Hypermethylation is an epigenetic change that blocks the promoter region of a gene and results in gene silencing. In breast cancer, tumor-related genes may be silenced by hypermethylation; many hypermethylated genes have been reported, and silencing of these genes plays important roles in carcinogenesis and tumor progression 123. Identification of epigenetic changes and their correlation with other clinical factors could lead to improvements in cancer diagnosis and treatment.

In patients with breast cancer, estrogen receptor (ER) status is an important treatment and prognostic factor. Breast cancer patients with ER-positive tumors generally have a more favorable prognosis than do those who have ER-negative tumors. These breast cancer patients can be treated successfully with hormonal therapies such as tamoxifen and aromatase inhibitors 456789. Epidemiological studies revealed that patients with ER-positive tumors have risk profiles different from patients with ER-negative tumors. Parity and timing of births were inversely associated with ER-positive tumors, but not with ER-negative tumors, and body mass index after menopause was more strongly associated with ER-positive than with ER-negative tumors 10. In addition, rates of ER-positive breast cancer incidence have been shown to increase after age 50 to 54 years, whereas the rates of ER-negative breast cancer incidence do not 11.

Similarly, differences in the gene expression patterns of ER-positive and ER-negative tumors have been documented in microarray expression studies, which identified profile differences in breast tumor subtypes 1213. These findings suggest that ER status of breast cancer represents distinct phenotypes. However, few studies have determined epigenetic changes in tumor-related genes in relation to ER status in matched-paired breast cancers.

To investigate the epigenetic differences between ER-positive and ER-negative breast cancer, we assessed the methylation frequency of several breast tumor-related genes that are known to undergo hypermethylated changes in breast cancer, and that play important roles in tumorigenesis and cancer progression. The objective of the study was to determine the association between ER status and epigenetic changes in these tumor-related genes.

The genes assessed were as follows: RASSF1A (RAS association domain family 1A; location: 3p21.3; GenBank: AF132675), CCND2 (cyclin D2; location: 12p13; GenBank: AF518005), GSTP1 (glutathione S-transferase P1; location: 11q13; GenBank: U12472), TWIST (human basic helix-loop-helix DNA binding protein; location: 7p21.2; GenBank: U80998), APC (adenomatous polyposis coli; location: 5q21-q22; GenBank: M74088), NES1 (normal epithelial cell-specific 1 or kallikrein 10; location: 19q13.3-q13.4; GenBank: AF024605), RARβ2 (retinoic acid receptor-β2; location: 3p24; GenBank: X07282), and CDH1 (E-cadherin; GenBank: L08599). RASSF1A is a putative tumor-suppressor gene that is frequently inactivated epigenetically rather than in a mutational event 14. A direct correlation between promoter region methylation and loss of expression has been shown in many tumor cell lines, including breast cancer 15161718. RASSF1A can exert a tumor-suppressing effect by blocking oncogene-mediated c-Jun amino-terminal kinase activation 19. It also associates with microtubules and contributes to the maintenance of genomic stability 20. Loss of CCND2 expression caused by methylation is an early event in breast cancer tumorigenesis 21. Methylation of CCND2 has been correlated with poor prognosis, implying that CCND2 has a tumor-suppressor function 22. GSTP1 is one of a family of enzymes that detoxifies hydrophobic electrophiles, and may be part of a protection system from environmental or dietary carcinogens 23. Our group has previously found that GSTP1 methylation correlates with increased tumor size and increased likelihood of sentinel lymph node metastases 24. TWIST induces E-cadherin mediated cell-cell adhesion and induction of cell motility. Increased expression of TWIST correlates with tumor invasion and metastasis 2526. APC gene germline mutations have been shown to be associated with familial adenomatous polyposis. Hypermethylation of the APC promoter is also associated with breast cancer, especially lobular breast cancer 272829. NES1 is a putative tumor suppressor gene found to be downregulated in breast cancer 3031. RARβ2 is postulated to be a tumor suppressor gene. RARβ2 methylation correlates with breast cancer metastasis 32. It is through retinoic acid receptors that retinoids can prevent primary tumor progression 33. CDH1 expression reduction is regarded as one of the main molecular events involved in dysfunction of the cell-cell adhesion system, triggering cancer invasion and metastasis. Mahler-Araujo and coworkers 34 reported a correlation between negative or reduced CDH1 expression and lack of ER expression in tumors from 245 breast cancer patients.

