Abstract
Base excision repair and nucleotide excision repair are vital responses to multiple types of DNA damage, including damage from tobacco exposure. Single-nucleotide polymorphisms (SNP) in these pathways may affect DNA repair capacity and therefore influence risk for cancer development. We performed a clinic-based, case-control study comprising 481 consecutive patients with confirmed pancreatic adenocarcinoma and 625 healthy controls. Allele and genotype frequencies for 16 SNPs in DNA repair genes ERCC1, XPD/ERCC2, XPC, XPF/ERCC4, OGG1, and XRCC1 were compared after adjusting for age, sex, and smoking history. Subgroup analysis by sex and smoking history was performed. Carriers of one or two XPF/ERCC4 minor alleles at R415Q had decreased risk of pancreatic adenocarcinoma compared with those who had two major alleles [odds ratio (OR), 0.59; 95% confidence interval (95% CI), 0.40–0.85]. Heavy smokers (>40 pack-years) had increased risk for cancer if they were carriers of at least one minor allele for XPD/ERCC2 at D312N (OR, 2.78; 95% CI, 1.28–6.04) or D711D (OR, 2.19; 95% CI, 1.01–4.73). No other significant differences in risk were identified. Minor alleles in DNA repair genes XPF/ERCC4 and XPD/ERCC2 were associated with altered risk for pancreatic cancer. [Cancer Res 2008;68(12):4928–35]
Introduction
DNA repair is a key human cellular response to DNA-damaging stimuli. Because mutations in DNA repair genes such as BRCA1 and BRCA2 increase risk for pancreatic adenocarcinoma (1, 2), genetic predisposition to pancreatic cancer has become a subject of intense research. However, high-penetrance tumor suppressor genes explain only a small proportion (0.5–1%) of cases (3).
The base excision repair (BER) pathway detects and repairs damage from stimuli such as reactive oxygen species, alkylating agents, and ionizing radiation (4, 5). 8-Hydroxyguanine DNA glycosylase (OGG1) initiates the process by cleaving the damaged base, leaving an unmatched base on the opposite strand. Apurinic/apyrimidinic endonuclease (APE1) then cleaves the associated sugar-phosphate chain. At this point, two pathways are possible: a single-nucleotide repair pathway (major pathway) or a long-patch repair pathway of a few base pairs (minor pathway). In the major pathway, polymerase β interacts with X-ray repair cross-complementing group 1 (XRCC1) in heterodimers with DNA ligase III to complete the repair process. In the minor pathway, a flap of several bases is constructed by polymerase δ/ε; the extraneous flap is removed by flap endonuclease I, and DNA ligase I completes the repair by using proliferating cell nuclear antigen as a scaffold (5–7).
Of the single-nucleotide polymorphisms (SNP) in the BER pathway genes, those in OGG1 and XRCC1 are among the best studied. Minor alleles at the OGG1 polymorphism S326C are associated with increased risk for lung cancer overall (8), squamous lung cancer (9), and prostate cancer (10). However, many reports have shown no association of this polymorphism with other cancer types (11).
In 2002, Duell and colleagues (12) reported an analysis of polymorphism R399Q of the BER gene XRCC1; they performed a population-based, case-control study of 309 pancreatic cancer cases and 964 healthy controls from the San Francisco Bay Area. A comparison of the overall frequency of the polymorphism (cases versus controls) showed no statistically significant differences. However, minor allele genotypes (R/Q or Q/Q) were overrepresented among heavy smokers (≥41 pack-years); for men, the odds ratio (OR) was 2.4 [95% confidence interval (95% CI), 1.1–5.0], and for women, the OR was 7.0 (95% CI, 2.4–20.7). The authors suggested that smoking-induced DNA damage might be repaired less efficiently in minor allele carriers, thereby increasing risk for pancreatic cancer.
A clinic-based case-control study in Houston, Texas, examined three polymorphisms in DNA repair genes [XRCC1 (R194W, R399Q) and APE1 (D148E)] in 384 pancreatic cancer cases and 357 controls (13). No significant increase in risk was identified for any individual minor allele. However, risk was identified for carriers of both XRCC1 R194W and the homozygous major allele APE1 D148D (OR, 4.98; 95% CI, 1.61-15.4). Interestingly, the XRCC1 R399Q minor allele was associated with increased risk (OR, 2.06; 95% CI, 1.01-4.18) for light smokers (<1 pack per day) but not with heavier smokers.
The nucleotide excision repair (NER) pathway identifies the site of damage, unwinds the DNA duplex around the site, cuts the DNA upstream and downstream of the damaged area, and repairs the gap (14, 15). NER most notably is involved in detection and repair of UV light–induced lesions and damage from tobacco-related carcinogens and other chemicals (e.g., pyrimidine dimers, bulky adducts; refs. 16–19). High-penetrance defects in NER genes such as XPA, ERCC3, XPC, XPD/ERCC2, XPE, XPF/ERCC4, and ERCC5, have been implicated in xeroderma pigmentosum (20, 21), resulting in up to a 1,000-fold increased risk for cutaneous malignancy after damage from UV light.
The effect of NER gene polymorphisms on pancreatic cancer risk is not quite as well studied as that of BER polymorphisms. One Chinese study (101 cases, 337 controls) observed a protective effect from an XPC intronic poly-AT polymorphism (OR, 0.30; 95% CI, 0.10–0.76; ref. 22). A clinic-based study (from Houston, Texas) of the XPD/ERCC2 polymorphisms D312N and Q751K did not detect an association between these polymorphisms and pancreatic cancer overall, but reported a protective effect of carrying at least one minor allele at D312N for ever-smokers (OR, 0.46; 95% CI, 0.24–0.83; ref. 23). No further characterization by smoking status was reported, and pack-years were not included in the analysis.
Polymorphisms in the NER pathway also have been associated with increased risk for melanoma (14), head and neck cancer (24), lung cancer (25), and basal cell skin cancer (26). Homozygous minor alleles for XPF/ERCC4 R415Q were associated with increased risk for breast cancer in a clinic-based case-control study (253 cases, 268 controls; ref. 27). To our knowledge, studies of this polymorphism in pancreatic cancer have not been reported (5).
To further characterize genetic risk of pancreatic cancer, we assessed the effect of selected BER and NER polymorphisms on pancreatic cancer risk. This study explored the gene-environment interaction of 16 SNPs in six DNA repair genes on pancreatic cancer risk.
