Purpose: To test the hypothesis that polymorphisms of ESR1, the gene encoding estrogen receptor α (ERα), are associated with susceptibility, clinical phenotype, and progression of breast cancer.

Patients and Methods: A case-control study was done on 940 patients with incident breast cancer and 1,547 healthy female controls. Fifteen single-nucleotide polymorphisms (SNP) selected from chr6:152,170,379-152,466,100 (exons 1–8 of the ESR1 gene, excluding flanking sequences), reflecting major polymorphisms of this gene, were genotyped. Frequencies of SNPs were compared between cases and controls to identify SNPs associated with cancer susceptibility and between cases with different clinical phenotypes to determine the role of ESR1 polymorphism in cancer progression.

Results: SNPs located in one cluster in intron 1 and one haplotype, based on these SNPs, showed a significant association with breast cancer susceptibility. The tumorigenic contribution of these intron 1 SNPs was more obvious in combination with reproductive risk factors (P for interaction <0.05). One of these intron 1 SNPs was also significantly associated with low ERα expression in tumors. Interestingly, the same intron 1 SNPs showed a correlation with worse clinical phenotypes, including poor differentiation of tumor cells and a late stage. These intron 1 SNPs also showed a significant association with the 5-year breast cancer–specific survival rate of patients, but had opposite effects in ERα-negative and ERα-positive early-stage patients.

Conclusions: Our findings provide support for diverse roles of ESR1 polymorphism in determining susceptibility in different stages of breast cancer. The differences between the important ESR1 SNPs identified in Chinese women in this study and those identified in studies on Western women with breast cancer suggest different roles of ERα in these two populations. Clin Cancer Res; 16(13); 3473–84. ©2010 AACR.

Translational Relevance

Genetic analysis of cancer-associated genes allows for the better assessment of cancer and provides framework to improve risk classification, outcome prediction, and development of targeted therapies. Given the well-documented role of estrogen receptor α (ERα) in breast cancer development and progression, this study explored whether polymorphisms of ESR1, the gene encoding ERα, are associated with susceptibility, clinical phenotype, and progression of breast cancer. Interestingly, diverse associations between ESR1 polymorphism and breast cancer development and progression were identified. Three polymorphisms of ESR1 showed an association with a decreased breast cancer risk and, coincidentally, with low ERα expression in tumors. However, the same polymorphism showed a significant correlation with worse clinical phenotypes, including poor differentiation of tumor cells and a late stage. Furthermore, these polymorphisms showed an association with the 5-year breast cancer–specific survival rate of the patients, but had opposite effects in ERα-negative and ERα-positive early-stage patients. Our findings provide support for diverse roles of ESR1 polymorphism in determining susceptibility in different stages of breast cancer development, and offer potential help to clinicians in determining more adequate therapeutic protocols.

The contribution of the reproductive hormone, estrogen, to the development of breast cancer has been well documented in epidemiologic and molecular and cell biology studies (14). The idea that estrogen is involved in breast tumorigenesis is based on the general concept that cell division is crucial for cancer development and estrogen increases mitotic activity in the breast epithelium, resulting in increased cancer risk (5). Estrogen triggers cell growth by binding to the ligand-binding domain of the ER (Fig. 1A) in the nucleus (3, 6, 7). Mammals express two ERs, one of which, ERα (encoded by ESR1), is expressed in normal and neoplastic breast tissue. The role of ERα in initiating breast cancer is well understood (3, 79). ERα binds estrogen and other coactivator proteins to form the estrogen-bound ER complex, which functions as a transcriptional regulator (10, 11). The DNA-binding domain of ERα (Fig. 1A) binds to estrogen response elements in the promoter region of estrogen-responsive genes, initiating their transcription (3, 7). ERα can also be activated in an estrogen-independent manner, in which specific amino acids in the AF-1 domain of ERα (Fig. 1A) are phosphorylated by activation of a signaling cascade downstream of receptor tyrosine kinases. The phosphorylated and thus activated ERα regulates the transcription of target genes by protein-protein interaction with other transcription factors, such as SP1 (3, 7, 11). Many proteins encoded by these ERα-regulated genes are involved in regulating the cell cycle, angiogenesis, and survival (3, 7, 12, 13). ERα is also an important biomarker for determining breast cancer progression, as its levels predict possible responses of patients to hormone therapy (6, 1418). Patients with ERα-positive breast cancer are seven to eight times more likely to benefit from endocrine therapy than ERα-negative patients (6).

Fig. 1.

Schematic diagrams of ERα and the estrogen receptor 1 gene (ESR1). A, functional domains of ERα. B, exon structure of ESR1. The lengths of the light blue boxes (exons) and black lines (introns) are proportional to the size (i.e., base pair) of the exon or intron; however, to reflect the relation between the exonic regions (B) and their corresponding domains (A), the scales for the exons and introns are different. C, SNPs of ESR1 associated with breast cancer risk; SNPs showing a significant effect (P < 0.05) are shown in red, and those showing a borderline effect (0.05 ≤ P < 0.10) are in orange. D, LD between the genotyped SNPs of ESR1 in Taiwanese control women. The strength of the LD between SNPs, as indicated by the color scheme, was measured using a combination of the statistic D′ and the logarithm of odds (LOD) score (dark red shading D′ = 1 and LOD score ≥2; light red shading D′ < 1 and LOD score ≥2).

Fig. 1.

Schematic diagrams of ERα and the estrogen receptor 1 gene (ESR1). A, functional domains of ERα. B, exon structure of ESR1. The lengths of the light blue boxes (exons) and black lines (introns) are proportional to the size (i.e., base pair) of the exon or intron; however, to reflect the relation between the exonic regions (B) and their corresponding domains (A), the scales for the exons and introns are different. C, SNPs of ESR1 associated with breast cancer risk; SNPs showing a significant effect (P < 0.05) are shown in red, and those showing a borderline effect (0.05 ≤ P < 0.10) are in orange. D, LD between the genotyped SNPs of ESR1 in Taiwanese control women. The strength of the LD between SNPs, as indicated by the color scheme, was measured using a combination of the statistic D′ and the logarithm of odds (LOD) score (dark red shading D′ = 1 and LOD score ≥2; light red shading D′ < 1 and LOD score ≥2).

Close modal

The human ESR1 gene contains eight exons spanning ∼295 kb (Fig. 1B) and is located on chromosome 6 at 6q25.1 (19), a susceptibility locus identified as particularly important for breast cancer in Chinese women in a recent genome-wide association study (20). Genotypic polymorphism of ESR1 has long been reported to be important in determining breast cancer susceptibility (2126). However, the results have been inconsistent and the molecular mechanisms by which these polymorphisms affect the function of ERα remain unclear. Given the scarcity of genetic research on ESR1 in Chinese women and the fact that Asian women have, on average, 20% lower serum levels of estrogen than Western women (27), a factor playing a unique role in modifying cancer risk associated with genetic polymorphism of ESR1, we performed a case-control study to explore whether single-nucleotide polymorphisms (SNP) of ESR1 are associated with susceptibility, clinical phenotype, and progression of breast cancer in this population.

Study subjects

This case-control study is part of an ongoing cooperative study aimed at understanding the causes of breast cancer in Taiwan, which is characterized by low incidence, early tumor onset, hormone dependency, and novel genomic alterations (2830). We studied 940 female breast cancer women with pathologically confirmed incident primary breast cancer seen at the Tri-Service General Hospital or the Changhua Christian Hospital between March 2002 and August 2005. The 1,547 healthy female controls were selected from women attending the health examination clinics of the same hospitals during the same period. The characteristics of these study participants have already been described in detail (29, 31, 32). Informed consent was obtained from all study participants before collection of epidemiologic data by personal interview. Considerations regarding methodologic issues in the present study (such as study design, sampling scheme, and potential bias) have been described in detail (29, 31, 32).

