Myeloperoxidase Genotype, Fruit and Vegetable Consumption, and Breast Cancer Risk Free

Both endogenous processes and exogenous exposures produce reactive oxygen species (ROS) in vivo. These ROS can cause oxidative damage to DNA and other macromolecules, resulting in genetic alterations, a process held in check by the existence of multiple antioxidant systems that alter the balance between prooxidant cellular activity and antioxidant defenses (1, 2, 3, 4). The role of oxidative stress in carcinogenesis has been widely demonstrated in in vitro studies and small human studies, and there is increasing evidence that oxidative stress is involved in the pathogenesis of breast cancer (5). DNA from invasive ductal carcinomas contains extensive modifications linked to oxidative stress, including 8-hydroxydeoxyguanosine, single- and double-strand breaks (6, 7, 8), and DNA-malondialdehyde adducts (9), markers of lipid peroxidation/oxidative stress. Higher levels of malondialdehyde have also been noted in urine of women with mammographic breast dysplasia (10), and damage resulting from hydroxyl radicals has been directly linked to the progression of human breast cancer (8). Oxidative damage has been reported to be higher in women with breast cancer, compared with controls, although studies to date remain small (11, 12) and, in a study of women at high risk for breast cancer (first-degree relative with the disease), levels varied with the consumption of meats, vegetables, and fruits (13).

Myeloperoxidase (MPO) generates ROS endogenously by functioning as an antimicrobial enzyme, catalyzing hydrogen peroxide-dependent oxidation of chloride to generate hypochlorous acid (HOCl), a potent oxidizing agent. HOCl reacts with other biological molecules to generate secondary oxidation products (14). MPO is present in neutrophils, monocytes, and some reactive microglial macrophages, which invade inflamed tissues, including the breast, to combat infection and, presumably, to protect breast milk during lactation (15). In addition to its presence in human breast milk, MPO has also been detected by immunohistochemistry in breast tissue from women with cancer (16), and MPO gene amplification with c-erB-2 has been observed in human breast carcinomas (17).

Although evidence is limited, there are indications that steroid hormones regulate MPO expression. For example, circulating variations in MPO are positively associated with estrogen levels during the menstrual cycle (18), and hormone replacement therapy increases MPO release from neutrophils in postmenopausal women (19). Thus, it is plausible to hypothesize that MPO could play a role in breast carcinogenesis, and that factors that would increase lifetime exposures to estrogens, such as age at menarche, could modify this relationship.

Numerous endogenous and exogenous factors, including dietary antioxidants, serve to reduce levels of ROS. Although fruits and vegetables contain numerous anticarcinogenic substances, they are also a rich source of antioxidant nutrients, such as carotenoids, vitamin C, and vitamin E. Although epidemiologic data are conflicting regarding fruit and vegetable consumption and breast cancer risk (20), analysis of data from the Long Island Breast Cancer Study Project (LIBCSP) showed that higher fruit and vegetable consumption was associated with decreased breast cancer risk among postmenopausal women [odds ratio (OR) for the highest quintile compared with the lowest, 0.72; 95% confidence interval (95% CI), 0.53–0.99], with weaker associations among premenopausal women (21).

A frequently occurring single nucleotide polymorphism in the human population in the promoter region of the MPO gene is a −463G→A substitution, located in the consensus binding site of an SP1 transcription factor in the 5′ upstream region of the gene (22). Piedrafita et al. (23) suggested that the MPO A variant allele confers lower transcriptional activation than the −463G (common) allele in vitro because of disruption of the SP1 binding site, and the G allele has been associated with increased MPO mRNA and protein levels in myeloid leukemia cells (24). However, a direct relationship between this polymorphism and MPO expression in human neutrophils has not been established. The A allele has been associated with decreased lung cancer risk in several (25, 26, 27, 28, 29), but not all (30, 31), studies, as well as with decreased risk of Alzheimer’s disease (32).

To date, there has been no report on the relation between MPO genotypes and breast cancer risk. In this study, we evaluated the association between the −463 MPO genotype and risk of breast cancer, and assessed potential modifying influences of fruit and vegetable intake and specific dietary antioxidants, as well as hormonal and reproductive risk factors, on risk relationships in the LIBCSP.

The LIBCSP, a population-based case–control study of breast cancer, has been described previously (33). In brief, the cases were English-speaking women over 20 years of age with newly diagnosed, primary in situ or invasive breast cancer who resided in Nassau and Suffolk Counties in Long Island, New York. Cases were ascertained between August 1, 1996, and July 31, 1997, using a rapid reporting network developed among the hospitals and physicians known to treat or diagnose breast cancer among Long Island women. Controls were identified with Waksberg’s method of random-digit dialing (RDD; ref. 34) for women under the age of 65 years, and by Health Care Finance Administration (HCFA) rosters for women who were 65 years or older, and frequency matched to the expected age distribution of case women by 5-year age groups. All respondents signed informed consent forms before the study interview.

