Background:BRCA1 and BRCA2 mutation carriers have a lifetime breast cancer risk of 40% to 80%, suggesting the presence of risk modifiers. We previously identified significant associations in genetic variants in the insulin-like growth factor (IGF) signaling pathway. Here, we investigate additional IGF signaling genes as risk modifiers for breast cancer development in BRCA carriers.

Methods: A cohort of 1,019 BRCA1 and 500 BRCA2 mutation carriers were genotyped for 99 single-nucleotide polymorphisms (SNP) in 13 genes. Proportional hazards regression was used to model time from birth to diagnosis of breast cancer for BRCA1 and BRCA2 carriers separately. For linkage disequilibrium (LD) blocks with multiple SNPs, an additive genetic model was used. For an SNP analysis, no additivity assumptions were made.

Results: Significant associations were found between risk of breast cancer and LD blocks in IGF2 for BRCA1 and BRCA2 mutation carriers (global P values of 0.009 for BRCA1 and 0.007 for BRCA2), HTRA1 for BRCA1 carriers (global P value of 0.005), and MMP3 for BRCA2 carriers (global P = 0.0000007 for BRCA2).

Conclusions: We identified significant associations of genetic variants involved in IGF signaling. With the known interaction of BRCA1 and IGF signaling and the loss of PTEN in a majority of BRCA1 tumors, this suggests that signaling through AKT may modify breast cancer risk in BRCA1 carriers.

Impact: These results suggest potential avenues for future research targeting the IGF signaling pathway in modifying risk in BRCA1and BRCA2 mutation carriers. Cancer Epidemiol Biomarkers Prev; 20(8); 1690–702. ©2011 AACR.

Women who carry mutations in the BRCA1 or BRCA2 (BRCA) genes have a substantially increased risk of developing breast and ovarian cancers as compared with the general population. Estimates of the age-specific risk attributable to mutations at these loci vary depending on the ascertainment scheme. The cumulative risks of breast cancer by age 70 years have been estimated between 40% and 87% for BRCA1 and BRCA2 mutation carriers (1–5). Among BRCA mutation carriers, there is considerable variability in both age at diagnosis and incidence of breast and ovarian cancers, even among women who carry the same BRCA mutation (6–8). These observations of large variation in risk are consistent with the hypothesis that the breast cancer risk in mutation carriers is modified by other genetic and/or environmental factors. To test this hypothesis, we have focused on genetic variation in the insulin-like growth factor (IGF) signaling pathway, as it regulates both cellular proliferation and apoptosis (reviewed in refs. 9–11). We previously reported significant associations of single-nucleotide polymorphisms (SNP) in insulin receptor substrate (IRS) 1 and IGF1R in BRCA1 mutation carriers and in IGF1R and IGFBP2 in BRCA2 mutation carriers (12).

For this study, we investigated the association of variants in additional genes involved in IGF signaling. We focused on 13 genes in this pathway, which were closely related to genes with significant associations with breast cancer risk in our previous work (12). These genes included the IGF1 receptor (IGF1R) ligands IGF2 and insulin (INS); the IGF binding proteins (IGBFP) IGFBP3 (and its partners in binding IGF, IGFALS), IGFBP4, and IGFBP6; the IGFBP proteases kallikrein–related peptidase 3 (KLK3, aka prostate-specific antigen; PSA), cathepsin D (CTSD), matrix metalloproteinases 3 and 7 (MMP3 and MMP7), HTRA serine peptidase 1 (HTRA1, aka PRSS11); the IGF1R docking protein IRS2; and its phosphorylation target phosphoinositide 3-kinase 110-kb catalytic subunit B (PIK3CB), as potential disease modifiers in deleterious mutation carriers of BRCA1 and BRCA2.

Participants

For the primary investigation, women with germline, deleterious mutations in BRCA1 or BRCA2 were identified in 14 centers in the United States, 1 center in Canada, and 1 center in Austria, including Baylor University Medical Center-Dallas, Dallas, TX, Beth Israel in Boston, City of Hope, Creighton University, Omaha, NE, Dana Farber, Boston, MA, Fox Chase Cancer Center, Philadelphia, PA, Georgetown University, Washington DC, the Mayo Clinic, Scottsdale, AZ, Medical University of Vienna, Vienna, Austria, North Shore University HealthSystem, Chicago, IL University of California, Los Angeles, CA, University of California, Irvine, CA, University of Chicago, Chicago, IL, University of Pennsylvania, Philadelphia, PA, and Women's College Hospital, Toronto, Ontario. The majority of subjects were recruited from the Medical University in Vienna, Creighton University in Nebraska, the University of Pennsylvania, and the University of California Irvine (previously at the University of Utah and now at the Beckman Research Institute of the City of Hope). All centers are part of the Modifiers and Genetics in Cancer (MAGIC) Consortium. All participants were enrolled under Institutional Review Boards or ethics committee approval at each participating site. Women were identified through either clinical or research high-risk programs, generally because of a family history of breast and/or ovarian cancer. The current study uses data on a total of 1,519 adult, female non-Hispanic Caucasian mutation carriers, including 1,019 BRCA1 carriers (381 cases and 638 unaffected controls) and 500 BRCA2 carriers (219 cases and 281 unaffected controls). The BRCA mutation status of all subjects was confirmed by direct mutation testing, with full informed consent under protocols approved by the human subjects review boards at each institution. Women were eligible for entry into the study cohort if they tested positive for a known deleterious mutation in BRCA1 or BRCA2. Women with deleterious mutations in both BRCA1 and BRCA2 were excluded. Women were excluded if they were missing information on year of birth, parity, menopausal status, and oral contraceptive use or had a diagnosis of breast or ovarian cancer more than 3 years prior to study entry. Information about breast cancer, ovarian cancer, prophylactic mastectomy, and prophylactic oophorectomy was obtained from medical records, and information on reproductive history and lifestyle habits was obtained by questionnaire.

SNP genotyping

The Genome Variation Server (GVS; ref. 13) at the University of Washington was used to identify SNPs in the genes and select haplotype tagging SNPs (htSNP) using the program LDSelect (14), which is integrated with the GVS. A total of 99 SNPs were selected from 13 genes to mark the common genetic variation in each gene and minimize the genotyping costs. However, because some of the genes had SNPs from the pooled HapMap race/ethnicity samples, we could not discern race and so we recalculated linkage disequilibrium (LD) blocks specifically for the European (EU) population in this article as described later.

