Skip to main content
NCBI home page
Carcinogenesis logoLink to Carcinogenesis
. 2012 May 11;33(9):1692–1698. doi: 10.1093/carcin/bgs170

A Population-based Study of DNA Repair Gene Variants in Relation to Non-melanoma Skin Cancer as a Marker of a Cancer-prone Phenotype

Ingo Ruczinski 1,†, Timothy J Jorgensen 2,†, Yin Yao Shugart 3, Yvette Berthier Schaad 4,5, Bailey Kessing 5, Judith Hoffman-Bolton 4,6, Kathy J Helzlsouer 7, WHLinda Kao 4, Lee Wheless 8, Lesley Francis 8, Rhoda M Alani 9, Paul T Strickland 4,10, Michael W Smith 11, Anthony J Alberg 8,*
PMCID: PMC3514896  PMID: 22581838

Abstract

For unknown reasons, non-melanoma skin cancer (NMSC) is associated with increased risk of other malignancies. Focusing solely on DNA repair or DNA repair-related genes, this study tested the hypothesis that DNA repair gene variants contribute to the increased cancer risk associated with a personal history of NMSC. From the parent CLUE II cohort study, established in 1989 in Washington County, MD, the study consisted of a cancer-free control group (n 5 2296) compared with three mutually exclusive groups of cancer cases ascertained through 2007: (i) Other (non-NMSC) cancer only (n 5 2349); (ii) NMSC only (n 5 694) and (iii) NMSC plus other cancer (n 5 577). The frequency of minor alleles in 759 DNA repair gene single nucleotide polymorphisms (SNPs) was compared in these four groups. Comparing those with both NMSC and other cancer versus those with no cancer, 10 SNPs had allelic trend P-values <0.01. The two top-ranked SNPs were both within the thymine DNA glycosylase gene (TDG). One was a non-synonymous coding SNP (rs2888805) [per allele odds ratio (OR) 1.40, 95% confidence interval (CI) 1.16–1.70; P-value 5 0.0006] and the other was an intronic SNP in high linkage disequilibrium with rs2888805 (rs4135150). None of the associations had a P-value <6.6310−5, the threshold for statistical significance after correcting for multiple comparisons. The results pinpoint DNA repair genes most likely to contribute to the NMSC cancer-prone phenotype. A promising lead is genetic variants in TDG, important not only in base excision repair but also in regulating the epigenome and gene expression, which may contribute to the NMSC-associated increase in overall cancer risk.

Introduction

Non-melanoma skin cancer (NMSC) is by far the most common human cancer (1). NMSC is usually treated by local excision and is rarely fatal, but a personal history of NMSC is associated with increased risk of other malignancies (2). In a meta-analysis of prospective studies with individual-level data, a prior NMSC diagnosis was associated with a 50% greater risk of developing another type of cancer (2). NMSC was associated with a broad spectrum of malignancies (2), suggesting that NMSC may be a marker of a cancer-prone phenotype. NMSC cases comprise two major histologic types, basal cell carcinoma (BCC) and squamous cell carcinoma (SCC), and the overall increased risk of other malignancies associated with NMSC has been consistently observed for both BCC and SCC. However, it is not known why a personal history of NMSC is associated with increased risk of non-cutaneous malignancies, and there has been no previous hypothesis-driven research on this topic. The fact that a personal history of NMSC affects overall cancer risk across a broad spectrum of malignancies potentially makes NMSC a cross-cutting marker of overall cancer risk that could potentially provide valuable clues to human carcinogenesis.

Major DNA repair defects cause cancer-prone phenotypes. For example, mutated DNA repair genes cause cancer-prone syndromes that involve multiple cancers, such as xeroderma pigmentosum (XP) (3,4) and Bloom’s syndrome (5,6). XP is an autosomal recessive condition caused by rare, high-penetrant mutations in nucleotide excision repair (NER) genes. XP directly demonstrates that defective DNA repair can be a potent cause of genetic predisposition to both NMSC and other malignancies. The DNA repair deficiency in XP is characterized by extreme sensitivity to ultraviolet radiation, causing t2000-fold increased skin cancer rates before age 20 years (4). XP patients also have a marked excess of internal malignancies (4). The fact that rare mutations in DNA repair genes cause major cancer susceptibility in XP patients suggests that even common, low-penetrant DNA repair gene variants may jointly affect susceptibility to NMSC and other malignancies in the general population (7,8). However, except for rare, autosomal recessive cancer syndromes caused by inherited mutations in DNA repair genes, the contribution of common allelic variants of DNA repair genes to cancer risk in the general population remains incompletely characterized (e.g. ref. 7).

DNA repair is accomplished through multiple interacting biochemical pathways entailing overlapping functions and cross-talk; collectively, these comprise t140 genes (Table I, column 1). All of these pathways have been implicated in carcinogenesis. The NER pathway is emphasized for NMSC because NER removes cyclobutane pyrimidine dimers and 6-4 photoproducts, carcinogenic DNA lesions unique to ultraviolet radiation exposure. Ultraviolet radiation-induced DNA lesions are not restricted to pyrimidine dimers but also entail oxidative lesions (8,9), demonstrating that even protection against skin cancer involves multiple DNA repair pathways. Thus, in the present study, all known human DNA repair genes were considered candidates for association with the NMSC-associated cancer-prone phenotype.

Table I.

DNA repair genes studied and the number of SNPs per gene according to DNA repair or DNA repair-related pathway.

