Gene-Specific Promoter Methylation Status in Hormone-Receptor-Positive Breast Cancer Associates with Postmenopausal Body Size and Recreational Physical Activity

C l i n M e d International Library Citation: McCullough LE, Chen J, White AJ, Xu X, Cho YH, et al., (2015) Gene-Specific Promoter Methylation Status in Hormone-Receptor-Positive Breast Cancer Associates with Postmenopausal Body Size and Recreational Physical Activity. Int J Cancer Clin Res 2:013. doi.org/10.23937/2378-3419/2/1/1013 Received: January 24, 2015: Accepted: February 06, 2015: Published: February 09, 2015 Copyright: © 2015 McCullough LE. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. McCullough et al. Int J Cancer Clin Res 2015, 2:1


Introduction
Breast cancer remains the leading cause of cancer-related illness in the United States (US), and may be influenced by a number of environmental [1], reproductive and lifestyle [2] factors.There is abundant research showing that elevated body mass and physical inactivity are associated with increased risk of postmenopausal breast cancer [3,4], but the mechanisms driving these associations are unresolved [5].Given the large proportion of women who are inactive in the US [6] and the steadily increasing rates of obesity [7], understanding the underlying mechanism for the observed association between these lifestyle factors and breast carcinogenesis is of paramount importance.
DNA methylation is an epigenetic alteration that can modify gene expression [8] and is known to been related to breast carcinogenesis [9,10].Specifically, hypermethylation of tumor suppressor genes has been associated with clinical/pathological factors for breast cancer, as well as mortality in our study population [11].Some investigators have hypothesized that elevated body mass and/or physical inactivity may affect DNA methylation through increased estrogen [12,13] and chronic inflammation [14,15]; but to date, only three studies have examined associations between body mass and gene-specific methylation in breast tumors [16][17][18].This previous research was limited by examining a very small (<5) subset of genes [16,17] and some studies did not stratify by menopausal status [17,18].No previous study has considered associations between physical activity and gene methylation of breast tumors.
The goals of our study were two-fold.First, we aimed to assess the potential association between body mass index (BMI) or recreational physical activity (RPA) in relation to promoter methylation status, assessed in a panel of 13 breast cancer-related genes measured in tumor tissue (APC, BRCA1, CCND2, CDH1, DAPK1, ESR1, GSTP1, HIN1, CDKN2A, PGR, RARβ, RASSF1A and TWIST1).These genes may play an important role in breast carcinogenesis and their promoter regions have been frequently methylated in breast tumors [19].Second, we explored whether associations between BMI/RPA and breast cancer subtypes, defined by estrogen and progesterone receptor (ER/PR) status, were modified by gene promoter methylation.
We hypothesized that: (1) breast tumors from postmenopausal women with elevated body size/physical inactivity would have a greater prevalence of methylation than tumors from postmenopausal women with lower body mass/high physical activity; and (2) elevated body size/physical inactivity would differentially associate with ER+PR+ breast cancer when we also consider the gene-promoter methylation status of the tumor (methylated vs. unmethylated).

Materials and Methods
We utilized case-only resources from the case-control component of the Long Island Breast Cancer Study Project (LIBCSP), a population-based study.Details of the parent study have been reported previously [20].Institutional Review Board approval was obtained by all participating institutions.

Study population
Case women were English-speaking female residents of Nassau and Suffolk counties, Long Island, New York (NY), newly diagnosed with a first primary breast cancer between August 1, 1996 and July 31, 1997.Participants were identified using rapid case ascertainment via daily or weekly contact with pathology departments of all 28 hospitals on Long Island, and three tertiary care hospitals in New York City.At diagnosis, participants were aged 20-98 years and 67% were postmenopausal.Approximately 94% of study participants selfreported their race as white, 4% as black, and 2% as other, which was consistent with the underlying racial/ethnic distribution in these two NY counties at the time of data collection.