Genotyping was done using the SNPlex assay from Applied Biosystems, Inc. (ABI), for all but 3 SNPs in IGFBP3 for which alleles were genotyped by ABI TaqMan exonuclease assays. Genotyping by SNPlex was done according to ABI's protocols. The SNPlex assay is based on an oligonucleotide ligation assay (OLA), followed by multiplex PCR amplification using a universal primer pair, hybridization of reporter probes to amplicons, and capillary electrophoresis to rapidly identify eluted reporter probes (ABI). On the basis of the list of SNPs sent to ABI, 3 SNPlex pools were designed and ordered for genotyping. We obtained genotype data for 99 SNPs following the manufacturer's protocol (ABI). Briefly, 120 ng genomic DNA for each sample was fragmented by denaturing at 99°C for 10 minutes. Phosphorylation of probes and allele-specific ligation between fragmented DNA and probes were done by running an OLA program on a PTC225 thermal cycler. After purification of ligation product by exonuclease digestion, PCR was carried out using universal primers with the reverse primer biotinylated. During hybridization, the captured biotinylated PCR strands were hybridized with fluorescently labeled ZipChute probes, with a unique sequence corresponding to the unique Zip code sequences at the 5′ end of the allelic-specific probes used in OLA reactions. The ZipChute probes were separated by the ABI 3130xl capillary sequencer, and genotypes were called by ABI Genemapper 4.0 program. The laboratory technician was blinded as to whether samples were duplicates, cases, or controls. Each 96-well plate contained 87 unique samples, 2 positive controls, 1 negative control, and 6 ladders. Genotyping call rates ranged from 67% to 99%. SNPs that were monomorphic or with minor allele frequencies (MAF) less than 0.03 (n = 10 SNPs) or with genotyping call rates less than 80% (n = 5 SNPs) were excluded from further analysis. Because the duplicate concordance rates of the 40 duplicated samples were higher than 99.9%, we felt confident that the data were accurate even for the 10 SNPs with 82% to 89% call rates.

Determination of LD blocks

To define LD blocks relative to the current study population, all LD blocks were computed using the current analytic data set. Because of the relatively small proportion of non-EU subjects within the study sample, and the possibility that LD structure may differ between EU and non-EU ancestries, we restricted the analysis to only non-Hispanic Caucasians. SNPs were grouped according to their adjacent pairwise LD coefficient, D′, which was computed between all adjacent marker pairs within each candidate gene. To account for within-family correlation, multiple outputation (15) was used to estimate D′. In this case, a single member from each family was randomly sampled to create a single bootstrap sample, from which D′ was computed. This process was repeated to obtain 200 bootstrap samples, yielding an empirical distribution of D′. An LD block was defined as a set of contiguous SNPs having D′ values exceeding 0.90 between each contiguous pair of SNPs. The boundary of an LD block would be defined by a marker pair with D′ ≤ 0.9. Given that individuals of Ashkenazi Jewish ancestry make up approximately 24% of the analytic sample, we further explored differential LD structure between Ashkenazi Jewish ancestry and non-Ashkenazi Jewish ancestry. No statistically significant differences were found (as defined by nonoverlapping 95% CIs for D′ between ancestry strata). However, to guard against grouping SNPs within an ancestry that may or may not be in linkage, we used the minimum D′ between the non-Jewish and the Jewish groups when defining the LD blocks. The LD blocks for the SNPs within each gene are shown in Supplementary Table S1 along with the MAFs.

Statistical analysis

For all analyses, subjects were considered at risk for breast cancer from birth until the first occurrence of breast cancer diagnosis, death, or loss to follow-up. Subjects were censored in the event that they underwent a bilateral prophylactic surgery of the breasts more than 1 year preceding the diagnosis of breast cancer. Bilateral prophylactic surgery of the breasts occurring within a year of breast cancer was considered an event to avoid potential biases resulting from informative censoring. Descriptive breast cancer rates were calculated as the observed number of breast cancers per total patient time at risk and standardized to the age distribution of the study cohort at the time of interview (16). Covariates that vary with time (ovarian cancer and prophylactic ovarian surgery) were treated as time-dependent in the calculation of rates. Thus, a subject who had a diagnosis of ovarian cancer contributed time at risk in the nonovarian cancer group prior to the diagnosis and then time at risk in the ovarian cancer group following the diagnosis. Because subjects were ascertained primarily from high-risk clinics, selection bias is likely to result in an overrepresentation of cases in the analysis sample relative to the general population. To account for potential bias in cumulative risk estimates due to nonrandom sampling from the general population, Kaplan–Meier estimates of the cumulative probability of breast cancer diagnosis were computed using age-specific sampling weights for cases and controls as obtained from Antoniou and colleagues (17).

Cox proportional hazards regression was used to estimate the association between time from birth to diagnosis of breast cancer and haplotypes in genes within the IGF pathway. In this model, the hazard or instantaneous rate of breast cancer diagnosis is modeled as a function of the predictor covariates. The relative risk or HR is then interpreted for each covariate as the proportionate change in the instantaneous rate of diagnosis for 2 individuals differing by a single unit of that covariate. When analyzing LD blocks with multiple SNPs, an additive haplotype effect was assumed where the haplotype composed of the major allele for each individual SNP within the LD block was used as the referent group for comparisons (i.e., 0000). However, when an LD block consisted of an SNP, a general genetic model making no additivity assumption was used. To account for phase uncertainty in haplotype analysis, we used a 2-step approximation to the semiparametric maximum likelihood estimator of Lin and Zeng (18). This method used the expectation–maximization algorithm to compute posterior estimates of the probability of all potential haplotypes for a subject given their known genotype, and these probabilities were used to weight the individual's contribution to the partial likelihood. A similar approach has previously been applied to logistic regression models for analyzing case–control data and shown to provide robust inference for relatively common haplotypes with little phase ambiguity (19). To account for hierarchical clustering at the individual level (multiple records per individual were analyzed according to the number of potential diplotypes consistent with the individual's genotype) and the family level (matched controls were often selected from the family of a case), the sandwich estimator of Lin and Wei (20) was used in combination with multiple outputation (15) to obtain robust inference about haplotype associations. All estimates were adjusted for birth cohort (to account for frequency matching of cases and controls), ancestry, age at first pregnancy, and region of center [North American (US) vs. EU]. Ashkenazi Jewish individuals were considered a separate group because the carriers had only 1 of 3 founder mutations. Age at first pregnancy, prophylactic oophorectomy, and ovarian cancer status were treated as time-dependent covariates in the analysis. To account for potential selection bias through recruitment from high-risk clinics, all regression models employed age-specific sampling probability weights for cases and controls as obtained from Antoniou and colleagues (17). For LD blocks exhibiting significant associations with the time to breast cancer diagnosis, secondary analyses of individual SNPs making up the LD block were conducted. In addition, we considered sensitivity analyses to assess the assumption of additivity for haplotype models and found no quantifiable differences.