DNA repair pathway Genes (no. of SNPs)
1. Nucleotide excision repair: 28 genes, 188 SNPs CCNH (2), CDK7 (3), DDB1 (2), DDB2 (6), ERCC1 (4), ERCC2 (8), ERCC3 (7), ERCC4 (5), ERCC5 (8), ERCC6 (8), ERCC8 (5), GTF2H1 (5), GTF2H3 (1), GTF2H4 (5), LIG1 (8), MMS19 (4), MNAT1 (8), POLD1 (5), POLE (10), POLH (2), RAD23A (1), RAD23B (16), RPA1 (12), RPA2 (6), RPA3 (34), XAB2 (3), XPA (2), XPC (8)
2. Base excision repair: 14 genes, 56 SNPs APEX1 (1), LIG3 (3), MBD4 (2), MPG (2), MUTYH (4), NEIL1 (1), NEIL2 (14), NTHL1 (2), OGG1 (3), PNKP (2), SMUG1 (3), TDG (5), UNG (5), XRCC1 (9)
3. Non-homologous end joining: 7 genes, 78 SNPs DCLRE1C (11), LIG4 (6), NHEJ1 (6), PRKDC (9), XRCC4 (17), XRCC5 (27), XRCC6 (2)
4. Homologous recombinational repair: 21 genes, 139 SNPs BRCA1 (9), BRCA2(16), BRIP1(12), DMC1(OTOP3) (6), EME1(4), EME2(1), MRE11A (8), MUS81 (2), NBN (11), PALB2 (FANCN) (6), RAD50 (3), RAD51 (9), RAD51C (5), RAD51L3 (5), RAD52 (9), RAD54B (5), RAD54L (3), SHFM1(LOC402679) (5), XRCC2 (5), XRCC3 (7), ZNF350 (8)
5. Mismatch repair: 9 genes, 101 SNPs MLH1 (4), MLH3 (4), MSH2 (25), MSH3 (25), MSH4 (6), MSH5 (6), MSH6 (3), PMS1 (19), PMS2 (9)
6. DNA–DNA crosslink repair: 9 unique genes, 1 shared gene, 47 unique SNPs FANCA (9), FANCC (11), FANCD2 (4), FANCE (3), FANCF (2), FANCG (2), FANCI (one SNP shared with POLG; see POLG below), FANCL (9), FANCM (6)
7. DNA–protein crosslink repair: 1 gene, 3 SNPs TDP1 (3)
8. Direct reversal repair: 3 genes, 23 SNPs ABH2 (ALKBH2) (1), ABH3 (ALKBH3) (1), MGMT (21)
9. PARP family: 3 genes, 14 SNPs PARP1 (9), PARP2 (1), PARP4 (4)
10. DNA damage signal transduction: 12 genes, 60 SNPs ATR (8), CDKN1A (6), CDKN2A (11), CHEK1 (1), CHEK2 (10), HUS1 (1), PPP1R13L (gene overlaps with ERCC1) (4), RAD1 (3), RAD17 (3), RAD9A (1), TELO2 (5), TP53 (7)
11. Hereditary disease genes with deficient DNA repair phenotypes: 4 genes, 24 SNPs ATM (6), BLM (4), REQL4 (2), WRN (12)
12. Putative DNA repair functions: 7 genes, 26 SNPs CHAF1A (8), DCLRE1B (2), H2AFX (2), NEIL3 (1), PCNA (4), POLB (3), POLG (6)
Open in a new tab

Some genes may be associated with more than one pathway; if so, the gene was arbitrarily assigned to the most recognized pathway.

Consistent with other cohort studies, we previously reported that a prior NMSC diagnosis was associated with significantly elevated risk of other malignancies (10). Building on this finding, the present study focused solely on DNA repair and DNA repair-related genes and employed the same cohort to test the novel hypothesis that germ-line polymorphisms in DNA repair genes contribute to the cancer-prone phenotype characterized by the diagnosis of both NMSC and another type of cancer in the same individual.

Materials and Methods

This prospective, population-based study was embedded within the CLUE II (named after the campaign slogan of ‘Give us a clue to cancer and heart disease’) cohort. CLUE II is a community-based cohort study established in Washington County, MD in 1989 (11). In this study, the independent variables were single nucleotide polymorphisms (SNPs) in DNA repair genes. The dependent variable was the occurrence of cancer classified according to the joint outcome of NMSC and other cancer, with a cancer-free group to compare with those with cancer other than NMSC only, NMSC only, and both NMSC plus other cancer. The study of all non-NMSC cancers combined is consistent with the prior observation that the excess risk of other malignancies observed in those with a personal history of NMSC is not confined to a few specific malignancies but rather seems to apply to an overall heightened cancer risk.

Study population

The CLUE II cohort was established when baseline data were collected in May through November 1989 from volunteers who were mostly residents of Washington County, MD. The campaign was designed to collect blood samples from as many adult residents as possible in the Washington County and surrounding (30 mile radius) area. Brief medical histories and blood pressures were taken, and 20ml of blood was drawn into heparinized vacutainers. Specimens were refrigerated at once and processed within 24h. Buffy coats were placed in storage at −70°C. During the baseline data collection, participants completed a questionnaire that included information on age, race, sex, cigarette smoking, height, weight and years of schooling. This study was approved by the Institutional Review Boards of the Johns Hopkins University and Medical University of South Carolina.

From the entire cohort of 30 726 participants, 6589 were selected for inclusion in the present study. This included all those with a confirmed cancer diagnosis as of 30 September 2007 (plus 96 cases added 1 April 2008) and a cancer-free comparison group that was a 10% age-stratified random sample of adult CLUE II participants, plus 250 controls added for a lung cancer substudy. Once selected, NMSC or other cancer diagnoses that occurred by 31 December 2010 contributed to the categorization of study endpoints.

Of the 6589 subjects selected for inclusion, genotyping could not be attempted for 312, and 263 with ≥5% genotyping failures were excluded. Among the remaining 6014 subjects, genetic heterogeneity was assessed using principal components. Comparison with the three HapMap2 populations identified 98 (1.6%) subjects of non-European ethnic ancestry, who were excluded due to concerns about population stratification and differential risks of NMSC. To account for any residual ancestral differences, the first three principal components were adjusted for in all analyses. The final study population of 5916 was classified into four categories: (i) cancer-free group (n 5 2296) for comparison to those with a pathologically confirmed diagnosis of (ii) any cancer other than NMSC (‘other cancer only,’ n 5 2349); (iii) NMSC with no other cancer diagnosis (‘NMSC only,’ n 5 694) and (iv) NMSC plus another type of cancer (‘NMSC plus other cancer,’ n 5 577).

Cancer diagnoses were ascertained via linkage to the Washington County Cancer Registry, augmented (except for NMSC) by linking to the Maryland Cancer Registry. Pathologically confirmed NMSC diagnoses were ascertained by the Washington County Cancer Registry, including NMSC cases diagnosed both before and after study baseline in 1989.

SNP Selection

The SNP selection algorithm was previously described (12). In brief, 118 human DNA-repair genes and DNA-repair-related genes (13) were assigned to one of the five central DNA repair pathways (Table I, rows 1–5) or to one of seven ancillary DNA repair-related pathways (Table I, rows 6–12) based on putative activities, functions, sequence homology or physical associations with other DNA repair proteins. These pathways were prioritized in the order listed in Table I based on evidence of association with NMSC. All known non-synonymous coding SNPs in genes from these 12 pathways were selected for genotyping, regardless of minor allele frequency (MAF). Tagging SNPs were selected using a general MAF cut-off of 0.05. This was relaxed to 0.01 for top priority pathway genes in the SNP selection process. Despite the inclusion of rare variants in the SNP selection, for the data analyses all SNPs with MAF < 0.05 were excluded, due to the high prevalence of monomorphic and quasi-monomorphic SNPs and the lack of precision associated with the low-MAF SNPs that were not monomorphic. The CEPH population (European ancestry) data from the HapMap (CEU) were used to estimate SNP frequencies and linkage disequilibrium, as the Clue II cohort comprised largely northern European ancestry. R 2 was set to t0.80, although some SNPs could be in high linkage disequilibrium (LD) due to the approach of selecting functional and tagging SNPs. SNPs were excluded if they had a genotyping platform design score <0.60. HapMap data were accessed using Tagger™ software (14), with 1337 DNA repair gene SNPs identified that met these criteria.