Data collection
Interviews were completed for 82.1% (n=1508) of eligible cases, and occurred within 3 months of diagnosis (before initiation of chemotherapy) for most case participants [20].Tumor tissue was excised prior to the initiation of chemotherapy or radiation for all case participants.Written informed consent was obtained from all study participants prior to the study interview.
For LIBCSP cases, study investigators obtained archived pathology blocks for the first primary breast cancer from the 31 hospitals in Long Island and adjacent areas.Tumor blocks were successfully retrieved for 962 women [21] and tumor tissue from 532 postmenopausal participants were available for this study.Cases with tumor blocks available for methylation analysis (vs.those without tumor tissue available) were more likely be older (mean age 59.6 vs. 57.9years), postmenopausal (70.7% vs. 64.6%),and have an invasive tumor (87.8% vs. 80.1%).Other demographic and clinical/ pathological characteristics were similar between the two groups [19].

Body size, physical activity and covariate assessment
Body size and physical activity were assessed as part of the interviewer-administered structured questionnaire that was completed shortly after diagnosis.BMI in the year prior to diagnosis was calculated for each participant based on the following formula: weight (kg)/height (m 2 ).RPA was assessed using a modified instrument developed by Bernstein and colleagues [22].RPA from menopause to reference date was used to estimate postmenopausal RPA as previously described [23] and defined as inactive, low RPA (≤9.23 hrs/wk) and high RPA (>9.23 hrs/wk) based on the control median.
During the interview participants were also asked about their demographic characteristics; reproductive, environmental, and medical histories (including family history of breast cancer); cigarette smoking and alcohol use; and use of exogenous hormones.Menopausal status was derived using information on the last menstrual period and gynecologic surgeries, combined with data on pregnancy, lactation, and use of hormone replacement therapy as previously described [24].

Gene-specific promoter DNA methylation assessment
DNA extraction from the archived tumor tissue was performed as previously described [25].For methylation analysis, a panel of 13 genes known to be involved in breast carcinogenesis was selected.Promoter methylation of ESR1, PGR and BRCA1 was determined by methylation-specific (MSP)-PCR as described previously [25,26].The Methy Light assay was used for determining the methylation status of the remaining genes [27,28].The percentage of methylation was calculated by the 2 -ΔΔCT method, where ΔΔC T = (C T,Target -C T,Actin ) sample -(C T,Target -C T,Actin ) fully methylated DNA [29] and multiplying by 100.The MSP-PCR assay for ESR1, PGR and BRCA1 promoter methylation generated dichotomous outcomes (i.e.methylated vs. unmethylated).Conversely, Methy Light assay yielded percentage of methylation for gene promoters that were subsequently dichotomized into methylated or unmethylated cases using a 4% cut-off as reported in previous literature [30].The numbers of assayed samples and corresponding methylation frequencies for the selected genes are summarized in Xu et al. [19].The main reason for missing methylation data was insufficient DNA, primarily due to small tumor size.

Hormone Receptor (HR) subtype assessment
We abstracted data recorded on the medical record to ascertain breast cancer subtype defined by HR status [20].ER/PR status of the first primary breast cancer was available from the medical record for 65.6% of cases (N=990), of which 67.7% (N=670) were postmenopausal and included in these analyses.