We also explored genetic susceptibility to progression of breast cancer associated with polymorphism of ESR1. We have established a reliable follow-up system to actively and regularly trace the progression of our patients, and this was used to obtain information concerning clinical and pathological features and survival status for 799 patients (ERα status: positive, 62.7%, negative, 37.3%; stage I, 33.8%, stage II, 43.8%, stage III/IV, 22.4%). In addition, we determined the cause of death from death certificates. No significant difference in SNP profile was found between patients for whom clinical information was available and those for whom this information was not available. Patients whose deaths were clearly documented to be due to breast cancer were considered to have died of breast cancer; other causes of deaths were considered censored events. Of these 799 patients, 5.4% died from breast cancer and 1.9% died from other causes during the follow-up period.

Questionnaire and clinical and pathological features

An experienced research nurse was assigned to administer a structured questionnaire to both cases and controls. The information collected has been described, and the validity of the questionnaire has been addressed and confirmed in our previous studies (29, 31, 32). Information concerning clinical and pathological features was obtained from the tumor registries of our hospitals.

The degree of differentiation (grade) of each tumor was categorized as I, II, or III according to the Nottingham modification of the Scarff-Bloom-Richardson system (33), in which tubular formation, nuclear polymorphism, and mitotic count are the indices of differentiation; all gradings were done by a single pathologist. The stage of the tumor at diagnosis was determined using the tumor-node-metastasis staging system proposed by the American Joint Committee on Cancer and was categorized as I, II, III, or IV (34). The ERα status of the breast tumors was determined immunohistochemically on paraffin sections (antibody clone ID5, 1/100 dilution; Dako). Positive expression of ERα was defined by the protein being detectable in >10% of nuclei. To ensure a valid comparison of protein expression levels, ERα status was determined by more than one pathologist, each of whom was unaware of the assessment made by the others, and the same slides were repeatedly assessed by the same pathologists to ensure consistency of results. Our laboratory is certified to perform ER status examinations and report ER results.

Specimen collection and SNP selection and genotyping

At the end of the interview, blood was taken for DNA isolation and genotyping. All samples were examined blind by laboratory personnel. DNA was extracted from peripheral blood samples of patients and controls using DNA purification kits (Promega). We initially used HapMap data (http://www.hapmap.org/), the National Center for Biotechnology Information database (http://www.ncbi.nlm.nih.gov/sites/entrez), and Haploview software to select SNPs to ensure that the SNPs selected were able to capture major genetic polymorphisms. As a result, 68 SNPs were required using the thresholds of ≥5% minor allele frequency and r2 ≥ 0.8. However, this resulted in the identification of too many SNPs for our purpose, which was to perform a candidate gene–based study. The purpose of the genotyped SNPs is to briefly reflect the genotypic polymorphism of ESR1 in our population, which is not the same as that of SNPs in genome-wide association studies or fine-mapping studies. Because we have our own SNP database based on 1,000 Han Chinese randomly selected from different parts of Taiwan (35), the information in this SNP database was used for SNP selection. In addition, we found specific SNPs in the Chinese population in the ABI SNP database to be very useful and chose some SNPs from this database. In our experience, the information on minor allele frequency, and, more importantly, haplotype-block structure of SNPs in Chinese populations in the last two databases differs from that in HapMap, and is more useful and representative of our population that that in HapMap. Thus, specific SNPs (e.g., rs3778609) that are not in HapMap were selected from these two databases and used in the present study. All these considerations resulted in the selection of the 15 genotyped SNPs (Fig. 1C; Supplementary Fig. S1), all of which had a minor allele frequency of >0.05. Almost all major blocks in ESR1 in our population were represented by at least one SNP (Supplementary Fig. S1).

Genotyping was carried out using the Taqman allelic discrimination assay and the manufacturer's predesigned primer/probe sets, and the results were read on a GeneAmp PCR System 7000 Thermocycler (Applied Biosystems). All assays were done by individuals blinded to the case-control status of the samples, and each genotyping plate contained positive and negative controls. As a quality control, we repeated the genotyping on 10% of the samples, and all genotype scoring was done and checked separately by two reviewers unaware of the case-control status. The concordance rate for replicate samples was 100%.

Data analysis

We followed our previously established sequential steps (29, 31, 32) to analyze the data. Hardy-Weinberg equilibrium (36) was tested, and linkage disequilibrium (LD) plots of the D′ values for ESR1 were produced using the Haploview program (37). Univariate and multivariate analyses were done to determine putative high-risk genotypes of ESR1 for breast cancer in this series of study subjects. Haplotype estimation was restricted to individuals for whom complete genotype data were available across all polymorphic sites, and the highest probability haplotypes estimated using the expectation maximization (EM) algorithm of SAS/Genetics 9.1 (SAS Institute) were assigned to each study participant (3840). Logistic regression was used to estimate the odds ratio of breast cancer associated with harboring individual specific haplotypes, each haplotype being compared with a shared reference haplotype. Data on ERα status (positive/negative), histologic grade (well-differentiated to poorly differentiated), and clinical stage (I–IV) were collected from hospital medical records. We were therefore able to determine whether SNPs were associated with clinical/pathological features of breast cancer. Furthermore, to examine whether an SNP influenced prognosis, we calculated the 5-year breast cancer–specific survival rate using the Kaplan-Meier method, two-tailed log-rank test, and Wilcoxon test; death from breast cancer was defined as the event of interest (41, 42). All statistical analyses were done using SAS version 9.1 (SAS Institute), and all tests were based on a two-sided probability.

ESR1 polymorphisms and breast cancer susceptibility

The distributions of demographic characteristics and major risk factors for breast cancer were comparable in the cases and controls (Supplementary Table S1). To determine the breast tumorigenic contribution of ESR1, we examined whether the genotypic and allelic distribution of the gene differed between the 940 cases and 1,547 controls (921 cases and 1,533 controls had results of complete 15 SNPs). The frequencies of all 15 SNPs in the controls agreed with those expected under Hardy-Weinberg equilibrium, suggesting that genotyping error was relatively unlikely. Although genotype analysis for each of the 15 SNPs yielded no significant differences, allelic analysis, permutation test, and test for trend showed that the polymorphism of two SNPs, rs3778609 and rs827421, located in intron 1 of ESR1, was significantly associated with breast cancer susceptibility (Table 1 and Fig. 1C). Two other SNPs, rs12665044 and rs7739506, located in intron 1 and 4, respectively (Fig. 1C), showed associations that were of borderline significance (Table 1). To determine the relative importance of individual SNPs, we performed multiple logistic regression analysis on these four variants and found a significant association between rs3778609 and breast cancer risk (P = 0.01). Taken together, these results suggest that SNPs in intron 1 were important. To explore this finding, we checked the LD among SNPs of ESR1 in our controls and used pairwise LD to measure the LD between pairs of SNPs. The results showed that these 15 SNPs fell into four haplotype blocks in our population (Fig. 1D), and that rs3778609, rs12665044, and rs827421 in intron 1 were located in the same haplotype block, suggesting that a haplotype within the LD block may harbor causal variant(s) of ESR1 associated with breast cancer susceptibility. To confirm this, haplotype analysis was done based on these three SNPs. The results were consistent with those based on single SNP analysis, showing that, among the three common haplotypes of intron 1 polymorphism in our population, that carrying the minor alleles of these three SNPs (i.e., the TTG haplotype) was significantly associated with a decreased risk of breast cancer (Table 2). However, a closer look at the risk associated with individual haplotypes suggested that the first two SNPs (rs3778609 and rs12665044) were more important than the third, as the significant TTG haplotype carries both the T (minor) allele of rs3778609 and the T (minor) allele of rs12665044, whereas the haplotype CCG, which is not significantly associated with risk, carries the G (minor) allele of rs827421. These findings confirm the role of ERα in the development of breast cancer in the Chinese population in Taiwan. Furthermore, a genetic variant linked to the SNPs located in an intron 1 cluster may be responsible for this association, and, in particular, this association was mainly driven by rs3778609, demonstrating a significant effect based on logistic regression and haplotype analysis.

Table 1.

