Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2014 Sep 10.
Published in final edited form as: Hum Genet. 2011 Sep 30;131(3):479–490. doi: 10.1007/s00439-011-1088-4

Fine Mapping of 14q24.1 Breast Cancer Susceptibility Locus

Phoebe Lee 1,*, Yi-Ping Fu 1,*, Jonine D Figueroa 1, Ludmila Prokunina-Olsson 1, Jesus Gonzalez-Bosquet 2, Peter Kraft 3, Zhaoming Wang 1,4, Kevin B Jacobs 1,4, Meredith Yeager 1,4, Marie-Josèphe Horner 1, Susan E Hankinson 5, Amy Hutchinson 1,4, Nilanjan Chatterjee 1, Montserrat Garcia-Closas 1, Regina G Ziegler 1, Christine D Berg 6, Saundra S Buys 7, Catherine A McCarty 8, Heather Spencer Feigelson 9, Michael J Thun 10, Ryan Diver 10, Ross Prentice 11, Rebecca Jackson 12, Charles Kooperberg 11, Rowan Chlebowski 13, Jolanta Lissowska 14, Beata Peplonska 15, Louise A Brinton 1, Margaret Tucker 1, Joseph F Fraumeni Jr 1, Robert N Hoover 1, Gilles Thomas 16, David J Hunter 3,5,17, Stephen J Chanock 1,**
PMCID: PMC4159746  NIHMSID: NIHMS616599  PMID: 21959381

Abstract

In the National Cancer Institute Cancer Genetic Markers of Susceptibility (CGEMS) genome-wide association study of breast cancer, a single nucleotide polymorphism (SNP) marker, rs999737, in the 14q24.1 interval, was associated with breast cancer risk. In order to fine map this region, we imputed a 3.93MB region flanking rs999737 for Stages 1 and 2 of the CGEMS study (5,692 cases, 5,576 controls) using the combined reference panels of the HapMap 3 and the 1000 Genomes Project. Single-marker association testing and variable-sized sliding-window haplotype analysis were performed, and for both analyses the initial tagging SNP rs999737 retained the strongest association with breast cancer risk. Investigation of contiguous regions did not reveal evidence for an additional independent signal. Therefore, we conclude that rs999737 is an optimal tag SNP for common variants in the 14q24.1 region and thus narrow the candidate variants that should be investigated in follow-up laboratory evaluation.

Keywords: RAD51L1, breast cancer, genome-wide association study, fine-mapping, imputation

Introduction

Breast cancer is the most commonly diagnosed cancer in women after non-melanoma skin cancer, and is the second leading cause of cancer deaths among women in the United States (Jemal et al. 2010). The contribution of genetic susceptibility to breast cancer has been the subject of intense study for decades. Initially, highly penetrant mutations in BRCA1 and BRCA2 were identified in families with multiple affected individuals, but together these account for less than 5% of breast cancer overall in the United States (Hall et al. 1990; Pruthi et al. 2010). The risk of a first-degree relative with breast cancer approximately doubles compared to women with no known family history of breast cancer, suggesting a more complex genetic architecture could contribute to breast cancer risk in the general population (Lichtenstein et al. 2000; Pruthi et al. 2010).

In the last four years, genome-wide association studies (GWAS) have successfully discovered at least twenty distinct loci associated with risk for breast cancer (Ghoussaini and Pharoah 2009; Varghese and Easton 2010) and additional regions are expected to be discovered as meta-analyses proceed (Park et al. 2010). Since each region identified in breast cancer GWAS appears to contribute a small fraction to the overall risk, it is not surprising that the current set explains only a small fraction of the overall familial risk for breast cancer (Ghoussaini and Pharoah 2009). Nonetheless, the discoveries in GWAS point towards new biological insights into the underlying factors contributing to a complex disease, such as breast cnacer. The follow-up for each region conclusively identified by GWAS requires fine mapping of common variants not directly tested in the GWAS to determine if one or more additional variants could be directly responsible for the signal. Since linkage disequilibrium (LD) patterns for common variants can often reveal multiple surrogates for any given marker, fine-mapping is necessary to nominate the subset of variants of high interest for functional investigation of the underlying biological mechanisms responsible for the direct association with breast cancer.

The Cancer Genetic Markers of Susceptibility (CGEMS) GWAS reported a single nucleotide polymorphism (SNP), rs999737, within the 14q24.1 interval that maps to an intron of the RAD51-like 1 (RAD51L1) gene. Since the most significant tagging signal rs999737 localized to an intron in RAD51L1, we conducted a fine-mapping analysis of the 14q21region in order to better define the chromosomal boundaries of interest for subsequent biological experiments designed to explain the association signal. RAD51L1 is one of five paralogs to the gene RAD51, which encodes the eukaryotic DNA recombination repair protein (Thacker and Zdzienicka 2004; Thomas et al. 2009). An independent pooled analysis in the Breast Cancer Association Consortium, using data from over 40,000 cases and 40,000 controls, confirmed the association between the SNP rs999737 and breast cancer risk; the data also supported this SNP to be a risk factor for the major breast cancer subtypes, as defined by expression of estrogen- receptor, progesterone receptor, and Her2/neu (Figuerora et al. 2011 submitted). In contrast, this SNP did not modify risk for BRCA1 or BRCA2 mutation carriers (Antoniou et al. 2011). Molecular studies of the RAD51L1 protein suggest its main function is to promote homologous recombination repair; specifically, the protein contributes to genome integrity maintenance in response to UV-induced damages (Havre et al. 1998; Takata et al. 2000). In breast cell lines, it has been shown that both transcription and protein levels of RAD51L1 are down-regulated by over-expression of EZH2, a transcriptional repressor previously linked to aggressive and metastatic breast cancers; RAD51L1 may also partner with the protein EVL (Ena/Vasp-like) to stimulate inappropriate RAD51-mediated recombination in breast cancer cell lines (Takaku et al. 2009; Zeidler et al. 2005).

To fine map the region surrounding rs999737 on 14q24.1, we used Stage 1 and 2 datasets from the CGEMS breast cancer GWAS to conduct an imputation analysis of SNP markers that were not directly genotyped in CGEMS. The imputation used 1000 Genomes Project (Durbin et al. 2010) and HapMap3 CEU (The International HapMap3 Consortium 2010) data. Here, we report the results of our work: the SNP marker rs999737 retains the strongest association of all imputed and genotyped SNPs across the 3.93 MB region and the data point toward a single signal on 14q24.1.

Materials and Methods

Study population

As described previously (Thomas et al. 2009), all study subjects analyzed in the CGEMS Breast Cancer Genome-Wide Association Scan were of European ancestry. Genotype data from Stage 1 and 2 of the CGEMS scan were used in this study. In Stage 1, 1,145 postmenopausal breast cancer cases and 1,142 healthy controls were genotyped with Human Hap500 Infinium Assay (Illumina) from the Nurses’ Health Study (NHS) (Hunter et al. 2007). For Stage 2, a custom-designed iSelect assay from Illumina was used to genotype 8,981 subjects from four additional study groups: the Women’s Health Initiative (WHI) (2,395 cases/2,410 controls); the Prostate, Lung, Colon and Ovarian Screening Trial (PLCO) Breast Cancer Study (948 cases/975 controls); the American Cancer Society Cancer Prevention Study II Nutrition Cohorts (CPSII) (535 cases/543 controls); and the Polish Breast Cancer Study (PBCS1) (669 cases/506 controls). For the combination of Stage 1 and Stage 2, a total of 5,692 breast cancer cases and 5,576 controls were available for further imputation and the subsequent association testing.