Statistical Methods
All statistical analyses were performed using SAS statistical software version 9.1 (SAS Institute, Cary, NC).
We previously reported the relationship between genepromoter methylation with demographic and clinical-pathological characteristics of the LIBCSP breast cancer cases by menopausal status [11,31].The study reported here focuses on: (1) whether BMI and/or RPA are associated with gene methylation in postmenopausal breast tumors; and (2) whether the association between BMI and/or RPA and ER/PR subtype is differential with respect to gene methylation status.To address these aims, we employed a case-case approach, and thus we relied solely upon data collected among postmenopausal case participants of the LIBCSP (n=532) [32].
To assess whether BMI or RPA was associated with genespecific promoter methylation levels measured in case tumor tissue, we used logistic regression [32] to estimate odds ratios (ORs),and corresponding 95% confidence intervals (CIs) with case groups characterized by tumor methylation status (methylated vs. unmethylated for each marker).With this approach the ORs estimate the likelihood of a case possessing a methylated gene-promoter given their body size/physical activity status.
To determine whether the association between BMI or RPA and ER/PR receptor status was differential with respect to gene-specific promoter methylation, we used logistic regression to estimate ORs (95% CIs) with case groups characterized by both gene methylation status (methylated vs. unmethylated) and ER/PR status (ER+PR+ vs. all others: ER-PR-, ER+PR-, ER-PR+).With this approach the ORs estimate the likelihood of an ER+PR+ case given both gene methylation and body size/physical activity status.If the sample size in any strata of BMI/RPA and gene promoter methylation was less than ≤ 5, the OR (95% CI) was not estimated.In addition to comparing ER+PR+ breast cancer cases to all others, we also considered the comparison of ER+PR+ cases (primarily Luminal A and B subtypes) to ER-PR-cases (exclusively HER2 and triple negative subtypes) to better understand of potential associations with intrinsic subtypes.
We formally assessed evidence for multiplicative interaction using a likelihood ratio test [33], comparing multivariable models with and without cross-product terms to represent the interaction between BMI or RPA and a gene-specific methylation marker (a priori α=0.05).A significant interaction indicates that the odds of a case possessing the ER+PR+ breast cancer subtype, given BMI (or RPA) level, are statistically different across strata of gene-specific methylation.
We identified potential confounders based on the known epidemiology of breast cancer and analysis of causal diagrams [34].For all models, potential confounders included: race (white/black/ other); family history of breast cancer (yes/no); and history of benign breast disease (yes/no).Confounders were added in the model if they their inclusion changed the exposure estimate >10% [35].None of the covariates assessed resulted in a >10% change in estimate, therefore only 5-year age group remained in our final case-case models.

Results
Associations between postmenopausal BMI and gene promoter methylation for the 13 breast cancer-related genes are shown in Table 1.Women with BMI 25-29.9kg/m 2 were more likely to have methylated HIN1 breast tumors (OR=1.57,95%CI: 1.03-2.39).Although we observed elevated likelihood of methylated HIN1 in breast tumors among women with BMI ≥ 30kg/m 2 , the estimate was less pronounced and included the null (OR=1.44,91% CI: 0.94-2.23).The remaining methylated gene promoters did not appear to be associated with postmenopausal BMI.We observed no differences in the likelihood of gene promoter methylation breast cancer in association with postmenopausal RPA for any of the 13 genes examined (Table 2).
We hypothesized that postmenopausal BMI or RPA may differentially associate with ER+PR+ breast cancer in strata of genepromoter methylation.We found that obesity was associated with ER+PR+ breast cancer among women with unmethylated ESR1 (OR=2.63;95% CI: 1.32-5.25)(Table 3); the corresponding OR among cases with methylated ESR1 was 1.24 (95% CI: 0.62-2.48)(multiplicative p for interaction=0.004).Similarly, we found that high RPA women with methylated GSTP1 were more likely to have ER+PR+ breast cancer (OR=2.33;95% CI: 0.79-6.84)than high RPA women with unmethylated GSTP1 (OR=1.05;95% CI: 0.53-2.10)(Table 4).We observed a multiplicative interaction (p=0.03) between GSTP1 promoter methylation, postmenopausal RPA and ER+PR+ breast cancer, but given the small proportion of women with methylated GSTP1 our estimates were imprecise.We were unable to estimate the ORs, due to the low prevalence tumor methylation, in several markers (e.g.CDH1, p16, PR and RASSF1A).With the remaining gene promoters that we considered, we identified no differential associations between BMI or RPA and ER+PR+ breast cancer.
The associations between postmenopausal BMI and breast cancer, defined by ESR1 methylation and estrogen-receptor status, were robust, and remained significant (p=0.019),when we compared ER+PR+breast cancer to ER-PR-breast cancer only (Supplemental Table 1).For postmenopausal RPA and GSTP1 methylation, however, our estimates were less robust, and were of borderline statistical significance (p=0.068) when comparing ER+PR+ breast cancer to ER-PR-breast cancer (Supplemental Table 2).

