APE1 genotype and risk of bladder cancer: Evidence for effect modification by smoking†
Article first published online: 19 JAN 2006
Published 2006 Wiley-Liss, Inc.
International Journal of Cancer
Volume 118, Issue 12, pages 3170–3173, 15 June 2006
How to Cite
Terry, P. D., Umbach, D. M. and Taylor, J. A. (2006), APE1 genotype and risk of bladder cancer: Evidence for effect modification by smoking. Int. J. Cancer, 118: 3170–3173. doi: 10.1002/ijc.21768
This article is a US Government work and, as such, is in the public domain in the United States of America.
- Issue published online: 27 MAR 2006
- Article first published online: 19 JAN 2006
- Manuscript Accepted: 10 NOV 2005
- Manuscript Received: 6 JUL 2005
- Intramural Research Program of the NIH, National Institute of Environmental Health Sciences
- bladder neoplasms;
- apurinic endonuclease;
- case–control studies
Apurinic/apyrimidinic (AP) sites are common mutagenic and cytotoxic DNA lesions that arise from the loss of normal bases. APE1, the major AP endonuclease of human cells, plays a central role in the repair of AP sites through both its endonuclease and phosphodiesterase activities. A common APE1 polymorphism, a TG transversion (Asp 148 Glu), was previously shown to be associated with risk of lung cancer, an association that was modified by cigarette smoking. To explore the association between APE1 genotype, smoking and bladder cancer risk, we examined data from an existing case–control study of bladder cancer patients (n = 239) and control individuals (n = 215) recruited from urology clinics at 2 hospitals in North Carolina. Genotype at the polymorphic site was determined using allele-specific primer extension reactions, followed by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. We found no overall association between APE1 genotype and bladder cancer risk. In stratified analyses, however, a positive association with risk was observed with an increasing number of Glu alleles among never smokers, but not among smokers (p-value for interaction = 0.005). We can speculate that small allelic differences that are apparent in never smokers are obscured by the large amount of DNA damage found in smokers. Given the lack of established biological mechanisms, and suboptimal numbers of subjects in some exposure categories, our findings should be interpreted cautiously. Published 2006 Wiley-Liss, Inc.
Several etiological factors have been associated with the development of bladder cancer, cigarette smoking being among the most important.1 The precise mechanisms by which smoking causes bladder cancer have yet to be fully clarified. Tobacco smoke is known to contain many potential human carcinogens, including aromatic amines and N-nitrosamines,1, 2 compounds that have been linked to DNA adduct formation2 and bladder cancer development.1, 3 Cigarette smoke also generates large quantities of free radicals,4 which are highly reactive species that induce base modifications and strand breaks.5 Thus, smoking may induce bladder cancer through different types of DNA damage.
Apurinic/apyrimidinic (AP) sites are common mutagenic and cytotoxic DNA lesions that arise from the loss of normal bases, either through spontaneous processes, or excision of damaged bases by DNA glycosylases during the DNA repair process.6, 7 AP sites are repaired by a series of reactions involving the multiple repair proteins of the base excision repair (BER) pathway.6, 8 AP Endonuclease 1 (APE1), the major AP endonuclease of human cells, plays a central role in BER by hydrolyzing the phosphodiester backbone immediately 5′ to the AP site.9, 10 This incision generates a normal 3′-hydroxyl group and an abasic deoxyribose-5-phosphate, which is processed by enzymes in the subsequent steps of the BER pathway.11 APE1 also acts as a 3′-phosphodiesterase to initiate repair of single strand breaks,9, 12, 13 damage that may result from direct attack and fragmentation of deoxyribose residues in DNA by free radicals.9, 14, 15, 16 By hydrolyzing 3′-blocking fragments from oxidized DNA, APE1 produces normal 3′-hydroxyl nucleotide termini that are necessary for DNA repair synthesis and ligation at single- or double-strand breaks.9, 17
A common APE1 polymorphism, a TG transversion (Asp 148 Glu), was shown to have 94% of the endonuclease activity relative to the Asp allele, and 21% less binding activity (Kd, 20.3 ± 3.4 versus 25.8 ± 12.2 nM in Asp allele).10 In a small study on X-ray exposure to lymphocytes and DNA repair gene polymorphisms on chromosome aberrations, samples from individuals with Asp/Glu or Glu/Glu genotype showed higher levels of damage with respect to all of the studied measures: aberrant cells, chromatid breaks, chromatid exchanges, deletions and dicentrics.18 Prolonged cell cycle delay was also associated with the number of APE1 variant alleles.19 To date, 3 studies have examined the association between APE1 genotype and risk of lung cancer. The Glu/Glu genotype has been associated with increased risk of lung cancer in a recent case–control study in Japan20 and, conversely, with a statistically nonsignificant decreased risk of nonsmall cell lung cancer in a case–control study in Germany.21 Finally, APE1 genotype was not associated with lung cancer risk among male smokers in Finland.22 An interaction between APE1 genotype and smoking was seen in the Japanese study such that the genotypic effect was apparent primarily among “light” smokers (those with 40 pack-years of consumption or less), as compared with never smokers and “heavy” smokers. The German study found no statistically significant interactions with any of the examined DNA repair genotypes, including APE1, but did not report odds ratios (ORs).21 Given the inconclusive evidence, the interaction between APE1 genotype and smoking status in the Japanese study, the dearth of studies on APE1 genotype and cancer risk, and effect modification of the association between other genes and risk of lung and bladder cancer by smoking,23, 24 more evidence regarding the association between APE1 genotype and cancer risk is warranted. Therefore, we examined the associations between the APE1 genotype, cigarette smoking and bladder cancer risk, in a case–control study in the United States.