Genomic DNA extraction and genotyping

Buffy coats with no previous DNA extraction were stored at −70°C from collection until DNA extraction. Genomic DNA was extracted from buffy coats using standard phenol/chloroform extraction procedures, followed by ethanol precipitation to concentrations of 150 µg/mL. The precipitated DNA was resuspended in low-salt buffer to uniform DNA concentrations (50 µg/mL) for genotyping. When previously extracted DNA was available, that DNA was resuspended in low-salt buffer to 50 µg/mL. Genotyping was attempted for 1337 selected DNA repair SNPs using Illumina GoldenGate® arrays customized by the manufacturer. Among these, SNPs were excluded for the following reasons: 86 (6.4%) genotyping failure in ≥5% samples, 467 (34.9%) with MAF < 0.05 (247 of these were monomorphic or quasi-monomorphic), and 25 (1.9%) deviated from Hardy–Weinberg equilibrium (HWE) at P < 0.001. This left 759 SNPs for analyses. For quality control, the genotyping results were compared for n 5 24 SNPs genotyped in previous studies using different genotyping platforms for those who overlapped with the present study population (per SNP number of participants with duplicate genotype results ranged from 3943 to 4757); the overall concordance was 98%.

Statistical analyses

For each SNP, an exact test was used to test whether genotypic frequencies in the cancer-free controls departed from HWE (15). Logistic regression was used to estimate odds ratios, confidence intervals and P-values for the associations between each SNP and cancer risk. Odds ratios were estimated using the cancer-free group as the referent category. The hypothesis tested was that minor alleles in DNA repair genes would be most prevalent in the NMSC plus other cancer group and least prevalent in the cancer-free comparison group. Given this hypothesis, the SNP screening strategy focused solely on the comparison of the NMSC plus other cancer group to the cancer-free comparison group. The additive genetic model (SNPs coded as having 0, 1 or 2 copies of the minor allele) was used to screen the associations with P-values < 0.01 in the NMSC plus other cancer category. For the SNPs screened with the additive genetic model P-value < 0.01 for the NMSC plus other cancer group, the associations between these SNPs and the risk of other cancer only and NMSC only were also evaluated. The associations between these SNPs and the NMSC cancer-prone phenotype were then further characterized in greater detail by assessing the dominant and recessive phenotypic models of inheritance.

Analyses were carried out in the statistical environment R (http://cran.r-project.org/). All statistical tests were two-sided. Assuming a P-value of 0.05 represents a statistically significant association for a single comparison, a Bonferroni multiple comparison correction for the 759 DNA repair gene SNPs yielded a significance threshold of 0.05/75956.6 × 10−5.

Results

Baseline descriptive characteristics of the four study groups are summarized in Table II. The cancer-free group was significantly younger (43 years) than the three groups that included cancer, whose average ages ranged from 57 to 62 years. The NMSC plus other cancer group was 56% male, notably higher than the other study groups. The prevalence of ever-smokers was highest (56%) in the other cancer only group, and current smoking was less prevalent in study groups that included NMSC. Other than NMSC, the most common malignancies were breast, prostate, lung and colorectal cancer.

Table II.

Baseline characteristics and cancer outcomes during follow-up according to a personal history of cancer by the end of follow-up: CLUE II Cohort, Washington County, MD (1989–2010)

Characteristic No cancer Other cancer only NMSC only NMSC plus other cancer Total
Total (column %) 2296 100% 2349 100% 694 100% 577 100% 5916 100%
Age group
<25 y 380 16.6% 22 0.9% 3 0.4% 0 0.0% 405 6.8%
25–34 y 419 18.2% 122 5.2% 34 4.9% 8 1.4% 583 9.9%
35–44 y 426 18.6% 288 12.3% 68 9.8% 32 5.5% 814 13.8%
45–54 y 407 17.7% 526 22.4% 128 18.4% 85 14.7% 1146 19.4%
55–64 y 350 15.2% 679 28.9% 157 22.6% 179 31.0% 1365 23.1%
65–74 y 234 10.2% 543 23.1% 200 28.8% 206 35.7% 1183 20.0%
≥75 y 80 3.5% 169 7.2% 104 15.0% 67 11.6% 420 7.1%
Mean age (Q1,Q3) 43.1 (30,57) 56.7 (48,67) 60.0 (50,71) 62.4 (56,69) 52.3 (41,65)
Sex
Male 963 41.9% 1041 44.3% 337 48.6% 325 56.3% 2666 45.1%
Female 1333 58.1% 1308 55.7% 357 51.4% 252 43.7% 3250 54.9%
Mean BMI (Q1,Q3) 25.7 (22.0,28.4) 26.8 (23.6,29.3) 25.9 (23.0,28.0) 26.2 (23.4,28.3) 26.2 (22.9,28.7)
Education
<12 y 597 26.0% 681 29.0% 201 29.0% 153 26.5% 1632 27.6%
12 y 998 43.5% 991 42.2% 284 40.9% 243 42.1% 2516 42.6%
t12 y 701 30.5% 674 28.7% 209 30.1% 181 31.4% 1765 29.8%
Cigarette smoking
Never 1161 50.6% 1034 44.0% 354 51.1% 283 49.0% 2832 47.9%
Former 561 24.4% 830 35.3% 253 36.5% 215 37.3% 1859 31.4%
Current 574 25.0% 485 20.6% 86 12.4% 79 13.7% 1224 20.7%
Type of Cancer
Lung Not applicable 294 12.5% Not applicable 38 6.6 332 11.3%
Colorectal 227 9.7% 65 11.3 292 10.0%
Breast 434 18.5% 93 16.1 527 18.0%
Prostate 363 15.5% 119 20.6 483 16.5%
Melanoma 138 5.9% 59 10.2 197 6.7%
Pancreas 62 2.6% 10 1.7 72 2.5%
Bladder 93 4.0% 37 6.4 130 4.4%
Other 737 31.4% 156 27.0 893 30.5%
Open in a new tab

Column totals may vary due to missing data and column percentage may not sum exactly to 100% due to rounding.

