Breast Cancer Research and Treatment

, Volume 126, Issue 1, pp 37–45

A systematic review of the relationship between polymorphic sites in the estrogen receptor-beta (ESR2) gene and breast cancer risk

Authors

  • Ke-Da Yu
    • Department of Breast Surgery, Cancer Hospital/Cancer Institute, Breast Cancer InstituteFudan University
    • Department of Oncology, Shanghai Medical CollegeFudan University
  • Nan-Yan Rao
    • Department of Breast SurgerySecond Affiliated Hospital of Sun Yat-sen University
  • Ao-Xiang Chen
    • Department of Breast Surgery, Cancer Hospital/Cancer Institute, Breast Cancer InstituteFudan University
    • Department of Oncology, Shanghai Medical CollegeFudan University
  • Lei Fan
    • Department of Breast Surgery, Cancer Hospital/Cancer Institute, Breast Cancer InstituteFudan University
    • Department of Oncology, Shanghai Medical CollegeFudan University
  • Chen Yang
    • Department of Breast Surgery, Cancer Hospital/Cancer Institute, Breast Cancer InstituteFudan University
    • Department of Oncology, Shanghai Medical CollegeFudan University
    • Department of Breast Surgery, Cancer Hospital/Cancer Institute, Breast Cancer InstituteFudan University
    • Department of Oncology, Shanghai Medical CollegeFudan University
    • Institutes of Biomedical ScienceFudan University
Preclinical study

DOI: 10.1007/s10549-010-0891-2

Cite this article as:
Yu, K., Rao, N., Chen, A. et al. Breast Cancer Res Treat (2011) 126: 37. doi:10.1007/s10549-010-0891-2

Abstract

The estrogen signal is mediated by the estrogen receptor (ER). The specific role of ER-beta, a second ER, in breast carcinogenesis is not known. A number of association studies have been carried out to investigate the relationship between polymorphic sites in the ESR2 gene and breast cancer risk, however, the results are inconsistent. We searched PubMed, Medline, and Web of Science database (updated to 10 January 2010) and identified 13 relevant case–control studies, and approximately 28 single-nucleotide polymorphisms (SNPs) and one micro-satellite marker were reported in the literature. The median number of study subjects was 776 (range 158–13,550). Three genetic variants [(CA)n, rs2987983, and rs4986938] showed significant overall associations with breast cancer, and rs4986938 was reported twice. Because rs4986938 and rs1256049 were the most extensively studied polymorphisms, we subsequently conducted a meta-analysis to evaluate their relationship with breast cancer risk (9 studies of 10,837 cases and 16,021 controls for rs4986938; 8 studies of 11,652 cases and 15,726 controls for rs1256049). For rs4986938, the women harboring variant allele seemed to be associated with a decreased risk either in the dominant model [pooled OR = 0.944, 95% confidence interval (95% CI) 0.897–0.993, fixed-effects] or in the co-dominant model (AG vs. GG) (OR = 0.944, 95% CI 0.895–0.997, fixed-effects). rs1256049 was not associated with breast cancer risk in any model. Five studies had investigated the effect of haplotypes in the ESR2 gene on breast cancer risk, and four of them had positive outcomes. In summary, the present systematic review suggests that SNP rs4986938 as well as haplotypes in the ESR2 gene might be associated with breast cancer. The need for additional studies examining these issues seems of vital importance.

Keywords

ESR2PolymorphismBreast cancerSystematic reviewMeta-analysis

Introduction

Estrogen plays an important role in the development of breast cancer [1]. The estrogen signal is mediated by the estrogen receptor (ER), which is a transcription factor belonging to the steroid hormone receptor super-family. The second ER, ER-beta, was identified in 1996 [2], and since then the former ER has been called ER-alpha. ER-alpha and ER-beta are coded by two separate genes, ESR1 on chromosome 6 and ESR2 on chromosome 14 [3]. Both ER-alpha and ER-beta proteins are expressed in normal breast luminal epithelial cells as well as in breast tumors [4, 5].