Genotypic and allelic frequencies of the ESR1 gene in breast cancer cases and controls and the estimated odds ratio and 95% confidence interval of breast cancer risk

SNP* genotypeCases (%)Controls (%)OR (95% CI)P for trendP for permutation
rs3778609; 4502C>T    0.029 0.027 
    CC 516 (55.7%) 796 (51.8%) 1.00   
    CT 361 (38.9%) 632 (41.1%) 0.88 (0.74–1.05)   
    TT 50 (5.4%) 109 (7.09%) 0.71 (0.50–1.01)   
    C allele 1,393 (75.1%) 2,224 (72.4%) 1.00   
    T allele 461 (24.9%) 850 (27.6%) 0.87 (0.76–0.99)   
rs12665044; 21187C>T    0.068 0.074 
    CC 505 (54.4%) 781 (50.9%) 1.00   
    CT 365 (39.3%) 640 (41.7%) 0.88 (0.74–1.05)   
    TT 58 (6.3%) 114 (7.4%) 0.79 (0.56–1.10)   
    C allele 1,375 (74.1%) 2,202 (77.7%) 1.00   
    T allele 481 (25.9%) 868 (28.3%) 0.89 (0.78–1.01)   
rs827421; 28437A>G    0.043 0.044 
    AA 392 (41.9%) 591 (38.2%) 1.00   
    AG 428 (45.8%) 735 (47.5%) 0.88 (0.74–1.05)   
    GG 115 (12.3%) 221 (14.3%) 0.79 (0.61–1.02)   
    A allele 1,212 (64.8%) 1,917 (62.0%) 1.00   
    G allele 658 (35.2%) 1,177 (38.0%) 0.88 (0.79–1.00)   
rs9340807; 35601A>T    0.646 0.673 
    AA 677 (72.8%) 1,135 (73.8%) 1.00   
    AT 238 (25.6%) 377 (24.5%) 1.06 (0.88–1.28)   
    TT 15 (1.6%) 26 (1.7%) 0.97 (0.51–1.84)   
    A allele 1,592 (85.6%) 2,647 (86.1%) 1.00   
    T allele 268 (14.4%) 429 (13.9%) 1.04 (0.88–1.23)   
rs12204714; 106654C>T    0.919 0.941 
    CC 548 (59.1%) 884 (58.4%) 1.00   
    CT 324 (35.0%) 555 (36.7%) 0.94 (0.79–1.12)   
    TT 55 (5.9%) 75 (4.9%) 1.18 (0.82–1.70)   
    C allele 1,420 (76.6%) 2,323 (76.7%) 1.00   
    T allele 434 (23.4%) 705 (23.3%) 1.01 (0.88–1.15)   
rs9371564; 133428A>G    0.752 0.799 
    AA 465 (49.8%) 749 (48.8%) 1.00   
    AG 381 (40.8%) 646 (42.1%) 0.95 (0.80–1.13)   
    GG 87 (9.3%) 139 (9.1%) 1.01 (0.75–1.35)   
    A allele 1,311 (70.3%) 2,144 (69.9%) 1.00   
    G allele 555 (29.7%) 924 (30.1%) 0.98 (0.87–1.11)   
rs1801132; 136837C>G    0.735 0.753 
    CC 241 (25.8%) 402 (26.1%) 1.00   
    CG 468 (50.1%) 751 (48.6%) 1.04 (0.85–1.27)   
    GG 225 (24.1%) 391 (25.3%) 0.96 (0.76–1.21)   
    C allele 950 (50.9%) 1,555 (50.4%) 1.00   
    G allele 918 (49.1%) 1,533 (49.6%) 0.98 (0.87–1.10)   
rs3020377; 143713G>A    0.879 0.885 
    GG 616 (65.6%) 999 (64.7%) 1.00   
    AG 279 (29.7%) 492 (31.9%) 0.92 (0.77–1.10)   
    AA 44 (4.7%) 53 (3.4%) 1.35 (0.89–2.03)   
    G allele 1,511 (80.5%) 2,490 (80.6%) 1.00   
    A allele 367 (19.5%) 598 (19.4%) 1.01 (0.88–1.17)   
rs7739506; 157758A>G    0.075 0.086 
    AA 769 (83.7%) 1,186 (81.0%) 1.00   
    AG 144 (15.7%) 264 (18.0%) 0.84 (0.67–1.05)   
    GG 6 (0.6%) 15 (1.0%) 0.62 (0.24–1.60)   
    A allele 1,682 (91.5%) 2,636 (90.0%) 1.00   
    G allele 156 (8.5%) 294 (10.0%) 0.83 (0.68–1.02)   
rs1884054; 162881C>A    0.551 0.563 
    CC 467 (50.2%) 736 (48.2%) 1.00   
    AC 377 (40.5%) 656 (43.0%) 0.91 (0.76–1.08)   
    AA 86 (9.3%) 135 (8.8%) 1.00 (0.75–1.35)   
    C allele 1,311 (70.5%) 2,128 (69.7%) 1.00   
    A allele 549 (29.5%) 926 (30.3%) 0.96 (0.85–1.09)   
rs7753398; 174127C>T    0.615 0.619 
    CC 602 (65.0%) 956 (64.0%) 1.00   
    CT 288 (31.1%) 479 (32.0%) 0.96 (0.80–1.14)   
    TT 36 (3.9%) 60 (4.0%) 0.95 (0.62–1.46)   
    C allele 1,492 (80.6%) 2,391 (80.0%) 1.00   
    T allele 360 (19.4%) 599 (20.0%) 0.96 (0.83–1.11)   
rs9383601; 177544T>A    0.576 0.586 
    TT 604 (64.9%) 972 (63.5%) 1.00   
    TA 289 (31.0%) 496 (32.4%) 0.94 (0.79–1.12)   
    AA 38 (4.1%) 62 (4.1%) 0.99 (0.65–1.50)   
    T allele 1,497 (80.4%) 2,440 (79.7%) 1.00   
    A allele 365 (19.6%) 620 (20.3%) 0.96 (0.83–1.11)   
rs9322349; 205178T>C    0.343 0.366 
    TT 694 (74.2%) 1,113 (72.0%) 1.00   
    CT 219 (23.4%) 399 (25.8%) 0.88 (0.73–1.07)   
    CC 22 (2.4%) 33 (2.2%) 1.07 (0.62–1.85)   
    T allele 1,607 (85.9%) 2,625 (84.9%) 1.00   
    C allele 263 (14.1%) 465 (15.1%) 0.92 (0.79–1.09)   
rs9397482; 250854G>A    0.270 0.267 
    GG 373 (39.8%) 674 (41.9%) 1.00   
    AG 430 (45.9%) 691 (44.8%) 1.08 (0.91–1.29)   
    AA 134 (14.3%) 205 (13.3%) 1.13 (0.88–1.46)   
    G allele 1,176 (62.7%) 1,985 (64.3%) 1.00   
    A allele 698 (37.3%) 1,101 (35.7%) 1.07 (0.95–1.21)   
rs13192678; 277225C>T    0.092 0.093 
    CC 316 (34.8%) 575 (38.2%) 1.00   
    CT 437 (48.1%) 695 (46.2%) 1.14 (0.95–1.37)   
    TT 155 (17.1%) 234 (15.6%) 1.21 (0.94–1.54)   
    C allele 1,069 (58.9%) 1,845 (61.3%) 1.00   
    T allele 747 (41.1%) 1,163 (38.7%) 1.11 (0.98–1.25)   
SNP* genotypeCases (%)Controls (%)OR (95% CI)P for trendP for permutation
rs3778609; 4502C>T    0.029 0.027 
    CC 516 (55.7%) 796 (51.8%) 1.00   
    CT 361 (38.9%) 632 (41.1%) 0.88 (0.74–1.05)   
    TT 50 (5.4%) 109 (7.09%) 0.71 (0.50–1.01)   
    C allele 1,393 (75.1%) 2,224 (72.4%) 1.00   
    T allele 461 (24.9%) 850 (27.6%) 0.87 (0.76–0.99)   
rs12665044; 21187C>T    0.068 0.074 
    CC 505 (54.4%) 781 (50.9%) 1.00   
    CT 365 (39.3%) 640 (41.7%) 0.88 (0.74–1.05)   
    TT 58 (6.3%) 114 (7.4%) 0.79 (0.56–1.10)   
    C allele 1,375 (74.1%) 2,202 (77.7%) 1.00   
    T allele 481 (25.9%) 868 (28.3%) 0.89 (0.78–1.01)   
rs827421; 28437A>G    0.043 0.044 
    AA 392 (41.9%) 591 (38.2%) 1.00   
    AG 428 (45.8%) 735 (47.5%) 0.88 (0.74–1.05)   
    GG 115 (12.3%) 221 (14.3%) 0.79 (0.61–1.02)   
    A allele 1,212 (64.8%) 1,917 (62.0%) 1.00   
    G allele 658 (35.2%) 1,177 (38.0%) 0.88 (0.79–1.00)   
rs9340807; 35601A>T    0.646 0.673 
    AA 677 (72.8%) 1,135 (73.8%) 1.00   
    AT 238 (25.6%) 377 (24.5%) 1.06 (0.88–1.28)   
    TT 15 (1.6%) 26 (1.7%) 0.97 (0.51–1.84)   
    A allele 1,592 (85.6%) 2,647 (86.1%) 1.00   
    T allele 268 (14.4%) 429 (13.9%) 1.04 (0.88–1.23)   
rs12204714; 106654C>T    0.919 0.941 
    CC 548 (59.1%) 884 (58.4%) 1.00   
    CT 324 (35.0%) 555 (36.7%) 0.94 (0.79–1.12)   
    TT 55 (5.9%) 75 (4.9%) 1.18 (0.82–1.70)   
    C allele 1,420 (76.6%) 2,323 (76.7%) 1.00   
    T allele 434 (23.4%) 705 (23.3%) 1.01 (0.88–1.15)   
rs9371564; 133428A>G    0.752 0.799 
    AA 465 (49.8%) 749 (48.8%) 1.00   
    AG 381 (40.8%) 646 (42.1%) 0.95 (0.80–1.13)   
    GG 87 (9.3%) 139 (9.1%) 1.01 (0.75–1.35)   
    A allele 1,311 (70.3%) 2,144 (69.9%) 1.00   
    G allele 555 (29.7%) 924 (30.1%) 0.98 (0.87–1.11)   
rs1801132; 136837C>G    0.735 0.753 
    CC 241 (25.8%) 402 (26.1%) 1.00   
    CG 468 (50.1%) 751 (48.6%) 1.04 (0.85–1.27)   
    GG 225 (24.1%) 391 (25.3%) 0.96 (0.76–1.21)   
    C allele 950 (50.9%) 1,555 (50.4%) 1.00   
    G allele 918 (49.1%) 1,533 (49.6%) 0.98 (0.87–1.10)   
rs3020377; 143713G>A    0.879 0.885 
    GG 616 (65.6%) 999 (64.7%) 1.00   
    AG 279 (29.7%) 492 (31.9%) 0.92 (0.77–1.10)   
    AA 44 (4.7%) 53 (3.4%) 1.35 (0.89–2.03)   
    G allele 1,511 (80.5%) 2,490 (80.6%) 1.00   
    A allele 367 (19.5%) 598 (19.4%) 1.01 (0.88–1.17)   
rs7739506; 157758A>G    0.075 0.086 
    AA 769 (83.7%) 1,186 (81.0%) 1.00   
    AG 144 (15.7%) 264 (18.0%) 0.84 (0.67–1.05)   
    GG 6 (0.6%) 15 (1.0%) 0.62 (0.24–1.60)   
    A allele 1,682 (91.5%) 2,636 (90.0%) 1.00   
    G allele 156 (8.5%) 294 (10.0%) 0.83 (0.68–1.02)   
rs1884054; 162881C>A    0.551 0.563 
    CC 467 (50.2%) 736 (48.2%) 1.00   
    AC 377 (40.5%) 656 (43.0%) 0.91 (0.76–1.08)   
    AA 86 (9.3%) 135 (8.8%) 1.00 (0.75–1.35)   
    C allele 1,311 (70.5%) 2,128 (69.7%) 1.00   
    A allele 549 (29.5%) 926 (30.3%) 0.96 (0.85–1.09)   
rs7753398; 174127C>T    0.615 0.619 
    CC 602 (65.0%) 956 (64.0%) 1.00   
    CT 288 (31.1%) 479 (32.0%) 0.96 (0.80–1.14)   
    TT 36 (3.9%) 60 (4.0%) 0.95 (0.62–1.46)   
    C allele 1,492 (80.6%) 2,391 (80.0%) 1.00   
    T allele 360 (19.4%) 599 (20.0%) 0.96 (0.83–1.11)   
rs9383601; 177544T>A    0.576 0.586 
    TT 604 (64.9%) 972 (63.5%) 1.00   
    TA 289 (31.0%) 496 (32.4%) 0.94 (0.79–1.12)   
    AA 38 (4.1%) 62 (4.1%) 0.99 (0.65–1.50)   
    T allele 1,497 (80.4%) 2,440 (79.7%) 1.00   
    A allele 365 (19.6%) 620 (20.3%) 0.96 (0.83–1.11)   
rs9322349; 205178T>C    0.343 0.366 
    TT 694 (74.2%) 1,113 (72.0%) 1.00   
    CT 219 (23.4%) 399 (25.8%) 0.88 (0.73–1.07)   
    CC 22 (2.4%) 33 (2.2%) 1.07 (0.62–1.85)   
    T allele 1,607 (85.9%) 2,625 (84.9%) 1.00   
    C allele 263 (14.1%) 465 (15.1%) 0.92 (0.79–1.09)   
rs9397482; 250854G>A    0.270 0.267 
    GG 373 (39.8%) 674 (41.9%) 1.00   
    AG 430 (45.9%) 691 (44.8%) 1.08 (0.91–1.29)   
    AA 134 (14.3%) 205 (13.3%) 1.13 (0.88–1.46)   
    G allele 1,176 (62.7%) 1,985 (64.3%) 1.00   
    A allele 698 (37.3%) 1,101 (35.7%) 1.07 (0.95–1.21)   
rs13192678; 277225C>T    0.092 0.093 
    CC 316 (34.8%) 575 (38.2%) 1.00   
    CT 437 (48.1%) 695 (46.2%) 1.14 (0.95–1.37)   
    TT 155 (17.1%) 234 (15.6%) 1.21 (0.94–1.54)   
    C allele 1,069 (58.9%) 1,845 (61.3%) 1.00   
    T allele 747 (41.1%) 1,163 (38.7%) 1.11 (0.98–1.25)   