In total, the current analysis involved testing of 83 SNPs in 41 LD blocks across the 13 genes, which is likely to result in an inflation of the family-wise type I error rate for the study if unadjusted critical values are used for assessing LD block significance. Noting that this analysis represents a first stage in identifying variants in the IGF pathway that are associated with time to breast cancer diagnosis, we sought to control the family-wise type error rate at 10% to minimize the type II error rate, limiting the possibility of ruling out potentially important LD blocks from future investigation. Simulation was used to estimate the family-wise type I error rate, assuming a correlation of 0.75 across tests. On the basis of 100,000 simulations, it was estimated that an adjusted P value of 0.010 on any individual LD block test would result in a family-wise type I error rate of 10% for the study. Thus, an adjusted P value less than 0.010 was interpreted as a significant association.

The characteristics of study participants, study sites, and the observed incidence rate (per 1,000 women per year) of breast cancer diagnosis stratified by BRCA status were previously described (12). Briefly, the age-standardized incidence rate of breast cancer diagnosis was estimated to be 26.94 per 1,000 per year in BRCA1 carriers (95% CI: 19.79, 34.10) compared with 25.03 per 1,000 per year in BRCA2 carriers (95% CI: 18.71–31.36). The characteristics of the study participants, limited to non-Hispanic Caucasians, are shown in Table 1. The majority of study subjects in both strata were of non-Jewish ancestry (76%). Of the study subjects, 9% underwent bilateral prophylactic mastectomy and 41% underwent prophylactic bilateral salpingo-oophorectomy. Median age at diagnosis was estimated to be 57.0 years (95% CI: 54.1–62.2) among BRCA1 carriers and was 70.5 years (95% CI: 67.7–INF) among BRCA2 carriers. The results are presented grouped by genes in components of the IGF signaling pathway. As described in Materials and Methods, only P values of less than 0.010 for individual LD blocks were considered significant.

Table 1.

Characteristics of study participants by BRCA gene

CharacteristicBRCA1 carriersBRCA2 carriers
Total, n (%)Cases, n (%)Total, n (%)Cases, n (%)
Total 1,019 381 500 219 
Ethnicity     
 Caucasian (non-Jewish, non-Hispanic) 774 (76) 283 (74) 381 (76) 176 (80) 
 Caucasian (Jewish, non-Hispanic) 245 (24) 98 (26) 119 (24) 43 (20) 
Ovarian cancer     
 Yes 111 (11) 26 (7) 30 (6) 5 (2) 
 No 908 (89) 407 (93) 513 (94) 233 (98) 
Bilateral BPO     
 Yes, before breast cancer diagnosis 270 42 103 17 
 Yes, after breast cancer diagnosis 154 154 94 94 
 No BPO 594 184 303 108 
Prophylactic bilateral mastectomy     
 Yes 99  37  
 No 920  463  
CharacteristicBRCA1 carriersBRCA2 carriers
Total, n (%)Cases, n (%)Total, n (%)Cases, n (%)
Total 1,019 381 500 219 
Ethnicity     
 Caucasian (non-Jewish, non-Hispanic) 774 (76) 283 (74) 381 (76) 176 (80) 
 Caucasian (Jewish, non-Hispanic) 245 (24) 98 (26) 119 (24) 43 (20) 
Ovarian cancer     
 Yes 111 (11) 26 (7) 30 (6) 5 (2) 
 No 908 (89) 407 (93) 513 (94) 233 (98) 
Bilateral BPO     
 Yes, before breast cancer diagnosis 270 42 103 17 
 Yes, after breast cancer diagnosis 154 154 94 94 
 No BPO 594 184 303 108 
Prophylactic bilateral mastectomy     
 Yes 99  37  
 No 920  463  

Abbreviation: BPO, prophylactic oophorectomy.

IGFIR ligands: IGF2 and INS

In Figure 1, we present the estimated HR for time to diagnosis by LD block and BRCA status after adjustment for covariates (described in Materials and Methods). For both BRCA1 and BRCA2 mutation carriers, significant associations of LD blocks in IGF2 were observed. Among BRCA1 carriers, for IGF2 LD block 3 (defined by SNP rs1065687), women with at least 1 variant allele were estimated to experience a 76% higher relative risk of diagnosis than did women with no variant allele (per-allele HR 1.76; 95% CI: 1.15–2.70; P = 0.009). Among BRCA2 carriers, in LD block 5, composed of 4 SNPs, the global P value was significant for LD block 4 (P = 0.003) and LD block 5 (P = 0.0007). In LD block 4, women with haplotype 10 had a more than 2-fold increased risk of breast cancer (HR = 2.20; 95% CI: 1.4–3.5; P = 0.0008). In LD block 5, women with haplotypes 0001 (HR = 0.48; 95% CI: 0.30–0.76; P = 0.0017), 0100 (HR = 0.40; 95% CI: 0.26–0.61; P = 0.00002), and 0101 (HR = 0.39; 95% CI: 0.20–0.79; P = 0.009) were estimated to experience a significantly decreased risk of diagnosis when compared with the reference haplotype. Given the significant HRs for the haplotypes in IGF2 LD blocks 4 and 5, we investigated the HRs for the 6 individual SNPs (rs3213232, rs3213229, rs3213223, rs3213221, rs3213217, and rs3213216) within the LD blocks to examine whether the observed haplotype effect could be attributed to any one SNP. Of the 6 SNPs, there was only an effect of carrying 1 copy of the variant allele of rs3213221 (HR = 0.60; 95% CI: 0.40–0.92; P = 0.018). There was no significant association of INS and breast cancer risk among BRCA1 or BRCA2 carriers after adjusting for multiple testing.

Figure 1.