Whereas the specific functions of ER-beta in breast carcinogenesis are not known yet, in vitro studies suggest that ER-beta variations may influence the susceptibility to and development of breast cancer [5, 6]. Given the potential role of ER-beta variations in breast carcinogenesis, it is reasonable to speculate that polymorphic sites in the ESR2 gene might be associated with risk of breast cancer. The ESR2 genetic variants have been investigated for their associations with body weight [7], menstrual disorders [8], anorexia nervosa [9], Alzheimer’s disease [10], and prostate cancer [11]. With regard to breast cancer, a number of association studies have been carried out, though the results are inconsistent.

In this study, we carried out a systematic review and a meta-analysis focusing on the relationship between polymorphisms within the ESR2 gene and breast cancer risk.

Methods

Literature search

Relevant studies were selected by searching PubMed, Medline, and Web of Science database (updated to 10 January 2010) using the following search terms: (ESR2 or ESR-2 or “estrogen receptor beta” or ER-beta or ER-β) and (polymorphism* or variant* or variation*) and breast. Eligible studies were retrieved and examined carefully. Their references were checked as well for other relevant publications. Review articles were also inspected to find additional eligible studies. Only studies published in English were included for systematic review and meta-analysis; we did not define any minimum number of subjects to be included. For overlapping studies, all of them were described in systematic review, but only the one with the largest sample number was included for meta-analysis.

Eligible studies and data extraction

The identified articles were assessed independently by two of the authors (K.-D. Yu and A.-X. Chen), and any discrepancy in studies’ eligibility were adjudicated by Professor Z.-M. Shao. The inclusion criteria were as following: (i) evaluation of the association between polymorphic sites in the ESR2 gene and breast cancer risk (cancer patients vs. cancer-free controls), (ii) retrospective case–control studies or prospective cohort studies, (iii) having available odds ratio (OR) with its 95% confidence interval (95% CI) of polymorphisms or haplotypes, or with sufficient available genotyping data to estimate these parameters, and (iv) fulfilling Hardy–Weinberg equilibrium (HWE). Departure from HWE or not was evaluated in the control population with the same ethnicity (P < 0.01 indicating a departure from HWE), but a deviation from HWE in a mixed control population was allowed [12, 13]. Studies should meet (i)–(iii) criteria for systematic review, and should meet all the four criteria for meta-analysis. Any study with wrong data or inconsistent data was excluded. The following variables were extracted from each study if available: first author’s surname or study organization name, publication year, source of controls (hospital-based or population-based), ethnicity, genetic variants studied, and numbers of cases and controls in different genotypes of studied polymorphisms whenever possible. Information was carefully extracted from all the eligible publications, independently by two of the authors (K.-D. Yu and A. X. Chen). Disagreement was resolved by discussion between the authors. If they could not reach a consensus, another investigator (Z. M. Shao) adjudicated over the disagreement.

Statistical methods

Meta-analysis was mainly performed as described previously [14, 15]. Briefly, for each study, the OR with its 95% CI was calculated to assess the association strength between a certain polymorphism and breast cancer risk. The pooled OR was calculated by a fixed-effects model (using the Mantel–Haenszel method) or a random-effects model (using the DerSimonian and Laird method) according to the heterogeneity among studies. Heterogeneity assumption was checked by the Q test and a P value >0.10 indicated a lack of heterogeneity. If P < 0.10, the between-study heterogeneity was considered to be significant, and we chose the random-effects model to calculate the pooled OR; otherwise, the fixed-effects model was employed. Four different types of OR were calculated: (i) aa genotype versus AA genotype, (ii) Aa genotype versus AA genotype, (iii) aa + Aa genotypes versus AA genotype (the dominant model), and (iv) aa genotype versus Aa + AA genotypes (the recessive model). The potential publication bias was examined visually in a funnel plot of log [OR] against its standard error (SE), and the degree of asymmetry was tested using Egger’s test (P < 0.05 considered to be statistically significant). We also performed influence analysis by omitting each study to find potential outliers. All of the statistical analyses were performed using Stata/SE version 10.0 (Stata Corporation, College Station, TX, USA).