Abbreviations: OR, odds ratio; 95% CI, 95% confidence interval.

*The “rs” number is the National Center for Biotechnology Information (NCBI) dbSNP cluster ID for each SNP. The second number indicates the location of the SNP relative to the start codon ATG according to the NCBI genomic contig NC_000006 (region: 152170379–152466099).

Data were calculated by logistic regression, in which we used a set of dummy variables representing different genotypes of ESR1.

The P values were calculated based on 10,000 random permutations.

Table 2.

Estimated haplotype frequencies of intron 1 of the ESR1 gene in breast cancer cases and controls and the significance of the association between harboring a specific haplotype and breast cancer development

Haplotype of ESR1 intron 1*Cases (%)Controls (%)OR (95% CI)aOR (95% CI)
CCA 63.0 61.6 1.00 (Reference) 1.00 (Reference) 
CCG 11.0 10.0 1.08 (0.89–1.31) 1.04 (0.85–1.27) 
TTG 22.8 27.1 0.82 (0.72–0.94) 0.83 (0.72–0.96) 
Others 3.2 1.3 — — 
Haplotype of ESR1 intron 1*Cases (%)Controls (%)OR (95% CI)aOR (95% CI)
CCA 63.0 61.6 1.00 (Reference) 1.00 (Reference) 
CCG 11.0 10.0 1.08 (0.89–1.31) 1.04 (0.85–1.27) 
TTG 22.8 27.1 0.82 (0.72–0.94) 0.83 (0.72–0.96) 
Others 3.2 1.3 — — 

Abbreviation: aOR, adjusted odds ratio.

