2001; Chen et al., 2000), no studies have published the associations of ERCC1 and ERCC2 polymorphisms and primary malignant brain tumors in population-based studies. Recently, Sadetzki et al. (2005) reported an OR of 1.68 (95% CI, 1.00–2.84; P = 0.05) for ERCC2 K751Q QQ/KQ versus KK and meningioma. ERCC2 (XPD), discovered through its role in the UV radiation sensitivity disease xeroderma pigmentosum (XP), has been implicated in brain cancer development in XP patients (Giglia et al., 1999; Kraemer et al., 1984). Giglia et al. (1999) pointed out that glioma in an XP patient might be due to unrepaired DNA damage from free radicals. There have been several studies of ERCC2 variants with other cancers, most notably with head and neck cancers, lung cancer, and skin cancer (Benhamou and Sarasin, 2002; Goode et al., 2002; Hoeijmakers et al., 1989; Hou et al., 2002; Liang et al., 2003; Moller et al., 2000; Mort et al., 2003; Spitz et al., 2001; Sturgis et al., 2000; Xing et al., 2002; Zhou et al., 2002, 2003; Zhu et al., 2004). ERCC1 and ERCC2 are potentially relevant to cancer because of their involvement in the process of nucleotide excision repair (NER) (Goode et al., 2002). The NER pathway is thought to act predominately on bulky DNA lesions and UV damage. However, other forms of DNA damage that are possibly relevant to gliomagenesis may require NER-type repair. Recent studies show that both base excision repair and NER pathways are required to repair DNA lesions induced at low N-nitrosodiethylamine concentrations (Aiub et al., 2004). Other studies have shown that oxidative DNA damage can also be eliminated via the NER pathway in bacteria (Czeczot et al., 1991; Gellon et al., 2001; Lin and Sancar, 1989; Scott et al., 1999; Swanson et al., 1999). The NER pathway removes damaged DNA bases by introducing nicks 5′ and 3′ to an abasic site in vitro, and NER contributes to the release of 8-oxoguanine from DNA (Boiteux et al., 2002; Gellon et al., 2001). Variation in efficiency of these processes might influence either cancer development (defective or inefficient repair could lead to accumulation of deleterious mutations in the absence of apoptotic destruction of DNA-damaged cells) or cancer progression (by any of the previously mentioned mechanisms or by more efficient repair reducing effectiveness of chemotherapy aimed at DNA damage and resultant reduction of cancer cell killing). In addition to the direct role of ERCC2 in DNA repair, the complexes it forms with the transcription factor TFIIH and other molecules are involved in transcription activation or with cdk-activating kinases that lead to involvement with apoptosis and phosphorylation of nuclear receptors (Benhamou and Sarasin, 2002). Chen et al. (2003) recently demonstrated that downregulation of ERCC2 influences upregulation of cdk-activating kinases and mitosis while at the same time leading to transcriptional silencing.
Although there has been limited study of the functional consequences in the numerous polymorphic variants in ERCC2, the K751Q variant would be expected to have the greatest functional effects because of the electronic configuration of the amino acid change and the location of the amino acid in the TFIIH complex. Despite this expectation, mixed results have been reported about the effects of this ERCC2 variant on DNA repair. For example, among 100 healthy volunteers, homozygous QQ individuals had 15% reduced DNA repair capacity in UV-induced damage to lymphocytes compared to KK individuals (P = 0.06) (Qiao et al., 2002). In 65 healthy individuals, Affatato and colleagues (2004) found no difference in frequency of lymphocyte chromosome aberrations overall (OR for variant QK and KK versus QQ individuals, 1.06; 95% CI, 0.39–2.88), but nonstatistically significant ORs in excess of 2.0 or less than 0.5 were observed in substratifications that were based on current smoking status and age. In a study of 31 women (one with breast cancer and 15 at high risk of breast cancer based on family history), Lunn and coworkers (2000) reported suboptimal repair of X-ray induced DNA breaks and gaps in peripheral lymphocytes in women who were KK versus KQ and QQ (P = 0.01).
No meaningful or significant associations were observed between ERCC2 K751Q genotypes and case-control status studies of bladder cancer, basal cell carcinoma, melanoma, colorectal cancer, or lung cancer (Goode et al., 2002; Mort et al., 2003). However, significant associations of ERCC2 K751Q and/or D312N genotypes or interactions of genotypes with smoking history were observed for lung cancer and squamous cell carcinoma of the head and neck in some studies (Hou et al., 2002; Liang et al., 2003; Sturgis et al., 2000; Xing et al., 2002; Zhou et al., 2002, 2003). Several authors have pointed out that findings for the two above-mentioned ERCC2 variants that are in linkage disequilibrium may reflect an effect of the variants themselves or of a nearby gene. In the present study, there was no overall association of either ERCC2 variant K751Q or R156R with glioma case-control status, but there was a suggestive increased odds for QQ versus KK/KQ genotypes among glioma cases with histologies other than glioblastoma multiforme in the model in which all three ERCC variants were included (P = 0.06).