Discussion
In this population-based study, we found that women with postmenopausal BMI 25-29.9kg/m 2 , and perhaps BMI ≥30kg/m 2 , were more likely to have methylated HIN1 breast cancer.We also observed a two-fold increase in the likelihood of ER+PR+ breast cancer among postmenopausal obese women with unmethylated ESR1 tumors and among postmenopausal highly active women with methylated GSTP1 tumors.Our findings are biologically plausible, as discussed below.4: Age-adjusted odds ratios (ORs) and 95% confidence intervals (CIs) for the association between postmenopausal recreational physical activity (RPA) and ER+PR+ breast cancer (vs.all others: ER-PR-, ER+PR-, ER-PR+) considering gene-specific methylation status of the tumor (methylated vs. unmethylated), the Long Island Breast Cancer Study Project (1996-1997).
Inactivation of tumor suppressor genes by promoter hypermethylation is a common epigenetic alteration in breast carcinogenesis [36,37].These alterations are known to occur more frequently in breast tumor tissue than adjacent nonmalignant tissue [36,37] and have been associated with the clinicopathologic parameters of breast cancer [10].Gene-promoter hypermethylation may therefore be an important event in breast carcinogenesis.
Increased BMI and physical inactivity are risk factors for postmenopausal breast cancer [4,38], and their influence on endogenous estrogens are well-documented [39,40].In vivo and in vitro data suggest estrogen may induce aberrant DNA methylation, altering several genes implicated in breast carcinogenesis [41,42].Specifically, estrogens were reported to induce promoter hypermethylation of CDH1 and CDKN2A in non-malignant breast cells of humans [43].In addition to increased levels of estrogen, central adiposity has been associated with chronic low-grade inflammation [44].Several studies have shown greater frequency of promoter methylation in CDKN2A, CDH1, BRCA1, and MLH1 among patients with chronic inflammatory disease compared with patients without [14,15].Moreover, clinical data indicate that weight loss (≥ 5% initial body weight) was associated with significantly lower promoter methylation of TNF-α in peripheral blood [45].Physical activity has similarly been found to reduce levels of pro-inflammatory markers [46].
Hormonal and inflammatory mediators have the capacity to induce and maintain promoter methylation facilitating the growth and survival of tumors, but to our knowledge, few studies have examined associations between body size and methylation status of breast tumors [16][17][18].Consistent with our findings, Tao and colleagues [16] observed no association between body size and methylation of CDH1, CDKN2A, and RAR-β2 among postmenopausal case women; associations by ER/PR status were not reported.Naushad and colleagues [17] examined the association between BMI and methylation of Ec-SOD, RASSF1, BRCA1, and BNIP3.BMI was significantly positively associated with Ec-SOD, RASSF1 and BRCA1 methylation but inversely associated with BNIP3.Most recently, Hair and colleagues [18] reported significant associations between BMI and methylation of 2 loci among all breast tumors and 21 loci specific to ER+ tumors, but did not examine menopause-specific associations.The association between body size and breast cancer risk is known to vary by menopausal subgroups [47].It is therefore likely that obesityassociated methylation sites also differ by menopausal status.While we employed a biologically driven candidate gene approach, our study improves on the prior research by including a larger number of candidate genes, exploring associations by ER/PR status, and focusing on postmenopausal women.Further, it is the first study to consider the association between physical activity and gene methylation in postmenopausal breast tumors.
In our findings reported here, elevated postmenopausal BMI more strongly associated with ER+PR+ breast cancer among women with unmethylated ESR1.The ER protein is coded for by ESR1 and gene silencing of ESR1 by DNA methylation is often observed in breast tissues that do not express ER (e.g.ER-) [48].Estrogens have long been hypothesized to underlie the positive association between obesity and postmenopausal breast cancer risk [39].Our observation of stronger and more precise associations between postmenopausal obesity and ER+PR+ breast cancer among women where ESR1 is active (unmethylated) is biologically reasonable and suggests that methylation-mediated silencing of the ESR1 gene may alleviate the role of obesity-related estrogen in postmenopausal breast carcinogenesis.