This study was conducted to investigate epigenetic differences in specific tumor-related genes between ER-positive and ER-negative breast cancers. We hypothesized that ER-positive breast tumors have different epigenetic profiles of tumor-related genes during early stages of cancer progression. We examined the methylation status of eight genes suspected of being involved in regulation of breast cancer, and investigated the methylation status of those genes at different stages of tumor development. We compared the methylation status of these genes between ER-positive and ER-negative breast tumors in early and advanced stages to investigate whether epigenetic changes occur in early stages of primary tumor progression.

Human epidermal growth factor receptor (HER)2/neu is an important factor for treatment and prognosis. HER2/neu over-expression occurs in 15% to 25% of breast tumors and is associated with poor prognosis and resistance to hormonal therapy 353637. HER2/neu and ER expression have been reported to exhibit an inverse correlation 93839. Furthermore, previous reports have demonstrated the effect of estrogen on downregulation of HER2/neu production 40. We investigated the epigenetic differences between HER2/neu-positive and HER2/neu-negative breast tumors relative to ER status and further identified the epigenetic characteristics of ER-negative, HER2/neu-negative (double-negative) breast tumors.

During the past several years, new molecular biomarkers have been discovered that are important targets for the diagnosis and therapy of breast cancer 342. ER and HER2/neu are important prognostic biomarkers and therapeutic targets in primary breast cancer. ER-negative tumors appear to be more malignant 47843, resulting in a poorer prognosis than with ER-positive tumors 5944. The present study was conducted to identify differences in epigenetic events related to ER expression by infiltrating breast cancer. To date, few studies have rigorously assessed matched paired ER-negative and ER-positive primary breast tumors for epigenetic differences. We focused on the epigenetic differences between ER-positive and ER-negative breast cancers, and used tumor specimen pairs matched for patient age, size, nodal status, and presence or absence of distant metastases. This sampling enabled rigorous analysis, and the results imply that epigenetic features of ER-positive tumors are different from those of ER-negative tumors.

Widschwendter and coworkers 3 demonstrated that methylation of APC correlated with ER positivity. Our data are consistent with this previous report. Furthermore, we demonstrated a significant difference in methylation status of RASSF1A, CCND2, GSTP1, TWIST, and APC between the ER-positive and ER-negative groups. In contrast, Li and colleagues 45 reported that ER-positive patients exhibited a higher frequency of TWIST methylation and a lower frequency of CDH1 methylation than did ER-negative patients. They also found no significant differences in the methylation frequencies of RARβ2, CCND2, and CDH1 between ER-positive and ER-negative groups. The reason for the dissimilarity in study findings may be due to differences in methylation analysis; Li and colleagues assessed methylated PCR products by gel electrophoresis, which is more subjective and less sensitive than capillary array electrophoresis analysis of methylated PCR products. The discrepancy may also have resulted from differences in the approach to the particular specimens assessed. Because methylation and ER status change with tumor progression 25, care should be taken when sampling ER-positive and ER-negative tumors to evaluate epigenetic changes and clinical associations.

To clarify when differences in methylation status between ER-positive and -negative tumors occur, we compared differences in methylation status between T1c and T2 stage subgroups. The ER-positive group exhibited significantly more frequent hypermethylation of two genes (RASSF1A and CCND2), independent of T stage. Moreover, the ratio of methylation frequency does not differ between T1c and T2/T3 stage subgroups. This observation indicates that the differences in methylation patterns do not significantly change when breast tumors progress from T1c to T2/T3 stage. Similarly, no difference in the methylation frequency ratio was detected between LN metastasis negative and positive tumors. Previously, we reported that GSTP1 hypermethylation correlates with LN metastasis 24. Similarly in this study, only GSTP1 hypermethylation was found to be more frequent in the LN metastasis positive group than in the negative group. In both LN metastasis positive and negative groups, GSTP1 hypermethylation was found to be more common in ER-positive than in ER-negative tumors. Furthermore, the differences in methylation status of RASSF1A and CCND2 between the ER-positive and ER-negative groups can be recognized in early stages of cancer, such as the T1c or N0 stages. These findings suggest that ER expression may influence epigenetic changes in early stages of breast cancer.