*ESR1 haplotypes with the SNPs in intron 1 arranged in the order 5′ to 3′; that is, rs3778609 (4502C>T), rs12665044 (21187C>T), and rs827421 (28437A>G).

The adjusted odds ratio and 95% confidence interval were estimated in a multivariate logistic regression model containing age, a family history of breast cancer, menarche age, a history of full-term pregnancy, frequency of pregnancy, body mass index, menopause status, and a group of dummy variables to represent women harboring the three possible haplotypes of the ESR1 gene.

The protective genotypes of intron 1 polymorphism are associated with tumor ERα status, and a joint effect of these polymorphisms and reproductive risk factors is associated with cancer susceptibility

The above finding that minor alleles of intron 1 polymorphism of ESR1 are associated with a decreased cancer risk is intriguing. Based on the general concept that increased proliferation of breast epithelium leading to tumor formation is caused by increased mitotic potential triggered by activated ERα (4, 5), we speculated that these protective polymorphisms composed of the minor alleles might be related to decreased expression of ERα. Interestingly, the association identified in our cases was consistent with this, as tumors harboring the homozygous variant genotype (i.e., TT) of intron 1 SNP rs3778609 tended to show lower ERα expression (i.e., a higher frequency of ERα negativity of 51.3%) than tumors harboring the homozygous wild-type (i.e., CC) or heterozygous wild-type (i.e., CT) genotype (i.e., a lower frequency of ERα negativity of 37.0%; P = 0.04). There was no association between ERα status and either of the other two SNPs (rs12665044 and rs827421) in the intron 1 cluster or the haplotype based on these three intron 1 SNPs, reconfirming the previous suggestion that rs3778609 is relatively more important than the other two SNPs located in the same LD block.

The association identified between intron 1 polymorphism and tumor ERα status prompted us to explore whether the known interaction between estrogen and ERα could be confirmed in this study on breast cancer susceptibility. Reproductive risk factors were used as surrogates to estimate the estrogen exposure experienced by our study participants throughout their lifetime. The results were in line with our hypothesis, as a significant protective effect associated with the haplotype composed of the minor alleles of intron 1 SNPs (i.e., the TTG haplotype) was only seen in women who had experienced lower estrogen exposure or with lower susceptibility to estrogen exposure, reflected by each of the reproductive risk factors of an older age at menarche, a history of full-term pregnancy, a younger age at first full-term pregnancy, or being menopausal (Table 3). A stringent statistical test for interaction based on a multiplicative model confirmed that an interaction between ESR1 polymorphism and reproductive risk factors was linked to breast cancer risk (P < 0.05). The two risk-associated SNPs (rs3778609 and rs827421) of intron 1 yield the same significant interaction with reproductive risk factors in determining breast cancer susceptibility (data not shown).

Table 3.

Risk (adjusted odds ratio and 95% confidence interval) of breast cancer associated with individual combinations of the intron 1 haplotypes of ESR1 and each of the reproductive risk factors

Haplotype of ESR1 intron 1*CCACCGTTGP for interaction
Reproductive risk factor 
Age at menarche, y 
    <13 1.00 (Reference) 1.17 (0.73–1.86) 0.74 (0.52–1.04) 0.015 
    ≥13 0.89 (0.74–1.08) 0.95 (0.73–1.24) 0.75 (0.60–0.93)  
History of FFTP 
    No 1.00 (Reference) 1.10 (0.64–1.92) 0.90 (0.61–1.32) 0.002 
    Yes 0.87 (0.70–1.09) 0.94 (0.71–1.25) 0.71 (0.55–0.90)  
Age at first full-term pregnancy, y 
    ≥35 1.00 (Reference) 0.53 (0.11–2.47) 0.57 (0.21–1.56) 0.004 
    <35 0.54 (0.30–0.95) 0.59 (0.32–1.07) 0.44 (0.25–0.79)  
Menopause 
    No 1.00 (Reference) 1.17 (0.89–1.54) 0.83 (0.21–1.56) 0.009 
    Yes 0.69 (0.56–0.86) 0.58 (0.40–0.83) 0.52 (0.39–0.69)  
Haplotype of ESR1 intron 1*CCACCGTTGP for interaction
Reproductive risk factor 
Age at menarche, y 
    <13 1.00 (Reference) 1.17 (0.73–1.86) 0.74 (0.52–1.04) 0.015 
    ≥13 0.89 (0.74–1.08) 0.95 (0.73–1.24) 0.75 (0.60–0.93)  
History of FFTP 
    No 1.00 (Reference) 1.10 (0.64–1.92) 0.90 (0.61–1.32) 0.002 
    Yes 0.87 (0.70–1.09) 0.94 (0.71–1.25) 0.71 (0.55–0.90)  
Age at first full-term pregnancy, y 
    ≥35 1.00 (Reference) 0.53 (0.11–2.47) 0.57 (0.21–1.56) 0.004 
    <35 0.54 (0.30–0.95) 0.59 (0.32–1.07) 0.44 (0.25–0.79)  
Menopause 
    No 1.00 (Reference) 1.17 (0.89–1.54) 0.83 (0.21–1.56) 0.009 
    Yes 0.69 (0.56–0.86) 0.58 (0.40–0.83) 0.52 (0.39–0.69)  

NOTE: The adjusted odds ratio and 95% confidence intervals were estimated in a multivariate logistic regression model containing age, a family history of breast cancer, body mass index, and a group of dummy variables to represent women harboring different haplotypes of ESR1 and the status of reproductive risk factors.

*ESR1 haplotypes with the SNPs in intron 1 arranged in the order 5′ to 3′; that is, rs3778609 (4502C>T), rs12665044 (21187C>T), and rs827421 (28437A>G).

The P value for interaction associated with ESR1 intron 1 haplotype and reproductive risk factor was estimated in a multiple logistic regression model containing age, a family history of breast cancer, body mass index, and the product term of (haplotype × reproductive risk factor).

The protective genotypes of intron 1 polymorphism are, conversely, associated with poor grade and late stage of breast cancer

ERα is a key regulator not only of proliferation but also of differentiation of mammary epithelia and is a crucial prognostic indicator in breast cancer. However, it is well recognized in both breast cancer biology and clinical practice that ERα plays a seemingly paradoxical role at different stages during breast tumorigenesis, its expression being associated with an increased cancer risk but a decreased invasive/metastatic potential of tumor cells (7, 6, 1618). On this basis, the protective polymorphisms identified as associated with decreased breast cancer susceptibility (Table 1) and lower ERα expression (mentioned in the previous section) might show a negative correlation with unfavorable clinical/pathological features of breast cancer. To examine this, we determined whether the frequency distribution of the intron 1 SNPs differed in tumors with different grades and stages. The results showed that a higher frequency of the homozygous variant genotypes of two intron 1 SNPs (rs3778609 and rs12665044) was found in higher-grade (poor differentiation) breast cancers, with significant P values in the trend test (Fig. 2; Supplementary Table S2), in line with our expectation. A similar trend was also observed between stage and intron 1 polymorphisms, showing that the genotype composed of alleles presumably protective against breast cancer development was, conversely, more frequent in patients with advanced-stage cancer than in those with early-stage cancer (Fig. 2). No significant association was found between tumor grade/stage and the third intron 1 SNP (rs827421) or haplotypes composed of three intron 1 SNPs (Supplementary Table S2).

Fig. 2.

Genotypic polymorphism of SNPs in intron 1 in ESR1 is associated with the clinical features of breast cancer patients. The proportion of the homozygous variant genotype in breast cancer patients of different grades and stages is shown. The case numbers in individual subsets are 682 (A), 784 (B), 682 (C), and 786 (D).

Fig. 2.

Genotypic polymorphism of SNPs in intron 1 in ESR1 is associated with the clinical features of breast cancer patients. The proportion of the homozygous variant genotype in breast cancer patients of different grades and stages is shown. The case numbers in individual subsets are 682 (A), 784 (B), 682 (C), and 786 (D).