Imputation analysis

To infer potential genomic variants tagged by the SNP rs999737, we used the hidden Markov model program, IMPUTE2 (Howie et al. 2009), to impute genotypes based on HapMap Phase 3 CEU data (The International HapMap3 Consortium 2010) and 1000 Genomes Project (Durbin et al. 2010). Genotyped markers within a region of approximately 3.93MB flanking rs999737 and RAD51L1 (chr14q21: 66,173,322 – 70,099,944, UCSC genome build hg18) were available for analysis, including 770 SNPs from CGEMS Stage 1 data and 41 SNPs from Stage 2 in 11,268 individuals. Based on the combined reference panels of the 1000 Genomes Projects (2010 June release) and HapMap3 CEU data (2009 February release 2), an additional 11,047 putative SNPs were imputed for Stage 1 and 2 subjects across this region of 14q24.1. The average allele dosage was estimated with 100 iterations of the imputation algorithm conditional on a set of known haplotypes while simultaneously estimating the recombination map. The best-guessed genotypes (those with highest posterior probabilities) at each SNP for each subject were also generated using GTOOL (http://www.well.ox.ac.uk/~cfreeman/software/gwas/gtool.html). In addition to the average posterior probability for the best-guessed genotypes, the IMPUTE2-info score, which is associated with the imputed allele frequency estimate ranging from 1 (high confidence) to 0 (decreasing confidence), was also estimated for each SNP to evaluate imputation performance.

Statistical Analysis

Fisher’s exact tests for fitness for Hardy-Weinberg proportion (also known as Hardy-Weinberg Equilibrium, HWE) were evaluated across controls for all markers. Those that significantly deviated from HWE (p< 10−6) were discarded from further analysis. LD measures (D′ and r2) and LD plots were estimated using Haploview (Barrett et al. 2005) for the control samples. Based on the per marker quality measures generated from IMPUTE2, markers with maximum posterior genotype probability <0.9 were also filtered out to obtain reliable association results. The best-guessed SNP genotypes (those with highest posterior probabilities ≥0.9, imputation certainty ≥0.9, and did not significantly deviate from HWE) for all imputed markers were tested for association with breast cancer risk under an additive model using PLINK v1.07 (Purcell et al. 2007), adjusting for 10-year age groups, study sites, and 4 significant principal components to account for population heterogeneity as determined in the initial scan (Hunter et al. 2007; Thomas et al. 2009). A two-SNP model was used to determine whether an additional independent signal could be found by conditioning on the original CGEMS hit, rs999737. To explore the possible difference of the effect between different subtypes of breast cancer, we conducted a stratified analysis according to estrogen receptor (ER) and progesterone receptor (PR) status, under an additive genetic model adjusted for the already defined covariates. An analysis to account for imputation uncertainty using estimated allelic dosage was also performed with SNPTEST v2 (http://www.stats.ox.ac.uk/~marchini/software/gwas/snptest.html) and showed similar results (data not shown). A 2- to 8-SNP sliding-window haplotype analysis was performed across the region using PLINK (Purcell et al. 2007). Odds ratios and p-values were estimated for each haplotype versus all others, adjusted for the effects of same covariates. SequenceLDhot (Fearnhead 2006) was used to compute likelihood ratios (LR) for a set of putative recombination hotspots across the RAD51L1 region, repeated with five non-overlapping sets of 100 controls and using background recombination rates estimated by PHASE v2.1 (Stephens et al. 2001). Snp.plotter (Luna and Nicodemus 2007), an R-based package, was used to simultaneously display p-values from single-marker and haplotype analyses.

Results

Imputation and quality control

A total of 10,273 SNP markers within a 3.93MB flanking region (chr14:66,173,332–70,099,944) of rs999737 were imputed for 11,268 samples from the CGEMS Stages 1 and 2 of the breast cancer GWAS. Table 1 summarizes the genotyped and imputed SNPs across the 3.93Mb region of RAD51L1. More than 60% of the imputed SNPs had a minor allele frequency greater than 1%. The overall IMPUTE2-info score was calculated to be 0.4347, the median value of all imputed markers using a measure of the performance of imputation with values ranging from 0–1 in ascending confidence. We did not remove SNPs with minor allele frequencies less than 5% in order to retain the less common variants that could provide evidence for a synthetic association in the region (Dickson et al. 2010). The association analysis was restricted to 464 SNPs, the subset of SNPs with an IMPUTE2-info score of ≥0.9 (a rigorous metric for high quality imputed genotype calls), and no evidence for deviation from HWE (p> 10−6). Among these SNPs, 36 were directly genotyped in both Stage 1 and 2 samples, 31 were genotyped in Stage 1 and imputed in Stage 2 and 397 were imputed based on the 1000 Genome Project and HapMap 3 data. It is notable that a subset of the SNPs genotyped in Stage 1 could not be accurately imputed in Stage 2 using the public databases.

Table 1.

Summary of genotyped and imputed SNPs within 3.93Mb flanking region of 14q24.1.

N (%)
SNP source
 Genotyped CGEMS phase1 only 733 (6.64)
CGEMS phase2 only 4 (0.04)
CGEMS phase 1 & 2 37 (0.33)
 Imputed HapMap 3 only 282 (2.55)
1000 Genomes only 8874 (80.33)
HapMap 3 & 1000 Genomes
1117 (10.11)
Total 11,047
IMPUTE2-info score <0.5 5816 (56.61)
0.5–0.75 2635 (25.65)
0.76–0.9 1256 (12.23)
≥0.9
566 (5.51)
Total 10,273
p-value of HWE in controls >10−3 7726 (75.21)
10−3 – 10−4 494 (4.81)
10−4 – 10−5 334 (3.25)
10−5 – 10−6 270 (2.63)
<10−6 1449 (14.10)
Total 10,273
Overall MAF 0–0.01 3917 (38.13)
0.01–0.05 2407 (23.43)
0.05–0.10 1448 (14.10)
0.10–0.20 1063 (10.35)
0.20–0.30 646 (6.29)
0.30–0.40 442 (4.30)
0.40–0.5
350 (3.41)
Total 10,273
Open in a new tab