HER2/neu is another important prognostic marker for breast cancer. HER2/neu gene over-expression is identified in 15% to 25% of invasive breast carcinomas, and is related to metastatic potential and poor survival 3546. In our study, HER2/neu over-expression was identified in 27% of breast cancers, and was independent of ER status. A negative relation between ER status and HER2/neu over-expression has been documented by others 93839. One of the plausible explanations for the lack of difference found in the frequency of HER2/neu-positive tumors between the ER-negative and ER-positive groups is that patients included in the present study were relatively young (average age 51 years). According to Huang and coworkers 38, in women younger than 45 years the inverse association between ER and HER2/neu was not apparent. Of our sample population, 37% of patients were under 45 years old, and this age distribution may explain why we could not detect an inverse relation between ER and HER2/neu status. Regarding gene methylation and HER2/neu status, Fiegl and colleagues 47 reported that methylation of CDH13, MYOD1, PGR, and HSD17B4 exhibited a positive association with HER2/neu expression. We demonstrated that APC, GSTP1, and RASSF1A are more frequently methylated in the HER2/neu over-expressed group. These findings indicate that there are epigenetic differences between HER2/neu breast tumors, and logistic regression analysis showed that these differences are independent of ER status.

Clinically, HER2/neu-negative, ER-negative, and PR-negative breast cancers are referred to as triple-negative breast cancers. About 85% of triple-negative breast tumors phenotypes are deemed to be basal-like subtypes, and they are associated with poor clinical outcome 48. Because the progesterone receptor status of our samples was not well defined, we compared the methylation status of ER-negative and HER2/neu-negative breast tumors (double-negative) with breast cancers expressing either ER or HER2/neu. Double-negative breast tumors exhibited frequent hypermethylation in APC, GSTP1, RASSF1A, and TWIST, revealing the presence of epigenetic differences between double-negative breast tumors and breast cancers expressing either HER2/neu or ER. Furthermore, both ER and HER2/neu status contributed independently to the difference in expression of APC, GSTP1, and RASSF1A. There is evidence that gene expression heterogeneity occurs in basal-like tumors of triple negatives; these subgroups are present with different pathology and clinical properties. Triple negative and ER and HER2/neu-positive status and gene expression levels of tumors may be quite heterogeneous in tumor populations, thus contributing to different subclassifications. Other additional biomarkers may be needed to subclassify phenotypes better 4950.

Another interesting finding is the correlation of methylation status between the tumor-related genes. Methylation status of RASSF1A, CCND2, GSTP1, TWIST, and APC, which was significantly higher in the ER-positive group, was correlated with the methylation status of two or more of the four other genes. Nonrandom distribution of methylation in tumor-related genes has previously been reported; Nass and coworkers 51 and Li and colleagues 45 found coincident methylation of CDH1 and ESR1, and Parrella and coworkers 52 reported that ESR1 promoter hypermethylation status correlates with those of CDH1, GSTP1, CCND2, and TRb1. Our findings revealed that occurrence of promoter region hypermethylation of RASSF1A, CCND2, GSPT1, TWIST, and APC are associated, whereas CDH1 methylation exhibited no correlation with methylation of those five genes (data not shown). This apparent nonrandom distribution of promoter hypermethylation of some genes suggests the existence of specific factors causing selective promoter region hypermethylation of tumor-related genes.

In this study, we demonstrated – for the first time – that epigenetic differences between ER-positive and ER-negative breast tumors arise early in cancer development and persist during cancer progression. Furthermore, we reported epigenetic differences between HER2/neu-positive and HER2/neu-negative breast tumors, and between double-negative breast tumors and breast tumors expressing either HER2/neu or ER. Taking advantage of potential reversibility of DNA methylation, epigenetic therapies directed against various cancers have been in development; 5-azacytidine, 5-aza-2'-deoxycytidines, procainamide, and hydralazine are promising agents. Investigators are attempting to combine epigenetic therapy with other standard therapies 253. For breast cancer, one approach may to combine hormone therapy and antimethylation treatment. Integrated information regarding clinical factors influencing therapeutic principles and epigenetic features, such as promoter hypermethylation of tumor-related genes, will be important for combining epigenetics targeting drugs and standard chemotherapy.

APC, adenomatous polyposis coli; CCND2, cyclin D2_; CDH1, E-cadherin; ER, estrogen receptor; GSTP1, glutathione S-transferase P1; HER2/neu, human epidermal growth factor receptor 2; LN, lymph node; MSP, methylation-specific PCR; NES1, normal epithelial cell-specific 1 or kallikrein 10; RARβ2, retinoic acid receptor- β2; RASSF1A, RAS association domain family 1A; SLN, sentinel lymph node; TWIST, human basic helix-loop-helix DNA binding protein.

The authors have submitted a patent application relating to the content of this manuscript.

ES and MS performed methylation assays for the studies. M-SS was the biostatistician in the study. SLN performed assays and assembled the data. A-TV helped with assays. AEG performed the clinical review of the patients and reviewed the study design. DSNH was the principle investigator and designed and reviewed the study.