Breast cancer risk in BRCA1 and BRCA2 mutation carriers and polyglutamine repeat length in the AIB1 gene
Article first published online: 17 MAY 2005
Copyright © 2005 Wiley-Liss, Inc.
International Journal of Cancer
Volume 117, Issue 2, pages 230–233, 1 November 2005
How to Cite
Hughes, D. J., Ginolhac, S. M., Coupier, I., Barjhoux, L., Gaborieau, V., Bressac-de-Paillerets, B., Chompret, A., Bignon, Y.-J., Uhrhammer, N., Lasset, C., Giraud, S., Sobol, H., Hardouin, A., Berthet, P., Peyrat, J.-P., Fournier, J., Nogues, C., Lidereau, R., Muller, D., Fricker, J.-P., Longy, M., Toulas, C., Guimbaud, R., Yannoukakos, D., Mazoyer, S., Lynch, H. T., Lenoir, G. M., Goldgar, D. E., Stoppa-Lyonnet, D. and Sinilnikova, O. M. (2005), Breast cancer risk in BRCA1 and BRCA2 mutation carriers and polyglutamine repeat length in the AIB1 gene. Int. J. Cancer, 117: 230–233. doi: 10.1002/ijc.21176
- Issue published online: 16 AUG 2005
- Article first published online: 17 MAY 2005
- Manuscript Accepted: 16 FEB 2005
- Manuscript Received: 16 NOV 2004
- Programme Hospitalier de Recherche Clinique AOR01082, Programme Incitatif et Coopératif: Génétique et Biologie de Cancer du Sein, Institut Curie
- U.S. Department of Defense Breast Cancer Research Program. Grant Number: DAMD17-02-1-0421
- Fondation de France
- AIB1 repeats;
- BRCA1/2 mutation carriers;
- breast cancer;
- risk modification
Marked variation in phenotypic expression among BRCA1 and BRCA2 mutation carriers may be partly explained by modifier genes that influence mutation penetrance. Variation in CAG/CAA repeat lengths coding for stretches of glutamines in the C-terminus of the AIB1 protein (amplified in breast cancer 1, a steroid receptor coactivator) has been proposed to modify the breast cancer risk in women carrying germline BRCA1 mutations. We genotyped the AIB1 repeat length polymorphism from the genomic DNA of a group of 851 BRCA1 and 324 BRCA2 female germline mutation carriers to estimate an association with breast cancer risk modification. Hazard ratios (HR) were calculated using a Cox proportional hazards model. For BRCA1 and BRCA2 mutation carriers, analyzed separately and together, we found that women who carried alleles with 28 or more polyglutamine repeats had no increased risk of breast cancer compared to those who carried alleles with fewer repeats (HR for BRCA1/2 carriers = 0.88, 95% CI [confidence interval] = 0.75–1.04). Analyzing average repeat lengths as a continuous variable showed no excess risk of breast cancer (BC) in BRCA1 or BRCA2 mutation carriers (HR for average repeat length in BRCA1/2 carriers = 1.01, 95% CI = 0.92–1.11). These results strongly suggest that contrary to previous studies, there is no significant effect of AIB1 genetic variation on BC risk in BRCA1 mutation carriers and provide an indication that there is also no strong risk modification in BRCA2 carriers. © 2005 Wiley-Liss, Inc.
A high overall lifetime risk of breast and ovarian cancer is conferred by germline alterations of the BRCA1 and BRCA2 genes. However, a great deal of variability in cancer risk estimates has been observed between analyses of high-risk families1, 2, 3, 4 and less selected families or population-based cases.5, 6, 7, 8, 9, 10, 11, 12 Hormonal exposures, mainly associated with a woman's reproductive life, were found to modify breast and ovarian cancer expression in mutation carriers.13, 14, 15 Mutation penetrance in the BRCA genes may also vary depending on the presence of specific alleles at other loci.11, 16, 17 In particular, genes involved in endocrine signaling have been hypothesized as modifiers of breast cancer (BC) risk in carriers of BRCA1/2 mutations.18
Steroid receptor signaling is a complex event that involves multiple interacting cofactors.19, 20 The AIB1 gene located on chromosome 20q12 encodes the AIB1 protein (also known as SRC-3). AIB1 is a steroid hormone receptor coactivator (OMIM 601937) from the SRC1 family of transcriptional coactivators involved in the control of estrogen-dependent transcription.21, 22 Therefore, changes in the expression levels of AIB1 may influence the progression of cancers associated with steroid hormones. In both breast and ovarian cancer, amplification and overexpression of AIB1 have been detected, especially in estrogen receptor (ER)-positive tumours.21, 22 A study by Louie et al.23 showed that AIB1 is required for estrogen-stimulated BC cell proliferation and that AIB1 overexpression renders ER-positive cells completely resistant to anti-estrogens. An isoform of AIB1 with a spliced exon 3 was found to cause a significant increase in the efficacy of 17β-estradiol at both ER-α and ER-β in ovarian, breast and endometrial cancer cell lines.24AIB1 amplification in other cancer cell lines25, 26 and mRNA overexpression in ER- and progesterone receptor (PR)-negative breast tumors20 have also been reported, suggesting that there are other pathways in which AIB1 acts.