Table III lists results for the 10 SNPs with additive model P-values < 0.01 for the cancer-prone phenotype (‘both NMSC plus other cancer’) versus the ‘no cancer’ comparison. The smallest P-value was 6 × 10−4, greater than the Bonferroni threshold of 6.6 × 10−5; thus, none of the observed associations were statistically significant after correction for multiple comparisons. The 10 SNPs represented the following 6 pathways and 8 genes: 2 nucleotide excision repair (ERCC8, ERCC3), 2 homologous recombinational repair (PALB2, DMC1) and 1 each from base excision repair (TDG), direct reversal repair (MGMT), DNA damage signal transduction (CHEK2) and mismatch repair (MSH6). Both TDG and PALB2 had two SNPs associated with the cancer-prone phenotype; each of these SNP pairs comprised one non-synonymous coding SNP and one intronic SNP that were in high LD and had identical associations with the NMSC cancer-prone phenotype. The results for all 759 SNPs are summarized in Supplementary Table 1, available at Carcinogenesis Online.

Table III.

DNA repair SNPs (n 5 10) with P-value < 0.01 for the allelic trend test comparing the ‘both NMSC plus other cancer’ to the ‘no cancer’ category, ordered by P-value

SNP Gene Pathway1 Type2 MAF3 P(HWE)4 Other cancer only NMSC only Both NMSC plus other cancer
OR 95% CI OR 95% CI OR 95% CI P-value
rs2888805 TDG BER Nonsyn 0.10 0.25 0.98 (0.85–1.12) 1.07 (0.88–1.30) 1.40 (1.16–1.70) 0.0006
rs4135150 TDG BER Intron 0.10 0.38 0.98 (0.86–1.13) 1.08 (0.89–1.31) 1.40 (1.15–1.70) 0.0007
rs1038144 ERCC8 NER Intron 0.09 0.23 1.07 (0.93–1.24) 0.96 (0.77–1.19) 1.38 (1.12–1.71) 0.0030
rs152451 PALB2 HR Nonsyn 0.09 0.64 0.95 (0.83–1.10) 1.01 (0.82–1.25) 1.35 (1.10–1.66) 0.0043
rs447529 PALB2 HR Intron 0.09 0.64 0.95 (0.82–1.10) 1.01 (0.82–1.24) 1.35 (1.10–1.65) 0.0046
rs11016857 MGMT DR Intron 0.43 0.43 0.96 (0.88–1.04) 1.04 (0.92–1.18) 0.83 (0.73–0.95) 0.0061
rs4150454 ERCC3 NER Intron 0.39 0.44 1.01 (0.92–1.11) 0.89 (0.77–1.03) 0.82 (0.70–0.95) 0.0088
rs11650024 DMC1 HR Intron 0.46 0.50 1.00 (0.92–1.09) 0.99 (0.88–1.12) 0.84 (0.73–0.96) 0.0088
rs2347443 CHEK2 Transduc Intron 0.16 0.92 1.09 (0.98–1.22) 1.01 (0.85–1.19) 1.25 (1.06–1.48) 0.0090
rs797690 MSH6 MMR 5′ UTR 0.26 0.43 1.10 (1.00–1.21) 1.13 (0.98–1.30) 1.21 (1.05–1.40) 0.0096
Open in a new tab

Allelic trend test odds ratios (ORs) and 95% CI are summarized for each SNP for comparisons of the following groups with the comparison group with no cancer: (i) other cancer only; (ii) NMSC only and (iii) both NMSC plus another cancer; all results adjusted for residual confounding by ancestry by including first three principal components.

1BER, base excision repair; NHEJ, non-homologous end-joining; NER, nucleotide excision repair; DR, direct reversal; HR, homologous recombination; Transduc, DNA damage signal transduction; MMR, mismatch repair.

2Nonsyn, non-synonymous coding; 5′ UTR, 5′ untranslated region.

3MAF, minor allele frequency.

4P(HWE), P-value for Hardy–Weinberg equilibrium.

The top ranked SNP (rs2888805) was a non-synonymous SNP in the coding region of the thymine DNA glycosylase gene (TDG) [per minor allele odds ratio for the ‘both’ category (ORBoth) 1.40, 95% confidence interval (CI) 1.16–1.70]. The associations were null for the other cancer only (OROther cancer only 0.98; 95% CI 0.85–1.12) and NMSC only (ORNMSC only 1.07; 95% CI 0.88–1.30) comparisons. This pattern of associations was mimicked by most of the other SNPs, with the risk association largely confined to the both NMSC plus other cancer category.

A measure of the internal consistency of this finding was that the second ranked SNP was an intronic TDG SNP (rs4135150) in high LD (r 2 5 0.99) with rs2888805 and with virtually identical results, indicating that both SNPs were equally informative about the strength of the allelic association because they are physically linked to each other. A total of six SNPs were successfully genotyped that tagged regions of TDG, rs2888805 and rs4135150 plus four more SNPs: rs4964435, rs6539116, rs4135113 and rs2583274. None of these other four were in high LD with one another (all r 2 < 0.10), and none were associated with the NMSC cancer-prone phenotype except rs4964435, which was nominally significantly inversely associated with cancer risk (per minor allele ORBoth 0.81; 95% CI 0.67–0.98, P-value 0.029). This pattern of associations suggests that the TDG haplotype blocks tagged by these other four SNPs (rs4964435, rs6539116, 4135113, rs2583274) are not likely to contribute to the functional changes in TDG that may be affecting the risk of NMSC plus another cancer. Rather, to the extent the observed associations with rs2888805 and rs4135150 are genuine, the results point toward a true functional variant residing within the haplotype block tagged by these two SNPs but unlikely to reside in the other haplotype blocks.

As with TDG, PALB2 also had two SNPs in high LD (r 2 5 0.99) and consequently nearly identical ORs. The two PALB2 SNPs were rs152451 (ORBoth 1.35; 95% CI 1.10–1.66, non-synonymous coding, rank no. 4) and rs447529 (ORBoth 1.35; 95% CI 1.10–1.65, intronic, rank no. 5). The remaining associated SNPs comprised one SNP each from ERCC8 (ORBoth 1.35, rank no. 3), ERCC3 (ORBoth 0.82, rank no. 7), DMC1 (ORBoth 0.84, rank no. 8), CHEK2 (ORBoth 1.25, rank no. 9) and MSH6 (ORBoth 1.21, rank no. 10). To account for the remote possibility that these associations between genetic markers and cancer could potentially be confounded by demographic or lifestyle characteristics, additional adjustments for age, gender, education, body mass index, and cigarette smoking status at baseline were tested. The adjustments made only negligible differences, and these slight differences tended to be in the direction of slightly strengthening the observed associations for ORBoth (Supplementary Table 2, available at Carcinogenesis Online). Due to the younger age distribution of the cancer-free comparison group, the associations were also assessed with age restricted to those ⪚35 years at baseline, which excluded 799 (35%) from the cancer-free group but only 8 (1%) from the NMSC plus other cancer group. The associations observed in this age-restricted analysis were very similar to the overall results (Supplementary Table 3, available at Carcinogenesis Online).