Results

Systematic review

We identified 13 eligible studies for systematic review [1628] (Table 1). One study consisting of 1,011 cases and 615 controls provided ORs of haplotypes but not ORs of single polymorphisms [18]. Approximately 28 single-nucleotide polymorphisms (SNPs) and one micro-satellite marker were studied in the literature. Most reported studies were in a relative small sample size: the median number of study subjects was 776 (range 158–13,550). To present results in a clear and comparable manner, we gave the genotypic risks in a dominant model and only showed the significant results. If some results were not reported properly, we estimated the ORs with 95% CIs from the published raw data.
Table 1

Characteristics of all the eligible studies regarding polymorphisms in the ESR2 gene and breast cancer risk

No.

Control source

Year

Author

Ethnicity

Polymorphic sites

OR (95% CI)

Factors of OR adjustment

Cases/controls

Included in the meta-analysis

Risk factors distribution in cases versus controls

1

HB

2003

Försti et al. [16]

Caucasian (Finnish)

nt805(del21); 846G > A; rs1256049; rs4986938; (CA)n; 1505-4 A > G

NS

No

219/248

Yes

Case: post-menopausal women, mean age of 63 years (range 50–76) and did not belong to breast cancer families.

Controls: from same geographic areas and matched for ethnicity.

2

PB

2003

Zheng et al. [17]

Asian (Chinese)

rs1271572; −11943G > A; rs3829768; Int5(16-bp down of exon4); rs1256049; rs1256054; 50766A > G; 50995G > A

NS

Adjusted for menopausal status, age, age at first birth, age at menarche, BMI, and family history

1,134/1,235

Yes

Age 47.79 ± 7.99 versus 47.24 ± 9.11 years (the whole population of Shanghai Breast Cancer Study).

Age at menarche ≤14 years: 52.84% versus 48.26%.

First degree of relatives with breast cancer: 3.7% versus 2.44%.

Age at menopause ≥52 years: 20.96% versus 16.01%.

Age at first live birth ≤27 years: 67.07% versus 71.33%.

3

PB

2004

Gold et al. [18]a

Mixed

rs1255998; rs928554; rs1152579; rs4986938; rs1256049; E2EX4CorT; rs1256030; rs1271572

NS

Stratified by ethnicity and age

1,011/615

No

N.M.

4

PB

2005

Maguire et al. [19]a

Caucasian (Swedish)

rs1256049; rs4986938; rs928554

NS

Stratified by family history

723/480

Yes

Of the breast cancer cases: 323 sporadic and 400 familial.

5

PB

2006

Gallicchio et al. [20]

Caucasian

rs4986938; rs928554; 5,696-bp 3′of STP-A > G; rs8018687

NS

Adjusted for age and menopause

91/1,347

No

N.M.

6

PB

2006

Iobagiu et al. [21]

Caucasian

(CA)n

NS

Adjusted for three microsatellites’ combination

139/145

No

Age: median 60 years versus 48 years

7

HB

2007

Tsezou et al. [22]

Caucasian (Greek)

(CA)n

0.010 (0.003-0.036)c

Adjusted for BMI, age, age at menarche, menopause, and family history

79/155

No

Age: mean 57.6 years versus 70.9 years

Age of menarche: 12.75 years versus 12.89 years

Age of menopause: 47.6 years versus 48.9 years

BMI: 27.4 versus 26.2

8

PB

2008

BPC3 [23]a

Mixed

rs1256049; rs1256031; rs3020450

NS

No

5,789/7,761

Yes

N.M.

rs4986938

0.93 (0.86–0.99)d

No

9

PB

2009

Treeck et al. [24]

Caucasian

rs2987983

1.99 (1.23–3.23)

No

318/318

No

N.M.

rs3020450; rs3020449

NS

No

318/318

10

PB

2009

Sonestedt et al. [25]b

Caucasian (Swedish)

rs915057; rs1269056; rs1256033; rs3020450; rs3020443

NS

Stratified by enterolactone concentration

542/1076

No

Age: 56.6 years (50.9-62.4) versus 56.6 years (50.8–63)

Nonsmokers: 42 versus 46%

Current hormone therapy use: 33 versus 21%

11

HB

2009

Surekha et al. [26]

Asian (India)

rs4986938

0.41 (0.16–0.95)d

No

250/250

Yes

N.M.