Estimated HRs associated with haplotype presence for IGF2 (A) and INS (B). Linkage blocks (C) were defined as pairwise D′ ≥ 0.90. Estimates were stratified by BRCA1/2 mutation status (BRCA1 in left column and BRCA2 in right column) and adjusted for birth cohort and ancestry (non-Jewish or Jewish) as well as first pregnancy, prophylactic oophorectomy, and diagnosis of ovarian cancer as time-dependent covariates. C, SNPs within each of the LD blocks are shown for each gene.

Figure 1.

Estimated HRs associated with haplotype presence for IGF2 (A) and INS (B). Linkage blocks (C) were defined as pairwise D′ ≥ 0.90. Estimates were stratified by BRCA1/2 mutation status (BRCA1 in left column and BRCA2 in right column) and adjusted for birth cohort and ancestry (non-Jewish or Jewish) as well as first pregnancy, prophylactic oophorectomy, and diagnosis of ovarian cancer as time-dependent covariates. C, SNPs within each of the LD blocks are shown for each gene.

Close modal

IGFBPs: IGFALS (in the complex with IGFBP3), IGFBP3, IGFBP4, and IGFBP6

Estimated HRs for the haplotypes of the IGBFPs are shown in Figure 2. No significant global tests of associations were observed for any haplotypes of the 3 IGFBP genes. The SNP in IGFALS was also not associated with risk (see Supplementary Table S1).

Figure 2.

Estimated HRs associated with haplotype presence for IGFBP3 (A), IGFBP4 (B), and IGFBP6 (C). Linkage blocks (D) were defined as pairwise D′ ≥ 0.90. Estimates were stratified by BRCA1/2 mutation status (BRCA1 in left column and BRCA2 in right column) and adjusted for birth cohort and ancestry as well as first pregnancy, prophylactic oophorectomy, and diagnosis of ovarian cancer as time-dependent covariates.

Figure 2.

Estimated HRs associated with haplotype presence for IGFBP3 (A), IGFBP4 (B), and IGFBP6 (C). Linkage blocks (D) were defined as pairwise D′ ≥ 0.90. Estimates were stratified by BRCA1/2 mutation status (BRCA1 in left column and BRCA2 in right column) and adjusted for birth cohort and ancestry as well as first pregnancy, prophylactic oophorectomy, and diagnosis of ovarian cancer as time-dependent covariates.

Close modal

IGFBP proteases: HTRA1, MMP3, MMP7, KLK3, and CTSD

Estimated HRs for the haplotypes of the IGFBP proteases are shown in Figures 3 and 4. A highly significant association in HTRA1 LD block 3, composed of 3 SNPs, was observed in BRCA1 carriers (global P = 0.005) with no association in BRCA2 carriers (global P = 0.147). Within LD block 3, the rare haplotypes grouped together were significantly associated with a 2-fold risk to develop breast cancer (HR = 2.12; 95% CI: 1.30–3.46; P = 0.0028). When looking at individual SNPs within HTRA1, no SNP within LD block 3 was associated with risk. There were no significant associations of LD blocks in HTRA1 and risk among BRCA2 carriers. However, among BRCA2 carriers, carrying the variant allele in rs12571363 in LD block 1 conferred a 2-fold increased risk (per allele HR = 2.03; 95% CI: 1.29–3.21; P = 0.002). Haplotypes in KLK3 and CTSD were not associated with any modification of risk for developing breast cancer. For MMP3 in BRCA2 carriers, a highly significant association (global P = 0.0000003) in LD block 3 was observed because of a combination of rare haplotypes associated with a 5.1-fold higher risk (HR: 5.08; CI: 2.67–9.67; P = 0.0000007; Fig. 4A). None of the 3 SNPs constituting this haplotype, rs476762, rs679620, or rs3025058, could explain the association. For MMP7, after adjusting for multiple testing, there were no significant associations of any LD blocks with risk in BRCA1 or BRCA2 carriers.

Figure 3.

Estimated HRs associated with haplotype presence for HTRA1 (A), KLK3 (B), and CTSD (C). Linkage blocks (D) were defined as pairwise D′ ≥ 0.90. Estimates were stratified by BRCA1/2 mutation status (BRCA1 in left column and BRCA2 in right column) and adjusted for birth cohort and ancestry as well as first pregnancy, prophylactic oophorectomy, and diagnosis of ovarian cancer as time-dependent covariates.

Figure 3.

Estimated HRs associated with haplotype presence for HTRA1 (A), KLK3 (B), and CTSD (C). Linkage blocks (D) were defined as pairwise D′ ≥ 0.90. Estimates were stratified by BRCA1/2 mutation status (BRCA1 in left column and BRCA2 in right column) and adjusted for birth cohort and ancestry as well as first pregnancy, prophylactic oophorectomy, and diagnosis of ovarian cancer as time-dependent covariates.

Close modal
Figure 4.

Estimated HRs associated with haplotype presence for MMP3 (A) and MMP7 (B). Linkage blocks (C) were defined as pairwise D′ ≥ 0.90. Estimates were stratified by BRCA1/2 mutation status (BRCA1 in left column and BRCA2 in right column) and adjusted for birth cohort and ethnicity as well as first pregnancy, prophylactic oophorectomy, and diagnosis of ovarian cancer as time-dependent covariates.

Figure 4.

Estimated HRs associated with haplotype presence for MMP3 (A) and MMP7 (B). Linkage blocks (C) were defined as pairwise D′ ≥ 0.90. Estimates were stratified by BRCA1/2 mutation status (BRCA1 in left column and BRCA2 in right column) and adjusted for birth cohort and ethnicity as well as first pregnancy, prophylactic oophorectomy, and diagnosis of ovarian cancer as time-dependent covariates.

Close modal

IGF1R downstream docking protein and phosphorylation target: IRS2 and PIK3CB

There was no significant association of LD blocks in either IRS2 or PIK3CB and breast cancer risk (Fig. 5).

Figure 5.

Estimated HRs associated with haplotype presence for IRS2 (A) and PIK3CB (B). Linkage blocks (C) were defined as pairwise D′ ≥ 0.90. Estimates were stratified by BRCA1/2 mutation status (BRCA1 in left column and BRCA2 in right column) and adjusted for birth cohort and ancestry as well as first pregnancy, prophylactic oophorectomy, and diagnosis of ovarian cancer as time-dependent covariates.

Figure 5.