Materials and Methods
Recruitment of Subjects
Cases. From October 1, 2000, through December 31, 2003, patients with pancreatic adenocarcinoma were recruited to a prospective registry (ultrarapid recruitment) during their visit to Mayo Clinic. Patients were approached by a study coordinator or contacted by mail. Written informed consent and provision of biospecimens were obtained from each subject for participation in this study, and this study was approved by the Mayo Clinic Institutional Review Board. We recruited 481 patients with histologically proved pancreatic adenocarcinoma of all stages (participation rate, 69.4%). Upon enrollment, a risk factor questionnaire and a family history questionnaire were completed by the patient. Peripheral blood was collected for DNA analysis.
Controls. We identified 625 individuals through a colon cancer screening study that was performed from June 1, 2000, through May 31, 2004. These subjects (noncancer controls) were recruited to our study if they had negative results from a routine colonoscopy (screening for colon cancer). Although this control group was a convenience sample and therefore not selected to match the pancreas cancer cases, we did not include subjects who reported a history of any type of cancer. Subjects completed a risk factor questionnaire with the same questions regarding smoking history as the pancreatic cancer cases. Self-reported height, weight, and diabetes mellitus status were not available for these individuals. Peripheral blood was collected for DNA analysis.
Smoking History
Study participants provided information about age at initiation and cessation of smoking and the number of packs smoked per day. If no smoking data were available from the self-completed questionnaire, smoking information was extracted from the patient's medical record (data were extracted for 24% of controls and 23% of cases). Smoking data were available for 99.7% of study participants.
Total number of pack-years was calculated by multiplying the typical number of packs smoked daily with the number of years smoked. Pack-years was used as a measure of smoking exposure. Subjects were categorized as “never smokers” and “ever smokers” (>100 cigarettes in their lifetime). Ever smokers were further stratified by the number of pack-years (≤20 pack-years, >20-40 pack-years, and >40 pack-years).
SNP Genotyping
We selected nonsynonymous BER and NER SNPs and also selected those previously reported to be associated with cancer risk. DNA was extracted from fresh peripheral blood using the Gentra AutoPure LS Purgene salting-out method (Gentra). The XRCC1 polymorphisms R194W (reference SNP accession number, rs1799782), R280H (rs25489), and R399Q (rs25487), OGG1 polymorphisms R229Q (rs1805373) and S326C (rs1052133), ERCC1 polymorphisms N118N (rs11615) and base 123 of intron 1 (I1 + 123; rs2298881), XPC polymorphisms L16V (rs1870134), A499V (rs2228000), and K939Q (rs2228001), XPD/ERCC2 polymorphisms R156R (rs238406), D312N (rs1799793), D711D (rs1052555), and Q751K (rs13181), and XPF/ERCC4 polymorphisms R415Q (rs1800067) and S662Q (rs2020955) were identified. Synonymous and nonsynonymous coding SNPs were chosen for analysis because of prior reports of their associations with pancreatic cancer and other tumor types, and we also used dbSNP to identify coding SNPs with reported minor allele frequency (MAF) of ≥0.05 (28).
Assays were performed in the Mayo Clinic Genotyping Shared Resource Core Facility using SNPstream (Beckman Coulter, Inc.; ref. 29) or Pyrosequencing (Biotage; ref. 30) methods (Pyrosequencing was used when SNPstream failed). For the SNPstream protocol, 2 ng of DNA were used per panel. For the Pyrosequencing protocol, template PCR primers were designed, and reaction conditions were optimized for each SNP using 20 ng of DNA. Pyrosequencing primers were designed using the Biotage Web site (31), and assays were performed in accordance with the recommended protocol. Case and control samples were randomized on 96-well plates, and genotyping analysis was blinded to subject status (case versus control). Genotyping quality was monitored by inclusion of two control DNA samples and a template-free control, each in duplicate on every plate.
Statistical Analysis
To evaluate the association between pancreatic cancer and the demographic and genetic markers of interest, we compared cases and controls and performed subgroup analysis on the basis of sex, age, and smoking exposure. Demographic variables are presented as mean ± SD for continuous variables and as percentages for categorical variables.
Before analysis of disease marker associations was performed, we used χ2 tests to determine whether the genotype distributions for each SNP showed Hardy-Weinberg equilibrium under Mendelian biallelic expectations. Univariate associations of allele and genotype (treating each chromosome as a unit and each person as a unit) with disease were evaluated using the contingency table methods (SAS software, version 8.2). Allele associations were assessed using the Pearson χ2 or Fisher exact test (when sample sizes were small), and genotype associations were assessed using the Cochran-Armitage trend test.
Haplotype-disease association was evaluated for each gene using Haplo.score (32), which accounted for ambiguous linkage phase. This method uses an expectation-maximization algorithm to infer haplotypes and accounts for ambiguity in haplotype assignment when comparing cases and controls. It further allows adjustment for nongenetic covariates, which are often critical when analyzing genetically complex phenotypes. The expectation-maximization method also provides global tests for association, as well as haplotype-specific tests, which give a meaningful advantage when attempting to understand the roles of different haplotypes. Haplotype ORs and 95% CIs were calculated using Haplo.glm (33). Haplotype analyses were performed using HaploStats version 1.1.1. For each polymorphism, we defined major and minor alleles as the most and least common alleles in controls, respectively. Multivariate analysis was adjusted for age, sex, smoking status, and pack-years and compared carriers of 0 versus one or two copies of minor alleles.
We assessed combined effects of the SNPs by comparing the total number of minor alleles between cases and controls. ORs and 95% CIs were calculated using an unconditional multivariate logistic regression model that adjusted for age, sex, and smoking status. Because type I and type II errors were of concern with these analyses, we made no adjustments for multiple comparisons (34, 35). However, when interpreting our results, all tests were two-sided, and we considered a P value between 0.05 and 0.01 as “suggestive,” a P value between 0.01 and 0.001 as “statistically significant,” and a P value of <0.001 as “highly significant.” Lohmueller and colleagues (36) observed that when a first genetic association study had a P value of <0.001 or when two replication studies had P values of <0.01, there was a high likelihood of replicating the original findings with meta-analysis. ORs and 95% CIs were used to describe associations in subgroups. Given the reduced power of these subanalyses, all associations were considered hypothesis-generating only.
