Association of +331G/A PgR Polymorphism with Susceptibility to Female Reproductive Cancer: Evidence from a Meta-Analysis

The progesterone receptor (PgR), a sex steroid hormone receptor that binds progesterone is critical for normal breast development. The PgR (+331G/A, rs10895068) promoter polymorphism is associated with cancer risk possibly by altering the expression of progesterone receptor B isoform. Previous studies have provided inconsistent results. To validate the association between the PgR +331G/A polymorphism and female reproductive cancer risk (breast, endometrial and ovarian cancer), we performed a meta-analysis of 19 studies (19,978 cases and 24,525 controls) by using the CMA Version 2 software. Odds ratios (ORs) and 95% confidence intervals (CIs) were used to assess the strength of the associations. The overall results indicated that the variant allele and genotypes were associated with a mild increase in overall female reproductive cancer risk (A vs. G: OR = 1.063, 95% CI = 1.001–1.129; AA+AG vs. GG: OR = 1.067, 95% CI = 1.002–1.136). The results suggest that the PgR +331G/A polymorphism might be associated with an increased female reproductive cancer risk.


Introduction
It is now well established that cancer is a multifactorial disease with an orchestrated relationship between genetic and environmental factors [1,2]. Despite intense study, the number of cancer cases continues to dramatically increase globally. According to American Cancer Society in 2012, more than 500,000 Americans are expected to die from cancer with more than 1500 deaths per day [3].
The progesterone receptor (PgR) is a member of a large ligandactivated nuclear receptor family characterised by a central DNA binding domain, a carboxyl-terminal ligand binding domain and multiple activation (AF) and inhibitory (IF) functional elements. The PgR, which is located on chromosome 11q22, and the oestrogen receptor (ER) have diverse reproductive functions associated with the establishment and maintenance of pregnancy, alveolar development and mammary epithelial development [4,5,6]. Progesterone, a steroid hormone, binds to the PgR, causing dimerization and thereby activating its target genes to transcribe through its response elements (PRE). The binding and activation are responsible for the mammary epithelial proliferation [7]. The PgR gene is transcribed by two alternative promoters that translate into two isoforms, progesterone receptor A (PgRA) and progesterone receptor B (PgRB) through which the PgR mediates its physiological functions [6]. The ratio of the expression of the two isoforms is important for the normal physiological function of progesterone signalling, which gets abrogated in cancer [8]. Both isoforms are identical except for the absence of 164 amino acids in N-terminal end of PgRA, which are present in the PgRB. This makes PgRB unique from PgRA because of the presence of AF-3 apart from AF-1 and AF-2, which are normally present in PgRA [9]. Six variable sites, four polymorphism and five common haplotypes have been detected in the PgR gene [10]. The most widely studied polymorphism in the promoter region of the PgR gene is the G to A substitution at position +331 (progesterone receptor, +331 G/A, rs10895068, +331 from ATG start codon). Interestingly, a unique GATA-5 binding site that is adjacent to the +331G/A PgR polymorphism has been found that leads to increased transcriptional activity of PgRB isoforms. In vitro overexpression of GATA-5 with hPgR luciferase construct showed greater transcriptional activity with +331AA than with +331GG included in the luciferase construct thereby contributing to breast tumorigenesis [11].
Recently, many studies have investigated the role of the PgR +331G/A polymorphism in the etiology of various types of cancer, i.e., breast, ovary and endometrial cancer. However, the results of these studies remain inconclusive. To clarify the role of +331G/A PgR polymorphism in female reproductive cancers, we performed a meta-analysis of all of the eligible case-control studies to derive overall cancer risk associated this polymorphism ( Figure S1).

Identification and eligibility of relevant study
Recent publications were identified through a literature search using the keywords ''Progesterone receptor polymorphism,'' ''PgR polymorphism'' or ''PR polymorphism'', and ''+331G.A'', or ''+331G/A'' and ''cancer'' in Pubmed and cross-references were checked that were not available in Pubmed or Science Direct (published before July 2012). Only those articles written in English were selected for the study. The following criteria were used for inclusion of identified articles for our meta-analysis: (a) evaluation of +331 G/A and cancer risk, (b) case-control study or cohort design and (c) studies that contained available genotype frequencies. The major reason of exclusion of some investigations was the absence of usable data reported in those studies. Finally, the data were extracted from 19 case-control studies totalling to 19,978 cases and 24,525 controls for +331G/A PgR polymorphism.

Data extraction
Two investigators (SC and DRM) independently extracted all of the data and jointly reached the consensus for all of the articles that were included in the study. Disagreement between the authors was resolved after discussion amongst the authors. The following information was sought from each article: first author's name, year of publication, source of control, country, ethnicity and numbers of cases and controls, genotype frequencies and allele frequencies for each case and control.