Estimated HRs associated with haplotype presence for IRS2 (A) and PIK3CB (B). Linkage blocks (C) were defined as pairwise D′ ≥ 0.90. Estimates were stratified by BRCA1/2 mutation status (BRCA1 in left column and BRCA2 in right column) and adjusted for birth cohort and ancestry as well as first pregnancy, prophylactic oophorectomy, and diagnosis of ovarian cancer as time-dependent covariates.

Close modal

The IGF pathway signals cells to grow, differentiate, and survive through activation of the phosphoinositol-3-kinase/protein kinase AKT (PI3K/AKT) pathway and the RAS/mitogen-activated protein kinase (RAS/MAPK) pathway. The role of the IGF pathway in breast cancer was reviewed, and a figure of the IGF system and its downstream effectors was presented by Hamelers and Steenbergh (21). This study is a continuation of our investigation of the role of genetic variants in IGF signaling as modifiers of breast cancer risk in women who carry deleterious mutations in BRCA1 and BRCA2. We previously investigated only a small number of the genes involved in IGF signaling and reported significant HRs associated with genetic variants in IGF1R and IRS1 in BRCA1 mutation carriers and in IGFBP2 and IGFBP5 in BRCA2 mutation carriers (12). In this investigation of an additional 13 genes in the pathway, there were significant associations of variants in IGF2 in both BRCA1 and BRCA2 mutation carriers and HTRA1 in BRCA1 mutation carriers and MMP3 in BRCA2 carriers.

There have been a limited number of epidemiologic studies of the association of sporadic breast cancer risk and genetic variation in the genes in the IGF pathway investigated in this study. In a large study in the NCI breast and prostate cancer cohort consortium of 6,292 breast cancer cases and 8,135 unaffected controls, they reported no significant associations (P < 0.00005) of SNPs in IGF2, IGFALS, IGFBP3, IGFBP4, IGFBP6, and IRS2 and breast cancer risk (22). Their results were inconsistent with our observation of significant associations with haplotypes in IGF2, which may reflect different SNPs and different population characteristics including ages at diagnosis and BRCA1 and BRCA2 carrier status. There have been several studies of the −202 A>C SNP in the IGFBP3 promoter and breast cancer risk, with most being negative (reviewed in ref. 23). In association studies of SNPs and breast cancer risk in the Cancer Prevention Study II, 3 of 11 SNPs in IRS2 were associated with breast cancer risk (haplotype OR = 0.76; 95% CI: 0.63–0.92; ref. 24) and the INS rs3842752 SNP was not associated with breast cancer risk (25). The INS SNP was the same, whereas none of the 11 IRS2 SNPs were the same as investigated in our study. Two studies have reported significant associations of SNPs in MMP7 and breast cancer survival, although with different SNPs than investigated here (26, 27) As yet, there are no reported association studies in breast cancer investigating SNPs in HTRA1, PIK3CB, or MMP3.

IGF1 and IGF2 are the primary ligands for binding to the IGF1R, and IGF2 levels are higher than IGF1 levels in cells. Interestingly, in a study of women of 43 women with localized breast cancer and 38 unaffected controls, there were significantly elevated circulating free IGF2 and IGF1 levels yet significantly reduced total IGF2 levels and reduced total IGF1 levels in the breast cancer cases as compared with controls (28). They also found that IGFBP2 levels were significantly reduced in the cases with no differences in IGFBP1, 3, 4 or 6 levels and suggested that the increase in free IGF levels was because there was less IGFBP2 to bind IGF2. IGFBP2 and IGFBP5 preferentially bind IGF2 over IGF1. With higher amounts of free IGFs, there can be more signaling activity through IGF1R. The significant association with variants in IGF2 fits with our previous results in which we had observed a significant association with variants at the IGFBP2_IGFBP5 locus in BRCA2 carriers and at IGF1R in BRCA1 carriers.

Proteolysis of IGFBPs occurs through the action of IGFBP proteases that include serine proteases (KLKs and HTRA1), MMPs, and CTS (29). With proteolysis of the IGFBPs, there is then more bioavailability of the IGFs for signaling through IGF1R. In BRCA1 mutation carriers, an HTRA1 haplotype in LD block 3 was significantly associated with breast cancer (global P = 0.0005). HTRA1, located at 10q25.3-q26.2, was identified in 1996 as a serine protease specific for IGFBPs (30). HTRA1 is also now thought to function as a tumor suppressor gene (31), possibly through regulation of TGF-B signaling (32). None of the 4 SNPs in the LD block were significantly associated with breast cancer, suggesting that the associated haplotype only marks the causal variant in the region. Further investigation of this region is needed to understand how it may be modifying risk of developing breast cancer in BRCA1 mutation carriers. A group of rare haplotypes (combined frequency of 2%) in MMP3, another IGFBP protease, was significantly associated with breast cancer in BRCA2 carriers (HR = 4.33). For MMP3, many more carriers need to be investigated to determine the actual associated haplotype within the grouped haplotypes, as neither of the 2 SNPs in the LD block showed association.

PIK3CB is a 110-kDa catalytic subunit of PI3K and is activated through the IGF1R as follows: Binding of IGF induces receptor autophosphorylation of the β-subunit of IGF1R, resulting in activation of the tyrosine kinase activity and subsequent tyrosine phosphorylation of IRS1. The activated IRS1 binds the p85 α-subunit PI3K, which, in turn, activates PIK3CB for downstream signaling. PTEN regulates PI3K signaling by dephosphorylating PI3K, so in the absence of PTEN, the pathway is activated. In cells without PTEN, experiments have shown that PIK3CB is necessary for continued PI3K signaling (33). PIK3CB may be particularly relevant to cancer in BRCA1 carriers because a majority of BRCA1 and other basal-like breast cancers have loss of PTEN expression (34). Furthermore, this suggests a possible therapeutic target in BRCA1 carriers. Clinical trials are ongoing with agents targeting the inhibition of the IGF1R and mTOR, although none are targeted to BRCA mutation carriers. In BRCA1 carriers, there was a marginal association of genetic variation in PIK3CB and breast cancer. The intronic SNP rs9878820 was associated with breast cancer (per allele HR = 0.83, CI: 0.70–0.99, P = 0.04; homozygous recessive HR = 0.69, P = 0.04), and was more strongly associated in the larger data set including Hispanics, African Americans, and others (P = 0.008).