Close modal

Different associations between intron 1 polymorphisms and 5-year breast cancer–specific survival rate of different ERα status

Consistent with the usual observation in breast cancer clinics, the stage of the tumor at diagnosis was the most important prognostic factor determining the 5-year breast cancer–specific survival rate in this cohort of patients (Fig. 3). Death from breast cancer was used as the event of interest in these analyses. Originally, we performed Cox regression analysis to estimate the hazard ratio associated with SNPs. However, as in any breast cancer cohort, both ER status and stage at tumor diagnosis played dominant roles in determining progression in our patients, making it difficult to examine low-risk effects, such as that contributed by SNPs, and we therefore decided to stratify our patients based on these dominant factors and address the importance of SNPs in different strata. With regard to individual SNPs in intron 1 block, no single SNP showed a significant association with survival in tumors of different stages and ERα status. We then used haplotype to reflect intron 1 polymorphism and found that this polymorphism played no role in the progression of advanced-stage (stage III/IV) tumors (Fig. 3C and D); however, the small number of stage III/IV observations is a limitation of this analysis. In contrast, a significant association between intron 1 haplotype and 5-year breast cancer–specific survival rate was found in early-stage (stage I/II) tumors (Fig. 3A and B; P < 0.05 estimated using the log-rank test); however, interestingly, the direction of the association was dependent on ERα status. ER-positive patients carrying the wild-type (CCA) haplotype (i.e., the haplotype composed of the major alleles of the three intron 1 SNPs) had a significantly better 5-year breast cancer–specific survival rate than those carrying the variant TTG haplotype (Fig. 3A). In contrast, in ER-negative tumors, in which ERα presumably plays a minor role due to low ERα expression, intron 1 polymorphism was still significantly associated with tumor progression; however, patients with the CCA haplotype had a much worse survival than those carrying the TTG haplotype (Fig. 3B). No evidence for a significant association of other SNPs of ESR1 with breast cancer progression was found (data not shown).

Fig. 3.

Genetic polymorphism of SNPs in intron 1 in ESR1 is associated with breast cancer progression. Kaplan-Meier estimates of 5-year breast cancer–specific survival rate associated with stage, ERα status, and ESR1 intron 1 haplotype. All P values were estimated by the log-rank test. Death from breast cancer was used as the event of interest in these analyses. The case numbers in individual subsets are 357 [A, ER(+) and stage I/II], 207 [B, ER(−) and stage I/II], 103 [C, ER(+) and stage III/IV], and 65 [D, ER(−) and stage III/IV].

Fig. 3.

Genetic polymorphism of SNPs in intron 1 in ESR1 is associated with breast cancer progression. Kaplan-Meier estimates of 5-year breast cancer–specific survival rate associated with stage, ERα status, and ESR1 intron 1 haplotype. All P values were estimated by the log-rank test. Death from breast cancer was used as the event of interest in these analyses. The case numbers in individual subsets are 357 [A, ER(+) and stage I/II], 207 [B, ER(−) and stage I/II], 103 [C, ER(+) and stage III/IV], and 65 [D, ER(−) and stage III/IV].

Close modal

Our case-control study provided several lines of evidence that polymorphism of ESR1 is associated with breast cancer susceptibility and progression. The evidence includes epidemiologic associations (i.e., clustering of significant SNPs and logistic regression), a genotype-phenotype association (i.e., ESR1 SNPs and ERα expression), genetic epidemiologic inference (i.e., haplotype analysis), and the clinical impact of SNPs on specific types of breast cancer and cancer patient survival. However, in considering whether our findings represent a true association between the SNPs of ESR1 and breast cancer, the most important issue is the interpretation of the identified association between SNPs and the trait. Because the SNPs analyzed are in introns, they do not affect amino acid coding and therefore probably do not directly affect protein function; thus, the observed associations between breast cancer and SNPs should be interpreted as the presence of LD between these SNPs and other SNPs in exons (resulting in functional polymorphism) or in regulatory regions (affecting the expression of these genes). It is notable that ESR1 is located at 6q25.1, the locus for SNP rs2046210, and that rs2046210 showed strong and consistent association with Chinese breast cancer in a recent multiple-stage genome-wide association study (20). Whether the observed association in our study is actually due to a link between the ESR1 SNPs and rs2046210 is an issue that needs to be clarified before further exploration of the meaning of our findings. We certainly do not exclude the possibility that the risk of breast cancer is associated with rs2046210. However, rs2046210 is quite distant from ESR1, and this SNP is located 29 kb upstream of the first untranslated exon and 180 kb upstream of the transcription starting site of exon 1 of the ESR1 gene. More importantly, based on the newly published haplotype map of the human genome, there are many recombination hotspots between rs2046210 and ESR1, and ESR1 and this SNP are located in totally different haplotype blocks (Supplementary Fig. S2), the values of pairwise r2 between rs2046210 and the intron 1 SNPs being ∼0.10 to 0.20 (based on the information of HapMap). We therefore suggest that our finding of an association between ESR1 polymorphism and breast cancer is more likely to be independent of rs2046210. Furthermore, genetic heterogeneity is less of a concern in Taiwan than in the United States (43); as a result, potential bias due to population stratification is less likely to be significant in our study, and the chance that the targeted genomic region or functional variants reflected by the same genotyped SNPs are different in cases and controls due to differences in the genetic background of the two groups is small. However, we recognize that the sequencing of the region reflected by intron 1 SNPs (promoter region, exon 1, intron 1, and exon 2) in our population is the definitive approach to identifying all of the important sequence variants, and that a large-scale evaluation of these variants and functional assessments are needed to address this question.

The possibility of false positives (type I errors) must be considered, and multiple testing is a major concern of the present study. Genotype analysis for each of the 15 SNPs yielded no significant association with breast cancer based on a nominal distribution; however, using a trend test, the decreasing trends of cancer risk associated with two polymorphisms in intron 1 were significant. However, we performed 15 separate tests for trend significance and could therefore expect one significant result at the P = 0.05 level by chance. Similarly, allelic analysis showed that these two intron 1 SNPs were significant; however, considering the number of statistical tests done, both would be expected to become insignificant after the Bonferroni correction. However, based on 10,000 random permutations, the association between cancer risk and these two SNPs remained significant (Table 1). The overly conservative correction of the Bonferroni approach might result in power reduction; therefore, it has been suggested that simultaneous consideration of multiple SNPs within a defined region provides more information. Given this, it is notable that the two significant SNPs were in the same cluster. All these considerations prompted us to propose that the intron 1 region may harbor genetic variants determining breast cancer susceptibility, and the result of haplotype analysis is in line with this suggestion. Interestingly, the importance of intron 1 variants of ESR1 in determining breast cancer risk has been reported in previous studies in specific populations (24, 44, 45).