Association Analysis of Surrogates for rs999737

Figure 1a shows results from single-marker analysis of all imputed and genotyped markers within chromosomal region chr14: 67795579–68450029. Among them, 46 SNPs (including, the sentinel tagging SNP, rs999737) had IMPUTE2-info scores ≥0.9 and did not deviate from Hardy-Weinberg equilibrium; these were also nominally associated with breast cancer risks at padj <0.05, yet none of the values approached genome-wide significance (Table 2). Eleven of the significantly associated SNPs were genotyped, while the remaining 35 were inferred from IMPUTE2 using a combined HapMap3 and 1000 Genome reference panel. Within this group, the SNP with the strongest association with breast cancer risk was the initial GWAS SNP rs999737 (padj = 8.23×10−6; ORadj = 0.87, 95% Cl = 0.81–0.92), as reported for Stages 1 and 2, but not Stage 3, which conducted single SNP analysis (Thomas et al. 2009). We performed a sequential analysis for the other 45 SNPs conditioned on rs999737 (padjRS999737); 32 SNPs remained nominally associated with breast cancer risk (padjRS999737 <0.05; Table 3). Five of the SNPs were genotyped in the original GWAS report, while the remaining 28 SNPs were inferred from IMPUTE2. To determine whether the list of 45 SNPs showing moderate association with breast cancer risk were independent signals from rs999737, we examined whether any SNPs showing significant association with breast cancer risks fell across the gene RAD51L1, or shared significant pair-wise LD with the initial GWAS SNP marker, rs999737. We discovered that although 13 of the 45 SNPs generated from imputation that were statistically significant in their association with breast cancer risk were located within the coding region of RAD51L1, only 1 SNP showed nominal LD correlation with rs999737 (rs10130694; r2= 0.107), while others shared minimal pair-wise LD with rs999737 (Table 2). To evaluate whether SNPs across the RAD51L1 coding region (e.g., in introns and exons) were strongly associated with breast cancer risk, we examined the group of 464 genotyped and imputed SNPs (including rs999737). We discovered that 92 SNPs were located within the RAD51L1 gene and that 12 showed minimally significant association with breast cancer risk (padj <0.05). However, upon conditioning on rs999737, none remained significantly associated with breast cancer risk, thus suggesting that this set of markers tag a single locus.

Fig. 1. Association results, recombination and linkage disequilibrium plots for chromosome 14q21 region.

Fig. 1

Open in a new tab

a. Single-marker analysis of CGEMS GWAS SNPs (red triangle) and imputed SNPs (blue circles) are shown in the top panel with −log10 p-values (left y axis). Overlaid on the panel are the likelihood ratio statistics (right y axis) to estimate putative recombination hotspot across the region, based on the 5 sets of 100 randomly selected control samples (connected lines in red, green, blue, purple, and cyan). Pairwise r2 values based on control samples are displayed at the bottom panel for all SNPs in this analysis. Genomic coordinates are based on NCBI Human Genome Build 36.3.

b. Variable-sized sliding-window haplotype analysis (blue dots and connected lines) and single-marker analysis results of the SNPs (red triangles) are shown with −log10 p-values (left y axis) in a 9.6kb region (yellow highlight in Fig. 1a) that includes SNP rs999737. Genomic coordinates are based on NCBI Human Genome Build 36.3.

Table 2. Forty-six genotyped and imputed SNPs demonstrating nominal significant association with breast cancer risk (padj <0.05).

Allele 1 marks the minor allele while allele 2 marks the major allele at each locus. Pair-wise linkage disequilibrium (r2 with rs999737) with rs999737 was also calculated and no significant LD was shared between rs999737 and any of these SNPs. OR (95%CI) and p-values were estimated from logistic regression with an additive genetic effect, adjusted for age (in10-year categories), study sites, and 4 significant principle components. Three clusters of SNPs are found within in silico elements possibly related to breast cancer, and are marked with additional notation.

SNP Positiona Sourceb A1/A2c MAF CS:CNd ORadj (95%CI)e padj r2 with rs999737
rs17105269 67795579 GWAS-P2 A/T 0.1711:0.1581 1.10 (1.02 – 1.19) 1.74E-02 0.0069
rs11626138 67856746 1kG A/G 0.2138:0.2025 1.07 (1.01 – 1.15) 3.62E-02 0.0208
rs4902567 67870353 1kG T/C 0.2028:0.1922 1.07 (1.00 – 1.15) 4.07E-02 0.0186
rs4902587 67975441 1kG A/G 0.1884:0.1749 1.10 (1.03 – 1.17) 7.75E-03 0.045
rs2331703 67976519 HM3,1kG C/T 0.1966:0.1861 1.08 (1.01 – 1.15) 3.16E-02 0.0312
rs2331704 67976586 GWAS-P1P2 G/A 0.1991:0.1887 1.08 (1.01 – 1.15) 3.26E-02 0.0295
rs4902588 67978211 1kG C/T 0.1976:0.1864 1.08 (1.01 – 1.06) 2.21E-02 0.0302
rs2753403 67982938 HM3,1kG A/G 0.1955:0.1841 1.08 (1.01 – 1.06) 2.00E-02 0.0331
rs2525507 68008117 HM3,1kG G/A 0.2076:0.1959 1.08 (1.01 – 1.15) 3.38E-02 0.0395
rs2842320 68013826 GWAS-P1 G/A 0.1863:0.1739 1.10 (1.0 2 – 1.17) 1.10E-02 0.0345
rs999737 68104435 GWAS-P1P2 T/C 0.2127:0.2380 0.87 (0.81–0.92) 8.23E-06 --
rs8009944 68109341 GWAS-P1P2 C/A 0.2834:0.2676 1.08 (1.02– 1.15) 1.14E-02 0.0947
rs10130694 68109831 1kG T/C 0.2787:0.2634 1.08 (1.01 – 1.15) 1.73E-02 0.1068
rs1290999 68115629 GWAS-P1P2 A/G 0.2133:0.1987 1.09 (1.02 – 1.17) 8.03E-03 0.063
rs10483818 68189896 GWAS-P1P2 A/G 0.0495:0.0414 1.21 (1.07 – 1.37) 2.91E-03 0.004
rs2043672 68258829 1kG G/C 0.0469:0.0559 0.84 (0.74 – 0.94) 2.88E-03 0
rs2025053 68258941 GWAS-P1P2 A/G 0.0569:0.0665 0.85 (0.74 – 0.95) 2.91E-03 0.0002
14-68260645 68260645 1kG T/C 0.0475:0.0563 0.84 (0.75 – 0.95) 3.89E-03 0.0001
rs8013316 68261417 1kG C/A 0.0478:0.0564 0.84 (0.75 – 0.95) 4.52E-03 0.0001
rs55761709 68263324 1kG T/C 0.0469:0.0560 0.84 (0.74 – 0.94) 2.79E-03 0.0001
14-68264094 68264094 1kG G/A 0.047:0.0559 0.84 (0.74 – 0.94) 2.86E-03 0.0001
rs10483821 68264360 HM3,1kG G/C 0.0541:0.0625 0.86 (0.77 – 0.96) 8.89E-03 0
14-68264618 68264618 1kG G/T 0.0466:0.0558 0.83 (0.74 – 0.94) 2.59E-03 0.0001
rs8015998 68264784 1kG C/T 0.0534:0.0621 0.86 (0.77 – 0.96) 6.22E-03 0
rs1274944 68395187 GWAS-P1P2 G/A 0.1758:0.1655 1.07 (1.00 – 1.15) 4.90E-02 0.0016
rs440317 68399592 HM3,1kG T/C 0.3122:0.2943 1.09 (1.02 – 1.15) 6.15E-03 0.0012
rs1742500 68399654 HM3,1kG A/G 0.3297:0.3120 1.08 (1.02 – 1.15) 7.16E-03 0.001
rs453964 68399988 HM3,1kG T/G 0.3364:0.3188 1.08 (1.02 – 1.14) 7.01E-03 0.0012
rs445863 68400930 HM3,1kG T/C 0.3324:0.3122 1.10 (1.04 – 1.16) 1.57E-03 0.001
rs61987505 68402055 1kG A/G 0.3159:0.2966 1.09 (1.03 – 1.16) 2.03E-03 0.0012
rs1742890 68402891 1kG C/G 0.3314:0.3144 1.08 (1.02 – 1.15) 1.03E-02 0
rs1742886 68404595 HM3,1kG G/C 0.3142:0.2941 1.10 (1.04 – 1.16) 1.20E-03 0.0013
rs1742498 68404615 1kG G/A 0.3001:0.2779 1.12 (1.05 – 1.19) 6.95E-04 0.0006
rs10139065 68405176 HM3,1kG G/A 0.3147:0.2950 1.10 (1.04 – 1.16) 1.45E-03 0.0012
rs10139069 68405191 1kG G/A 0.3148:0.2950 1.10 (1.04 – 1.16) 1.41E-03 0.0012
rs10151103 68405195 1kG T/C 0.3148:0.2950 1.10 (1.04 – 1.16) 1.41E-03 0.0012
rs7401911 68405317 HM3,1kG T/C 0.3147:0.2950 1.10 (1.04 – 1.16) 1.45E-03 0.0012
rs4902660 68405884 1kG T/C 0.3333:0.3119 1.10 (1.04 – 1.17) 1.29E-03 0.0008
rs6573861 68406898 1kG G/A 0.3148:0.2950 1.10 (1.04 – 1.16) 1.41E-03 0.0012
rs4902661 68406934 1kG C/T 0.3271:0.3090 1.09 (1.03 – 1.15) 5.35E-03 0.0014
rs4902662 68406986 1kG G/C 0.3182:0.2993 1.09 (1.03 – 1.16) 2.35E-03 0.0005
rs7149938 68407529 1kG G/A 0.3303:0.3110 1.09 (1.03 – 1.16) 2.93E-03 0.0019
rs1475035 68407977 GWAS-P1P2 T/C 0.3145:0.2952 1.10 (1.04 – 1.16) 1.67E-03 0.0012
rs4312253 68408187 1kG C/T 0.3149:0.2951 1.10 (1.04 – 1.16) 1.42E-03 0.0012
68410352 1kG T/C 0.3951:0.4122 0.93 (0.88 – 0.99) 1.52E-02 0.0033
rs1475034 68410513 GWAS-P1 T/G 0.3951:0.4122 1.09 (1.02 – 1.16) 8.57E-03 0.0005
Open in a new tab
a