12

PB

2009

MARIE-GENICA Consortium [27]a

Caucasian (German)

rs944050; rs4986938; rs1255998; rs1271572; rs1256049; rs928554

NS

Adjusted for estrogen/progesterone mono-therapy, tibolone, unknown hormone, type of menopause, number of births, breastfeeding, smoking, number of mammograms, benign breast disease, family history, and BMI

3,149/5,489

Yes

With benign breast disease: 41.3 versus 34.4%

Smokers: 45.4 versus 48.8%

Ever user of oral contraceptives: 62.1 versus 63.9%.

First degree relative with breast cancer: 16.1 versus 11.7%.

13

HB

2009

Iwasaki et al. [28]

Japanese

rs4986938; rs1256049

NS

Adjusted for menopausal status, number of births, family history of breast cancer, smoking status, moderate physical activity in the past 5 years, and vitamin supplement use

388/388

Yes

Control matched for each case by age (within 5 years) and ethnicity.

Jap.Brazilian

NS

79/79

Yes

Non-Jap.Brazilian

NS

379/379

Yes

BPC3 Breast and prostate cancer cohort Consortium, NS no significance, OR odds ratio, CI confidence interval, HB hospital-based, PB population-based, Jap. Japanese, N.M. not mentioned

aStudy including haplotypes-based analysis and showing statistic significance

bStudy including haplotypes-based analysis but showing no statistic significance

cAdjusted OR = 0.002 (95% CI 0.000–0.022)

dCalculated according to raw data

Although many polymorphisms in the ESR2 gene had been studied, few investigators reported significant results for a single variant. In this systematic review, the associations that had been reported as significant by at least one study were discussed below. Regarding the overall risk estimate, three genetic variants had showed significant relationships with breast cancer in four studies [2224, 26]; rs4986938 was reported twice [23, 26]. In the Breast and Prostate Cancer Cohort Consortium (BPC3) study [23] with the largest sample size, the investigators indicated that the A-allele carriers had a decrease in breast cancer risk (OR = 0.93, 95% CI 0.86–0.99), though the trend P value of unadjusted logistic regression did not reach significance. In another study, Surekha et al. [26] found that India women harboring the A allele of rs4986938 also had a reduced risk for breast cancer (OR = 0.41, 95% CI 0.16–0.95). Treeck et al. [24] found a significantly increased risk of breast cancer in variant carriers of SNP rs2987983 in Caucasian women (OR = 1.99, 95% CI 1.23–3.23); Tsezou et al. [22] revealed an association between micro-satellite (CA)n and breast cancer in Greek women (OR = 0.010, 95% CI 0.003–0.036; adjusted OR = 0.002, 95% CI 0.000–0.022). An additional study reported by Maguire et al. [19] implied that rs1256049 might decrease the breast cancer risk in Sweden women with a borderline significance (OR = 0.63, 95% CI 0.37–1.07).

Regarding gene–environment interaction, some investigators had reported interactions of ESR2 genetic variants with endogenous/exogenous exposure factors. Zheng et al. [17] found an increased breast cancer risk in rs1256054 variant allele carriers with long duration (>34 years) of menstruation (OR = 2.37, 95% CI 1.18–4.77). In a recent study by MARIE-GENICA Consortium [27], the investigators found that three SNPs (rs4986938, rs1271572, rs928554) could modify the relationship between postmenopausal breast cancer risk and estrogen monotherapy use. For rs4986938, its relationship with breast cancer risk could also be modified by isoflavone intake in different populations (Japanese Brazilian and non-Japanese Brazilian subjects) [28]. However, no further studies attempting to reconfirm these complex interactions had been published yet.