Material and methods
Bladder cancer patients (n = 239) and control individuals (n = 215) were enrolled from the Urology Clinics at Duke University Medical Center and the University of North Carolina Hospitals, as described previously.25 The association between smoking, variation in other genes and bladder cancer risk have previously been reported for this study.25, 26, 27 Briefly, cases were urology clinic patients with histologically confirmed transitional cell carcinoma. Controls were urology clinic patients without a history of cancer, frequency matched to cases based on ethnicity, sex and age (10-year intervals). The most common diagnoses among controls were benign prostatic hypertrophy and impotence. Case patients and control subjects were interviewed by trained nurse-interviewers using a structured questionnaire that detailed their smoking, occupational and other exposure histories. Smoking was assessed as qualitative (never, former, current) and quantitative (cigarettes per day, years smoked) measures. In analyses of genotype stratified by smoking status, individuals were dichotomized into never or ever smokers.
DNA was extracted from peripheral blood lymphocytes by standard methods and frozen until used. Genotyping assays were conducted by BioServe Biotechnologies (Laurel, MD) using the Sequenom (San Diego CA) MassARRAY matrix-assisted laser desorption/ionization mass spectrometry (MALDI-TOF) system. A 119-bp region was PCR-amplified using primers 5′-ACGTTGGATGTGCTTTCCCTTTTCTTATAG-3′ and 5′-ACGTTGGATGAAATTCAGCCACAATCACCC-3′ with 5 ng of template DNA, followed by a shrimp alkaline phosphatase reaction that dephosphorylated the remaining dNTPs. Allele-specific products were created using a mass extend primer 5′-GGCCTTCCTGATCATGCTCCTC-3′ and a termination mix of dideoxy CGT. These products were diluted and applied to a SpectroCHIP array and analyzed via MALDI-TOF. SpectroTYPER software was used to detect peaks corresponding to predicted masses for each genotype. Each set of study assays included 4 positive and 4 negative control samples.
Individuals homozygous for the minor allele, heterozygous or homozygous for the common allele were clearly distinguished in 230 (96%) of 239 cases, and 207 (96%) of 215 control subjects, with peaks correctly corresponding to the predicted masses. Genotype information for the remaining 9 cases and 8 control subjects was not obtainable, either because the peak intensity on MALDI-TOF was below the minimum threshold value for reliably determining alleles or because the ratio of the mass intensities for the 2 alleles did not fit within the ranges defined for each of the 3 possible genotypes. The requirements of MALDI-TOF high throughput genotyping did not allow individual samples with inadequate results to be repeated. Seventeen subjects were missing genotype data for APE1: 9 cases and 8 control subjects. In addition, information on smoking status was not available for 1 individual, a case. Thus, 229 cases and 207 control subjects were available for the analysis of APE1 genotype.
Unconditional logistic regression models were used to estimate ORs and 95% confidence intervals (CIs). Multivariable models included age (continuous variable), sex (female, male), race (white, others) and smoking (duration in 10-year categories). To evaluate effect modification by smoking, separate indicator terms were created for APE1 genotype within categories of smoking status. Tests for interaction were likelihood ratio tests that compared models with and without product terms representing the variables of interest. All statistical tests were 2-sided. Analyses were performed using SAS statistical software, version 8.02 (SAS Institute, Cary, NC).
Participants were mostly males (77% of cases and 81% of controls) and white (91% of cases and 93% of controls) (Table I). The mean age was 65.7 (SD = 10.7) years for cases and 63.3 (SD = 10.3) years for controls. As previously reported for this study population, bladder cancer risk was positively associated with cigarette smoking, particularly smoking of long duration.25, 26, 27 We detected no deviations from Hardy–Weinberg equilibrium in APE1 genotype distribution among controls (p > 0.80). The frequency of the variant allele in cases was 49%; in controls, it was 44%, which is similar to previous reports.10, 20
|Cases (n = 239)1||Controls (n = 215)||OR2||p3|
|Mean Age||65.7 (10.7)4||63.3 (10.3)4||0.007|
|1–34 years||92||38.8||93||43.3||2.7 [1.6–4.5]|
|>34 years||105||44.3||43||20.0||6.1 [3.5–10.8]|
|Mean Years||29.3 (19.9)||16.6 (16.8)||<0.001|
We found no clear association between APE1 genotype and bladder cancer risk in the total study population. Relative to individuals with the Asp/Asp genotype, those with the Asp/Glu genotype showed some indication of increased risk (OR = 1.7, 95% CI = 1.1–2.8), although a significantly elevated OR was not observed for those with the Glu/Glu genotype (OR = 1.4, 95% CI = 0.8–2.6); thus, carriers of 2 copies of the variant did not exhibit greater risk than carriers of a single copy. The OR for having 1 or 2 copies of the Glu allele compared with no copies was 1.5 (1.0–2.4; p-value = 0.05). Thus, the data may suggest a slightly elevated bladder cancer risk in subjects with at least 1 Glu allele.