Genomic coordinates are based on NCBI Human Genome Build 36.3.

b

SNP marker sources. GWAS-P1: actual genotyped SNPs in CGEMS GWAS Phase 1; GWAS-P2: actual genotyped SNPs in CGEMS GWAS Phase 2; 1kG: imputed SNPs based on 1000 Genomes Project data; HM3: imputed SNPs based on HapMap Phase 3 data.

c

A1 marks the minor allele; A2 marks the major allele.

d

Minor allele frequency (MAF) for cases (CS) and controls (CN).

e

Estimates from logistic regression with an additive genetic effect, adjusted for age (in 10-year categories), study sites, and 4 significant principle components.

Table 3. Thirty-two imputed and genotyped SNPs with nominally significant association after sequential conditioning on rs999737 (padjrs99 <0.05).

OR (95%CI) and p-values were estimated from logistic regression with an additive genetic effect, adjusted for age (in10-year categories), study sites, 4 significant principle components and rs999737, when applicable.

SNP Positiona Sourceb A1/A2c ORadj (95%CI)d padjd ORadjRS99(95%CI)e padjRS99e r2 with rs999737 Notef
rs1742498g 68404615 1kG G/A 1.12 (1.05 – 1.19) 6.95E-04 1.11 (1.04 – 1.18) 1.01E-03 0.0006 rs71824488 (indel); heterozygous deletion for K562 cells
rs4902660 68405884 1kG T/C 1.10 (1.04 – 1.17) 1.29E-03 1.10 (1.04 – 1.17) 1.96E-03 0.0008 heterozygous deletion for K562 cells
rs1742886g 68404595 HM3,1kG G/C 1.10 (1.04 – 1.16) 1.20E-03 1.09 (1.03 – 1.16) 2.04E-03 0.0013 rs71824488 (indel)
14-68264618 68264618 1kG G/T 0.83 (0.74 – 0.94) 2.59E-03 0.83 (0.74 – 0.93) 2.05E-03 0.0001 heterozygous deletion for K562 cells
rs2025053 68258941 GWAS-P1P2 G/A 0.85 (0.74 – 0.95) 2.91E-03 0.84 (0.76 – 0.94) 2.14E-03 0.0002 LINE element; heterozygous deletion for K562 cells
rs55761709 68263324 1kG C/T 0.84 (0.74 – 0.94) 2.79E-03 0.83 (0.74 – 0.94) 2.21E-03 0.0001 heterozygous deletion for K562 cells
14-68264094 68264094 1kG G/A 0.84 (0.74 – 0.94) 2.86E-03 0.83 (0.74 – 0.94) 2.25E-03 0.0001 heterozygous deletion for K562 cells
rs10139069g 68405191 1kG G/A 1.10 (1.04 – 1.16) 1.41E-03 1.09 (1.03 – 1.15) 2.32E-03 0.0012 rs71824488 (indel); heterozygous deletion for K562 cells
rs10151103g 68405195 1kG T/C 1.10 (1.04 – 1.16) 1.41E-03 1.09 (1.03 – 1.15) 2.32E-03 0.0012 rs71824488 (indel); heterozygous deletion for K562 cells
rs6573861 68406898 1kG G/A 1.10 (1.04 – 1.16) 1.41E-03 1.09 (1.03 – 1.15) 2.32E-03 0.0012 DNA element; heterozygous deletion for K562 cells
rs4312253h 68408187 1kG C/T 1.10 (1.04 – 1.16) 1.42E-03 1.09 (1.03 – 1.15) 2.34E-03 0.0012 mammary gland EST AK022224
rs2043672 68258829 1kG C/G 0.84 (0.74 – 0.94) 2.88E-03 0.83 (0.74 – 0.94) 2.38E-03 0.0000 LINE element; heterozygous deletion for K562 cells
rs445863 68400930 HM3,1kG T/C 1.10 (1.04 – 1.16) 1.57E-03 1.09 (1.03 – 1.16) 2.38E-03 0.0010 LTR element; Dnase sensitive; heterozygous deletion for K562 cells
rs10139065g 68405176 HM3,1kG G/A 1.10 (1.04 – 1.16) 1.45E-03 1.09 (1.03 – 1.16) 2.39E-03 0.0012 rs71824488 (indel); heterozygous deletion for K562 cells
rs7401911g 68405317 HM3,1kG T/C 1.10 (1.04 – 1.16) 1.45E-03 1.09 (1.03 – 1.16) 2.39E-03 0.0012 rs71824488 (indel); heterozygous deletion for K562 cells
rs1475035h 68407977 GWAS-P1P2 T/C 1.10 (1.04 – 1.16) 1.67E-03 1.09 (1.03 – 1.16) 2.79E-03 0.0012 mammary gland EST AK022224, transcription factor binding sites
14-68260645 68260645 1kG C/T 0.84 (0.75 – 0.95) 3.89E-03 0.84 (0.75 – 0.94) 3.16E-03 0.0001 transcription binding site; DNAse sensitivity; heterozygous deletion for K562 cells
rs4902662h 68406986 1kG G/C 1.09 (1.03 – 1.16) 2.35E-03 1.09 (1.03 – 1.15) 3.17E-03 0.0005 mammary gland EST AK022224
rs61987505 68402055 1kG G/A 1.09 (1.03 – 1.16) 2.03E-03 1.09 (1.03 – 1.15) 3.30E-03 0.0012 within LTR element; heterozygous deletion for K562 cells
rs8013316 68261417 1kG A/C 0.84 (0.75 – 0.95) 4.52E-03 0.84 (0.75 – 0.94) 3.63E-03 0.0001 SINE element; heterozygous deletion for K562 cells
rs7149938h 68407529 1kG G/A 1.09 (1.03 – 1.16) 2.93E-03 1.09 (1.03 – 1.15) 5.25E-03 0.0019 SINE element; mammary gland EST AK02224
rs8015998 68264784 1kG C/T 0.86 (0.77 – 0.96) 6.22E-03 0.85 (0.76 – 0.95) 5.36E-03 0.0000 transcription factor binding site; heterozygous deletion for K562 cells
rs10483818 68189896 GWAS-P1P2 A/G 1.21 (1.07 – 1.37) 2.91E-03 1.19 (1.05 – 1.35) 6.61E-03 0.0040 heterozygous deletion for K562 cells
rs10483821 68264360 HM3,1kG G/C 0.86 (0.77 – 0.96) 8.89E-03 0.86 (0.77 – 0.96) 7.77E-03 0.0000 transcription binding site; DNAse sensitivity; heterozygous deletion for K562 cells
rs440317i 68399592 HM3,1kG T/C 1.09 (1.02 – 1.15) 6.15E-03 1.08 (1.02 – 1.15) 9.07E-03 0.0012 mammary gland EST DA647031
rs4902661 68406934 1kG C/T 1.09 (1.03 – 1.15) 5.35E-03 1.08 (1.02 – 1.15) 9.14E-03 0.0014 DNA element; heterozygous deletion for K562 cells
rs1742890 68402891 1kG C/G 1.08 (1.02 – 1.15) 1.03E-02 1.08 (1.02 – 1.15) 9.35E-03 0.0000 SINE element; rs71824488 (indel); heterozygous deletion for K562 cells
rs1742500i 68399654 HM3,1kG A/G 1.08 (1.02 – 1.15) 7.16E-03 1.08 (1.02 – 1.14) 9.92E-03 0.0010 SINE element; mammary gland EST DA647031
rs453964i 68399988 HM3,1kG T/G 1.08 (1.02 – 1.14) 7.01E-03 1.08 (1.02 – 1.14) 1.01E-02 0.0012 EST from mammary gland DA647031; transcription factor binding sites; heterozygous deletion for K562 cells
rs1475034h 68410513 GWAS-P1 T/G 1.09 (1.02 – 1.16) 8.57E-03 1.08 (1.02 – 1.15) 1.15E-02 0.0005 mammary gland EST AU121625 and AK022224; heterozygous deletion for K562 cells
rs181456h 68410352 1kG T/C 0.93 (0.88 – 0.99) 1.52E-02 0.94 (0.89 – 0.99) 2.89E-02 0.0033 mammary gland EST AU121625 and AK022224; heterozygous deletion for K562 cells; CpG islands
rs17105269 67795579 GWAS-P2 A/T 1.10 (1.02 – 1.19) 1.74E-02 1.09 (1.01 – 1.18) 3.50E-02 0.0069 heterozygous deletion for K562 cells
Open in a new tab
a