Although this study provides an important continued step in identifying potential genetic modifiers of risk among BRCA1 and BRCA2 mutation carriers, there are some limitations. First, the sample size was limited, particularly for BRCA2 carriers, so that there may be associations that we could not detect and the point estimates for the associations we did detect may be imprecise, particularly for SNPs and haplotypes of low frequency. Second, the loci that we identified in this study that had significant HRs for risk were not previously identified as the top candidates in the genome-wide association study (GWAS) of BRCA1 (35) or BRCA2 mutation carriers (36). However, even known associations can be missed in GWAS, as only the top hits are further investigated through replication and imputation of additional SNPs in those regions. One example is the previous finding that a promoter SNP in RAD51C (37) is a modifier of risk in BRCA2 carriers with an HR of 3.2 among the rare homozygotes, yet it was not identified in the GWAS (36). Therefore, it is possible that one or more of the associations identified here were missed in the GWAS. Third, although the IGF pathway was a priori hypothesized as a source for potential modifiers, multiple LD blocks were considered for association testing and such testing could lead to inflation of the overall type I error rate for the study. With this being said, we studied only a relatively small number of the genes in IGF signaling that we a priori deemed would potentially play a role in the time to diagnosis and considered significant only those associations that decreased beyond a multiple comparison-adjusted critical value. In addition, we view the goal of the current research as hypothesis refinement where, based upon the results from this well-defined set of genes, we will further validate the results using an independent sample. As with all observational studies, there is the potential for selection bias and unmeasured confounding. However, we have adjusted for those factors that are most likely to influence the risk of breast cancer diagnosis within this cohort, thus lowering the potential for unadjusted confounding.

We and others have investigated putative risk factors, and a number of published studies have implicated loci as modifiers of breast or ovarian cancer penetrance in women who carry germline BRCA1 or BRCA2 mutations (12, 35–42). The majority of modifiers of BRCA1 cancer risk identified to date have been genetic variants in proteins that interact with BRCA1 in DNA repair. Our results suggest that IGF signaling also modifies breast cancer penetrance in BRCA1 and BRCA2 mutation carriers. IGF signaling and BRCA1 directly interact with multiple lines of evidence, including that IGF-1 enhances BRCA1 activity (43), BRCA1 regulates IGF1R and IRS1 activity (44, 45), and BRCA1 protein expression and localization are dependent on AKT activation (46).

In conclusion, this is the second study to investigate the role of genetic variation in IGF signaling and breast cancer risk in women carrying deleterious mutations in BRCA1 and BRCA2. In the previous study, we identified significant associations in variants in IGF1R and IRS1 for BRCA1 carriers and in IGF1R, IGFBP2, and IGFBP5 in BRCA2 carriers. On the basis of those results, and given the known interaction of BRCA1 and IGF signaling, we investigated additional variants in genes in this pathway and identified significant breast cancer risk modifiers with haplotypes in IGF2 for both BRCA1 and BRCA2 mutation carriers and in HTRA1 and MMP3 for BRCA1 and BRCA2 mutation carriers, respectively. The identification of causal mechanisms within these loci is needed to validate and better understand these associations. We hope that with further identification of genetic factors that modulate age at diagnosis and overall incidence of breast cancer, and better elucidation of the actual mechanisms of modifying risk, these data can be used in the future to provide more precise risk estimates to develop cancer and design targeted therapeutics for carriers.

T.R. Rebbeck is the Editor-in-Chief of Cancer Epidemiology, Biomarkers & Prevention. In keeping with the AACR's Editorial policy, the manuscript was peer reviewed and a member of the AACR's Publications Committee rendered the decision concerning acceptability. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the State of Nebraska or the Nebraska Department of Health and Human Services.

The MAGIC Consortium includes the following centers and individuals: Baylor-Charles A. Sammons Cancer Center (J.L. Blum, MD, PhD; Estelle Brothers, RN; G. Ethington), Beckman Research Institute of the City of Hope (formerly at the University of California, Irvine, S.L. Neuhausen, PhD; L. Steele; J.N. Weitzel, MD; S. Sand), Beth Israel Deaconess Medical Center (N. Tung, MD), Creighton University (C. Snyder, BA; H.T. Lynch, MD; P. Watson, PhD), Dana Farber Cancer Institute (K. Stoeckert; J.E. Garber, MD MPH), Fox Chase Cancer Center (M.B. Daly, MD, PhD; A.K. Godwin, PhD), Georgetown University (C. Isaacs, MD), Jonsson Comprehensive Cancer Center at the University of California, Los Angeles (J. Seldon, MSGC; P.A. Ganz, MD), Mayo Clinic (F. Couch, PhD), Northshore University HealthSystem Center for Medical Genetics (C. Selkirk, S.M. Weissman, W.S. Rubinstein, MD, PhD), University of Chicago (S. Cummings; O. Olopade, MD), University of Pennsylvania Health System: (S. Domchek, MD; K. Nathanson, MD; T. Friebel, MPH; T. Rebbeck, PhD), University of Texas, San Antonio (G. Tomlinson, MD), University of Vienna (C. Singer, MD), Women's College Hospital (S.A. Narod, MD).

Support was received from NIH grants 5UO1 CA86389 (to H.T. Lynch), R01-CA083855 and R01-CA74415 (to S.L. Neuhausen), R03-CA135691 (to D.L. Gillen), R01-CA102776 and R01-CA083855 (to T.R. Rebbeck), RC4-CA153828 (to J.N. Weitzel), R01CA140323 and 5U01CA113916 (to A. Godwin), R01-CA128978 and R01-CA116167 (to F. Couch), and U01CA69631 (to M.B. Daly) and a grant from the Breast Cancer Research Foundation (to F. Couch). This publication was supported in part by revenue from Nebraska cigarette taxes awarded to Creighton University by the Nebraska Department of Health and Human Services. S.L. Neuhausen is partially supported by the Morris and Horowitz Families Endowment.