An understanding of the etiologic role of ESR1 in breast tumorigenesis will require studies that evaluate both multiple functions of ERα in regulating cell growth and preventing metastasis and the extent to which estrogen exposure modifies the associations of this gene with breast cancer. This understanding is incomplete, as previous epidemiologic research has usually been based on single-SNP analysis without considering reproductive risk factors. Thus, the discrepancy in the results regarding the degree and nature of breast cancer–related genetic polymorphisms of ESR1 is not surprising. On the basis of a multi-SNP approach and considering the structure of the gene to reflect most haplotype blocks of ESR1, our examination of the association between genetic variants of this gene and breast cancer provides an opportunity for exploring which genetic variants of ESR1 play a more significant role. Our identification of an interaction between intron 1 SNPs and risk factors reflecting estrogen exposure in determining breast cancer susceptibility is of particular interest, as a recent comprehensive SNP-tagging association study on ESR1 identified a totally different region, SNP rs3020314, tagging a region of ESR1 intron 4, as being associated with an increase in breast cancer susceptibility in European populations (26), whereas the present study and other studies on Asian populations (26) showed no such association. Furthermore, this SNP is located in a different haplotype block to the significant intron 1 SNPs identified in the present study. The r2 value between rs3020214 and rs827421 is only 0.12 in the HapMap data and the r2 between each of our three intron 1 SNPs and rs1801132 (an intron 4 SNP genotyped in the present study close to rs3020314) ranged from 0.14 to 0.23, all of which are low values. We propose that this racial difference in significant SNPs may be partially explained by endogenous estrogen levels in the different populations. Given that Western women have, on average, 20% higher serum estradiol levels than Asian women (27), we postulate that the effect of SNPs tagging the ligand binding domain (i.e., the AF-2 domain; Fig. 1A and B) would be more obvious in Western women. The rationale underlying this is derived from the clues provided by many epidemiologic observations that substrate dosage might affect the associations between cancers and polymorphisms of enzyme-encoding genes. For example, in subjects with the high-risk genotype of carcinogen-metabolizing genes, the greatest incremental lung cancer risk is seen in heavy smokers, whereas light smokers have a lower risk (46). Thus, it is biologically reasonable to find that, in Taiwanese women, ESR1 plays an equally important role to that in Western women, but the important domain of ERα reflected by significant SNPs (i.e., intron 1 SNPs) is AF-1, a region mostly involved in ligand-independent function of ERα (3). However, the failure to include SNPs in the promoter is a major limitation of our study. In our population, the promoter sequence is in the same block as the SNPs of exon 1/intron 1 of ESR1. Thus, it is possible that causal variants in the promoter affect the level of expression through transcriptional regulation and determine cancer susceptibility. The determination of the effects of ethnicity-specific SNPs of ESR1 on breast cancer risk might yield valuable clues about the association between breast tumorigenesis and estrogen, underscoring the importance of comprehensive examination of SNPs throughout the entire gene in a genotype-based association study.

Our finding that alleles of intron 1 SNPs protective against breast cancer development were, conversely, associated with poorly differentiated/late-stage tumor phenotypes (Fig. 2) is mechanistically plausible. This agrees with the diverse roles of ERα during different stages of breast cancer development. By triggering cell outgrowth, ER expression is a risk factor for breast cancer initiation (2, 4, 5); however, it is also related to a favorable progression, as ER is able to activate a signaling pathway resulting in the expression of cell adhesion molecule E-cadherin (47, 48), which controls the epithelial-to-mesenchymal transition in human breast tumors, leading to well-differentiated and less invasive tumors and fewer metastic tumors (47, 49). Similarly, the established causal link between ER and E-cadherin can be used to explain the association between ESR1 polymorphism and breast cancer progression observed in early-stage ER-positive patients (Fig. 3A). In these subjects, ERα is a promising therapeutic target and the use of the antiestrogen drug tamoxifen in targeted therapy can activate an invasion suppressor phenotype by restoring E-cadherin function (5052). To this end, ERα may work in a ligand-independent manner, contributing to the therapeutic benefit in breast cancer patients carrying the haplotype composed of the wild-type (supposedly high expression) alleles of ESR1. The only finding that cannot be currently explained is that in the group of patients with low ERα expression (i.e., the ER-negative patients), the association between intron 1 polymorphism and worse survival (Fig. 3B) is in sharp contrast to that between the same haplotype and poor survival seen in ER-positive patients. We are currently exploring whether this is an effect at the transcriptional level; for example, whether the intron 1 region tagged by the ESR1 SNPs contains a cis-regulatory element or micro-RNA coding sequences that regulates unique tumorigenic functions. In conclusion, the present study provides an important insight into the role of the ESR1 gene in breast tumorigenesis and offers potential help to clinicians in determining more adequate therapeutic protocols.

No potential conflicts of interest were disclosed.

Grant Support: NSC 96-2314-B-345-001-MY2 from the National Science Council, Taiwan (S-l. Ding), and TSGH-C96-3-S03 and TSGH-C97-7-S03 from the Tri-Service General Hospital (J.C. Yu).

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.