Genomic coordinates are based on NCBI Human Genome Build 36.3.

b

SNP marker sources. GWAS-P1: actual genotyped SNPs in CGEMS GWAS Phase 1; GWAS-P2: actual genotyped SNPs in CGEMS GWAS Phase 2; 1kG: imputed SNPs based on 1000 Genome Project data; HM3: imputed SNPs based on HapMap Phase 3 data.

c

A1 marks the minor allele; A2 marks the major allele.

d

Estimates from logistic regression with an additive genetic effect, adjusted for age (in 10-year categories), study sites, and 4 significant principle components.

e

Estimates from logistic regression with an additive genetic effect, adjusted for age (in 10-year categories), study sites, 4 significant principle components, and rs999737.

f

SNP notation and associated genomic elements.

g

Genomic interval containing 6 SNPs located in an polymorphic insertion/deletion element.

h

Genomic interval containing 6 SNPs located in an mammary gland EST AU121625.

i

Genomic interval containing 3 SNPs located in an mammary gland EST DA647031.

In addition to the region proximal to rs999737 within RAD51L1, we observed several short regions with clusters of SNPs showing nominal association with breast cancer risk; most importantly, each interval contained in silico genomic elements that could be related to breast cancer (Table 2; see footnotes). The first cluster was 3,527 bp in length (chr14:68,406,986 – 68,410,513), and contained 6 SNPs that localize to a mammary gland-specific EST, AK02224. The second cluster of 722 bp (chr14:68,404,595 – 68,405,317), contained 6 SNPs within a polymorphic insertion/deletion, estimated to be 2,841bp. The third cluster was the smallest at 396bp (chr14:68,399,592- 68,399,988), and contained only 3 SNPs that mapped to a mammary gland-specific EST DA647031. The clusters of SNPs were separated from rs999737 by recombination hotspots, which we estimated based on available controls (Figure 1a), and the two regions shared minimal LD correlations (Table 2).

Variable-sized sliding window haplotype analysis

Using 5 non-overlapping sets of 100 controls, randomly selected from the entire control set, we calculated and plotted the recombination pattern and LD structure for CGEMS data (Figure 1a). To further explore whether one or more haplotype can explain the signal, we applied a variable-sized sliding-window haplotype analysis across the imputed region (Table 4; Figure 1b). Haplotypes inclusive of rs999737 were significantly associated with breast cancer; the strongest association was a 2-SNP haplotype between rs999737 and 14-68103993 (p-value = 8.39×10−6). For rs12588940, a SNP that is 3.6kb away from rs999737, a 7-SNP haplotype inclusive of these two SNPs remained significantly associated with breast cancer (p-value = 2.13×10−5). We examined haplotypes within the short clusters of interest, and found that they were moderately associated with breast cancer risk (chr14:68,406,986 – 68,410,514 interval, smallest p-value = 1.36×10−3; chr14:68,404,595 – 68,405,317 interval, smallest p-value = 5.17×10−4; chr14:68,399,592– 68,399,988 interval, smallest p-value = 3.72×10−3). Substantially larger independent follow-up studies will be required to determine if these signals indicate a distinct locus.

Table 4.

Top signals from haplotype analysis associated with breast cancer risk.