On receiving physician and participant consent, 1,508 cases (82%) and 1,556 controls (63%) were interviewed in their homes by a trained interviewer. The questionnaire focused on known and suspected risk factors for breast cancer, including reproductive, hormonal, medical, and lifestyle histories. To assess diet for the 12 months before the interview, 98% of participants completed a self-administered modified Nation Cancer Institute-Block food frequency questionnaire, which has been previously validated (35). The food frequency questionnaire included 10 questions related to fruit intake, 2 for juices, and 21 for vegetables, in addition to questions about supplement use. Grams of total fruit and vegetable intake were calculated and units of carotenoids, vitamin C, and vitamin E were computed from food composition data as described previously (21).

Blood samples were drawn from 73% of all participants. DNA was extracted from buffy coat by standard RNase/proteinase K and phenol/chloroform treatment, and genotyping was performed by BioServe Biotechnologies (Laurel, MD) with Sequenom’s high-throughput matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry (36). The PCR was performed in a total volume of 5 μL, containing 10× Buffer B (Solis Biodyne), 0.5 μL; DNA (diluted to 2.5 ng/μL), 2 μL; primers (5′-ACGTTGGATGTCTTGGGCTGGTAGTGC-3′ and 5′-TGGATGTATTTTTAGTAGATACAGGGTTTCA-3′), at 2 pmol/μL and, 0.5 μL each; MgCl₂ (25 mmol/L, 0.5 μL); dNTPs (2.5 mmol/L each, 0.25 μL); Taq polymerase (Qiagen), 0.02 μL; and water, 0.73 μL. Controls for genotype and two non-template controls were included on each plate. In addition, 170 sets of blinded controls (8%) were distributed throughout the DNA samples for quality control purposes. Laboratory staff were blinded to case/control status.

MPO genotype data were available for 1,011 women with breast cancer and 1,067 population-based controls. As previously published (33), an increase in breast cancer among women on Long Island was found to be associated with lower parity, late age at first birth, little or no breastfeeding, a family history of breast cancer, and increasing income and education. Results were similar when the analyses were restricted to respondents who donated blood, or for those with DNA available for these analyses (33). Factors found to be associated with a decreased likelihood that a respondent, regardless of case–control status, would donate blood (33) included: increasing age (1% decrease for each yearly increase in age) and past active smoking (25% decrease); factors associated with an increased probability include Caucasian or other race (65% and 74% increase, respectively, versus African-American), alcohol use (28% increase), ever breastfed (47% increase), ever hormone replacement therapy use (63% increase), ever oral contraceptive use (21% increase), and ever had a mammogram (51% increase). Case–control status and fruit and vegetable consumption were not predictors of blood donation.

Among those with DNA available for these analyses, 94% of cases and 93% of controls were Caucasian; 4% of cases and 4% of controls were African-Americans. The racial distribution in cases and controls was not statistically different (P = 0.12). Age range of cases and controls were 25.1 to 98.1 years (mean, 58.7; median, 57.8), and 20.3 to 95.5 years (mean, 56.1; median, 55.6), respectively. Thirty-three percent of cases and 34% of controls were premenopausal women. Overall mean and SD of fruit and vegetable consumption were 31.2 servings per week (17.2) for cases, and 32.3 servings per week (18.2) for controls.

Distributions of putative risk factors in controls by MPO genotypes were assessed for differences with ANOVA. The measure of observer agreement of genotype between 8% of randomly selected duplicates that were included for quality control purposes was assessed with kappa statistic. Unconditional logistic regression (37) was used to calculate ORs and corresponding 95% CIs for breast cancer, in relation to genotype. Risk associated with the low-risk AA genotype, as well as with the heterozygous GA genotype, was computed in reference to the common GG genotype.

Multivariate models were simultaneously adjusted for age at reference date (defined as date of diagnosis for cases and date of identification for controls, continuous), family history of breast cancer (first degree relative: yes/no), and number of pregnancies (continuous). Factors found not to confound the associations of interest included the following: race, body-mass index at reference date (continuous), age at first pregnancy (centered, continuous), passive/active cigarette smoking (nonsmoker, residential passive smoking only, active smoking only, both active and passive smoking), age at menarche (continuous), hormone replacement therapy use (yes/no), fertility problem (yes/no), benign breast disease (yes/no), and lifetime alcohol intake (grams per day; 0, <15 g/d, 15–30 g/d, and >30 g/d). The final multivariate-adjusted models shown include those factors that either changed the estimated effect by 10% or more, or that remained in a best–fitting model, which was developed by starting with a full model and then excluding covariates that did not improve the overall fit, as measured by the −2log likelihood ratio test (37). Confounding was evaluated separately for premenopausal and postmenopausal women.