We then reviewed the effects of haplotypes in the ESR2 gene on breast cancer risk. Five studies investigated this issue and four studies found significantly increased risks in carriers of some haplotypes [18, 19, 23, 25, 27]. Of note, two studies with the largest sample sizes (MARIE-GENICA Consortium study [27] and BPC3 study [23]) observed an obvious genetic contribution of haplotypes in the ESR2 gene to breast cancer.

Meta-analysis

Thus far, rs4986938 and rs1256049 had been the most extensively studied polymorphisms. The small effect of each single polymorphism on risk and the relatively small sample size in each published study made the results inconclusive and controversial. We subsequently meta-analyzed the association between breast cancer and rs4986938 or rs1256049. The eligible studies for meta-analysis were marked in Table 1. Table 2 lists the additional characteristics of each study. No study had a deviation from HWE in controls with the same ethnicity.
Table 2

Characteristics of studies included in meta-analysis of the two SNPs

SNP

Year

Author

Cases

Controls

HWE

MAF

AA

Aa

aa

AA

Aa

aa

rs4986938

2003

Försti

95

99

25

105

103

30

0.547

0.3424

 

2005

Maguire

298

315

83

175

190

56

0.697

0.3587

 

2006

Gallicchio

26

43

19

470

612

190

0.688

0.3899

 

2008

BPC3

2,513

2,382

705

3,229

3,304

984

0.003

0.3507

 

2009

Surekha

21

95

132

9

81

159

0.738

0.8012

 

2009

MARIE-GENICA Consortium

1,277

1,431

432

2,169

2,557

752

0.971

0.3707

 

2009

Iwasaki-Japanese

289

94

5

281

102

5

0.205

0.1443

 

2009

Iwasaki-Jap.Brazilian

59

17

3

60

17

2

0.555

0.1329

 

2009

Iwasaki-non-Jap.Brazilian

169

163

47

176

154

49

0.100

0.3325

   

10,837

16,021

  

rs1256049

2003

Försti

189

30

0

198

39

1

0.528

0.0861

 

2003

Zheng

480

506

127

537

541

131

0.762

0.3321

 

2005

Maguire

628

52

1

356

41

1

0.874

0.054

 

2008

BPC3

4,987

610

50

6,751

734

70

<0.001

0.0578

 

2009

MARIE-GENICA Consortium

2,921

224

1

5,086

389

5

0.384

0.0364

 

2009

Iwasaki-Japanese

203

161

24

182

178

28

0.079

0.3015

 

2009

Iwasaki-Jap.Brazilian

47

26

6

48

30

1

0.119

0.2025

 

2009

Iwasaki-non-Jap.Brazilian

342

36

1

345

32

2

0.194

0.0475

   

11,652

15,726

  

HWE Hardy–Weinberg equilibrium, MAF minor allele frequency, A major allele, a minor allele

Table 3 presents the results of meta-analyses from seven publications (9 studies 10,837 cases and 16,021 controls) of rs4986938 and six publications (8 studies 11,652 cases and 15,726 controls) of rs1256049. For rs4986938, the women harboring variant allele seemed to be associated with a reduced breast cancer risk either in the dominant model (pooled OR = 0.944, 95% CI 0.897–0.993, fixed-effects) or in the co-dominant model (AG vs. GG) (OR = 0.944, 95% CI 0.895–0.997, fixed-effects). For rs1256049, however, it was not associated with breast cancer risk in any model (co-dominant, dominant, or recessive). In the subgroup analysis by ethnicity, no significantly increased risk was found for variant carriers of rs4986938 or rs1256049 among Caucasians. Because there were very limited studies of non-Caucasian populations, we did not perform subgroup analysis in Asians or Africans.
Table 3

Pooled ORs of the two SNPs in different genetic models

Analysis model

rs4986938 OR (95% CI)

rs1256049 OR (95% CI)

Overall (n = 9; 10,837:16,021)a

Caucasians (n = 4; 4,143:7,409)

Overall (n = 8; 11,652:15,726)a

Caucasians (n = 3; 4,046:6,116)

Codominant (Het. vs. Common Hom.)