We similarly found that the odds of being an ER+PR+ breast cancer case was enhanced among women engaging in high postmenopausal RPA in the presence of GSTP1 methylation.GSTP1 is involved in a wide range of detoxification reactions which protect cells from carcinogens [49].The 5' region of GSTP1 is rich in CpG islands and its methylation has been associated with loss of GSTP1 expression [50], breast carcinogenesis [51] and ER+PR+ case status [52].The immediate systemic response to physical activity is an increase in reactive oxygen species production; it is therefore biologically plausible that reduced GSTP1 expression via DNA methylation may enhance risk of breast cancer, specifically ER+PR+ breast cancer.
Strengths of our epidemiologic study include: (1) our novel examination of the potential role of physical activity, as well as obesity, in the association between tumor methylation and breast cancer; (2) restricting eligibility to postmenopausal breast cancer, where the associations with obesity and physical activity are most pronounced; (3) our population-based design, which enhances generalizability and facilitates quantification of any study bias due to subject selection; (4) relatively large sample size, which facilitates examining subgroup associations as we did here; (5) detailed exposure assessment of our anthropometic measures, which reduces the likelihood of random measurement error; (6) our case-case approach, which substantially reduces the likelihood of recall bias given that both the "case" group and our "comparison" group had breast cancer (and it is highly unlikely that misreporting of anthropometric-related information is differential with respect to methylation or HR status [53]); and (7) we only considered associations for which we had a priori strong biologic rationale, mitigating concerns regarding multiple comparisons.
There are also several limitations to consider when examining methylation in tumors in an epidemiologic study.First, we  (1996)(1997).
were unable to obtain archived tumor tissue for all LIBCSP case participants, which may result in selection bias; however, we were able to identify and consider potential sources of this error.Second, we were underpowered to explore potential variation by intrinsic subtype (Luminal A, Luminal B, HER2 and triple negative) given our study population primarily consisted of postmenopausal white women with low proportion of HER2-tumors.Third, genepromoter methylation analyses were constrained by sample size for several of the genes we considered, and thus future studies should consider enlarging study enrollment.Fourth, we had a limited panel of 13 biologically relevant genes for analyses.Although this is four times that of the one previous investigation focused on obesity, gene methylation and postmenopausal breast cancer [16], we were unable to explore all the mechanistic pathways that may be involved in this association.Finally, classification of methylation status is not universally defined and our cutoff of 4% may not be biologically relevant for all the genes assessed.
In summary, using data from a large population-based sample, we found that BMI may associate with HIN1 methylation status of postmenopausal breast tumor tissue.Notably, we also observed that both postmenopausal body size and physical activity may increase the likelihood of ER+PR+ breast cancer (which is the most common subtype diagnosed among American women [54]) in the absence and presence of ESR1 and GSTP1 methylation, respectively.While our results require confirmation in larger studies of postmenopausal women with greater number of genes, they suggest that DNA methylation may play an important role in understanding mechanisms underlying the associations between body size, physical activity and postmenopausal breast cancer.Given the plasticity of epigenetic marks in response to cancer-related exposures, additional research is needed to clarify these mechanisms and identify specific changes likely to be involved in the pathogenesis of breast cancer.

Table 1 :
Age-adjusted odds ratios (ORs) and 95% confidence intervals (CIs) for the association between postmenopausal body mass index (BMI) and breast cancer, as defined by gene-specific promoter methylation (comparing methylated vs. unmethylated cases), Long Island Breast Cancer Study Project(1996)(1997).

Table 2 :
Age-adjusted odds ratios (ORs) and 95% confidence intervals (CIs) for the association between postmenopausal recreational physical activity (RPA) and breast cancer, defined by tumor gene-specific promoter methylation (comparing methylated vs. unmethylated cases), in the Long Island Breast Cancer Study Project(1996)(1997).
a Low RPA ≤ 9.23 hours/week, High RPA > 9.23 hours/week Table

Table 2 :
Age-adjusted odds ratios (ORs) and 95% confidence intervals (CIs) for the associations between postmenopausal recreational physical activity (RPA) and ER+PR+ breast cancer (vs.ER-PR-breast cancer) considering the GSTP1 methylation status of the tumor (methylated vs. unmethylated), the Long Island Breast Cancer Study Project