To explore whether the main effect of genotype was modified by hormonal and reproductive characteristics, data were stratified by parous versus nulliparous women, earlier versus later age at menarche (divided at the median values of controls), and hormone replacement therapy (yes/no) among postmenopausal women, with associations between MPO and risk examined within each stratum. For dietary factors, total fruit and vegetable consumption and specific dietary antioxidants were dichotomized at the median in controls to categorize women into high and low consumption on each factor, and effects of genotype on risk were evaluated within these high and low strata. For tests of interaction, genotypes were collapsed into categories: GG and GA+AA. To test statistical interactions on a multiplicative scale, a cross-product term of the ordinal score for each genotype and the risk factor variables (e.g., genotype × dietary antioxidants) was included in multivariate models. Relationships were evaluated in the entire data set, and separately by menopausal status. Because only a small proportion of women in the study were African-American (42 cases, 49 controls), we also restricted analyses to only Caucasian women to determine whether race altered risk relationships. All of the analyses were conducted with SAS version 8.2 (SAS Institute, Inc., Cary, NC).

The MPO A allele was present in 22 and 24% of the breast cancer cases and the population-based controls, respectively, with the allele frequencies within the range described previously for Caucasians (0.20–0.26), and genotype distribution was comparable with that observed in other published studies (25, 26, 27, 28, 29, 30, 31). The GG, GA, and AA genotypes were present in 59.2, 33.9, and 6.8% of controls, although distributions of MPO in the control group deviated significantly from Hardy-Weinberg equilibrium (P = 0.03). The frequencies of the GA and AA genotypes were lower in women with breast cancer. There were no significant differences between putative risk factors by MPO genotypes among control women, although the AA genotype frequency tended to be higher among Caucasians than among other races (data not shown).

The associations between MPO genotypes and breast cancer risk are shown in Table 1. Having at least one A allele (GA and AA genotypes) was associated with a 13%, nonsignificant reduction in risk of breast cancer, after adjusting for possible confounding factors. After stratification by menopausal status, premenopausal women with GA or AA genotypes had decreased risk of breast cancer in comparison with women with GG genotypes, with homozygotes for A alleles with a 43% reduction in risk (adjusted OR, 0.57; CI, 0.32–1.05) in comparison with women with GG genotypes. The association between genotype and breast cancer among postmenopausal women was null (Table 1). When we restricted these analyses to Caucasians only (953 cases, 989 controls), results were similar to those observed among the entire study population (adjusted OR, 0.86; and CI, 0.72–1.04, for GA and AA genotypes).

Because of the potential role for estrogens in regulating MPO, we also evaluated relationships between MPO and breast cancer risk within strata of hormonally related risk factors. Among postmenopausal women, there were no substantial differences in risk for breast cancer in relation to MPO among users and nonusers of hormone replacement therapy (HRT; adjusted OR, 0.72; CI, 0.49–1.05, for HRT users; adjusted OR, 1.00; CI, 0.76–1.32 for non-HRT users for GA and AA combined). There was some suggestion of an inverse association between MPO and risk among parous women (adjusted OR, 0.82; CI, 0.69–0.99), but not among nulliparous women, and among those with later age at menarche but not among those with earlier age at menarche (data not shown), although this association was not statistically significant. There were no differences observed by use of oral contraceptives among premenopausal women.

Consumption of fruits and vegetables clearly modified the associations between MPO genotype and breast cancer risk (Table 2), with a statistically significant interaction between MPO genotypes and fruit and vegetable consumption (P on a multiplicative scale = 0.04). When women were dichotomized at the median into lower and higher consumers of dietary fruits and vegetables and specific sources of dietary antioxidants, we observed that inverse associations were most pronounced among women who consumed higher amounts of fruit and vegetables and had A alleles (adjusted ORs for GA and AA genotype combined, 0.75; CI, 0.58–0.97). Inverse associations between GA or AA genotype combined and risk were not noted among women who had diets lower in fruits and vegetables (adjusted OR for GA and AA combined, 1.07; CI, 0.83–1.38). Similar findings were noted for fruits and vegetables separately and for sources of specific antioxidants, e.g., carotenoids and ascorbic acid, with the MPO polymorphism conferring lower risk primarily among women with diets high in these antioxidants. Similar results were noted when use of antioxidant supplements was included in computation of dietary intake of each vitamin (data not shown).