Results
Characteristics of cases and controls are shown in Table 1. Cases were older and more likely to be men, to live outside the Upper Midwest, and to smoke. Because nearly all cases and controls were White, we decided to exclude the small number of minority patients from our analyses to limit any possible confounding effects. Genotype frequencies of the 16 polymorphisms were determined for cases and controls (Table 2). We determined that all genotypes were in Hardy-Weinberg equilibrium, except for XPF/ERCC4 S662Q, which had only very rare minor alleles. After adjusting for age, sex, and smoking status (Table 3), only the XPF/ERCC4 R415Q genotype differed in frequency when comparing cases with controls (OR, 0.59; 95% CI, 0.40–0.85). The minor allele of this polymorphism was associated with a decreased risk for pancreatic cancer, and the effect was seen equally among never smokers (OR, 0.60; 95% CI, 0.35–1.05) and ever smokers (OR, 0.61; 95% CI, 0.38–0.96). However, the protective effect seemed to dissipate among heavier smokers (Table 3). Interestingly, when heavy smokers carrying the major allele were analyzed using nonsmokers carrying the minor allele as the reference group, a significant relationship was found (OR, 2.99; 95% CI, 1.50–5.96). However, when we assessed risk for heavy smokers versus nonsmokers, regardless of genotype, the OR was similar (OR, 3.12; 95% CI, 0.96–10.16); thus, the difference likely can be explained by smoking alone, although an interaction is possible.
Characteristics of study subjects
Variable . | Cases (n = 481) . | Controls (n = 625) . | P . | |||
---|---|---|---|---|---|---|
Age, mean ± SD (y) | 65.7 ± 10.5 | 59.7 ± 12.1 | <0.001 | |||
Female, no. patients (%) | 194 (40) | 297 (48) | 0.02 | |||
Race, no. patients (%) | <0.001 | |||||
White | 473 (98) | 615 (98) | ||||
Black* | 5 (1.0) | 2 (0.3) | ||||
Hispanic* | 2 (0.4) | 1 (0.2) | ||||
Asian/Pacific Islander* | 0 (0) | 3 (0.5) | ||||
Other* | 1 (0.2) | 2 (0.3) | ||||
Unknown* | 0 (0) | 2 (0.3) | ||||
Geographic location, no. patients (%) | ||||||
Minnesota | 119 (25) | 355 (57) | <0.0001 | |||
Minnesota, Wisconsin, Iowa | 241 (50) | 452 (72) | <0.0001 | |||
Upper Midwest† | 396 (82) | 544 (87) | 0.03 | |||
Smoking history, no. patients (%) | 0.01 | |||||
Never smoker | 196 (41) | 303 (49) | ||||
Ever smoker | 284 (59) | 320 (51) | ||||
Unknown | 1 (0.2) | 2 (0.3) | ||||
Pancreatic cancer stage, no. patients (%) | ||||||
I | 32 (7) | — | ||||
II | 164 (34) | — | ||||
III | 90 (19) | — | ||||
IV | 189 (39) | — | ||||
Pancreatic cancer surgery, no. patients (%) | 119 (25) | — |
Variable . | Cases (n = 481) . | Controls (n = 625) . | P . | |||
---|---|---|---|---|---|---|
Age, mean ± SD (y) | 65.7 ± 10.5 | 59.7 ± 12.1 | <0.001 | |||
Female, no. patients (%) | 194 (40) | 297 (48) | 0.02 | |||
Race, no. patients (%) | <0.001 | |||||
White | 473 (98) | 615 (98) | ||||
Black* | 5 (1.0) | 2 (0.3) | ||||
Hispanic* | 2 (0.4) | 1 (0.2) | ||||
Asian/Pacific Islander* | 0 (0) | 3 (0.5) | ||||
Other* | 1 (0.2) | 2 (0.3) | ||||
Unknown* | 0 (0) | 2 (0.3) | ||||
Geographic location, no. patients (%) | ||||||
Minnesota | 119 (25) | 355 (57) | <0.0001 | |||
Minnesota, Wisconsin, Iowa | 241 (50) | 452 (72) | <0.0001 | |||
Upper Midwest† | 396 (82) | 544 (87) | 0.03 | |||
Smoking history, no. patients (%) | 0.01 | |||||
Never smoker | 196 (41) | 303 (49) | ||||
Ever smoker | 284 (59) | 320 (51) | ||||
Unknown | 1 (0.2) | 2 (0.3) | ||||
Pancreatic cancer stage, no. patients (%) | ||||||
I | 32 (7) | — | ||||
II | 164 (34) | — | ||||
III | 90 (19) | — | ||||
IV | 189 (39) | — | ||||
Pancreatic cancer surgery, no. patients (%) | 119 (25) | — |
These patients were excluded from the genotyping analyses.
Patients were residents of Illinois, Indiana, Iowa, Michigan, Minnesota, North Dakota, South Dakota, or Wisconsin.
Genotype frequencies
SNP (major-minor allele) . | Genotype frequency* . | . | . | Significance for genotype differences (cases versus controls), P† . | Hardy-Weinberg equilibrium, P . | . | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
. | Major-major . | Major-minor . | Minor-minor . | . | Goodness of fit‡ . | Exact test§ . | ||||||
OGG1 | ||||||||||||
R229Q (G/A) | 0.06 | 0.76 | >0.99 | |||||||||
Cases (n = 463) | 459 (99) | 4 (1) | 0 (0) | |||||||||
Controls (n = 585) | 570 (97) | 15 (3) | 0 (0) | |||||||||
S326C (C/G) | 0.62 | 0.90 | 0.91 | |||||||||
Cases (n = 469) | 268 (57) | 178 (38) | 23 (5) | |||||||||
Controls (n = 599) | 339 (57) | 223 (37) | 37 (6) | |||||||||
XRCC1 | ||||||||||||
R194W (C/T) | 0.86 | 0.53 | 0.76 | |||||||||
Cases (n = 466) | 400 (86) | 64 (14) | 2 (0.4) | |||||||||
Controls (n = 602) | 520 (86) | 80 (13) | 2 (0.3) | |||||||||
R280H (G/A) | 0.16 | 0.59 | >0.99 | |||||||||
Cases (n = 468) | 432 (92) | 34 (7) | 2 (0.4) | |||||||||
Controls (n = 585) | 525 (90) | 59 (10) | 1 (0.2) | |||||||||
R399Q (G/A) | 0.27 | 0.52 | 0.55 | |||||||||
Cases (n = 473) | 196 (41) | 211 (45) | 66 (14) | |||||||||
Controls (n = 612) | 228 (37) | 296 (48) | 88 (14) | |||||||||