Meta-analysis
The statistical analysis for the current meta-analysis study was performed by using the comprehensive meta-analysis (CMA) V2 software (Biostat, USA). Meta-analysis is a powerful tool which combines results of independent similar studies and derive a definitive conclusion [12]. CMA V2 has several advantages over other software available for computing meta-analyses (http:// www.meta-analysis.com/pages/comparisons.html). The strength of the association between the PgR +331G/A polymorphism and risk of cancer was measured by odd ratios (ORs) with 95% confidence intervals (CIs). We examined the association between the allele A of the PgR +331G/A polymorphism and cancer risk, and made comparisons with the dominant genetic model (AA+AG vs. GG). A recessive genetic model was not performed as the data were not available for the homozygous mutant (AA) in the studied cases. The heterogeneity assumption was evaluated with an I 2based Cochran's Q statistic test. A significant p value (,0.10) was used to indicate heterogeneity among studies, and the combined OR was calculated by using a random effect model [13]. In contrast, a fixed effect model [14] was used for the calculation of the combined OR for homogeneity among the studies. In addition, the I 2 statistic was used to quantify inter-study variability. This statistic ranged from 0 to 100%, where a value of 0% indicated no observed heterogeneity, and as the values increased the degree of heterogeneity increased (cut-off points include: I 2 = 0-25%, no heterogeneity; I 2 = 25-50%, moderate heterogeneity; I 2 = 50-75%, large heterogeneity; I 2 = 75-100%, extreme heterogeneity) [15]. The funnel plot was employed to examine the publication bias. Egger's regression analysis was used for reevaluation of publication bias, and a P value less than 0.10 was considered to be significant. Funnel plots and Egger's linear regression tests were used to provide a diagnosis of the potential publication bias.

Study characteristics
For the association between PgR +331G/A polymorphism and cancer susceptibility, articles were retrieved based on the established search criteria. Nineteen studies met the inclusion criteria, the characteristics of which are reported in Table 1. Of these 19 studies, 10 were breast cancer (BC) studies in which all of the patients were of European ethnicity [11,[16][17][18][19][20][21][22][23][24]; six were ovarian cancer (OC) studies that only included patients of European origin [10,17,20,25,26,27] and three studies were endometrial cancer (EC) studies, which enrolled patients of different ethnic groups [28,29,30].

Publication bias
Begg's Funnel plot and Egger's test was conducted to assess whether there was any publication bias in the studies included in our meta-analysis. The shape of the funnel did not elucidate any obvious asymmetry in all of the comparison models. Thereafter, Egger's test was used to provide statistical evidence of the funnel plot symmetry and did not show any publication bias (

Meta-analysis result
Our meta-analysis identified an evident mild association between PgR (+331G/A) polymorphism and an increased cancer risk. Compared with the wild-type G allele, the overall variant A allele was associated with a mild increased cancer risk