When data were stratified by menopausal status (Table 3), the modifying effects of diet on risk relationships were strongest among premenopausal women, with substantially reduced risk among women with A alleles who consumed higher amounts of total fruits and vegetables, although cell sizes were small and risk estimates somewhat unstable. Premenopausal women with high consumption of vegetables had an approximate 70% reduction in risk (adjusted OR for AA genotype, 0.29; CI, 0.11–0.77). The effects of A alleles were weaker among premenopausal women who had diets lower in fruits and vegetables. Conversely, among postmenopausal women, those homozygous for the AA alleles had a nonsignificant increased risk, particularly those with lower consumption of fruits and vegetables. Similar odds ratios were noted for sources of specific antioxidant vitamins, e.g., carotenoids, ascorbic acid, and vitamin E, with the MPO polymorphism conferring lower risk primarily among premenopausal women with diets high in these antioxidants.

In this large population-based study, we found that MPO genetic polymorphisms were associated with breast cancer risk. Having at least one variant A allele, resulting in reduced MPO transcription, was related to reduced risk of breast cancer, and risk reduction was greatest among women with higher consumption of fruits and vegetables and specific dietary antioxidants. Inverse associations between risk, MPO genotype, and fruit and vegetable consumption were most notable among premenopausal women. These findings could indicate that the effects of ROS are greater in a higher hormonal milieu, and that MPO expression may play a role in the complex relationships between oxidative stress, hormonal factors, and breast cancer risk.

These findings support our biologically plausible hypothesis, that MPO, a major enzyme involved in generating ROS, would be associated with breast cancer risk, and that women heterozygous or homozygous for the A allele, associated with reduced transcription of the gene, would have reduced risk of breast cancer. The −463 A allele in the promoter region has been shown to be associated with lower transcriptional activation than the common G allele (22, 23). In epidemiologic studies, five (25, 26, 27, 28, 29) of seven (25, 26, 27, 28, 29, 30, 31) case–control studies have found significantly reduced lung cancer risk associated with the A alleles, postulating an oxidative stress mechanism. A recent meta-analysis of lung cancer studies noted a 14% reduction in risk of lung cancer with the A allele (26), similar to the effect observed in our breast cancer study. Moreover, the most pronounced inverse associations in our study were observed among women who consumed higher amounts of fruits and vegetables and specific sources of dietary antioxidants, which is also consistent with the hypothesis that ROS are better neutralized in an environment high in antioxidants. These findings emphasize the importance of fruit and vegetable consumption in reduction of breast cancer risk.

In earlier findings regarding a main effect for fruit and vegetable consumption on breast cancer risk (21) in this study population, results were strongest among postmenopausal women. This is in contrast to our observation that MPO risk reduction was greatest among premenopausal women who consumed higher amounts of fruits and vegetables. The statistical interaction, on a multiplicative scale, between MPO, fruit and vegetable intake, and breast cancer risk may infer the importance of oxidative stress in breast cancer etiology, particularly in premenopausal women, with a higher estrogenic milieu.

There is support in the literature for differential findings for MPO genotypes and breast cancer risk among pre- and postmenopausal women, although the evidence comes from disparate, unrelated reports and is not yet conclusive. Nonetheless, there are numerous indications that steroid hormone levels affect MPO. For example, lachrymal fluid peroxidase activity is positively correlated with 17-β estradiol plasma level (38), and intracellular MPO activity in neutrophils was higher in premenopausal women than postmenopausal women (19). In addition, associations between MPO and several diseases have been shown to be gender dependent (32, 39), which could be attributed to the possible effects of sex hormones on MPO gene expression. Rutgers et al. (40) demonstrated that genotype-dependent differences in MPO activity were observed only in women. Most importantly, Porter et al. (41) showed that Sp1, which is a transcription factor binding preferentially to the G allele, can enhance binding by the estrogen receptor to a nearby halfsite, suggesting that Sp1 might facilitate estrogen receptor binding to the G allele moreso than to the A allele (23). Norris et al. has suggested that the Alu-encoded hormone response element (AluHRE), including the −463 polymorphic site, can serve as an estrogen response element (42). On the basis of this evidence, Reynolds et al. (32) proposed that there may be differential regulation of MPO gene expression by estrogen by genotype. MPO expression from the G allele might be most notable among women in a high estrogen environment, such as before menopause, but additional studies of the mechanisms of MPO regulation are very much needed. Several epidemiologic studies have also noted differential relationships by menopausal status in studies related to genes involved in oxidative stress (43, 44), although the underlying biological mechanisms remain to be elucidated.