Table IV summarizes associations for the same 10 SNPs according to genotype (two degrees of freedom) and dominant and recessive modes of inheritance. Nine SNPs (all except rs797690) were more consistent with the dominant than recessive model based on the deviances in the respective logistic regression models. For the top-ranked SNPs in TDG (rs2888805: ORBoth 1.53; rs4135150: ORBoth 1.54), ERCC8 (rs1038144: ORBoth 1.44) and PALB2 (rs152451: ORBoth 1.40, rs447529: ORBoth 1.39), the ORs for the dominant model ranged from 1.4 to 1.5 with P-values ranging from 0.0001 to 0.0036. For the TDG SNPs, in the model with two degrees of freedom, the significance of the parameter for the one versus zero variant allele comparison was near (rs2888805: 8.2 × 10−5) or below (rs4135150: 6.1 × 10−5) the Bonferroni significance threshold.

Table IV.

Odds ratios (OR) and 95% Confidence Intervals (CI) for the association between SNPs with additive model P-value < 0.01 and risk of both NMSC plus other cancers compared with comparison group with no cancer, according to genotype and one or more versus zero minor alleles (dominant model) and two versus less than two minor alleles (recessive model)

SNP Gene No. of Minor alleles
(no. of both/no cancer)1		 Genotype
OR (95% CI)2		 Dominant model2,3
OR (95% CI) P-value		 Recessive Model2,4
OR (95% CI) P-value
rs2888805 TDG 0 (432/1920)
1 (136/379)
2 (8/32)		 1.0 (referent)
1.57 (1.26–1.96)
1.07 (0.49–2.35)		 1.0 (referent)
1.53 (1.23–1.90) 0.0001		 1.0 (referent)
0.98 (0.45–2.14) 0.96
rs4135150 TDG 0 (433/1886)
1 (137/411)
2 (7/31)		 1.0 (referent)
1.58 (1.27–1.98)
0.97 (0.42–2.21)		 1.0 (referent)
1.54 (1.24–1.91) 0.0001		 1.0 (referent)
0.88 (0.39–2.01) 0.76
rs1038144 ERCC8 0 (452/1924)
1 (121/354)
2 (4/15)		 1.0 (referent)
1.45 (1.15–1.83)
1.10 (0.36–3.33)		 1.0 (referent)
1.44 (1.15–1.81) 0.0017		 1.0 (referent)
1.03 (0.34–3.11) 0.96
rs152451 PALB2 0 (444/1893)
1 (126/384)
2 (6/19)		 1.0 (referent)
1.40 (1.12–1.75)
1.33 (0.53–3.35)		 1.0 (referent)
1.40 (1.12–1.74) 0.0033		 1.0 (referent)
1.25 (0.50–3.13) 0.64
rs447529 PALB2 0 (445/1892)
1 (126/384)
2 (6/19)		 1.0 (referent)
1.39 (1.11–1.75)
1.33 (0.53–3.34)		 1.0 (referent)
1.39 (1.11–1.74) 0.0036		 1.0 (referent)
1.24 (0.49–3.13) 0.64
rs11016857 MGMT 0 (216/714)
1 (268/1159)
2 (92/321)		 1.0 (referent)
0.76 (0.62–0.93)
0.72 (0.55–0.95)		 1.0 (referent)
0.75 (0.62–0.91) 0.0034 		1.0 (referent)
0.85 (0.66–1.08) 0.18
rs4150454 ERCC3 0 (229/631)
1 (276/1126)
2 (72/338)		 1.0 (referent)
0.82 (0.67–1.01)
0.67 (0.49–0.92)		 1.0 (referent)
0.80 (0.65–0.98) 0.0278		 1.0 (referent)
0.77 (0.58–1.02) 0.07
rs11650024 DMC1 0 (194/640)
1 (277/1184)
2 (102/471)		 1.0 (referent)
0.77 (0.63–0.95)
0.72 (0.55–0.94)		 1.0 (referent)
0.76 (0.62–0.92) 0.0054		 1.0 (referent)
0.84 (0.66–1.06) 0.15
rs2347443 CHEK2 0 (379/1646)
1 (179/588)
2 (18/62)		 1.0 (referent)
1.33 (1.09–1.62)
1.26 (0.74–2.16)		 1.0 (referent)
1.32 (1.09–1.60) 0.0051		 1.0 (referent)
1.16 (0.68–1.98) 0.58
rs797690 MSH6 0 (305/1312)
1 (222/855)
2 (50/129)		 1.0 (referent)
1.11 (0.92–1.35)
1.68 (1.18–2.38)		 1.0 (referent)
1.19 (0.99–1.43) 0.0654		 1.0 (referent)
1.61 (1.14–2.26) 0.0063
Open in a new tab

1Numbers do not sum to 5916 due to genotyping failures.

2Adjusted for residual confounding by ancestry by including first three principal components.

3One or two minor alleles versus zero minor alleles.

4Two minor alleles versus zero or one minor allele.

Discussion

The present study was carried out to test the hypothesis that germ-line variants in DNA repair genes were associated with the NMSC cancer-prone phenotype. SNPs were genotyped in known DNA repair genes in a population-based study to determine if germ-line polymorphisms contribute to joint susceptibility to both NMSC and other cancers. The results revealed several variant alleles that may contribute to a cancer-prone phenotype characterized by susceptibility to both NMSC and other malignancies.

An interesting observation was that carrying a non-synonymous coding SNP (rs2888805) in the base excision repair gene TDG was shown to be associated with 50% increased risk of developing NMSC plus another cancer. This polymorphic SNP leads to a methionine residue at position 367, or rarely leucine, rather than the more common valine (16). This residue does not fall within the known active sites for the protein or targets of acetylation, and its relevance to overall protein function is currently unknown. In the absence of direct experimental data, bioinformatics databases were queried to assess for the putative functionality and, consistent with the information above, TDG SNP rs2888805 was predicted to be benign in the PolyPhen-2, Condel and SIFT bioinformatics databases. For the SNP in high LD with rs2888805, rs4135150, the most current alignment has this SNP residing in GLT8D2, not TDG. rs4135150 is in intron 8 (of 8) in GLT8D2, but was selected as a tagging SNP for the 3′ UTR of TDG. rs2888805 is in exon 10 (of 10) in TDG. The TDG and GLT8D2 genes run in opposite directions, so the two SNPs are only approximately 3000bp apart. The most current alignment also has three other TDG tagging SNPs (rs4964435, rs6539116, rs2583274) physically residing in GLT8D2.