1
Kelsey
JL
,
Gammon
MD
,
John
EM
. 
Reproductive factors and breast cancer
.
Epidemiol Rev
1993
;
15
:
36
47
.
2
Key
T
,
Appleby
P
,
Barnes
I
,
Reeves
G
. 
Endogenous sex hormones and breast cancer in postmenopausal women: reanalysis of nine prospective studies
.
J Natl Cancer Inst
2002
;
94
:
606
16
.
3
Ali
S
,
Coombes
RC
. 
Endocrine-responsive breast cancer and strategies for combating resistance
.
Nat Rev Cancer
2002
;
2
:
101
12
.
4
Yager
JD
,
Davidson
NE
. 
Estrogen carcinogenesis in breast cancer
.
N Engl J Med
2006
;
354
:
270
82
.
5
Pike
MC
,
Spicer
DV
,
Dahmoush
L
,
Press
MF
. 
Estrogens, progestogens, normal breast cell proliferation, and breast cancer risk
.
Epidemiol Rev
1993
;
15
:
17
35
.
6
Osborne
CK
,
Schiff
R
. 
Estrogen-receptor biology: continuing progress and therapeutic implications
.
J Clin Oncol
2005
;
23
:
1616
22
.
7
Musgrove
EA
,
Sutherland
RL
. 
Biological determinants of endocrine resistance in breast cancer
.
Nat Rev Cancer
2009
;
9
:
631
43
.
8
McKenna
NJ
,
Lanz
RB
,
O'Malley
BW
. 
Nuclear receptor coregulators: cellular and molecular biology
.
Endocr Rev
1999
;
20
:
321
44
.
9
Osborne
CK
,
Schiff
R
,
Fuqua
SA
, et al
. 
Estrogen receptor: current understanding of its activation and modulation
.
Clin Cancer Res
2001
;
7
:
4338
42s
.
10
Kato
S
,
Endoh
H
,
Masuhiro
Y
, et al
. 
Activation of the estrogen receptor through phosphorylation by mitogen-activated protein kinase
.
Science
1995
;
270
:
1491
4
.
11
Pearce
ST
,
Jordan
VC
. 
The biological role of estrogen receptors α and β in cancer
.
Crit Rev Oncol Hematol
2004
;
50
:
3
22
.
12
Welboren
WJ
,
Stunnenberg
HG
,
Sweep
FC
, et al
. 
Identifying estrogen receptor target genes
.
Mol Oncol
2007
;
1
:
138
43
.
13
Safe
S
,
Kim
K
. 
Non-classical genomic estrogen receptor (ER)/specificity protein and ER/activating protein-1 signaling pathways
.
J Mol Endocrinol
2008
;
41
:
263
75
.
14
Kahn
SA
,
Rogers
MA
,
Khurana
KK
, et al
. 
Estrogen receptor expression in benign breast epithelium and breast cancer risk
.
J Natl Cancer Inst
1998
;
90
:
37
42
.
15
Ali
SH
,
O'Donnell
AL
,
Balu
D
, et al
. 
Estrogen receptor-α in the inhibition of cancer growth and angiogenesis
.
Cancer Res
2000
;
60
:
7094
8
.
16
Osborne
CK
,
Zhao
H
,
Fuqua
SAW
. 
Selective estrogen receptor modulators: structure, function, and clinical use
.
J Clin Oncol
2000
;
18
:
3172
86
.
17
Fisher
B
,
Costantino
JP
,
Wickerham
DL
, et al
. 
Tamoxifen for prevention of breast cancer: report of the National Surgical Adjuvant Breast and Bowel Project P-1 study
.
J Natl Cancer Inst
1998
;
90
:
1371
88
.
18
Jordan
VC
,
O'Malley
BW
. 
Selective estrogen-receptor modulators and antihormonal resistance in breast cancer
.
J Clin Oncol
2007
;
25
:
5815
24
.
19
Menasce
LP
,
White
GR
,
Harrison
CJ
, et al
. 
Localization of the estrogen receptor locus (ESR) to chromosome 6q25.1 by FISH and a simple post-FISH banding technique
.
Genomics
1993
;
17
:
263
5
.
20
Zheng
W
,
Long
J
,
Gao
YT
, et al
. 
Genome-wide association study identifies a new breast cancer susceptibility locus at 6q25.1
.
Nat Genet
2009
;
41
:
324
8
.
21
Cai
Q
,
Shu
XO
,
Jin
F
, et al
. 
Genetic polymorphisms in the estrogen receptor α gene and risk of breast cancer: results from the Shanghai breast cancer study
.
Cancer Epidemiol Biomarkers Prev
2003
;
12
:
853
9
.
22
Modugno
F
,
Zmuda
JM
,
Potter
D
, et al
. 
Association of estrogen receptor a polymorphisms with breast cancer risk in older Caucasian women
.
Int J Cancer
2005
;
116
:
984
91
.
23
Conway
K
,
Parrish
E
,
Edmiston
SN
, et al
. 
The estrogen receptor-α A908G (K303R) mutation occurs at a low frequency in invasive breast tumors: results from a population-based study
.
Breast Cancer Res
2005
;
7
:
R871
80
.
24
Colson
NJ
,
Lea
RA
,
Quinlan
S
, et al
. 
No role for estrogen receptor 1 gene intron 1 Pvu II and exon 4 C325G polymorphisms in migraine susceptibility
.
BMC Med Genet
2006
;
7
:
12
.
25
Mavaddat
N
,
Dunning
AM
,
Ponder
BAJ
, et al
. 
Common genetic variation in candidate genes and susceptibility to sub-types of breast cancer
.
Cancer Epidemiol Biomarkers Prev
2009
;
18
:
255
9
.
26
Dunning
AM
,
Healey
CS
,
Baynes
C
, et al
. 
Association of ESR1 gene tagging SNPs with breast cancer risk
.
Hum Mol Genet
2009
;
18
:
1131
9
.
27
Bernstein
L
,
Yuan
JM
,
Ross
RK
, et al
. 
Serum hormone levels in pre-menopausal Chinese women in Shanghai and white women in Los Angeles: results from two breast cancer case-control studies
.
Cancer Causes Control
1990
;
1
:
51
8
.
28
Shen
CY
,
Yu
JC
,
Lo
YL
, et al
. 
Genome-wide search for loss of heterozygosity using laser capture microdissected tissue of breast carcinoma: an implication for mutator phenotype and breast cancer pathogenesis
.
Cancer Res
2000
;
60
:
3884
92
.
29
Ding
SL
,
Yu
JC
,
Chen
ST
, et al
. 
Genetic variants of BLM interact with RAD51 to increase breast cancer susceptibility
.
Carcinogenesis
2009
;
30
:
43
9
.
30
Ding
SL
,
Sheu
LF
,
Yu
JC
, et al
. 
Abnormality of the DNA double-strand-break checkpoint/repair genes, ATM, BRCA1 and TP53, in breast cancer is related to tumor grade
.
Br J Cancer
2004
;
90
:
1995
2001
.
31
Ding
SL
,
Yu
JC
,
Chen
ST
, et al
. 
Genetic variation in the premature aging gene WRN: a case-control study on breast cancer susceptibility
.
Cancer Epidemiol Biomarkers Prev
2007
;
16
:
263
9
.
32
Yu
JC
,
Ding
SL
,
Chang
CH
, et al
. 
Genetic susceptibility to the development and progression of breast cancer associated with polymorphism of cell-cycle and ubiquitin ligase genes
.
Carcinogenesis
2009
;
30
:
1562
70
.
33
Le Doussal
V
,
Tubiana-Hulin
M
,
Friedman
S
, et al
. 
Prognostic value of histologic grade nuclear components of Scarff-Bloom-Richardson (SBR). An improved score modification based on a multivariate analysis of 1262 invasive ductal breast carcinomas
.
Cancer
1989
;
64
:
1914
21
.
34
Greene
F
,
Page
D
,
Fleming
I
,
Fritz
A
,
Balch
C
,
Haller
D
,
Morrow
M
, editors.
AJCC cancer staging manual. 6
.
Philadelphia
:
Springer
; 
2002
.
35
Pan
WH
,
Fann
CSJ
,
Wu
JY
, et al
. 
Han Chinese Cell and Genome Bank in Taiwan: purpose, design and ethical considerations
.
Hum Hered
2006
;
61
:
27
30
.
36
Sellers
TA
. 
Genetic ancestry and molecular epidemiology
.
Cancer Epidemiol Biomarkers Prev
2004
;
13
:
499
500
.
37
Barrett
JC
,
Fry
B
,
Maller
J
, et al
. 
Haploview: analysis and visualization of LD and haplotype maps
.
Bioinformatics
2005
;
21
:
263
5
.
38
Dempster
AP
,
Laird
NM
,
Rubin
D
. 
Maximum likelihood from incomplete data via the EM algorithm
.
J R Stat Soc B
1977
;
39
:
1
38
.
39
Excoffier
L
,
Slatkin
M
. 
Maximum-likelihood estimation of molecular haplotype frequencies in a diploid population
.
Mol Biol Evol
1995
;
12
:
921
7
.
40
Stephens
M
,
Smith
NJ
,
Donnelly
P
. 
A new statistical method for haplotype reconstruction from population data
.
Am J Hum Genet
2001
;
68
:
978
89
.
41
Kaplan
E
,
Meier
P
. 
Non-parametric estimation from incomplete observations
.
J Am Stat Assoc
1958
;
53
:
457
81
.
42
Mantel
N
. 
Evaluation of survival data and two new rank order statistics arising in its consideration
.
Cancer Chemother Rep
1966
;
50
:
163
70
.
43
Yang
HC
,
Lin
CH
,
Hsu
CL
, et al
. 
A comparison of major histocompatibility complex SNPs in Han Chinese residing in Taiwan and Caucasians
.
J Biomed Sci
2006
;
13
:
489
98
.
44
Yaich
L
,
Dupont
WD
,
Cavener
DR
, et al
. 
Analysis of the PvuII restriction fragment-length polymorphism and exon structure of the estrogen receptor gene in breast cancer and peripheral blood
.
Cancer Res
1992
;
52
:
77
83
.
45
Shin
A
,
Kang
D
,
Nishio
H
, et al
. 
Estrogen receptor α gene polymorphisms and breast cancer risk
.
Breast Cancer Res Treat
2003
;
80
:
127
31
.
46
Spivack
SD
,
Hurteau
GJ
,
Fasco
MJ
, et al
. 
Phase I and II carcinogen metabolism gene expression in human lung tissue and tumors
.
Clin Cancer Res
2003
;
9
:
6002
11
.
47
Fearon
ER
. 
Connecting estrogen receptor function, transcriptional repression, and E-cadherin expression in breast cancer
.
Cancer Cell
2003
;
3
:
307
10
.
48
Fujita
N
,
Jaye
DL
,
Kajita
M
, et al
. 
MTA3, a Mi-2/NuRD complex subunit, regulates an invasive growth pathway in breast cancer
.
Cell
2003
;
113
:
207
19
.
49
Oka
H
,
Shiozaki
H
,
Kobayashi
K
, et al
. 
Expression of E-cadherin cell adhesion molecules in human breast cancer tissues and its relationship to metastasis
.
Cancer Res
1993
;
53
:
1696
701
.
50
Helguero
LA
,
Lindberg
K
,
Gardmo
C
, et al
. 
Different roles of estrogen receptors α and β in the regulation of E-cadherin protein levels in a mouse mammary epithelial cell line
.
Cancer Res
2008
;
68
:
8695
704
.
51
Cardamone
MD
,
Bardella
C
,
Gutierrez
A
, et al
. 
ERα as ligand-independent activator of CDH-1 regulates determination and maintenance of epithelial morphology in breast cancer cells
.
Proc Natl Acad Sci U S A
2009
;
106
:
7420
5
.
52
Bracke
ME
,
Charlier
C
,
Bruyneel
EA
, et al
. 
Tamoxifen restores the E-cadherin function in human breast cancer MCF-7/6 cells and suppresses their invasive phenotype
.
Cancer Res
1994
;
54
:
4607
9
.