ERCC1 | ||||||||||||
N118N (C/T) | 0.83 | 0.24 | 0.24 | |||||||||
Cases (n = 472) | 197 (42) | 202 (43) | 73 (15) | |||||||||
Controls (n = 603) | 244 (40) | 279 (46) | 80 (13) | |||||||||
I1 + 123 (G/T) | 0.57 | 0.74 | 0.73 | |||||||||
Cases (n = 473) | 384 (81) | 83 (18) | 6 (1) | |||||||||
Controls (n = 612) | 503 (82) | 104 (17) | 5 (1) | |||||||||
XPD/ERCC2 | ||||||||||||
R156R (T/G) | 0.62 | 0.92 | 0.95 | |||||||||
Cases (n = 460) | 140 (30) | 233 (51) | 87 (19) | |||||||||
Controls (n = 588) | 178 (30) | 287 (49) | 123 (21) | |||||||||
D312N (G/A) | 0.28 | 0.06 | 0.06 | |||||||||
Cases (n = 473) | 198 (42) | 203 (43) | 72 (15) | |||||||||
Controls (n = 612) | 269 (44) | 265 (43) | 78 (13) | |||||||||
D711D (G/A) | 0.31 | 0.09 | 0.09 | |||||||||
Cases (n = 473) | 207 (44) | 200 (42) | 66 (14) | |||||||||
Controls (n = 602) | 275 (46) | 257 (43) | 70 (12) | |||||||||
Q751K (T/G) | 0.44 | 0.70 | 0.70 | |||||||||
Cases (n = 473) | 186 (39) | 211 (45) | 76 (16) | |||||||||
Controls (n = 611) | 241 (39) | 291 (48) | 79 (13) | |||||||||
XPC | ||||||||||||
L16V (C/G) | 0.78 | 0.39 | >0.99 | |||||||||
Cases (n = 440) | 418 (95) | 22 (5) | 0 (0) | |||||||||
Controls (n = 550) | 520 (94) | 30 (6) | 0 (0) | |||||||||
A499V (T/C) | 0.16 | 0.64 | 0.68 | |||||||||
Cases (n = 457) | 246 (54) | 182 (40) | 29 (6) | |||||||||
Controls (n = 582) | 339 (58) | 211 (36) | 32 (6) | |||||||||
K939Q (A/C) | 0.42 | 0.77 | 0.79 | |||||||||
Cases (n = 468) | 161 (34) | 231 (49) | 76 (16) | |||||||||
Controls (n = 598) | 221 (37) | 285 (48) | 92 (15) | |||||||||
XPF/ERCC4 | ||||||||||||
R415Q (G/A) | 0.003 | 0.17 | 0.23 | |||||||||
Cases (n = 470) | 411 (87) | 59 (13) | 0 (0) | |||||||||
Controls (n = 596) | 481 (81) | 111 (19) | 4 (1) | |||||||||
S662Q (T/C) | 0.64 | <0.001 | 0.005 | |||||||||
Cases (n = 468) | 467 (100) | 1 (0) | 0 (0) | |||||||||
Controls (n = 593) | 590 (100) | 2 (0) | 1 (0) |
SNP (major-minor allele) . | Genotype frequency* . | . | . | Significance for genotype differences (cases versus controls), P† . | Hardy-Weinberg equilibrium, P . | . | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
. | Major-major . | Major-minor . | Minor-minor . | . | Goodness of fit‡ . | Exact test§ . | ||||||
OGG1 | ||||||||||||
R229Q (G/A) | 0.06 | 0.76 | >0.99 | |||||||||
Cases (n = 463) | 459 (99) | 4 (1) | 0 (0) | |||||||||
Controls (n = 585) | 570 (97) | 15 (3) | 0 (0) | |||||||||
S326C (C/G) | 0.62 | 0.90 | 0.91 | |||||||||
Cases (n = 469) | 268 (57) | 178 (38) | 23 (5) | |||||||||
Controls (n = 599) | 339 (57) | 223 (37) | 37 (6) | |||||||||
XRCC1 | ||||||||||||
R194W (C/T) | 0.86 | 0.53 | 0.76 | |||||||||
Cases (n = 466) | 400 (86) | 64 (14) | 2 (0.4) | |||||||||
Controls (n = 602) | 520 (86) | 80 (13) | 2 (0.3) | |||||||||
R280H (G/A) | 0.16 | 0.59 | >0.99 | |||||||||
Cases (n = 468) | 432 (92) | 34 (7) | 2 (0.4) | |||||||||
Controls (n = 585) | 525 (90) | 59 (10) | 1 (0.2) | |||||||||
R399Q (G/A) | 0.27 | 0.52 | 0.55 | |||||||||
Cases (n = 473) | 196 (41) | 211 (45) | 66 (14) | |||||||||
Controls (n = 612) | 228 (37) | 296 (48) | 88 (14) | |||||||||
ERCC1 | ||||||||||||
N118N (C/T) | 0.83 | 0.24 | 0.24 | |||||||||
Cases (n = 472) | 197 (42) | 202 (43) | 73 (15) | |||||||||
Controls (n = 603) | 244 (40) | 279 (46) | 80 (13) | |||||||||
I1 + 123 (G/T) | 0.57 | 0.74 | 0.73 | |||||||||
Cases (n = 473) | 384 (81) | 83 (18) | 6 (1) | |||||||||
Controls (n = 612) | 503 (82) | 104 (17) | 5 (1) | |||||||||
XPD/ERCC2 | ||||||||||||
R156R (T/G) | 0.62 | 0.92 | 0.95 | |||||||||
Cases (n = 460) | 140 (30) | 233 (51) | 87 (19) | |||||||||
Controls (n = 588) | 178 (30) | 287 (49) | 123 (21) | |||||||||
D312N (G/A) | 0.28 | 0.06 | 0.06 | |||||||||
Cases (n = 473) | 198 (42) | 203 (43) | 72 (15) | |||||||||
Controls (n = 612) | 269 (44) | 265 (43) | 78 (13) | |||||||||
D711D (G/A) | 0.31 | 0.09 | 0.09 | |||||||||
Cases (n = 473) | 207 (44) | 200 (42) | 66 (14) | |||||||||
Controls (n = 602) | 275 (46) | 257 (43) | 70 (12) | |||||||||
Q751K (T/G) | 0.44 | 0.70 | 0.70 | |||||||||
Cases (n = 473) | 186 (39) | 211 (45) | 76 (16) | |||||||||
Controls (n = 611) | 241 (39) | 291 (48) | 79 (13) | |||||||||
XPC | ||||||||||||
L16V (C/G) | 0.78 | 0.39 | >0.99 | |||||||||
Cases (n = 440) | 418 (95) | 22 (5) | 0 (0) | |||||||||
Controls (n = 550) | 520 (94) | 30 (6) | 0 (0) | |||||||||
A499V (T/C) | 0.16 | 0.64 | 0.68 | |||||||||
Cases (n = 457) | 246 (54) | 182 (40) | 29 (6) | |||||||||
Controls (n = 582) | 339 (58) | 211 (36) | 32 (6) | |||||||||
K939Q (A/C) | 0.42 | 0.77 | 0.79 | |||||||||
Cases (n = 468) | 161 (34) | 231 (49) | 76 (16) | |||||||||
Controls (n = 598) | 221 (37) | 285 (48) | 92 (15) | |||||||||
XPF/ERCC4 | ||||||||||||
R415Q (G/A) | 0.003 | 0.17 | 0.23 | |||||||||
Cases (n = 470) | 411 (87) | 59 (13) | 0 (0) | |||||||||
Controls (n = 596) | 481 (81) | 111 (19) | 4 (1) | |||||||||
S662Q (T/C) | 0.64 | <0.001 | 0.005 | |||||||||
Cases (n = 468) | 467 (100) | 1 (0) | 0 (0) | |||||||||
Controls (n = 593) | 590 (100) | 2 (0) | 1 (0) |
Data are shown as number of patients (%).