Within the carboxyl-terminal region of AIB1 lies a polymorphic stretch of glutamine residues (poly-Q).27 These glutamines are encoded by 2 polymorphic stretches of CAGs separated by a CAA repeat. Although the biologic function of these repeats is unclear, the analogous region of SRC1 directly interacts with the androgen receptor (AR) to enhance its signaling,28 and the AR CAG repeat length has a direct bearing on the function of the encoded protein.29 In a study of women with BRCA1 mutations,18 long alleles of the AR poly-CAA/CAG repeat showed an association with earlier age of BC diagnosis (HR 1.8). These data suggested to Rebbeck et al.30 that the poly-Q region of AIB1 may also functionally affect signaling of steroid hormones and thus modify BC risk. They typed the glutamine repeats in a matched case-control sample of 448 women with BRCA1/2 mutations (278 BC cases and 170 controls). They found that women carrying alleles with at least 28 or 29 poly-Q repeats were at a higher BC risk compared to women with shorter alleles (OR = 1.59, 95% CI 1.03–2.47 and OR = 2.85, 95% CI 1.64– 4.96, respectively). A significant association of these repeats with BC risk in carriers of BRCA1 mutations (RR = 1.25) was also reported in a study of 310 BRCA1/2 carrier women.31 However, there appears to be no association between poly-Q length and BC development in the general population32 nor in familial cases.33
Our aim for this study was to examine whether the AIB1 poly-Q genotype is a modifier of BC risk in our cohort of 1,175 BRCA1/2 mutation carriers, as suggested by Rebbeck et al.30 and Kadouri et al.31 Due to the relatively small sample sizes, these 2 earlier studies were only able to reliably examine the effect of AIB1 genotype on BC risk in women with BRCA1 mutations but not those with BRCA2 mutations. Our large BRCA1 mutation carrier cohort (n = 851) allows us to clarify whether the AIB1 gene is a genetic risk factor for BC in BRCA1 mutation carriers. Our number of BRCA2 carriers, though still modest (n = 324), was approximately triple the number analyzed in each of the previous studies, enabling a preliminary study on the effect of AIB1 genotype on BC risk in women with BRCA2 mutations.
Material and methods
All cases in our study were included after giving informed consent. Our study includes 1,175 women belonging to 678 families recruited and identified as carriers of BRCA1 (n = 851) or BRCA2 (n = 324) germline mutations in the framework of research and counseling programs on hereditary breast and ovarian cancer in France (n = 865), Greece (n = 32) and USA (n = 278). Of these 1,175 women, 642 have been diagnosed with breast cancer, 84 with breast-ovarian cancer and 449 were breast cancer-free at the time of last follow-up. The mean age of diagnosis was 41 years for a first breast cancer (20–75 years) and the mean age at death or last follow-up exam of the unaffected carriers was 44 years (16–100). The median values were 40 and 43 years, respectively. Information available on study subjects included clinical characteristics, date of birth, age at last follow-up exam or age at death, age at diagnosis of breast and/or ovarian cancer, age at prophylactic surgery (oophorectomy or mastectomy) and parity.