The allele frequencies of the more common ERCC2 allele K751 of 0.61 and 0.63 among white cases and controls, respectively, are very similar to those observed in other studies of Caucasians ranging from 0.63 to 0.67 (David-Beabes et al., 2001; Hou et al., 2002; Spitz et al., 2001; Sturgis et al., 2000; Zhou et al., 2002). This similarity suggests that our genotyping methods for this polymorphism achieved results comparable to those published by other laboratories.
The functional consequences of the ERCC1 variant studied here are not well defined, and no other published studies have examined this variant in relation to cancer. However, antisense to ERCC1 (ASE-1) was discovered to be a unique autoantigen in human serum with reactivity to components of cellular mitotic apparatus; its function is not precisely known, but it appears to co-localize with rDNA (i.e., DNA that codes for ribosomal RNA) (Whitehead et al., 1997). It is found in fibrillar centers in the interphase nucleolus and the nucleolus organizer regions of chromosomes during mitosis. Whitehead and colleagues (1997) pointed out that the overlap of ASE-1 and ERCC1 might lead to co-regulation through transcriptional interference. With regard to disease implications of ASE-1, we found only one report. Edworthy et al. noted that autoantibodies to ASE-1 are more common in sera of patients with systemic lupus erythematosus than in sera from people with other systemic rheumatic diseases (Edworthy et al., 2000). In the present study, there was no overall association of ERCC1 (ASE-1) variant genotypes with glioma, although there was a suggestive increased OR for AA versus AC/CC genotypes among glioblastoma cases in both series (overall OR, 1.67; 95% CI, 0.93–3.02; P = 0.09), the magnitude of which remained the same after restricting analyses to whites and further adjustment for other ERCC variants. We did not observe significant variation in ERCC1 variant genotypes by histologic type of glioma or molecular type of glioblastoma multiforme, but among the small number of cases in which the tumors had MDM2 amplification, there was a suggestive increase of people with the more common ERCC1 allele, CC (P = 0.09). We also observed a significantly increased OR for glioma among people who were homozygous for variants in both ERCC1 C8092A (AA) and ERCC2 K751Q (QQ).
Since most of the subjects in series 1 were included in the two previous reports (Caggana et al., 2001; Chen et al., 2000), results for series 1 are similar to those previously reported. In particular, we previously noted a decreased odds of glioma with K751Q QQ/KQ (CC/CA) versus KK (AA) and increased risk of glioma with R156R AA versus AC/CC (Caggana et al., 2001), with similar results seen for the complete set of series 1 subjects reported here in Table 2. However, these results were not replicated in series 2. We also previously reported significantly elevated ORs of oligoastrocytoma for those with ERCC1 CC versus AC/AA genotype (this would correspond to a decreased OR for AC/AA versus CC genotypes) (Chen et al., 2000). In the complete set of series 1 subjects reported here, we found significantly decreased ORs for nonglioblastoma histologies (which includes oligoastrocytoma) for those with AC/AA versus CC genotypes. However, these results were not replicated in series 2. The different results of the two series could be due to false positives in series 1 or false negatives in series 2. Without results from additional subjects, it is not possible to decide if any lack of consistency between series is due to random variation or to some differences between the subjects in the two series. However, it is worth noting the higher participation rates for genotyping in the second series (94% for controls and 71% for cases) versus the first series (37% for controls and 39% for cases), a consequence of having obtained funding for blood draws for polymorphism studies only partway through the first series. Also of note is that among the white controls, the K allele frequency was 0.65 for series 2 and 0.58 for series 1, indicating that K allele frequency in series 2 controls was more similar to those observed in other studies (range, 0.63–0.67) (David-Beabes et al., 2001; Hou et al., 2002; Spitz et al., 2001; Sturgis et al. 2000; Zhou et al., 2002). Whether this is just due to chance (P = 0.15 comparing genotype frequencies in series 1 and 2 white controls) or to some unknown selection bias operating in series 1 controls is difficult to determine.