Calculated using the Cochran-Armitage trend test. For OGG1 R229Q, XPC L16V, and XPF/ERCC4 S622Q, minor allele carriers were grouped together, and the Fisher exact test was used.
Asymptomatic P value of the χ2 goodness-of-fit statistic (1 df).
P value for the exact test of the Hardy-Weinberg equilibrium.
Gene-tobacco interaction effect on pancreatic cancer risk for selected DNA repair gene polymorphisms
SNP (major-minor allele) . | Pancreatic cancer risk, OR (95% CI)* . | . | . | . | . | . | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
. | All subjects (473 cases, 615 controls)† . | All never smokers (193 cases, 295 controls)‡ . | All ever smokers (279 cases, 315 controls)‡ . | ≤20 pack-year smokers (82 cases, 155 controls)‡ . | >20–40 pack-year smokers (77 cases, 83 controls)‡ . | >40 pack-year smokers (70 cases, 54 controls)‡ . | ||||||
OGG1 | ||||||||||||
R229Q (G/A) | 0.29 (0.08–1.08) | 0.35 (0.04–3.11) | 0.37 (0.10–1.43) | — | — | 0.86 (0.07–10.40) | ||||||
S326G (C/G) | 0.95 (0.73–1.25) | 1.18 (0.80–1.74) | 0.78 (0.55–1.09) | 1.23 (0.70–2.16) | 0.52 (0.26–1.04) | 0.66 (0.31–1.40) | ||||||
XRCC1 | ||||||||||||
R194W (C/T) | 1.08 (0.74–1.57) | 1.12 (0.64–1.95) | 0.97 (0.60–1.56) | 0.94 (0.43–2.08) | 0.74 (0.32–1.71) | 2.63 (0.77–9.01) | ||||||
R280H (G/A) | 0.76 (0.48–1.22) | 0.64 (0.32–1.30) | 0.65 (0.37–1.20) | 1.01 (0.44–2.33) | 1.38 (0.40–4.75) | 0.34 (0.07–1.60) | ||||||
R399Q (G/A) | 0.85 (0.65–1.12) | 0.89 (0.60–1.31) | 0.95 (0.67–1.33) | 0.62 (0.35–1.09) | 0.88 (0.46–1.70) | 1.42 (0.65–3.11) | ||||||
ERCC1 | ||||||||||||
N118N (C/T) | 0.96 (0.73–1.25) | 0.98 (0.66–1.45) | 0.92 (0.66–1.30) | 0.78 (0.44–1.38) | 0.84 (0.44–1.60) | 1.65 (0.76–3.56) | ||||||
I1+123 (G/T) | 1.19 (0.84–1.67) | 1.19 (0.74–1.93) | 1.04 (0.67–1.60) | 1.21 (0.63–2.34) | 1.24 (0.47–3.32) | 1.05 (0.32–3.44) | ||||||
XPD/ERCC2 | ||||||||||||
R156R (T/G) | 1.06 (0.80–1.42) | 1.05 (0.69–1.59) | 0.98 (0.67–1.41) | 1.02 (0.56–1.88) | 1.75 (0.81–3.82) | 0.59 (0.25–1.38) | ||||||
D312N (G/A) | 1.08 (0.83–1.41) | 1.24 (0.84–1.82) | 1.00 (0.72–1.40) | 0.71 (0.41–1.25) | 0.64 (0.34–1.22) | 2.78 (1.28–6.04)§ | ||||||
D711D (G/A) | 1.06 (0.82–1.39) | 1.35 (0.92–1.99) | 0.93 (0.66–1.30) | 0.61 (0.35–1.06) | 0.68 (0.36–1.31) | 2.19 (1.01–4.73)§ | ||||||
Q751K (T/G) | 0.96 (0.73–1.25) | 1.20 (0.81–1.77) | 0.91 (0.65–1.29) | 0.61 (0.35–1.08) | 0.59 (0.30–1.14) | 1.78 (0.82–3.86) | ||||||
XPC | ||||||||||||
L16V (C/G) | 1.04 (0.56–1.94) | 0.96 (0.39–2.38) | 1.14 (0.52–2.50) | 1.11 (0.35–3.55) | 3.26 (0.58–18.28) | — | ||||||
A499V (T/C) | 1.27 (0.97–1.67) | 1.18 (0.80–1.74) | 1.19 (0.84–1.67) | 1.00 (0.55–1.79) | 1.61 (0.83–3.12) | 1.81 (0.82–4.00) | ||||||
K939Q (A/C) | 1.03 (0.78–1.36) | 1.01 (0.68–1.50) | 1.06 (0.75–1.51) | 1.29 (0.72–2.32) | 0.71 (0.37–1.38) | 1.36 (0.60–3.08) | ||||||
XPF/ERCC4 | ||||||||||||
R415Q (G/A) | 0.59 (0.40–0.85)§ | 0.60 (0.35–1.05) | 0.61 (0.38–0.96)§ | 0.38 (0.16–0.88) | 0.58 (0.24–1.39) | 1.17 (0.40–3.40) | ||||||
S662Q (T/C) | 0.36 (0.03–3.98) | — | 0.49 (0.04–5.74) | — | — | — |
SNP (major-minor allele) . | Pancreatic cancer risk, OR (95% CI)* . | . | . | . | . | . | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
. | All subjects (473 cases, 615 controls)† . | All never smokers (193 cases, 295 controls)‡ . | All ever smokers (279 cases, 315 controls)‡ . | ≤20 pack-year smokers (82 cases, 155 controls)‡ . | >20–40 pack-year smokers (77 cases, 83 controls)‡ . | >40 pack-year smokers (70 cases, 54 controls)‡ . | ||||||
OGG1 | ||||||||||||
R229Q (G/A) | 0.29 (0.08–1.08) | 0.35 (0.04–3.11) | 0.37 (0.10–1.43) | — | — | 0.86 (0.07–10.40) | ||||||
S326G (C/G) | 0.95 (0.73–1.25) | 1.18 (0.80–1.74) | 0.78 (0.55–1.09) | 1.23 (0.70–2.16) | 0.52 (0.26–1.04) | 0.66 (0.31–1.40) | ||||||
XRCC1 | ||||||||||||
R194W (C/T) | 1.08 (0.74–1.57) | 1.12 (0.64–1.95) | 0.97 (0.60–1.56) | 0.94 (0.43–2.08) | 0.74 (0.32–1.71) | 2.63 (0.77–9.01) | ||||||