0.944 (0.895–0.997)

0.966 (0.887–1.053)

1.035 (0.959–1.117)

0.945 (0.811–1.100)

P/Ph

0.037/0.721

0.433/0.686

0.376/0.259

0.463/0.303

Codominant (Rare Hom. vs. Common Hom.)

0.941(0.868–1.019)

0.985 (0.870–1.115)

1.001 (0.820–1.222)

0.392 (0.089–1.737)

P/Ph

0.137/0.217

0.811/0.240

0.990/0.576

0.218/0.959

Dominant (Rare Hom. + Het. vs. Common Hom.)

0.944 (0.897–0.993)

0.970 (0.895–1.052)

1.031 (0.958–1.110)

0.935 (0.804–1.088)

P/Ph

0.026/0.379

0.464/0.470

0.418/0.254

0.387/0.291

Recessive (Rare Hom. vs. Het. + Common Hom.)

0.960 (0.892–1.034)

1.004 (0.895–1.126)

1.001 (0.825–1.214)

0.398 (0.090–1.759)

P/Ph

0.281/0.343

0.948/0.339

0.992/0.628

0.224/0.955

Het. heterozygous, Hom. homozygous, PhP value of Q test for heterogeneity test, OR odds ratio, CI confidence interval

aStudy number; numbers of cases and controls

In addition, we evaluated the influence of any individual study on the overall OR for rs4986938 and rs1256049, respectively (Fig. 1a, b). For rs4986938, it was likely that BPC3 study dominated the result, since once we omitted the BPC3 study, the OR (95% CI) was 0.97 (0.90–1.04) and the weight of BPC3 study in the pooled estimate was high up to 54.0%. For rs1256049, omission of any single study did not affect the null result. Moreover, Begg’s funnel plot and Egger’s test were performed to evaluate the publication bias of literature, and no significant publication bias was observed either in the dominate model (Fig. 1c, P = 0.748 for rs4986938; Fig. 1d, P = 0.078 for rs1256049) or in other genetic models (data not shown).
https://static-content.springer.com/image/art%3A10.1007%2Fs10549-010-0891-2/MediaObjects/10549_2010_891_Fig1_HTML.gif
Fig. 1

Influence analyses and publication bias plots for association between breast cancer risk and rs4986938 or rs1256049. a, b The influence of individual studies on the summary OR of rs4986938 and rs1256049, respectively. The vertical axis indicates the overall OR and the two vertical axes indicate its 95% CI. Every hollow round indicates the pooled OR when the left study is omitted in this meta-analysis. The two ends of every broken line represent the respective 95% CI. c, d The Begg’s funnel plots of studies included in the meta-analysis for rs4986938 and rs1256049 in the dominant model, respectively. The vertical axis represents log [OR] and the horizontal axis means the standard error of log [OR]. Horizontal line and sloping lines in funnel plot represent random-effect summary OR and expected 95% CI for a given standard error, respectively. Area of each circle represents contribution of the study to the pooled OR

Discussion

The present meta-analysis systematically reviewed the association between the genetic variants in the ESR2 gene and breast cancer risk. Our results indicate that one SNP, rs4986938, is likely to be a low-penetrant risk factor for developing breast cancer. Our systematic review also suggests that some haplotypes in the ESR2 gene are probably related to breast cancer.