Genomic DNA was extracted by standard methods, and 20 ng was used as template for the PCRs. The AIB1 polyglutamine tract at positions 1244–1272 of the 1420 amino-acid protein is encoded by (CAG)n CAA (CAG)n (CAA CAG)4 CAG CAA (CAG)2 CAA (GenBank accession no. AF012108). The total repeat length is usually 29, as the 2 (CAG)n regions usually have 6 and 9 repeats, between nucleotides 3930–3947 and 3951–3977, respectively.27 The primers used were those described by Rebbeck et al.:30 forward, 5′-AGT CAC ATT AGG AGG TGG GC-3′; and reverse, 5′-TTC CGA CAA CAG AGG GTG G-3′. The PCR reactions were performed with 0.02 MBq (0.5 μCi) of P33-dATP. The PCR products were run on 6% polyacrylamide electrophoresis gels for 2 hr at 75 watts. The gels were prerun for 20 min with 1 ul of loading dye in alternate lanes to check for any leaking wells. Gels were then dried prior to overnight autoradiography with Biomax MR1 film (Kodak, Rochester, NY). Samples were amplified in 96-well PCR plates with 2 water controls and 2 reactions from the same CEPH sample (1347.02). Several samples were sequenced on an ABI 3100 sequencer using Big Dye™ deoxy terminators (Perkin-Elmer, Courtaboeuf, France) to provide sizing controls for each gel. Genotype calls were made by 2 persons blinded to the status of the samples and scored according to Rebbeck et al.30 and Haiman et al.32
To facilitate comparisons between studies, genotypes were grouped for analysis based on the length of the AIB1 polyglutamine repeats according to Rebbeck et al.30 and Kadouri et al.31 The data were analyzed as a retrospective cohort study by disease-free survival analysis in a Cox proportional hazards model using the STATA statistical analysis package (STATA Corporation, College Station, TX). For the estimation of cancer risk, the women were followed until the first diagnosis of breast cancer (726 women) and were censored at age of bilateral prophylactic mastectomy or oophorectomy (20 women), other cancer diagnosis (including ovarian, 153 women) or last follow-up exam or death (276 women). As the data set included multiple members of the same families, all analyses were performed using robust variance estimators to account for familial correlations in the AIB1 genotypes.34
Genotype and allele distributions of the poly-Q repeats are comparable to those previously reported in other populations.27, 30, 31, 32 The number of glutamine codons ranged from 22–35, and the most common alleles contained 26 (13%), 28 (40%) and 29 (46%) repeats. Other allele frequencies were less than 1%. The most prevalent genotypes were 28/28 (18%), 28/29 (36%) and 29/29 (22%).
To compare our results with those obtained by Rebbeck et al.,30 we grouped our samples into the same genotype classes of AIB1 polyglutamine repeats. They reported a significant association with BC risk in women carrying 28 or 29 or more repeats compared to all those with shorter repeat lengths. However, as shown in Table I, we found no evidence of such a link for BRCA1 mutation carriers alone or in the overall sample. In BRCA1 carriers, our estimated HR was 1.02 (95% CI 0.83–1.26) for individuals with both AIB1 alleles containing 28 or more repeats compared to those with genotypes containing at least 1 allele with 27 or fewer repeats. The HR was 1.02 (0.81–1.28) for women with 29 or more repeats compared to those carrying at least 1 allele with 28 or fewer repeats. The corresponding values for BRCA2 carriers were 0.67 (0.51–0.88, p = 0.005) and 1.16 (0.86–1.57), respectively, suggesting a small reduction in BC risk for BRCA2 carriers who harbor longer poly-Q repeats (28 or more repeats) on both copies of their AIB1 gene, although the carriers with the longest alleles (29 or more repeats) did not show any effect on risk.
|AIB1 variants||BRCA1 carriers (n = 851)HR (95% CI)||BRCA2 carriers (n = 324)HR (95% CI)||BRCA1/2 carriers (n = 1175)HR (95% CI)|
|AIB1 polyglutamine repeat length genotype groups1|
|≥28||1.02 (0.83–1.26)||0.67 (0.51–0.88) p=0.005||0.88 (0.75–1.04)|
|≥29||1.02 (0.81–1.28)||1.16 (0.86–1.57)||1.06 (0.88–1.27)|
|AIB1 polyglutamine repeat length as a continuous covariate3|
|Average repeat length||1.07 (0.95–1.20)||0.90 (0.73–1.09)||1.01 (0.92–1.11)|
|Shorter allele||1.02 (0.94–1.11)||0.93 (0.82–1.06)||0.99 (0.92–1.06)|
|Longer allele||1.10 (0.97–1.24)||0.94 (0.69–1.28)||1.06 (0.94–1.19)|
To compare our results with those obtained by Kadouri et al.,31 we analyzed, as a continuous variable, average, minimum, and maximum repeat lengths. This also showed no modification of BC risk in BRCA1 or BRCA2 carriers (Table I) as the HR by average poly-Q repeat length as a continuous variable was 1.07 (95% CI 0.95–1.20) and 0.90 (0.73–1.09) per repeat unit for BRCA1 and BRCA2 carriers, respectively. For each analysis adjustment for year of birth and parity, known modifiers of BC risk17 or stratifying by European and North American populations did not significantly alter the hazard ratio estimates.