R280H (G/A) | 0.76 (0.48–1.22) | 0.64 (0.32–1.30) | 0.65 (0.37–1.20) | 1.01 (0.44–2.33) | 1.38 (0.40–4.75) | 0.34 (0.07–1.60) | ||||||
R399Q (G/A) | 0.85 (0.65–1.12) | 0.89 (0.60–1.31) | 0.95 (0.67–1.33) | 0.62 (0.35–1.09) | 0.88 (0.46–1.70) | 1.42 (0.65–3.11) | ||||||
ERCC1 | ||||||||||||
N118N (C/T) | 0.96 (0.73–1.25) | 0.98 (0.66–1.45) | 0.92 (0.66–1.30) | 0.78 (0.44–1.38) | 0.84 (0.44–1.60) | 1.65 (0.76–3.56) | ||||||
I1+123 (G/T) | 1.19 (0.84–1.67) | 1.19 (0.74–1.93) | 1.04 (0.67–1.60) | 1.21 (0.63–2.34) | 1.24 (0.47–3.32) | 1.05 (0.32–3.44) | ||||||
XPD/ERCC2 | ||||||||||||
R156R (T/G) | 1.06 (0.80–1.42) | 1.05 (0.69–1.59) | 0.98 (0.67–1.41) | 1.02 (0.56–1.88) | 1.75 (0.81–3.82) | 0.59 (0.25–1.38) | ||||||
D312N (G/A) | 1.08 (0.83–1.41) | 1.24 (0.84–1.82) | 1.00 (0.72–1.40) | 0.71 (0.41–1.25) | 0.64 (0.34–1.22) | 2.78 (1.28–6.04)§ | ||||||
D711D (G/A) | 1.06 (0.82–1.39) | 1.35 (0.92–1.99) | 0.93 (0.66–1.30) | 0.61 (0.35–1.06) | 0.68 (0.36–1.31) | 2.19 (1.01–4.73)§ | ||||||
Q751K (T/G) | 0.96 (0.73–1.25) | 1.20 (0.81–1.77) | 0.91 (0.65–1.29) | 0.61 (0.35–1.08) | 0.59 (0.30–1.14) | 1.78 (0.82–3.86) | ||||||
XPC | ||||||||||||
L16V (C/G) | 1.04 (0.56–1.94) | 0.96 (0.39–2.38) | 1.14 (0.52–2.50) | 1.11 (0.35–3.55) | 3.26 (0.58–18.28) | — | ||||||
A499V (T/C) | 1.27 (0.97–1.67) | 1.18 (0.80–1.74) | 1.19 (0.84–1.67) | 1.00 (0.55–1.79) | 1.61 (0.83–3.12) | 1.81 (0.82–4.00) | ||||||
K939Q (A/C) | 1.03 (0.78–1.36) | 1.01 (0.68–1.50) | 1.06 (0.75–1.51) | 1.29 (0.72–2.32) | 0.71 (0.37–1.38) | 1.36 (0.60–3.08) | ||||||
XPF/ERCC4 | ||||||||||||
R415Q (G/A) | 0.59 (0.40–0.85)§ | 0.60 (0.35–1.05) | 0.61 (0.38–0.96)§ | 0.38 (0.16–0.88) | 0.58 (0.24–1.39) | 1.17 (0.40–3.40) | ||||||
S662Q (T/C) | 0.36 (0.03–3.98) | — | 0.49 (0.04–5.74) | — | — | — |
All ORs compared genotype frequencies of minor-minor plus major-minor versus major-major genotypes. Some study subjects did not provide smoking data; ORs were calculated using all available data (as shown by the number of cases and controls in each subgroup).
Adjusted for age, sex, and pack-year group.
Adjusted for age and sex.
Statistically significant values.
In the subset analysis of groups stratified by sex, smoking status, and pack-years, findings from the XPD/ERCC2 polymorphisms were particularly notable (Table 3). Among the heaviest smokers (>40 pack-years), carriers of the D312N minor allele (OR, 2.78; 95% CI, 1.28–6.04) and the D711D minor allele (OR, 2.19; 95% CI, 1.01–4.73) had increased risk of pancreatic cancer. Carriers of the Q751K minor allele had increased risk (OR, 1.78; 95% CI, 0.82–3.86) and those with the synonymous SNP R156R showed decreased risk (OR, 0.59; 95% CI, 0.25–1.38), but neither were statistically significant. Among the subgroup analysis by sex, the female heavy smokers (n = 32) had higher risk than males if they carried minor alleles for D312N (OR, 8.71; 95% CI, 1.64–46.24 versus OR, 2.06; 95% CI, 0.84–5.07), D711D (OR, 9.66; 95% CI, 1.69–55.27 versus OR, 1.43; 95% CI, 0.59–3.46), or Q751K (OR, 4.61; 95% CI, 0.88–23.99 versus OR, 1.35; 95% CI, 0.56–3.29).
Patients with young-onset pancreatic cancer (n = 136), previously defined by our group as a diagnosis established before age of 60 years (37), showed no difference in genotype frequencies for any SNPs when compared with all controls or with controls younger than 60 years (data not shown). Because cases and controls resided in different geographic areas, genotype frequencies in cases and controls combined were compared for all SNPs by location (inside or outside the tristate area of Minnesota, Wisconsin, and Iowa), with no significant differences seen (data not shown).
In overall haplotype analyses controlling for age, sex, and smoking status, only the XPF/ERCC4 haplotypes were associated with risk (P = 0.02); this association was attributable to the R415Q allele.