Discussion
Genetic epidemiological studies have suggested the relationship between different SNPs and disease. However, robust statistical powers with genotype-phenotype investigation are essential to detect the mild, moderate or strong association with the disease. The progesterone hormone mediates its physiological function through the PgR, a steroid receptor of class I nuclear receptor super-family. Two different isoforms of PgR (PgRA and PgRB), which are transcribed from two different promoters of the same gene [31,32] are co-expressed in equal levels in normal mammary epithelial [33], ovarian [34] and endometrial cells [35]. The relative expression of this two isoform is altered in malignant transformation. The PgR +331 G/A polymorphism in the promoter results in the introduction of a TATA box that selectively enhances the production of the PgRB isoform, causing an altered PgRA: PgRB protein ratio [36]. The PgR +331 G/A promoter polymorphism has been extensively studied because of its relationship in different cancers such as breast, endometrial and ovarian cancer. However, the results from these studies were ambiguous. Apart from the PgR +331 G/A polymorphism, there are additional Table 1. Characteristics of all studies included in meta-analysis. polymorphisms such as +44C/T, S344T, G393G, V600L, H770H and Alulns allele [37]. We evaluated the +331G/A PgR promoter polymorphism for two reasons: (1) its association with many cancers like endometrial, ovarian and breast cancer, although there are conflicting observations and (2) the functional attribute of +331G/A PgR has been delineated and has been found to play an important role in hormone related cancer.
In previous published pooled studies, Yang and colleagues have reported an association between PgR +331 G/A variant and breast cancer risk [38]. However, the other meta-analysis reviewed an additional 4 studies and did not find this association [39]. On the basis of availability of more studies on PgR +331 G/A polymorphism and female hormone dependent cancer, we did a meta-analysis to determine conclusively whether an association exists between them. We detected a mild association between PgR +331 G/A polymorphism with overall female cancer risk. As all the studies involved women of only Caucasian origin, we did not perform any subgroup analysis by ethnicity. The prevalence of a minor allele A is restricted to only the European population (5-10%) and is rare in women of African, Asian and Middle Eastern descent which might be the reason for mild association even if the included studies are high. [40]. The non-detection of association between the PgR +331 G/A polymorphism and breast cancer was consistent with the previous study [39].
A genome-wide association studies (GWAS) have been carried out for various cancers (breast, prostate, colorectal, and lung) and several susceptible alleles have been identified [41]. None of the GWAS have identified the +331 G/A PgR polymorphism as a risk factor for cancer predisposition despite this functional SNP has been established. In contrast, our meta-analysis results revealed an association of the PgR (+331 G/A) polymorphism with a susceptibility to cancer. Corroborating with current study, discrepancies between the GWAS and candidate gene association studies (CG) have been reported for different disease [42,43,44,45]. CG studies tend to have higher statistical power than the power in GWAS [46].
The progesterone hormone exerts its physiologic effect exclusively through the presence of the PgR as shown in mouse models [47,48]. The PgR isoforms, PgRA and PgRB, display different transactivation properties based on the cell type and target promoter [49]. PgRB functions as a strong transcriptional activator of the PgR-dependent promoters in PgRA-inactive cell types, but the agonist-bound PgRA under this condition can repress the PgRB transcriptional activity and other steroid receptors like ER a [50]. Progesterone acts as an antagonist of oestrogen (E2)-induced cell proliferation of uterine epithelium through PgRA [48]. Treating PRAKO 2/2 mice (deficient in PgRA) with E2 and progesterone resulted in a progesteronedependent cell proliferation, which was not detected in E2 and progesterone-treated wild-type or PRAKO mice, suggesting that PgRB has a role in uterine endometrium proliferation. Further-more, the PgRB isoform could also elicit normal proliferation and differentiation as observed in PRAKO 2/2 mammary epithelium through progesterone. During endometriosis, only the PgRA isoform is expressed and the absence of PgRB is suggested due to the inappropriate cycling of the endometrial gland [25]; however PgRA expression in normal cycling is predominant during the proliferative phase, which shifts towards PgRB expression during the early secretary phase [33]. The upregulation and proliferative nature of the PgRB isoform in the carriers of the +331A allele may result in different types of cancer [25]. With respect to this biological postulation, the +331G/A PgR may modulate cancer risk. However, the association of this polymorphism with overall cancer risk was mild (OR = 1.063, p = 0.048). It is important to note that expression of the PgR is associated with better disease-free survival [51]. An alteration in the ratio of the expression of the PgR isoforms precedes changes that may lead to endometrial carcinoma [52]. The increase in PgRB expression due to polymorphism in promoter leads to changes in the isoform ratio and is associated with an increased risk of developing endometrial cancer [53]. Additionally, a case-control study of endometrial cancer with the +331G/A PgR polymorphism and PgR expression predicted that the recurrence risk and clinical response to progesterone therapy was six times more likely in women with PR(+) than with PR(2) tumors. Progesterone therapy is effective against the developed endometrial cancer, which is dependent on oestrogen-mediated proliferation. Also, PR(+) endometrial cancer was an independent prognostic factor in disease-free survival [54]. The finding that PgRB could contribute to cell proliferation may provide clinical implications for hormonal management of endometrial and mammalian epithelium by rectifying relative expression of the PgRA: PgRB isoform, which is essential for appropriate reproductive tissue responses. Isoform-specific modulators like progestin, which could differentiate between PgRA and PgRB isoform, may also be of clinical value.
Although earlier reports on the association of PgR (+331G/A) polymorphism and cancer are inconsistent, the current metaanalysis has provided a definitive conclusion. Inconsistencies among the CG association studies could be attributed to small sample sizes, improper clinical categorisation and the inclusion of different ethnicities.

Conclusion
A mild association between PgR +331G/A polymorphism and female reproductive cancer risk was detected. As the eligible case-control studies cannot provide a casual relationship, large and well-designed genotype-phenotype studies on different cancer are necessary to derive a definitive role of this SNP with cancer.  Figure 1. Forest plot of overall cancer risk associated with +331G/A PgR polymorphism. The squares and horizontal lines correspond to the study-specific OR and 95% CI. The area of the squares reflects the weight of the respective study. The diamond represents the pooled OR and 95% CI. Forest plots evaluating the association of overall allele (I), and dominant (II) genetic model with cancer risk are presented. Breast cancer, endometrial cancer and ovarian cancer denoted as a, b, and c respectively. doi:10.1371/journal.pone.0053308.g001