Since the study used random-digit dialing to ascertain controls, it is worth exploring the possible impact that bias introduced by the greatly increased use of cell phones and caller ID during the 1990s might have on the characteristics of people ascertained through random-digit dialing. Although we could find no information on the influence of caller ID, Blumberg et al. (2004) recently reported that among 16,677 households interviewed in person by the U.S. Census Bureau for the National Health Interview Survey in the first half of 2003, only 4.4% of U.S. civilian households had no phone or only a wireless phone. Furthermore, the major factors associated with greater percentage of wireless-only service were age and income. Although approximately 6% of interviewees aged 18 to 24 and 6% of those with an annual income less than $20,000 reported only wireless telephone service, neither this age nor this household income group is prevalent among glioma cases. Although we cannot assess the bias resulting from who chooses to answer the telephone, the results of Blumberg et al. suggest that the vast majority of people in our study catchment area have landline phones.
Although we found no other published studies on associations of glioma with the genotype frequencies of the particular ERCC1 and ERCC2 polymorphisms studied here, Kollmeyer et al. (2004) reported that among 141 glioma cases and 108 controls, in seven single nucleotide polymorphisms near GLTSCR1, only polymorphisms in GLTSCR1-exon1 T and ERCC2-exon 22 C alleles were significantly associated with oligodendroglioma development.
In addition, as the complexity of genetic susceptibility to cancer is beginning to be revealed (Chen and Hunter, 2005; Spitz et al., 2005), it is becoming increasingly clear that multiple genes are likely to be involved in susceptibility to glioma and other cancers and that these genes may interact in various ways with myriad potentially carcinogenic environmental exposures. A major challenge in the future of molecular cancer epidemiology will be to integrate information on environmental exposures, constitutive genomics, and tumor genotypes and phenotypes to provide meaningful evaluation of cancer risk.
In conclusion, we do not find strong evidence for a role of these ERCC variants in adult glioma. However, given the relatively small numbers of subjects in some subgroups, the finding of Kollmeyer et al. noted above, and the likelihood that multiple genes each with a potentially small effect may be involved in cancer susceptibility, clarification about the relation of glioma, if any, to polymorphisms in this chromosomal region in additional series of cases and controls may be helpful. Given the proximity of ERCC1 (ASE-1) and ERCC2 to the putative glioma suppressor genes GLTSCR1 and GLTSCR2 (Smith et al., 2000), future studies might consider association of haplotype variation among these genes with glioma.
Thanks to Richard Davis for pathology review of series 1 cases, Peter Houghton of the St. Jude Children’s Research Hospital for providing Rh18 and Rh30, and the pathology departments of Alexian Brothers Medical Center, Alta Bates Summit Medical Center, Brook-side, California Pacific Medical Center, Doctors Medical Center Pinole/San Pablo, Eden Medical Center, El Camino Hospital, Good Samaritan Hospital, Alameda County Medical Center Highland Hospital, John Muir Medical Center, Kaiser Redwood City, Kaiser San Francisco, Kaiser Santa Teresa, Community Hospital of Los Gatos, Los Medanos Hospital, Marin General Hospital, Merrithew Memorial Hospital, Mills Peninsula Hospital, Mt. Diablo Medical Center, Mt. Zion Medical Center, Naval Hospital, O’Connor Hospital, Ralph K. Davies Medical Center, Saint Louise Regional Hospital, San Francisco General, San Jose Medical Center, San Leandro Hospital, San Mateo County Health Center, San Ramon Regional Medical Center, Santa Clara Valley Medical Center, Sequoia Hospital, Seton Medical Center, St. Francis Memorial Hospital, St. Luke’s Hospital, St. Rose Hospital, Stanford, University of California, San Francisco, Valley Livermore Memorial, Veterans Palo Alto, San Francisco VA Medical Center, and Washington Hospital for providing tumor specimens for review and molecular analyses.
1This work was supported by the National Institutes of Health through grants R01CA52689 and P50CA097257.
3Abbreviations used are as follows: ASE-1, antisense to ERCC1; CN, copy number; CT, cycle threshold; EGFR, epidermal growth factor receptor; ERCC1, excision repair cross-complementing rodent repair deficiency complementation group 1; ERCC2, excision repair cross-complementing rodent repair deficiency complementation group 2; GLTSCR, glioma tumor suppressor candidate region gene; K, lysine; MDM2, mouse double minute 2 (human homolog of p53 binding protein); NER, nucleotide excision repair; OR, odds ratio; PBS, phosphate-buffered saline; PCR, polymerase chain reaction; Q, glutamine; TFIIH, transcription factor IIH; var, variant; wt, wild type; XP, xeroderma pigmentosum.