In analyses limited to the heaviest smokers (>40 pack-years), persons carrying one XPD/ERCC2 haplotype with a minor allele at D312N, D711D, or Q751K had increased risk of pancreatic cancer when compared with those whose haplotype had no minor alleles (OR, 3.03; 95% CI, 1.11–8.25; Table 4). Persons carrying two haplotypes with a minor allele had an even higher risk (OR, 4.11; 95% CI, 1.25–13.54). Persons carrying a haplotype encoding a minor allele at all three SNPs also had increased risk (OR, 1.92; 95% CI, 1.02–3.61).
Haplotype analysis of XPD/ERCC2 polymorphisms in heavy smokers
Variable . | Haplotype frequencies . | . | . | Simulated P . | Pancreatic cancer risk, OR (95% CI) . | . | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
. | Overall . | Case . | Control . | . | Unadjusted . | Adjusted for age and sex . | ||||||
Individual haplotype | ||||||||||||
D312N (G/A), D711D (G/A), Q751K (T/G) | ||||||||||||
G, G, T | 0.567 | 0.534 | 0.609 | 0.19 | 1.00 | 1.00 | ||||||
G, G, G | 0.032 | 0.021 | 0.046 | 0.32 | 0.51 (0.11–2.36) | 0.59 (0.13–2.77) | ||||||
G, A, G | 0.042 | 0.030 | 0.057 | 0.34 | 0.56 (0.14–2.21) | 0.73 (0.18–2.90) | ||||||
A, G, T | 0.046 | 0.052 | 0.039 | 0.56 | 1.43 (0.42–4.84) | 1.71 (0.50–5.90) | ||||||
A, A, G | 0.312 | 0.362 | 0.248 | 0.07 | 1.63 (0.91–2.93) | 1.92 (1.02–3.61) | ||||||
No. haplotypes with at least one minor allele* | ||||||||||||
1 versus 0 | 2.70 (1.04–7.06) | 3.03 (1.11–8.25) | ||||||||||
2 versus 0 | 3.20 (1.04–9.85) | 4.11 (1.25–13.54) | ||||||||||
≥1 versus 0 | 2.84 (1.13–7.13) | 3.31 (1.27–8.67) |
Variable . | Haplotype frequencies . | . | . | Simulated P . | Pancreatic cancer risk, OR (95% CI) . | . | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
. | Overall . | Case . | Control . | . | Unadjusted . | Adjusted for age and sex . | ||||||
Individual haplotype | ||||||||||||
D312N (G/A), D711D (G/A), Q751K (T/G) | ||||||||||||
G, G, T | 0.567 | 0.534 | 0.609 | 0.19 | 1.00 | 1.00 | ||||||
G, G, G | 0.032 | 0.021 | 0.046 | 0.32 | 0.51 (0.11–2.36) | 0.59 (0.13–2.77) | ||||||
G, A, G | 0.042 | 0.030 | 0.057 | 0.34 | 0.56 (0.14–2.21) | 0.73 (0.18–2.90) | ||||||
A, G, T | 0.046 | 0.052 | 0.039 | 0.56 | 1.43 (0.42–4.84) | 1.71 (0.50–5.90) | ||||||
A, A, G | 0.312 | 0.362 | 0.248 | 0.07 | 1.63 (0.91–2.93) | 1.92 (1.02–3.61) | ||||||
No. haplotypes with at least one minor allele* | ||||||||||||
1 versus 0 | 2.70 (1.04–7.06) | 3.03 (1.11–8.25) | ||||||||||
2 versus 0 | 3.20 (1.04–9.85) | 4.11 (1.25–13.54) | ||||||||||
≥1 versus 0 | 2.84 (1.13–7.13) | 3.31 (1.27–8.67) |
Minor allele frequency for XPD/ERCC2 D312N, D711D, and Q751K were 1, 1, and 1, respectively.
The total number of minor alleles in the BER pathway was calculated by summing the number of minor alleles for all five tested SNPs. There was no significant association between the number of minor alleles and pancreatic cancer (OR, 0.92; 95% CI, 0.80–1.06). Similarly, there was no association between the total number of minor alleles in the NER pathway and pancreatic cancer (OR, 1.02; 95% CI, 0.96–1.09).
Gene level analyses using unconditional multivariate logistic regression models examined the total number of minor-allele variants in each gene, and a second set of analyses considered the number of haplotypes that contained minor-allele variants (0, 1, or 2). Analysis of ERCC1 and XPD/ERCC2 did not show a significant association between an increasing number of minor-allele variants and pancreatic cancer. Univariate analysis of XPC showed an association (OR, 1.22; 95% CI, 1.02–1.47), but it was not significant in multivariate analysis (OR, 1.18; 95% CI, 0.97–1.43). XPF/ERCC4 analysis suggested an association with pancreatic cancer (adjusted OR, 0.58; 95% CI, 0.40–0.83). However, as in the haplotype analysis, this association likely was driven by the R415Q SNP because we had only a limited number of patients with a variant allele (n = 5) for the S662P marker.
Discussion
Measurement of pancreatic cancer risk by analysis of SNPs in DNA repair genes is challenging. The probable low penetrance of minor genetic variations in DNA repair genes, in conjunction with many possible gene-gene interactions, makes large study sample sizes mandatory. However, pancreatic cancer research was lacking such resources until recently. Population-based studies may be affected by survival bias because of the short survival time after diagnosis and the time needed to identify, contact, and recruit patients to studies. Our prospective recruitment of patients with pancreatic cancer was clinic-based to achieve as representative a sample as possible.
An understanding of genetic risk for pancreatic cancer beyond that of classic family cancer syndromes is imperative. Currently, the only option for potential cure is early detection that allows surgical resection. Screening and prevention strategies to detect lesions at an earlier stage require identification of patients with high risk of pancreatic cancer development. Characterization of SNPs at key genomic locations has the potential to contribute to risk modeling.
Our results for the XRCC1 R399Q polymorphism in heavy smokers (Table 3) differed from those reported by Duell and colleagues (12). However, we did not exclude the possibility of an association because we measured an increased risk of cancer for heavy smokers with the minor allele (OR 1.42; 95% CI, 0.65–3.11). Nevertheless, the number of heavy smokers in our study (70 cases, 54 controls) was low. A future study with more subjects could better elucidate the interaction of the R399Q minor allele and smoking status with risk of pancreatic cancer.