1.
Brose
MS
,
Rebbeck
TR
,
Calzone
KA
,
Stopfer
JE
,
Nathanson
KL
,
Weber
BL
. 
Cancer risk estimates for BRCA1 mutation carriers identified in a risk evaluation program
.
J Natl Cancer Inst
2002
;
94
:
1365
72
.
2.
Easton
DF
,
Bishop
DT
,
Ford
D
,
Crockford
GP
. 
Genetic linkage analysis in familial breast and ovarian cancer: results from 214 families. The Breast Cancer Linkage Consortium
.
Am J Hum Genet
1993
;
52
:
678
701
.
3.
Risch
HA
,
McLaughlin
JR
,
Cole
DE
,
Rosen
B
,
Bradley
L
,
Kwan
E
, et al
Prevalence and penetrance of germline BRCA1 and BRCA2 mutations in a population series of 649 women with ovarian cancer
.
Am J Hum Genet
2001
;
68
:
700
10
.
4.
Struewing
JP
,
Hartge
P
,
Wacholder
S
,
Baker
SM
,
Berlin
M
,
McAdams
M
, et al
The risk of cancer associated with specific mutations of BRCA1 and BRCA2 among Ashkenazi Jews
.
N Engl J Med
1997
;
336
:
1401
8
.
5.
Chen
S
,
Parmigiani
G
. 
Meta-analysis of BRCA1 and BRCA2 penetrance
.
J Clin Oncol
2007
;
25
:
1329
33
.
6.
Thorlacius
S
,
Olafsdottir
G
,
Tryggvadottir
L
,
Neuhausen
S
,
Jonasson
JG
,
Tavtigian
SV
, et al
A single BRCA2 mutation in male and female breast cancer families from Iceland with varied cancer phenotypes
.
Nat Genet
1996
;
13
:
117
9
.
7.
Tonin
P
,
Weber
B
,
Offit
K
,
Couch
F
,
Rebbeck
TR
,
Neuhausen
S
, et al
Frequency of recurrent BRCA1 and BRCA2 mutations in Ashkenazi Jewish breast cancer families
.
Nat Med
1996
;
2
:
1179
83
.
8.
Ford
D
,
Easton
DF
,
Stratton
M
,
Narod
S
,
Goldgar
D
,
Devilee
P
, et al
Genetic heterogeneity and penetrance analysis of the BRCA1 and BRCA2 genes in breast cancer families. The Breast Cancer Linkage Consortium
.
Am J Hum Genet
1998
;
62
:
676
89
.
9.
Sachdev
D
,
Yee
D
. 
The IGF system and breast cancer
.
Endocr Relat Cancer
2001
;
8
:
197
209
.
10.
Jerome
L
,
Shiry
L
,
Leyland-Jones
B
. 
Deregulation of the IGF axis in cancer: epidemiological evidence and potential therapeutic interventions
.
Endocr Relat Cancer
2003
;
10
:
561
78
.
11.
Pollak
MN
,
Schernhammer
ES
,
Hankinson
SE
. 
Insulin-like growth factors and neoplasia
.
Nat Rev Cancer
2004
;
4
:
505
18
.
12.
Neuhausen
SL
,
Brummel
S
,
Ding
YC
,
Singer
CF
,
Pfeiler
G
,
Lynch
HT
, et al
Genetic variation in insulin-like growth factor signaling genes and breast cancer risk among BRCA1 and BRCA2 carriers
.
Breast Cancer Res
2009
;
11
:
R76
.
13.
SeattleSNPs Variation Discovery Resource
.
Available from:
http://gvs.gs.washington.edu/GVS/2008.
14.
Carlson
CS
,
Eberle
MA
,
Rieder
MJ
,
Yi
Q
,
Kruglyak
L
,
Nickerson
DA
. 
Selecting a maximally informative set of single-nucleotide polymorphisms for association analyses using linkage disequilibrium
.
Am J Hum Genet
2004
;
74
:
106
20
.
15.
Follmann
D
,
Proschan
M
,
Leifer
E
. 
Multiple outputation: inference for complex clustered data by averaging analyses from independent data
.
Biometrics
2003
;
59
:
420
9
.
16.
van Belle
G
,
Heagerty
PJ
,
Fisher
LD
,
Lumley
TS
. 
Biostatistics: a methodology for the health sciences
. 
Wiley series in probability and statistics
.
New York
:
Wiley
; 
2008
.
17.
Antoniou
A
,
Pharoah
PD
,
Narod
S
,
Risch
HA
,
Eyfjord
JE
,
Hopper
JL
, et al
Average risks of breast and ovarian cancer associated with BRCA1 or BRCA2 mutations detected in case Series unselected for family history: a combined analysis of 22 studies
.
Am J Hum Genet
2003
;
72
:
1117
30
.
18.
Lin
DY
,
Zeng
D
. 
Likelihood-based inference on haplotype effects in genetic association studies
.
J Am Stat Assoc
2006
;
101
:
89
104
.
19.
French
B
,
Lumley
T
,
Monks
SA
,
Rice
KM
,
Hindorff
LA
,
Reiner
AP
, et al
Simple estimates of haplotype relative risks in case-control data
.
Genet Epidemiol
2006
;
30
:
485
94
.
20.
Lin
DY
,
Wei
LJ
. 
The robust inference for the Cox proportional hazards model
.
J Am Stat Assoc
1989
;
84
:
1074
8
.
21.
Hamelers
IH
,
Steenbergh
PH
. 
Interactions between estrogen and insulin-like growth factor signaling pathways in human breast tumor cells
.
Endocr Relat Cancer
2003
;
10
:
331
45
.
22.
Canzian
F
,
Cox
DG
,
Setiawan
VW
,
Stram
DO
,
Ziegler
RG
,
Dossus
L
, et al
Comprehensive analysis of common genetic variation in 61 genes related to steroid hormone and insulin-like growth factor-I metabolism and breast cancer risk in the NCI breast and prostate cancer cohort consortium
.
Hum Mol Genet
2010
;
19
:
3873
84
.
23.
Wagner
K
,
Hemminki
K
,
Forsti
A
. 
The GH1/IGF-1 axis polymorphisms and their impact on breast cancer development
.
Breast Cancer Res Treat
2007
;
104
:
233
48
.
24.
Feigelson
HS
,
Teras
LR
,
Diver
WR
,
Tang
W
,
Patel
AV
,
Stevens
VL
, et al
Genetic variation in candidate obesity genes ADRB2, ADRB3, GHRL, HSD11B1, IRS1, IRS2, and SHC1 and risk for breast cancer in the Cancer Prevention Study II
.
Breast Cancer Res
2008
;
10
:
R57
.
25.
Wang
Y
,
McCullough
ML
,
Stevens
VL
,
Rodriguez
C
,
Jacobs
EJ
,
Teras
LR
, et al
Nested case-control study of energy regulation candidate gene single nucleotide polymorphisms and breast cancer
.
Anticancer Res
2007
;
27
:
589
93
.
26.
Beeghly-Fadiel
A
,
Shu
XO
,
Long
J
,
Li
C
,
Cai
Q
,
Cai
H
, et al
Genetic polymorphisms in the MMP-7 gene and breast cancer survival
.
Int J Cancer
2009
;
124
:
208
14
.
27.
Tapper
W
,
Hammond
V
,
Gerty
S
,
Ennis
S
,
Simmonds
P
,
Collins
A
, et al
The influence of genetic variation in 30 selected genes on the clinical characteristics of early onset breast cancer
.
Breast Cancer Res
2008
;
10
:
R108
.
28.
Espelund
U
,
Cold
S
,
Frystyk
J
,
Orskov
H
,
Flyvbjerg
A
. 
Elevated free IGF2 levels in localized, early-stage breast cancer in women
.
Eur J Endocrinol
2008
;
159
:
595
601
.
29.
Moschos
SJ
,
Mantzoros
CS
. 
The role of the IGF system in cancer: from basic to clinical studies and clinical applications
.
Oncology
2002
;
63
:
317
32
.
30.
Zumbrunn
J
,
Trueb
B
. 
Primary structure of a putative serine protease specific for IGF-binding proteins
.
FEBS Lett
1996
;
398
:
187
92
.
31.
Chien
J
,
Staub
J
,
Hu
SI
,
Erickson-Johnson
MR
,
Couch
FJ
,
Smith
DI
, et al
A candidate tumor suppressor HtrA1 is downregulated in ovarian cancer
.
Oncogene
2004
;
23
:
1636
44
.
32.
Oka
C
,
Tsujimoto
R
,
Kajikawa
M
,
Koshiba-Takeuchi
K
,
Ina
J
,
Yano
M
, et al
HtrA1 serine protease inhibits signaling mediated by TGFbeta family proteins
.
Development
2004
;
131
:
1041
53
.
33.
Wee
S
,
Wiederschain
D
,
Maira
SM
,
Loo
A
,
Miller
C
,
deBeaumont
R
, et al
PTEN-deficient cancers depend on PIK3CB
.
Proc Natl Acad Sci U S A
2008
;
105
:
13057
62
.
34.
Saal
LH
,
Gruvberger-Saal
SK
,
Persson
C
,
Lovgren
K
,
Jumppanen
M
,
Staaf
J
, et al
Recurrent gross mutations of the PTEN tumor suppressor gene in breast cancers with deficient DSB repair
.
Nat Genet
2008
;
40
:
102
7
.
35.
Antoniou
AC
,
Wang
X
,
Fredericksen
ZS
,
McGuffog
L
,
Tarrell
R
,
Sinilnikova
OM
, et al
A locus on 19p13 modifies risk of breast cancer in BRCA1 mutation carriers and is associated with hormone receptor-negative breast cancer in the general population
.
Nat Genet
2010
;
42
:
885
92
.
36.
Gaudet
MM
,
Kirchhoff
T
,
Green
T
,
Vijai
J
,
Korn
JM
,
Guiducci
C
, et al
Common genetic variants and modification of penetrance of BRCA2-associated breast cancer
.
PLoS Genet
2010
;
6
:
e1001183
.
37.
Antoniou
AC
,
Sinilnikova
OM
,
Simard
J
,
Leone
M
,
Dumont
M
,
Neuhausen
SL
, et al
RAD51 135G–>C modifies breast cancer risk among BRCA2 mutation carriers: results from a combined analysis of 19 studies
.
Am J Hum Genet
2007
;
81
:
1186
200
.
38.
Rebbeck
TR
,
Kantoff
PW
,
Krithivas
K
,
Neuhausen
S
,
Blackwood
MA
,
Godwin
AK
, et al
Modification of BRCA1-associated breast cancer risk by the polymorphic androgen-receptor CAG repeat
.
Am J Hum Genet
1999
;
64
:
1371
7
.
39.
Rebbeck
TR
,
Wang
Y
,
Kantoff
PW
,
Krithivas
K
,
Neuhausen
SL
,
Godwin
AK
, et al
Modification of BRCA1- and BRCA2-associated breast cancer risk by AIB1 genotype and reproductive history
.
Cancer Res
2001
;
61
:
5420
4
.
40.
Wang
WW
,
Spurdle
AB
,
Kolachana
P
,
Bove
B
,
Modan
B
,
Ebbers
SM
, et al
A single nucleotide polymorphism in the 5′ untranslated region of RAD51 and risk of cancer among BRCA1/2 mutation carriers
.
Cancer Epidemiol Biomarkers Prev
2001
;
10
:
955
60
.
41.
Antoniou
AC
,
Spurdle
AB
,
Sinilnikova
OM
,
Healey
S
,
Pooley
KA
,
Schmutzler
RK
, et al
Common breast cancer-predisposition alleles are associated with breast cancer risk in BRCA1 and BRCA2 mutation carriers
.
Am J Hum Genet
2008
;
82
:
937
48
.
42.
Antoniou
AC
,
Sinilnikova
OM
,
McGuffog
L
,
Healey
S
,
Nevanlinna
H
,
Heikkinen
T
, et al
Common variants in LSP1, 2q35 and 8q24 and breast cancer risk for BRCA1 and BRCA2 mutation carriers
.
Hum Mol Genet
2009
;
18
:
4442
56
.
43.
Maor
S
,
Papa
MZ
,
Yarden
RI
,
Friedman
E
,
Lerenthal
Y
,
Lee
SW
, et al
Insulin-like growth factor-I controls BRCA1 gene expression through activation of transcription factor Sp1
.
Horm Metab Res
2007
;
39
:
179
85
.
44.
Maor
S
,
Yosepovich
A
,
Papa
MZ
,
Yarden
RI
,
Mayer
D
,
Friedman
E
, et al
Elevated insulin-like growth factor-I receptor (IGF-IR) levels in primary breast tumors associated with BRCA1 mutations
.
Cancer Lett
2007
;
257
:
236
43
.
45.
Shukla
V
,
Coumoul
X
,
Cao
L
,
Wang
RH
,
Xiao
C
,
Xu
X
, et al
Absence of the full-length breast cancer-associated gene-1 leads to increased expression of insulin-like growth factor signaling axis members
.
Cancer Res
2006
;
66
:
7151
7
.
46.
Nelson
AC
,
Lyons
TR
,
Young
CD
,
Hansen
KC
,
Anderson
SM
,
Holt
JT
. 
AKT regulates BRCA1 stability in response to hormone signaling
.
Mol Cell Endocrinol
2010
;
319
:
129
42
.

Supplementary data