TDG was the first known mismatch-specific DNA repair gene; TDG repairs thymine mispaired with guanine, a mispairing that is both highly mutagenic and common due to the propensity of 5-methylcytosine to deaminate to thymine to create G–T mispairings that often lead to mutations (17). A role for TDG in preserving global genomic stability via maintenance of the epigenome and gene expression is supported by strong evidence that it plays an essential role in regulating CpG island methylation status and in DNA demethylation (18–20), to the extent that a knock-out model resulted in an embryonic lethal phenotype (18). Further, TDG’s involvement in gene expression/ transcription includes activation of CREB-binding protein, p300, SRC-1, RARα and retinoid × receptor α, ERα, and p53 family proteins (21). These pivotal functions support the notion that TDG variants could play a cross-cutting role in contributing to joint susceptibility to both NMSC plus other cancers.

Only a modicum of prior evidence exists concerning any TDG variants and human health outcomes, let alone for the specific TDG SNPs included in the present study. The outcomes that have been studied have been either cancer-related intermediate endpoints or cancer endpoints. TDG also excises uracil when it is mispaired with guanine, but in a study of eight TDG SNPs, none of which overlapped with the SNPs in the present study, none of the SNPS were significantly associated with DNA uracil concentrations (22). The most commonly studied TDG SNP has been rs4135113 (Gly199Ser), which was not strongly associated with cancer risk in the present study or in previous studies of lung (23) and rectal (24) cancer but was associated with significantly increased likelihood of micronuclei, a measure of genotoxicity, among vinyl-chloride exposed workers in China (25). In a DNA repair pathway study, no significant associations were observed for the eleven TDG tagging SNPs studied in relation to meningioma risk (26). Minor alleles of SNPs in TDG intron 1 (rs2374327) and intron 2 (rs167715) were significantly inversely associated with one-year transplant-related mortality after hematopoietic stem cell transplant (27). Two previous reports specifically of rs2888805 were identified. In one, in 94 cases with a family history of colorectal cancer that excluded those with known colorectal cancer susceptibility alleles, rs2888805 was heterozygous in t10 patients; there was no control group (16). In the other study, rs2888805 was not significantly associated with lung cancer risk in a study of 96 cases and 94 controls (23). Thus, the limited available evidence on TDG polymorphisms in relation to cancer or biomarkers of cancer is equivocal, neither clearly supporting nor refuting possible associations with cancer risk in different contexts. The present study focused on a study population that is hypothesized to be enriched for genetic susceptibility to cancer, a circumstance that may enhance the likelihood of detecting associations that may otherwise go undetected in studies of a single cancer.

In addition to the two TDG SNPs, eight other SNPs were associated with the NMSC cancer-prone phenotype with additive model P-values < 0.01, raising the index of suspicion that these SNPs may be markers for variants that also contribute to the phenotype. These eight SNPs were located in six pathways and seven genes: two nucleotide excision repair (ERCC8, ERCC3), two homologous recombinational repair (PALB2, DMC1) and one each from direct reversal repair (MGMT), DNA damage signal transduction (CHEK2) and mismatch repair (MSH6).

For only three of these eight SNPs were previously published reports identified, the two PALB2 SNPs (rs152451, rs447529) and the CHEK2 SNP (rs2347743). PALB2 (for Partner and Localizer of BRCA2) has been extensively studied in relation to hereditary breast cancer but is also associated with Fanconi anemia and pancreatic cancer (28). rs152451 is a non-synonymous coding SNP that may alter the PALB2 protein, but queries of the PolyPhen-2, Condel and SWIFT bioinformatics databases indicated it was likely to be benign. Studies of familial or bilateral breast cancer documented that the prevalence of rs152451 heterozygotes ranged from 9% to 22%, consistently ranking among the highest of the genetic markers studied (29–32). The other PALB2 SNP, rs447529, is an intronic SNP that was in high LD (r 2 5 0.99) with rs152451 in our study population; in a study in China, rs447529 was significantly associated with breast cancer risk (33). Specific CHEK2 mutations are known to increase susceptibility to breast cancer (34), and CHEK2 variants have been studied in relation to Li Fraumeni syndrome and a diverse suite of other malignancies with equivocal results (35). In a case–control study, the intronic SNP rs2347443 was not significantly associated with breast cancer risk (36).

No previous reports of the other five SNPs were ascertained, but a brief synopsis of the relevance of the genes is provided below. The two genes from the NER pathway were ERCC8 and ERCC3. ERCC8, also known as Cockayne syndrome A (CSA), is not only involved in NER but also in protecting DNA from damage via oxidative stress (37). Rare, high-penetrant mutations in ERCC8 are linked with Cockayne syndrome. ERCC3 encodes for XPB, which comprises an important element of NER as well as the general transcription factor IIH (TFIIH) complex, so it is critical not only to NER but also to gene transcription (38). ERCC3 mutations have been implicated in XP, Cockayne syndrome, and trichothiodystrophy (39). The DMC1 gene contributes to meiosis-specific homologous recombination repair (40), with the only previous report of DMC1 variants in relation to cancer risk finding an association with cervical cancer (41). MSH6 is a mismatch repair gene, with mutations implicated in Lynch syndrome (42) and endometrial cancer (43). O6-methylguanine-DNA methyltransferase (MGMT) directly repairs alkyl adducts that arise in the O6 position of guanine, preventing G:C to A:T transition mutations (44). Germ-line MGMT polymorphisms could potentially affect carcinogenesis by directly diminishing MGMT-mediated DNA repair, either by reducing methyltransferase activity via a functional coding change or by increasing MGMT’s susceptibility to epigenetic silencing of its transcription.

In summary, this list includes an intriguing set of candidate SNPs that warrant consideration as potential genetic markers of susceptibility to the NMSC cancer-prone phenotype. By way of caution, the very nature of the candidate gene pathway approach ensured that these genes would have strong a priori biologic plausibility based on their key roles in carcinogenic pathways.

Associations observed in the NMSC plus other cancer group were not sustained in the NMSC only and other cancer only groups. The confinement of the risk associations to the NMSC plus other cancer group may reflect a greater likelihood of genetically driven, rather than environmentally driven, cancer susceptibility. A high-risk subgroup is expected to be enriched for genetic susceptibility markers associated with a predisposition to cancer. In contrast, environmental exposures likely play a more prominent role in driving carcinogenesis in the NMSC only and other cancer only groups, exposures that are more likely cancer-type specific (e.g. sun exposure for NMSC and smoking for lung cancer). Hence, in assessing risk for multiple cancers that lack common environmental risk factors (e.g. NMSC and lung cancer), genetic risk factors would be expected to predominate. This is analogous to other situations for which cancers are also more strongly genetically driven than the average-risk population, such as phenotypic markers defined by early age of cancer onset or a positive family history of cancer. Studying the relationship between NMSC and risk of other malignancies thus provides a fertile research model for identifying markers of potential susceptibility to human cancer.