The protective effect of the XPF/ERCC4 R415Q minor allele was unexpected. This SNP previously was associated with increased risk of breast cancer in a study of 253 cases and 268 controls (27). However, a larger, two-stage study of >1,700 cases and 1,900 controls found no increased risk for patients with XPF/ERCC4 minor alleles and possibly a protective effect from an intronic SNP (38). Given the relatively low frequency of the A allele of this polymorphism (10% in our control group), our finding may have been spurious; a larger study may elucidate the significance of this allele.
With regard to the increased risk of the XPD/ERCC2 minor alleles among the heaviest smokers, the prior report by Jiao and colleagues (23) showed a protective effect of the N/N genotype at the D312N polymorphism in ever smokers. However, they did not report pack-year information, so specific effects attributable to the degree of smoking exposure from our study and theirs cannot be compared directly. Indeed, our study shows no association between the XPD/ERCC2 polymorphisms and ever smokers when all were pooled together (Table 3). The effect among the D312N, D711D, and Q751K minor alleles was strongest among female heavy smokers; furthermore, it seemed to depend on the number of haplotypes with SNPs containing at least one minor allele. Why the effect was seen only in heavy smokers is unclear. It is possible that pancreatic ductal cells with variant genotypes can withstand low amounts of DNA damage without increased risk for cancer, but prolonged high exposures of carcinogens in tobacco may make the excess risk evident. As stated in Materials and Methods, this finding should be interpreted as hypothesis generating and not as a conclusive association.
Zhou and colleagues (25) studied the D312N and Q751K polymorphisms of XPD/ERCC2 in the context of smoking exposure in patients with lung cancer. Unlike our findings, which showed elevated risk of pancreatic cancer for heavy smokers with minor alleles, their findings showed the opposite effect: patients who were homozygous for both minor alleles had a higher risk of lung cancer for nonsmokers (OR, 2.56; 95% CI, 1.3–5.0) but not among the heaviest smokers (OR, 0.69; 95% CI, 0.4–1.2). The authors theorized that observed gene-tobacco interactions in their study might explain inconsistent results from prior studies (11, 39–41). These differences potentially could be explained by the different carcinogens in the exposed lung versus the pancreatic ductal cells because aerosolized and blood-borne carcinogens may differ. In addition, the specific chemicals responsible for carcinogenesis in the lung may be markedly different from those that affect the pancreas (17).
As with any association study, our findings could be spurious, particularly because of the relatively small number of patients in each subgroup. Given that the common SNPs in XPD/ERCC2 are associated inconsistently with a decrease in DNA repair capacity (42) and are not highly conserved among species (43), it is also possible that the effects are not attributable to the SNPs themselves and perhaps arise from neighboring variants. In fact, the functional effect of many DNA repair SNPs is poorly understood, possibly because assays to measure DNA repair capacity are insufficiently sensitive to detect minor variations in function. Two computer-based programs [Sorting Intolerant from Tolerant (SIFT) and Polymorphism Phenotyping], designed to predict the effect of amino acid substitutions on protein function, calculated that 44% and 34%, respectively, of all nonsynonymous SNPs in DNA repair genes would adversely affect function (44). However, none of the nonsynonymous SNPs presented in this report were predicted by SIFT to have a high likelihood of markedly affecting protein function (SIFT score range, 0.11–1.0; scores of <0.05 predict intolerance; ref. 45). Using a newer algorithm (SNPs3D), reported to be more sensitive and specific than SIFT or Polymorphism Phenotyping (46), the XPF/ERCC4 R415Q substitution is predicted to negatively affect protein function (score, −0.91; negative scores predict damage), whereas the XPD/ERCC2 SNPs D312N and Q751K are not predicted to be damaging (scores, 2.03–2.88; ref. 47).
Our study was sufficiently powered (80%) to detect (a) an overall OR of 2.0 between cases and controls and an MAF of 0.051 among controls, (b) an OR of 1.8 and an MAF of 0.077 among controls, or (c) an OR of 1.6 and an MAF of 0.14 among controls. However, in subgroups, the power was insufficient to make definitive conclusions. Clearly, larger studies are needed to confirm findings from subgroup analyses. In addition, our controls were patients recruited into a colon cancer study after receiving normal results from screening colonoscopies; they were not specifically matched to our cases. Because of referral patterns, cases were less likely to be from Minnesota or the tristate area of Minnesota, Wisconsin, and Iowa. However, >80% of cases and controls were from the Upper Midwest.
The age and sex differences between cases and controls were statistically significant and may have introduced bias, although we accounted for these variables in our analyses. Our cases and controls were nearly exclusively White, so the generalizability of our findings may be limited. However, the homogeneity of racial background may reduce the possibility of bias by race-driven effects. Twenty-four percent of our cases did not complete the questionnaire. Although smoking data and family history data were readily available for most through the medical records, the difference in data collection methods has the potential to affect results. In addition, the reasons our controls underwent colonoscopies were unknown but may be related to risk of colon cancer. If the risks of colorectal and pancreatic cancer are each associated with DNA repair polymorphisms, our results could be biased downward owing to shared risk alleles.
We believe that our clinic-based case-control study has several advantages over population-based studies when researching pancreatic adenocarcinoma. Our participation rate of 70% was considerably higher than rates (<50% participation) reported in population-based studies (48, 49). The short survival of patients with pancreatic cancer—4 to 5 months for those with metastatic disease (50)—impedes the ability to identify, recruit, and obtain samples from affected patients and perhaps may introduce a survival bias to population-based studies. Because our patients were identified during their clinical visits, this bias was minimized.
Conclusions
The minor allele in XPF/ERCC4 at R415Q was associated with decreased overall risk of pancreatic cancer and was distributed approximately equally between never smokers and ever smokers. Among heavy smokers, and most notably for women, minor alleles in XPD/ERCC2 at D312N and D711D were associated with increased risk for pancreatic cancer. Full elucidation of the contributions of low-penetrance genetic variations to overall risk will require larger studies of patients with pancreatic cancer in population-based and clinic-based settings.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Note: Abstract presented at the American Association for Cancer Research Special Conference in Cancer Research: Approaches to Complex Pathways in Molecular Epidemiology, Santa Ana Pueblo, New Mexico, May 30 to June 2, 2007, at the annual meeting of the Lustgarten Foundation for Pancreatic Cancer Research, Chapel Hill, North Carolina, June 26 to June 27, 2006, and at the American Society of Clinical Oncology 41st annual meeting; May 13 to 17, 2005, Orlando, Florida.
Acknowledgments
Grant support: National Cancer Institute grants P50 CA102701, R01 CA97075, and K07 CA116303.
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.
Editing, proofreading, and reference verification were provided by the Section of Scientific Publications, Mayo Clinic.