The genomic regions surrounding GNRH1 and GNRHR are shown in Figure . GNRH1 consists of a single LD block, whereas GNRHR includes two LD blocks, one of them including exon 1 and the other exons 2 and 3. GNRH1 was tagged by 3 SNPs, which account for 94% of haplotype diversity. Block 1 of GNRHR was tagged by 3 SNPs and block 2 by 4 SNPs (98% and 95% of haplotype diversity, respectively). Frequency of common haplotypes ranged between 19 and 35% in controls for GNRH1 and 5% and 52% for GNRHR.
Figure 1 Haploview plot of the genomic region of GNRH1 (A) and GNRHR (B). From top to bottom: position of genes (boxes: exons, lines: introns), SNPs genotyped in the multiethnic panel, graphical representation of LD and block structure (darker color represents (more ...)
A total of 5,603 invasive breast cancer cases and 7,480 controls were available for genotyping from each of the participating cohorts. Samples not yielding a genotype were removed from individual SNP analyses, and samples not yielding at least one genotype were removed from haplotype analyses. Both between-center genotyping concordance and within-center blinded quality control concordance were above 99%. Genotype success rate among cases and controls in all cohorts was greater than 95%. No polymorphisms deviated from Hardy-Weinberg Equilibrium among the controls.
Detailed results of associations between serum concentrations of steroid hormones and SNPs are presented in Additional file 1
. SNP rs2630488 within GNRHR
showed a nominally significant association (p = 0.04) with estradiol levels in post-menopausal women. Presence of the minor allele at this polymorphism was associated with an increase in estradiol level (a 4% increase among homozygotes for the minor allele, as compared to homozygotes for the common allele). However, this effect was entirely driven by the association observed in post-menopausal breast cancer cases within EPIC (p = 0.012 in this subgroup), and was not observed in the other subgroups. A few borderline associations were observed between hormone levels and SNPs in pre-menopausal women, however sample size of this group of subjects was considerably smaller, and all subjects derived from only one cohort (EPIC). No associations between polymorphisms and hormone levels remained significant after correction for multiple testing.
Results of association analyses between htSNPs of GNRH1 and breast cancer risk are presented in Table , and results of haplotype analysis in Table . Results of analyses for GNRHR are presented in Tables and . No association was observed for any of the htSNPs of either gene. Haplotype analysis also showed no association, with global tests for comparison of haplotypes frequency in cases and controls resulting in non-significant results (Wald tests: d.f. = 4, p = 0.364 for GNRH1; d.f. = 5, p = 0.897 for GNRHR block 1 and d.f. = 6, p = 0.967 for GNRHR block 2). Analyses were unadjusted (conditional on matching criteria) or adjusted for known breast cancer risk factors, but results did not show any difference.
Association between GNRH1 htSNPs and breast cancer risk in the BPC3 study.
Association between GNRH1 haplotypes and breast cancer risk in the BPC3 study.
Association between GNRHR htSNPs and breast cancer risk in the BPC3 study.
Association between GNRHR haplotypes and breast cancer risk in the BPC3 study.
Analyses performed by stratifying cases by age at diagnosis (greater or lower than 55 years), localized or metastatic disease or estrogen/progesterone receptor status did not show significant differences. Stratification of subjects by cohort, country in EPIC or ethnicity in MEC showed only few results supported by p values ranging from 0.01 to 0.05, which were always based on a small number of subjects. For these, we performed heterogeneity tests, which in all cases were not statistically significant. For example, heterozygotes for SNP rs1812594 of GNRH1 had an odds ratio (OR) of 1.08 (95% confidence interval (CI) = 1.00–-1.16, p = 0.04). When we analyzed the results for each cohort, it resulted that the association was driven by EPIC data (OR = 1.16, 95% CI 1.00–1.34, p = 0.046), and within EPIC the only significant result came from the Spanish sub-cohort (OR = 1.59, 95% CI = 1.07–2.37, p = 0.021), which is based on 62 cases and 79 controls heterozygous for this SNP. However, heterogeneity tests for this genotype were not statistically significant for either the entire study (p = 0.226) or within EPIC (p = 0.147). Nor were homozygotes for the less common allele at this SNP significantly associated with increased risk in any subgroup.
No statistically significant interactions were observed between htSNPs or haplotypes and known breast cancer risk factors, including age at first full term pregnancy, number of pregnancies, never/ever menopausal hormone therapy, height, smoking status, or body mass index.