Starting SNP Starting SNP position No. of SNPs in sliding window Haplotype spanned length (bp) p-valuea
14-68101612 68101612 8 2823 6.09E-05
rs1468280 68101613 7 2822 1.22E-04
rs12588940 68101621 7 3557 2.13E-05
rs2842324 68102125 6 3053 8.96E-06
rs8016149 68102742 5 2436 8.83E-06
rs7153476 68102983 7 2897 1.63E-05
14-68103993 68103993 2 442 8.39E-06
rs999737 68104435 3 817 2.51E-05
rs8007194 68105178 6 2016 1.86E-03
14-68105252 68105252 4 1821 1.45E-04
14-68105840 68105840 3 1233 1.86E-04
rs17756147 68105880 2 1193 7.38E-05
rs11628293 68107073 2 121 2.52E-05
rs12100794 68107194 7 1605 2.35E-05
rs6573841 68107274 7 2067 3.40E-05
Open in a new tab
a

Estimates from haplotype-specific tests of each haplotype versus all others, adjusted for age (in 10-year categories), study sites, and 4 significant principle components.

Discussion

To follow-up on the original rs999737 signal detected in the CGEMS breast cancer study (Thomas et al. 2009), we have conducted a fine-mapping analysis of the genomic region surrounding rs999737, which localizes to the RAD51L1 gene in the 14q21 interval (Thomas et al. 2009). To explore the set of common and uncommon untested variants in this interval, we imputed SNPs based on CGEMS Stages 1 and 2 with data from the HapMap3 and 1000 Genome Project, and applied single-marker and haplotype analyses. Our study has shown that the original SNP, rs999737, retained the strongest association signal for breast cancer susceptibility in this region of 14q21, even after correcting for analysis on the discovery set.

In an analysis of SNPs from across a nearly 4 MB region on 14q24.1, we observed 46 SNPs (inclusive of rs999737) demonstrating a nominal association with breast cancer risk (padj < 0.05); however, the initial discovery SNP, rs999737 remained most significant in Stage 1 and 2 (padj = 8.23×10−6). To evaluate whether any of the signals contributed to breast cancer susceptibility independently from rs999737, we conducted a conditional analysis using the rs999737 discovery marker for the remaining 45 SNPs. While 32 SNPs retained nominal significance for breast cancer risk, in each case, the signal was attenuated. In an exploration of the correlation between each of the 45 SNPs and rs999737 in the control set, we estimated the pair-wise LD (r2 values). None of the pairs demonstrated a high correlation and we observed residual correlation with one signal marked by rs999737 (Table 2; Table 3). In a variable-sized sliding-window haplotype analysis (2–8 SNPs) for all genotyped and imputed SNPs, the results revealed that haplotypes containing rs999737 showed the highest association with breast cancer risk, with p-value = 8.39×10−6. Thus, both single-marker and haplotype analysis suggested that the genomic element tagged by rs999737 had the strongest association with breast cancer risk (Thomas et al. 2009).

To determine whether the imputed SNPs exhibited different effects for distinct breast cancer subtypes, specifically, ER and PR subtypes, we conducted a stratified analysis in those cases with reliable ER/PR status (approximately 80%). Supplementary Table 1 shows results for the top 46 SNPs (inclusive of rs999737) with nominal association with breast cancer (Table 2) in this study, stratified by each of ER and PR status. Our data suggest that compared with ER-negative tumors, a subset of SNPs had evidence for a stronger association with ER-positive breast cancer than in ER-negative tumors. This might be due to the predominance of ER-positive subjects among breast cancer cases. The analysis stratified by PR status was unremarkable.

As genome-wide association studies can detect disease susceptibility loci throughout multi-stage study design, there might be potential bias introduced by “winner’s curse” effect (Capen et al. 1971) in the SNP selection process of replication studies (Zhong and Prentice 2008). In our analysis, we conducted the fine-mapping in the same set used to discover the tag SNP, rs999737. We recognized that there could be a bias away from the null in favor of the discovered tag SNP (Zhong and Prentice 2008). Here we calculated the bias-reduced estimators and confidence intervals using a recently proposed method (Zhong and Prentice 2008, 2009) for top SNPs reported in the present study. The bias-adjusted OR and 95%CI were slightly weaker than the unadjusted estimates but maintained good concordance with unadjusted estimates (Supplementary Table 2). In our study of approximately 5600 cases and 5600 controls, we had sufficient power to detect variants with minor allele frequency >0.15 and odds ratio greater than 1.20 assuming a disease prevalence of 12% for the at risk alleles at the level of p < 5 × 10−6 (Supplementary Table 3).

Our results did not identify new strong signals across the large region extending well beyond the recombination hot spots confirmed in our analysis. In this regard, we did not identify a second locus or a substantially better tag SNP than the originally reported SNP, rs999737, suggesting that the common variant signal associated with breast cancer is most likely due to a single locus on chromosome 14q24.1. Functional analysis of the tag SNP and its highly correlated variants will be needed to define the underlying mechanism and to establish the variant(s) that confer the direct association.

In an exploration of a nearly 4 MB interval, we examined the genomic landscape for each of the 32 SNPs noted above in search of possible biological mechanisms that might be worth pursuit, by either laboratory investigation or by extending the replication effort (Table 3). Several clusters of variants map to regions previously related to breast cancer (Figure 1b; mammary gland specific EST AK022224: chr14:68,406,968–68,410,613; rs71824488: chr14:68,402,332–68,405,672; mammary gland specific EST DA647031: chr14:68,399,325–68,400,050). A cluster of 6 SNPs is within the mammary gland-specific EST, AK02224; a second cluster of 6 SNPs localizes within a polymorphic indel of 2.841bp on the + strand, marked by the SNP rs71824488; the third cluster of 3 SNPs localizes to another mammary gland- specific EST, DA647031. We also observed possible signals separate from the signal marked by rs999737, but additional studies will be needed to determine if any of these represent independent loci. It is likely that the additional signals either reflect residual LD across this region, still pointing to the single locus marked by rs999737, or that these observations are due to chance.

RAD51L1 is a paralog of the RAD51 gene, and it forms a complex with RAD51L2 in the homologous recombination repair pathway through the complex’s ability to catalyze DNA strand- exchange (Hussain et al. 2004; Miller et al. 2002; Sigurdsson et al. 2001). Interestingly, akin to RAD51L1, RAD51L2 was also identified as a breast cancer susceptibility gene (Meindl et al. 2010). In addition to the RAD51 paralogs, other homologous recombination repair pathways genes have also been associated with breast cancer susceptibility, such as ATM, CHEK2, BRCA1 and BRCA2 (Hollestelle et al. 2010). It is also noteworthy that RAD51L1 had previously been identified as a fusion gene in myeloid malignancies, thymoma and autoimmune diseases (Nicodeme et al. 2005; Odero et al. 2005). Together, these observations suggest that genetic variation, both germ-line and somatic, within this region could contribute to one or more oncogenic processes.

Supplementary Material

Supplemental Material

Acknowledgments

The Nurses’ Health Studies are supported by NIH grants CA 65725, CA87969, CA49449, CA67262, CA50385 and 5UO1CA098233. The authors thank Barbara Egan, Lori Egan, Helena Judge Ellis, Hardeep Ranu, and Pati Soule for assistance, and the participants in the Nurses’ Health Studies.

The WHI program is supported by contracts from the National Heart, Lung and Blood Institute, NIH. The authors thank the WHI investigators and staff for their dedication, and the study participants for making the program possible. A full listing of WHI investigators can be found at http://www.whi.org.

The ACS study is supported by UO1 CA098710. We thank Cari Lichtman for data management and the participants on the CPS-II.