The present study had multiple strengths, including addressing a novel question embedded within a well-established community-based cohort study. No previous study has tested a hypothesis to understand why a personal history of NMSC is associated with enhanced susceptibility to other cancers. The hypothesis that germ-line polymorphisms in DNA repair genes may contribute to this general cancer-prone phenotype was rigorously tested via thorough coverage of allelic variants in DNA repair pathway genes. Using the co-occurrence of NMSC and another cancer as a sentinel for a cancer-prone phenotype is novel and may enhance detection of common allelic variants that might otherwise escape notice as markers of cancer susceptibility. This is particularly true when one considers that in contrast to the complications usually encountered when studying multiple primary cancers, NMSC has the advantages that it is rarely fatal and therefore less susceptible to survival biases and, because most tumors are locally excised, its treatment is not associated with risk of other cancers.

Nevertheless, caution is warranted in drawing inferences. Even though a cohort of ~30 000 followed-up for almost two decades enabled inclusion of a total of 3620 cancer cases, the primary study inferences centered on the both NMSC plus another cancer category, which was limited to 577 cancer cases. A resulting weakness was limited statistical power, a likely reason that no associations survived conservative significance thresholds imposed by corrections for multiple comparison testing. The sample size constraints of the both NMSC plus other cancer category highlight challenges of this line of inquiry and emphasize the need for replication of our findings. These same sample size constraints precluded meaningful subgroup analyses, such as separate analyses for BCC and SCC, but the totality of the current evidence suggests that the NMSC cancer-prone phenotype is consistently observed for both histologic types.

A limitation of the genetic association study approach is that we have identified statistical associations, but confirmation of the specific SNPs that might be involved would require additional experiments not included in the present study. In this discovery phase research, the results identified the genetic regions of DNA repair genes where variants are most likely to be associated with the risk of NMSC plus another cancer. If the observed associations are genuine, the association of these SNPs with the NMSC cancer-prone phenotype allows us to infer that allelic variations in TDG and the other genes identified may affect cancer development but it does not allow us infer causality for the SNPs themselves, because they are simply markers for gene regions that contain many other sequence variations. Thus, a limitation of this study is that further research will be needed to identify and characterize the specific functional variants within these gene regions and the protein change in function that might modify cellular DNA repair capacity. In the absence of direct experimental data, bioinformatics databases were queried to assess for the putative functionality of the two non-synonymous SNPs. For both TDG SNP rs2888805 and PALB2 SNP rs152451, the scores for the PolyPhen-2, Condel and SIFT bioinformatics databases were consistent in indicating that these SNPs were expected to be benign.

In conclusion, the findings have generated new leads suggesting that common variants in DNA repair genes may affect the increased overall cancer risk experienced by individuals with a personal history of NMSC. The strongest association was for a non-synonymous coding SNP in TDG, suggesting that base excision repair, possibly via regulation of the epigenome, may be important in the NMSC cancer-prone phenotype.

Supplementary material

Supplementary Tables 1– 3 can be found at http://carcin.oxfordjournals.org/

Funding

National Cancer Institute (NCI) (R01 CA105069, HHSN26120080001E); Intramural Research Program of the NCI, Center for Cancer Research.

Acknowledgements

Cancer incidence data were provided by the Maryland Cancer Registry, Center for Cancer Surveillance and Control, Department of Health and Mental Hygiene, which is funded by the State of Maryland, the Maryland Cigarette Restitution Fund and the National Program of Cancer Registries of the CDC to make the cancer registry data available. This publication does not necessarily reflect the views or policies of the NCI, NIMH, NIH, US DHHS, the US government or the Maryland Cancer Registry, nor does mention of trade names, commercial products or organizations imply endorsement by the US government.

Conflict of Interest Statement: None declared.