Table II shows ORs for bladder cancer according to APE1 genotype and smoking status. Although the Asp/Asp never smokers would be the conventional referent group for these analyses, the small number of cases (n = 3) in this group makes it undesirable as a referent. Instead, we elected to set Asp/Asp smokers as the referent group. Among smokers, APE1 genotype was not associated with bladder cancer risk (p-value for trend = 0.80; Table II). In contrast, there was a positive association between the putative high risk allele and bladder cancer risk among never smokers (p-value for trend = 0.002). Specifically, we observed an approximately 9-fold difference between the OR of never smokers with Glu/Glu (OR = 0.52) compared to those with Asp/Asp (OR = 0.06), although the number of cases in this latter group is very small. Similar results were obtained, when we examined smokers and never smokers in separate multivariable models (data not shown). Using ever versus never smokers, a likelihood ratio test for interaction between APE1 genotype and smoking status in relation to bladder cancer risk was statistically significant (p = 0.005). Analyses using finer categories of smoking to examine dose-response effects were inconclusive due to low numbers of subjects in those exposure categories.
|Asp/Asp||3/28||0.1 (0.0–0.2)2||48/35||1.0 (referent)|
|Asp/Glu||24/34||0.4 (0.2–0.8)||109/70||1.2 (0.7–2.0)|
|Glu/Glu||12/13||0.5 (0.2–1.3)||33/27||0.9 (0.4–1.7)|
|p-value for trend||0.002||0.80|
The multivariate-adjusted ORs for bladder cancers occurring among never smokers (Asp/Glu and Glu/Glu compared to Asp/Asp) were 7.5 (95% CI = 1.9–29.7) and 10.6 (95% CI = 2.3–48.5), respectively. There were insufficient subjects to generate ORs for cancers occurring among smokers who had smoked for less than 10 years' duration. The corresponding ORs for cancers occurring among individuals who smoked 10 years or longer were 1.4 (95% CI = 0.8–2.5) and 0.9 (95% CI = 0.5–1.9), respectively. The results were similar between the current and former smokers. The results of our study were not altered appreciably by exclusion of the 37 nonwhites from the analyses. Analyses performed after excluding the 95 females from the analyses yielded unstable results due to the small number of remaining subjects in some of the risk categories.
We found no overall association between APE1 genotype and bladder cancer risk. In stratified analyses, a positive association was observed with increasing number of putative risk alleles among never smokers, but not among smokers. To our knowledge, our study is the first that examined effect modification of the association between APE1 genotype and bladder cancer risk by smoking. However, the category for never smokers with Asp/Asp genotype had only 3 cases. Thus, although these individuals had a significantly reduced risk of bladder cancer compared with never smokers carrying the Glu/Glu genotype, or with smokers carrying that genotype, small numbers compromised the reliability of the risk estimates.
A previous study found a significant positive association between APE1 Glu/Glu genotype and lung cancer among current smokers, particularly among those who were “light” smokers.20 A formal test for interaction between APE1 genotype and smoking was statistically significant in that study and indicated that APE1 genotype influenced risk only among light smokers, namely, those with 40 pack-years of consumption or less. This finding is similar to the effect-modification by smoking intensity and duration (or pack-years) in studies of MSPI and GSTM1 genotype and lung and bladder cancer risk,23, 24 where genotype associations have been demonstrated at low, but not at high, exposure levels. In light of these previous studies, the effect modification between APE1 genotype and smoking status in our data is interesting. Smoking induces a variety of oxidative DNA damage, the repair of which depends on the 3′-phosphodiesterase activity of APE1.12 Suh et al. found that, relative to its endonuclease activity, the 3′-phosphodiesterase activity of APE1 is rate-limiting.12 Thus, perhaps the small allelic differences in repair that appear to affect risk among never smokers are overwhelmed in smokers, where DNA damage is much greater. Given the dearth of studies of APE1 genotype and cancer risk, the lack of established biological mechanisms, and the small number of subjects in some of the genotype and smoking strata, our findings should be interpreted cautiously.
Finally, Asp 148 Glu is the only known common nonsynonymous APE1 coding region variant. However, we cannot rule out the possibility that variation in APE1 other than Asp 148 Glu may be associated with bladder cancer risk. We also cannot rule out possible interaction with variation in other BER genes,28 including those previously associated with bladder cancer risk.
The authors thank Dr. K. Meadows for helpful comments on an earlier version of the manuscript, and Dr. T. Lehman for help with genotyping.
- 3Urinary bladder cancer. In: AdamiHO, HunterD, TrichopoulosD, eds. Textbook of cancer epidemiology. New York: Oxford University Press, 2002. 446–66., .