The PLCO study is supported by the Intramural Research Program of the Division of Cancer Epidemiology and Genetics and contracts from the Division of Cancer Prevention, National Cancer Institute, NIH, DHHS. The authors thank Dr. Philip Prorok, Division of Cancer Prevention, National Cancer Institute; the Screening Center investigators and staff of the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial (PLCO). Most importantly, we acknowledge the study participants for their contributions to making this study possible.

References

  1. Antoniou AC, Kartsonaki C, Sinilnikova OM, Soucy P, McGuffog L, Healey S, Lee A, Peterlongo P, Manoukian S, Peissel B, Zaffaroni D, Cattaneo E, Barile M, Pensotti V, Pasini B, Dolcetti R, Giannini G, Laura Putignano A, Varesco L, Radice P, Mai PL, Greene MH, Andrulis IL, Glendon G, Ozcelik H, Thomassen M, Gerdes AM, Kruse TA, Birk Jensen U, Cruger DG, Caligo MA, Laitman Y, Milgrom R, Kaufman B, Paluch-Shimon S, Friedman E, Loman N, Harbst K, Lindblom A, Arver B, Ehrencrona H, Melin B, Nathanson KL, Domchek SM, Rebbeck T, Jakubowska A, Lubinski J, Gronwald J, Huzarski T, Byrski T, Cybulski C, Gorski B, Osorio A, Ramon YCT, Fostira F, Andres R, Benitez J, Hamann U, Hogervorst FB, Rookus MA, Hooning MJ, Nelen MR, van der Luijt RB, van Os TA, van Asperen CJ, Devilee P, Meijers-Heijboer HE, Gomez Garcia EB, Peock S, Cook M, Frost D, Platte R, Leyland J, Gareth Evans D, Lalloo F, Eeles R, Izatt L, Adlard J, Davidson R, Eccles D, Ong KR, Cook J, Douglas F, Paterson J, John Kennedy M, Miedzybrodzka Z, Godwin A, Stoppa-Lyonnet D, Buecher B, Belotti M, Tirapo C, Mazoyer S, Barjhoux L, Lasset C, Leroux D, Faivre L, Bronner M, Prieur F, Nogues C, Rouleau E, Pujol P, Coupier I, Frénay M, Hopper JL, Daly MB, Terry MB, John EM, Buys SS, Yassin Y, Miron A, Goldgar D, Singer CF, Tea MK, Pfeiler G, Catharina Dressler A, Hansen TV, Jønson L, Ejlertsen B, Bjork Barkardottir R, Kirchhoff T, Offit K, Piedmonte M, Rodriguez G, Small L, Boggess J, Blank S, Basil J, Azodi M, Ewart Toland A, Montagna M, Tognazzo S, Agata S, Imyanitov E, Janavicius R, Lazaro C, Blanco I, Pharoah PD, Sucheston L, Karlan BY, Walsh CS, Olah E, Bozsik A, Teo SH, Seldon JL, Beattie MS, van Rensburg EJ, Sluiter MD, Diez O, Schmutzler RK, Wappenschmidt B, Engel C, Meindl A, Ruehl I, Varon-Mateeva R, Kast K, Deissler H, Niederacher D, Arnold N, Gadzicki D, Schönbuchner I, Caldes T, de la Hoya M, Nevanlinna H, Aittomäki K, Dumont M, Chiquette J, Tischkowitz M, Chen X, Beesley J, Spurdle AB, Neuhausen SL, Chun Ding Y, Fredericksen Z, Wang X, Pankratz VS, Couch F, Simard J, Easton DF, Chenevix-Trench G CEMO Study Collaborators; Breast Cancer Family Registry; kConFab investigators; on behalf of CIMBA. Common alleles at 6q25.1 and 1p11.2 are associated with breast cancer risk for BRCA1 and BRCA2 mutation carriers. Hum Mol Genet. 2011;20:3304–3321. doi: 10.1093/hmg/ddr226. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics. 2005;21:263–5. doi: 10.1093/bioinformatics/bth457. [DOI] [PubMed] [Google Scholar]
  3. Capen EC, Clapp RV, Campbell WM. Competitive bidding in high-risk situations. J Petroleum Tech. 1971;23:641–653. [Google Scholar]
  4. Dickson SP, Wang K, Krantz I, Hakonarson H, Goldstein DB. Rare variants create synthetic genome-wide associations. PLoS Biol. 2010;8:e1000294. doi: 10.1371/journal.pbio.1000294. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Durbin RM, Abecasis GR, Altshuler DL, Auton A, Brooks LD, Gibbs RA, Hurles ME, McVean GA. A map of human genome variation from population-scale sequencing. Nature. 2010;467:1061–73. doi: 10.1038/nature09534. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Fearnhead P. SequenceLDhot: detecting recombination hotspots. Bioinformatics. 2006;22:3061–6. doi: 10.1093/bioinformatics/btl540. [DOI] [PubMed] [Google Scholar]
  7. Ghoussaini M, Pharoah PD. Polygenic susceptibility to breast cancer: current state-of-the-art. Future Oncol. 2009;5:689–701. doi: 10.2217/fon.09.29. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Hall JM, Lee MK, Newman B, Morrow JE, Anderson LA, Huey B, King MC. Linkage of early-onset familial breast cancer to chromosome 17q21. Science. 1990;250:1684–9. doi: 10.1126/science.2270482. [DOI] [PubMed] [Google Scholar]
  9. Havre PA, Rice MC, Noe M, Kmiec EB. The human REC2/RAD51B gene acts as a DNA damage sensor by inducing G1 delay and hypersensitivity to ultraviolet irradiation. Cancer Res. 1998;58:4733–9. [PubMed] [Google Scholar]
  10. Hollestelle A, Wasielewski M, Martens JW, Schutte M. Discovering moderate-risk breast cancer susceptibility genes. Curr Opin Genet Dev. 2010;20:268–76. doi: 10.1016/j.gde.2010.02.009. [DOI] [PubMed] [Google Scholar]
  11. Howie BN, Donnelly P, Marchini J. A flexible and accurate genotype imputation method for the next generation of genome-wide association studies. PLoS Genet. 2009;5:e1000529. doi: 10.1371/journal.pgen.1000529. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Hunter DJ, Kraft P, Jacobs KB, Cox DG, Yeager M, Hankinson SE, Wacholder S, Wang Z, Welch R, Hutchinson A, Wang J, Yu K, Chatterjee N, Orr N, Willett WC, Colditz GA, Ziegler RG, Berg CD, Buys SS, McCarty CA, Feigelson HS, Calle EE, Thun MJ, Hayes RB, Tucker M, Gerhard DS, Fraumeni JF, Jr, Hoover RN, Thomas G, Chanock SJ. A genome-wide association study identifies alleles in FGFR2 associated with risk of sporadic postmenopausal breast cancer. Nat Genet. 2007;39:870–4. doi: 10.1038/ng2075. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Hussain S, Wilson JB, Medhurst AL, Hejna J, Witt E, Ananth S, Davies A, Masson JY, Moses R, West SC, de Winter JP, Ashworth A, Jones NJ, Mathew CG. Direct interaction of FANCD2 with BRCA2 in DNA damage response pathways. Hum Mol Genet. 2004;13:1241–8. doi: 10.1093/hmg/ddh135. [DOI] [PubMed] [Google Scholar]
  14. Jemal A, Siegel R, Xu J, Ward E. Cancer statistics, 2010. CA Cancer J Clin. 2010;60:277–300. doi: 10.3322/caac.20073. [DOI] [PubMed] [Google Scholar]