Results from these analyses could be affected by sources of bias that are common to case–control studies (e.g., selection bias or recall bias), or to misclassification or selection bias related to genotyping. However, to affect these results, problems with recalling the details of past exposures would have to be associated with genotyping status, which is not likely. Distribution of AA genotypes differed somewhat between premenopausal and postmenopausal control women (i.e., premenopausal women, 9.7%, and postmenopausal women, 5.4%). Response rates to blood collection did vary by age (33), with older women having a lower response rate than younger women. However, when we compared AA genotype frequency by 10-year age groups among controls (data not shown), we did not observe differences in risk relationships by age. Thus, we cannot completely rule out the possibility of selection bias, although sampling variation may be a factor as well. Distributions of MPO in the control group also deviated significantly from Hardy–Weinberg equilibrium, which may inflate Type I error (45). However, it does not appear that this is due to errors in genotyping for a number of reasons; genotyping was performed with MALDI-TOF technology, which is a highly accurate genotyping method (36, 46); there was an excellent measure of observer agreement in the 8% of randomly selected duplicates that were included for quality control purposes (kappa statistic of 0.91), and there was <1% failure rate of the assay (7 of 1,098 controls). Most importantly, the distributions of genotypes and allele frequencies for MPO in our study are similar to those observed in other studies (25, 26, 27, 28, 29, 30, 31, 47), with several of the previously published MPO genotype studies also not in Hardy–Weinberg equilibrium among controls (25, 28, 47), indicating that factors other than error may be responsible for the departure from Hardy–Weinberg equilibrium in our and other studies. Finally, when we recalculated the ORs assuming controls were under Hardy–Weinberg equilibrium (with the expected frequency for control group), the crude OR remained approximately the same (OR for GA and AA genotype combined, 0.85; CI, 0.71–1.01).

In summary, genotypes associated with reduced MPO expression appear to be associated with reduced breast cancer risk, and our data indicate that consumption of fruits and vegetables may further decrease risk relationships between MPO genotypes and breast cancer. To our knowledge, this is the first study to evaluate MPO genotypes and breast cancer risk, and is based on data from a large population-based case–control study with adequate statistical power and in-depth interview assessments. Although there is a need for biochemical studies to evaluate underlying mechanisms, particularly relationships between MPO genotypes, oxidative stress, and estrogens, these data support a link between oxidative stress and breast cancer. Although genotype may not be changed, it is encouraging to note that the inverse associations with the MPO polymorphism were observed primarily among women who consumed high amounts of fruits and vegetables and other sources of dietary antioxidants. If these findings are replicated, they will not only further elucidate breast cancer etiology, but could also reinforce public health recommendations for the consumption of diets rich in fruits, vegetables, and other sources of antioxidants as a means of cancer prevention.

For their valuable contributions to the Long Island Breast Cancer Study Project the authors thank: members of the Long Island Breast Cancer Network; the 31 participating institutions on Long Island and in New York City, NY; our NIH collaborators, Gwen Collman, Ph.D., National Institutes of Environmental Health Sciences; G. Iris Obrams, M.D., Ph.D. formerly of the National Cancer Institute; members of the External Advisory Committee to the population-based case–control study: Leslie Bernstein, Ph.D., (Committee chair); Gerald Akland, M.S.; Barbara Balaban, MSW; Blake Cady, M.D.; Dale Sandler, Ph.D.; Roy Shore, Ph.D.; and Gerald Wogan, Ph.D.; as well as other collaborators who assisted with various aspects of our data collection efforts including Gail Garbowski, MPH; Mary S. Wolff, Ph.D.; Steven D. Stellman, Ph.D.; Maureen Hatch, Ph.D.; Geoffrey Kabat, Ph.D.; Jan Beyea, Ph.D.; Bruce Levin, Ph.D.; H. Leon Bradlow, Ph.D.; David Camann, B.S.; Martin Trent, B.S.; Ruby Senie, Ph.D.; Carla Maffeo, Ph.D.; Pat Montalvan; Gertrud Berkowitz, Ph.D.; Margaret Kemeny, M.D.; Mark Citron, M.D.; Freya Schnabel, M.D.; Allen Schuss, M.D.; Steven Hajdu, M.D.; and Vincent Vinceguerra, M.D. Finally, the authors thank Cutberto Garza, M.D., Ph.D., Cornell University, and Susan E. McCann, Ph.D., Roswell Park Cancer Institute, for their insightful input.