To date, studies of association between polymorphisms in the ESR2 gene and breast cancer risk have yet been inconclusive. Few individual studies were statistically significant and no significant observation has been reported by more than two studies. In some cases, this might be due to a lack of statistical power in most studies. We in this study attempted to reach a more powerful conclusion by combining results for meta-analysis, and found that rs4986938 seemed to be a low-penetrance breast cancer susceptibility locus in the dominant model. It is biologically plausible because rs4986938 is located at 3′UTR of the ESR2 gene and this SNP could affect pre-mRNA splicing, mRNA stability and translatability [29]. However, the meta-analysis results should to be treated with cautions. First, among nine study populations eligible for our meta-analysis, only two studies showed positive results of overall ORs (BPC3 study [23] and Surekha’s study [26]). The BPC3 study had the largest sample size, played a dominant role, and weighted 54.0% of the overall OR. When we performed influence analysis by omitting BPC3 study, the result showed that rs4986938 was not associated with breast cancer any more. Second, in ethnicity subgroup analysis, neither rs4986938 nor rs1256049 was associated with breast cancer risk in any genetic model. Although the null results might be caused by the relatively limited study number (only four studies for rs4986938 and three studies for rs1256049 in subgroup analysis specific to Caucasians) and small sample size, we could rule out the possibility that rs4986938 is actually not associated with breast cancer. Our overall positive results might be dominated by BPC3 study.

Several studies had reported the gene–environment interactions. Again, rs4986938 could influence breast cancer developing by modifying estrogen exposure [27] or isoflavone intake [28]. Some other SNPs such as rs1271572 and rs928554 could also act as modifiers of the relationship between estrogen exposure and breast cancer risk [27]. These results of subgroup analysis, however, should be treated with caution. Multiple hypothesis tests render the interpretation of positive results difficult. More studies with larger sample size should be required to elucidate those effects. Someone may argue that the effect of a single polymorphism within a gene might have a limited impact on breast cancer susceptibility, and a haplotype-based approach needs to be carried out for a more objective evaluation. Four studies (including two studies with the largest sample sizes: MARIE-GENICA Consortium study and BPC3 study [23, 27]) had observed the obvious haplotype effects on ESR2 genetic contribution to breast cancer. Regretfully, different studies constructed haplotype using different SNPs, which made previously significant results hard to replicate.

Some limitations of this systematic review should be acknowledged. First, in any systematic review, a major concern is the potential affect of publication bias. The most common case is the non-publication of negative studies, resulting in the outcome away from the null [30]. Though we found no publication bias among the studies for meta-analysis, the positive outcome of rs4986938 still needs further validation. Second, there is no consensus nomenclature for all the studied genetic polymorphisms: early reports often described the restriction enzyme site involved; recent studies usually identified the base substitution. But it could also be arbitrary, since some studies presented polymorphism’s name according to its position in DNA sequence, while some presented it according to RNA or protein position. The requirement for public databases of SNPs has been recognized, and the standardized description of a SNP is to use a reference number. Third, the current overall OR is based on individual unadjusted ORs; a more precise evaluation should be adjusted by other potentially suspected factors including age, menopausal status, estrogen exposure, and environmental factors. Lastly, though most controls were selected from healthy populations, some controls were derived from women with benign breast diseases [20], which might lead to misclassification bias because those women have potential risks of developing breast cancer [31].

Despite these concerns, we believe that some conclusions could be drawn according to current evidence. First, rs4986938 is likely to be related to breast cancer risk, and it might also act as a modifier of the relationship between breast cancer risk and some environmental factors. We encourage further evaluations on the contribution of rs4986938 to breast cancer risk in larger, more comprehensive and well-designed association studies. Second, haplotypes of the ESR2 gene is implied to be associated with breast cancer. Further work is clearly needed to address these issues.

Acknowledgments

This research is supported by grants from the National Basic Research Program of China (2006CB910501), 2009 Youth Foundation of Shanghai Public Health Bureau, 2009 Youth Foundation of Shanghai Medical College, and the National Natural Science Foundation of China (30971143, 30972936).

Conflict of interest statement

All authors declared no potential conflicts of interest.

Copyright information

© Springer Science+Business Media, LLC. 2010