  15. Lichtenstein P, Holm NV, Verkasalo PK, Iliadou A, Kaprio J, Koskenvuo M, Pukkala E, Skytthe A, Hemminki K. Environmental and heritable factors in the causation of cancer--analyses of cohorts of twins from Sweden, Denmark, and Finland. N Engl J Med. 2000;343:78–85. doi: 10.1056/NEJM200007133430201. [DOI] [PubMed] [Google Scholar]
  16. Luna A, Nicodemus KK. snp.plotter: an R-based SNP/haplotype association and linkage disequilibrium plotting package. Bioinformatics. 2007;23:774–6. doi: 10.1093/bioinformatics/btl657. [DOI] [PubMed] [Google Scholar]
  17. Meindl A, Hellebrand H, Wiek C, Erven V, Wappenschmidt B, Niederacher D, Freund M, Lichtner P, Hartmann L, Schaal H, Ramser J, Honisch E, Kubisch C, Wichmann HE, Kast K, Deissler H, Engel C, Muller-Myhsok B, Neveling K, Kiechle M, Mathew CG, Schindler D, Schmutzler RK, Hanenberg H. Germline mutations in breast and ovarian cancer pedigrees establish RAD51C as a human cancer susceptibility gene. Nat Genet. 2010;42:410–4. doi: 10.1038/ng.569. [DOI] [PubMed] [Google Scholar]
  18. Miller KA, Yoshikawa DM, McConnell IR, Clark R, Schild D, Albala JS. RAD51C interacts with RAD51B and is central to a larger protein complex in vivo exclusive of RAD51. J Biol Chem. 2002;277:8406–11. doi: 10.1074/jbc.M108306200. [DOI] [PubMed] [Google Scholar]
  19. Nicodeme F, Geffroy S, Conti M, Delobel B, Soenen V, Grardel N, Porte H, Copin MC, Lai JL, Andrieux J. Familial occurrence of thymoma and autoimmune diseases with the constitutional translocation t(14;20)(q24.1;p12.3) Genes Chromosomes Cancer. 2005;44:154–60. doi: 10.1002/gcc.20225. [DOI] [PubMed] [Google Scholar]
  20. Odero MD, Grand FH, Iqbal S, Ross F, Roman JP, Vizmanos JL, Andrieux J, Lai JL, Calasanz MJ, Cross NC. Disruption and aberrant expression of HMGA2 as a consequence of diverse chromosomal translocations in myeloid malignancies. Leukemia. 2005;19:245–52. doi: 10.1038/sj.leu.2403605. [DOI] [PubMed] [Google Scholar]
  21. Park JH, Wacholder S, Gail MH, Peters U, Jacobs KB, Chanock SJ, Chatterjee N. Estimation of effect size distribution from genome-wide association studies and implications for future discoveries. Nat Genet. 2010;42:570–5. doi: 10.1038/ng.610. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Pruthi S, Gostout BS, Lindor NM. Identification and Management of Women With BRCA Mutations or Hereditary Predisposition for Breast and Ovarian Cancer. Mayo Clin Proc. 2010;85:1111–20. doi: 10.4065/mcp.2010.0414. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D, Maller J, Sklar P, de Bakker PI, Daly MJ, Sham PC. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet. 2007;81:559–75. doi: 10.1086/519795. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Sigurdsson S, Van Komen S, Bussen W, Schild D, Albala JS, Sung P. Mediator function of the human Rad51B-Rad51C complex in Rad51/RPA-catalyzed DNA strand exchange. Genes Dev. 2001;15:3308–18. doi: 10.1101/gad.935501. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Stephens M, Smith NJ, Donnelly P. A new statistical method for haplotype reconstruction from population data. Am J Hum Genet. 2001;68:978–89. doi: 10.1086/319501. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Takaku M, Machida S, Hosoya N, Nakayama S, Takizawa Y, Sakane I, Shibata T, Miyagawa K, Kurumizaka H. Recombination activator function of the novel RAD51- and RAD51B-binding protein, human EVL. J Biol Chem. 2009;284:14326–36. doi: 10.1074/jbc.M807715200. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Takata M, Sasaki MS, Sonoda E, Fukushima T, Morrison C, Albala JS, Swagemakers SM, Kanaar R, Thompson LH, Takeda S. The Rad51 paralog Rad51B promotes homologous recombinational repair. Mol Cell Biol. 2000;20:6476–82. doi: 10.1128/mcb.20.17.6476-6482.2000. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Thacker J, Zdzienicka MZ. The XRCC genes: expanding roles in DNA double-strand break repair. DNA Repair (Amst) 2004;3:1081–90. doi: 10.1016/j.dnarep.2004.04.012. [DOI] [PubMed] [Google Scholar]
  29. The International HapMap3 Consortium. Integrating common and rare genetic variation in diverse human populations. Nature. 2010;467:52–58. doi: 10.1038/nature09298. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Thomas G, Jacobs KB, Kraft P, Yeager M, Wacholder S, Cox DG, Hankinson SE, Hutchinson A, Wang Z, Yu K, Chatterjee N, Garcia-Closas M, Gonzalez-Bosquet J, Prokunina-Olsson L, Orr N, Willett WC, Colditz GA, Ziegler RG, Berg CD, Buys SS, McCarty CA, Feigelson HS, Calle EE, Thun MJ, Diver R, Prentice R, Jackson R, Kooperberg C, Chlebowski R, Lissowska J, Peplonska B, Brinton LA, Sigurdson A, Doody M, Bhatti P, Alexander BH, Buring J, Lee IM, Vatten LJ, Hveem K, Kumle M, Hayes RB, Tucker M, Gerhard DS, Fraumeni JF, Jr, Hoover RN, Chanock SJ, Hunter DJ. A multistage genome-wide association study in breast cancer identifies two new risk alleles at 1p11.2 and 14q24.1 (RAD51L1) Nat Genet. 2009;41:579–84. doi: 10.1038/ng.353. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Varghese JS, Easton DF. Genome-wide association studies in common cancers--what have we learnt? Curr Opin Genet Dev. 2010;20:201–9. doi: 10.1016/j.gde.2010.03.012. [DOI] [PubMed] [Google Scholar]
  32. Zeidler M, Varambally S, Cao Q, Chinnaiyan AM, Ferguson DO, Merajver SD, Kleer CG. The Polycomb group protein EZH2 impairs DNA repair in breast epithelial cells. Neoplasia. 2005;7:1011–9. doi: 10.1593/neo.05472. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Zhong H, Prentice RL. Bias-reduced estimators and confidence intervals for odds ratios in genome-wide association studies. Biostatistics. 2008;9:621–34. doi: 10.1093/biostatistics/kxn001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Zhong H, Prentice RL. Correcting “winner’s curse” in odds ratios from genomewide association findings for major complex human diseases. Genet Epidemiol. 2009;34:78–91. doi: 10.1002/gepi.20437. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Material
NIHMS616599-supplement-Supplemental_Material.docx (19.6KB, docx)

RESOURCES