The studies of Rebbeck et al.30 and Kadouri et al.31 suggested that the length of a poly-Q coding repeat in the AIB1 gene modifies breast cancer risk in BRCA1/2 mutation carriers. Rebbeck et al.30 reported an increased BC risk of 1.59- and 2.85-fold in women with at least 28 or 29 repeat lengths, respectively, in BRCA1/2 carriers. They did not analyze BRCA2 carriers separately because they had only 78 of them in their cohort of 448 women. Kadouri et al.31 estimated the relative risk of BC in BRCA1 carriers to be 1.25 per glutamine repeat unit when considering the average poly-Q length as a continuous variable. However, both studies suffered from a relatively small number of mutation carriers (370 BRCA1 carriers, 78 BRCA2 carriers for Rebbeck et al.30 and 222 BRCA1, 88 BRCA2 for Kadouri et al.31), thus they could only provide an indication of the effect of AIB1 genotype on BC risk in women with BRCA1 mutations but not those with BRCA2 mutations. We set out to examine this proposed association in our cohort of 851 BRCA1 and 324 BRCA2 carriers and found no evidence to support any of these proposed modifications of BC risk as a function of AIB1 poly-Q lengths, for BRCA1 carriers alone or in combination with BRCA2 carriers (Table I). Our sample size enables us to conclude that there is no influence of AIB1 repeat length on BC risk in BRCA1 mutation carriers and suggests that there is also no striking risk modification in BRCA2 mutation carriers.
Based on our analysis of 1,175 carriers, we can exclude with 95% confidence an effect on BC risk modification greater than 1.04 for individuals having both alleles of at least 28 repeat units. This is in contrast to the point estimate of 1.59 reported by Rebbeck et al.,30 although it is difficult to precisely compare these estimates given the difference in study design. Similarly, their risk estimate for 29+ repeats is much higher than the upper confidence limit in our study (2.85 vs. 1.27). Treating the repeat length as a continuous variable, the relative risk of Kadouri et al.31 for BRCA1 carriers (RR = 1.25) falls outside our confidence limit in BRCA1 carriers alone, although the confidence intervals overlap (1.09–1.42 vs. 0.95–1.20). Here we can better make a direct comparison since a similar design and analysis was used in the 2 studies.
Analysis of our data according to the genotype classes used by Rebbeck et al.30 showed a reduction in BC risk for BRCA2 carriers when we compared all the BRCA2 carriers who harbored 2 alleles of 28 or more repeats against all those with genotypes containing at least 1 allele with 27 or fewer repeats. However, those BRCA2 carriers with the longest repeats (≥29) did not show any increased BC risk compared to individuals carrying at least 1 allele with 28 or fewer repeats. In addition, no increased risk for BRCA2 carriers was seen for any of the analyses where we treated the AIB1 poly-Q repeat length as a continuous covariate. Although the number of the BRCA2 mutation carriers in our study (324) was around 2-fold that of the previous 2 studies combined, we feel that our results would need to be clarified in a larger series.
Steroid hormones via their receptors regulate the expression of proteins involved in the development and proliferation of breast tissue. Steroid hormone receptors, including those for estrogen, progesterone and androgen, as mediators of hormone action, and their coactivators, such as AIB1, are important in breast carcinogenesis. The genes encoding these proteins are prime candidates as genetic modifiers of BC susceptibility. However, because the majority of BRCA1-associated tumors are ER negative,35 it is perhaps not surprising that AIB1 variants do not alter the risk of BC in BRCA1 mutation carriers. The AIB1 poly-Q variants also do not seem to act as low penetrance BC susceptibility alleles in general as there appears to be no association between repeat length and BC development in postmenopausal women (464 cases, 624 controls)32 or in familial BC cases (591 cases and 536 controls).33 The length of the CAG/CAA tract in AIB1 has been associated with prostate cancer progression in a population-based case-control study in China, and risk was increased in combination with short AR CAG repeats.36 Further studies are warranted to investigate any involvement of AIB1 alleles in hormonal-related cancers and to establish if the variation in AIB1 glutamine repeats has any functional significance.
In summary, we show here in a large series of carriers that there is no significant effect of the AIB1 poly-Q variants on BC risk modification in BRCA1 mutation carriers in contrast to the results of Rebbeck et al.30 and Kadouri et al.31 We conclude that these previously reported risk associations were probably due to the small sample sizes tested. We also report here no striking association of BC risk with AIB1 poly-Q repeats in BRCA2 mutation carriers but caution that this requires testing in a large BRCA2 study.
We thank Ms. C. Bonnardel for assistance in data collection and database management and Ms. M. Corbex for her help with statistical analysis. D.H. is funded by the U.S. Department of Defense Breast Cancer Research Program (DAMD17-02-1-0421). S.G. was the recipient of a fellowship from the Fondation de France.