Glossary

Abbreviations

BCC

basal cell carcinoma

LD

linkage disequilibrium

MAF

minor allele frequency

NER

nucleotide excision repair

NMSC

non-melanoma skin cancer

SNP

single nucleotide polymorphism

SCC

squamous cell carcinoma

TDG

thymine

DNA

glycosylase gene

XP

xeroderma pigmentosum

References

  • 1. Rogers H.W, et al. (2010). Incidence estimate of nonmelanoma skin cancer in the United States, 2006 Arch. Dermatol. 146 283–287 [DOI] [PubMed] [Google Scholar]
  • 2. Wheless L, et al. (2010). Nonmelanoma skin cancer and the risk of second primary cancers: a systematic review Cancer Epidemiol. Biomarker Prev. 19 1686–1695 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3. Cleaver J.E. (2000). Common pathways for ultraviolet skin carcinogenesis in the repair and replication defective groups of xeroderma pigmentosum. J. Dermatol. Sci. 23 1–11 [DOI] [PubMed] [Google Scholar]
  • 4. Cleaver J.E. (2005). Cancer in xeroderma pigmentosum and related disorders of DNA repair. Nat. Rev Cancer 5 564–573 [DOI] [PubMed] [Google Scholar]
  • 5. German J. (1997). Bloom’s syndrome. XX. The first 100 cancers. Cancer Genet. Cytogenet. 93 100–106 [DOI] [PubMed] [Google Scholar]
  • 6. Thomas E.R.A, et al. (2008). Surveillance and treatment of malignancy in Bloom’s Syndrome. Clin. Oncol. 20 375–379 [DOI] [PubMed] [Google Scholar]
  • 7. Vineis P, et al. (2009). A field synopsis on low-penetrance variants in DNA repair genes and cancer susceptibility J. Natl. Cancer Inst. 101 24–36 [DOI] [PubMed] [Google Scholar]
  • 8. Halliday G.M. (2005). Inflammation, gene mutation and photoimmunosuppression in response to UVR-induced oxidative damage contributes to photocarcinogenesis. Mutat. Res. 571 107–120 [DOI] [PubMed] [Google Scholar]
  • 9. Cadet J, et al. (2005). Ultraviolet radiation-mediated damage to cellular DNA. Mutat. Res. 571 3–17 [DOI] [PubMed] [Google Scholar]
  • 10. Chen J, et al. (2008). Nonmelanoma skin cancer and risk for subsequent malignancy. J. Natl. Cancer Inst. 100 1215–1222 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Gallicchio L, et al. Single nucleotide polymorphisms in obesity-related genes and all-cause and cause-specific mortality: a prospective cohort study. BMC Med. Genet. (2009);10:103. doi: 10.1186/1471-2350-10-103. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12. Jorgensen T.J, et al. (2009). Hypothesis-driven candidate gene association studies: practical design and analytical considerations. Am. J. Epidemiol. 170 986–993 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13. Wood R.D, et al. (2005). Human DNA repair genes 2005. Mutat. Res. 577 275–283 [DOI] [PubMed] [Google Scholar]
  • 14. de Bakker P.I, et al. (2005). Efficiency and power in genetic association studies. Nat. Genet. 37 12171223 [DOI] [PubMed] [Google Scholar]
  • 15. Wigginton J.E, et al. (2005). A note on exact tests of Hardy-Weinberg equilibrium. Am. J. Hum. Genet. 76 887893 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Broderick P, et al. Evaluation of NTHL1, NEIL1, MPG, TDG, UNG, and SMUG1 genes in familial colorectal cancer predisposition. BMC Cancer. (2006);6:243. doi: 10.1186/1471-2407-6-243. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17. Cortazar D, et al. (2007). The enigmatic thymine DNA glycosylase. DNA Repair 6 489504 [DOI] [PubMed] [Google Scholar]
  • 18. Cortázar D, et al. (2011). Embryonic lethal phenotype reveals a function of TDG in maintaining epigenetic stability. Nature 470 419423 [DOI] [PubMed] [Google Scholar]
  • 19. Cortellino S, et al. (2011). Thymine DNA glycosylase is essential for active DNA demethylation by linked deamination-base excision repair. Cell 146 6779 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20. He Y.F, et al. (2011). Tet-mediated formation of 5-carboxylcytosine and its excision by TDG in mammalian DNA. Science 333 13031307 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21. Kim E.-J, et al. (2008). Thymine-DNA glycosylase interacts with and functions as a coactivator of p53 family proteins. Biochem. Biophys. Res. Commun. 377 838842 [DOI] [PubMed] [Google Scholar]
  • 22. Chanson A, et al. (2009). Polymorphisms in uracil-processing genes, but not one-carbon nutrients, are associated with altered DNA uracil concentrations in an urban Puerto Rican population. Am. J. Clin. Nutr. 89 19271936 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23. Krzesniak M, et al. (2004). Polymorphisms in TDG and MGMT genes—epidemiological and functional study in lung cancer patients from Poland. Ann. Hum. Genet. 68 300312 [DOI] [PubMed] [Google Scholar]
  • 24. Curtin K, et al. (2011). Candidate pathway polymorphisms in one-carbon metabolism and risk of rectal tumor mutations. Int. J. Mol. Epidemiol Genet. 2 18 [PMC free article] [PubMed] [Google Scholar]
  • 25. Wen-Bin M, et al. (2009). Micronucleus occurrence related to base excision repair gene polymorphisms in Chinese workers occupationally exposed to vinyl chloride monomer. J. Occup. Environ Med. 51 578585 [DOI] [PubMed] [Google Scholar]
  • 26. Bethke L, et al. (2008). Comprehensive analysis of the role of DNA repair gene polymorphisms on risk of glioma. Hum. Mol. Genet. 17 800–805 [DOI] [PubMed] [Google Scholar]
  • 27. Thyagarajan B, et al. (2010). Association between genetic variants in the base excision repair pathway and outcomes after hematopoietic cell transplantations. Biol. Blood Marrow Transplant. 16 1084–1089 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28. Tischkowitz M, et al. (2010). PALB2/FANCN: recombining cancer and Fanconi anemia. Cancer Res. 70 7353–7359 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29. Bogdanova N, et al. (2011). PALB2 mutations in German and Russian patients with bilateral breast cancer. Breast Cancer Res. Treat. 126 545–550 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30. Papi L, et al. (2010). A PALB2 germline mutation associated with hereditary breast cancer in Italy. Fam. Cancer 9 181–185 [DOI] [PubMed] [Google Scholar]
  • 31. Garcia M.J, et al. (2009). Analysis of FANCB and FANCN/PALB2 Fanconi anemia genes in BRCA1/2-negative Spanish breast cancer families Breast Cancer Res. Treat. 113 545–551 [DOI] [PubMed] [Google Scholar]
  • 32. Erkko H, et al. (2007). A recurrent mutation in PALB2 in Finnish cancer families. Nature 446 316–319 [DOI] [PubMed] [Google Scholar]
  • 33. Cao A-Y, et al. (2010). Five common single nucleotide polymorphisms in the PALB2 gene and susceptibility to breast cancer in eastern Chinese population Breast Cancer Res. Treat. 123 133–138 [DOI] [PubMed] [Google Scholar]
  • 34. Narod S.A. (2010). Testing for CHEK2 in the cancer genetics clinic: ready for prime time? Clin. Genet. 78, 1––7 [DOI] [PubMed] [Google Scholar]
  • 35. Nevanlinna H, et al. (2006). The CHEK2 gene and inherited breast cancer susceptibility. Oncogene 25 5912–5919 [DOI] [PubMed] [Google Scholar]
  • 36.Baynes C, et al. Common variants in the ATM1, BRCA1, BRCA2, CHEK2, and TP53 cancer susceptibility genes are unlikely to increase breast cancer risk. Breast Cancer Res. (2007);9:R27. doi: 10.1186/bcr1669. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37. Laugel V, et al. (2010). Mutation update for the CSB/ERCC6 gene and CSA/ERCC8 genes involved in Cockaynesyndrome. Hum. Mutat. 31 113–126 [DOI] [PubMed] [Google Scholar]
  • 38. Oh K.-S, et al. (2006). Phenotypic heterogeneity in the XPB DNA helicase gene(ERCC3): Xeroderma pigmentosum without and with Cockaynesyndrome. Hum. Mutat. 27 1092–1103 [Google Scholar]
  • 39. Rybanska I, et al. (2010). Newly identified CHO ERCC3/XPB muations and phenotype characterization Mutagenesis 25 179–185 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40. Kagawa W, et al. (2010). From meiosis to postmeiotic events: uncovering the molecular roles of the meiosis-specific recombinase Dmc1. FEBS J 277 590–598 [DOI] [PubMed] [Google Scholar]
  • 41.Wang S.S, et al. Common genetic variants and risk for HPV persistence and progression to cervical cancer. PLoS One. (2010);5:e8667. doi: 10.1371/journal.pone.0008667. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42. Pineda M, et al. (2010). Detection of genetic alterations in heredity colorectal cancer screening. Mutat. Res. 693 19–31 [DOI] [PubMed] [Google Scholar]
  • 43. Bansal N, et al. (2009). The molecular biology of endometrial cancers and the implications for pathogenesis, classification, and targeted therapies. Cancer Control 16 8–13 [DOI] [PubMed] [Google Scholar]
  • 44. Tubbs J.L, et al. (2007). DNA binding, nucleotide flipping, and the helix-turn-helix motif in base repair by O6-alkylguanine-DNA- alkyltransferase and its implications for cancer chemotherapy. DNA Repair 6 1100–1115 [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from Carcinogenesis are provided here courtesy of Oxford University Press

RESOURCES