SEARCH

SEARCH BY CITATION

Keywords:

  • BRCA1;
  • BRCA2;
  • breast cancer;
  • ovarian cancer;
  • risk modifiers;
  • risk prediction

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methodological and analytical issues in assessing modifiers of cancer risk in mutation carriers
  5. Genetic modifiers of risk
  6. Genetic modifiers of ovarian cancer risk
  7. Environmental, hormonal and reproductive modifiers of risk
  8. Common alleles and cancer risks for mutation carriers
  9. Future challenges
  10. Conclusions
  11. Acknowledgements
  12. Conflict of interest statement
  13. References

Abstract.  Barnes DR, Antoniou AC (Centre for Cancer Genetic Epidemiology, University of Cambridge, Cambridge, UK). Unravelling modifiers of breast and ovarian cancer risk for BRCA1 and BRCA2 mutation carriers: update on genetic modifiers (Review). J Intern Med 2012; 271: 331–343

Pathogenic mutations in the tumour suppressor genes BRCA1 and BRCA2 confer increased risks for breast and ovarian cancer and account for approximately 15% of the excess familial risk of breast cancer amongst first-degree relatives of patients with breast cancer. There is considerable evidence indicating that these risks vary by other genetic and environmental factors clustering in families. In the past few years, based on the availability of genome-wide association data and samples from large collaborative studies, several common alleles have been found to modify breast or ovarian cancer risk for BRCA1 and BRCA2 mutation carriers. These common alleles explain a small proportion of the genetic variability in breast or ovarian cancer risk for mutation carriers, suggesting more modifiers remain to be identified. We review the so far identified genetic modifiers of breast and ovarian cancer risk and consider the implications for risk prediction. BRCA1 and BRCA2 mutation carriers could be some of the first to benefit from clinical applications of common variants identified through genome-wide association studies. However, to be able to provide more individualized risk estimates, it will be important to understand how the associations vary with different tumour characteristics and their interactions with other genetic and environmental modifiers.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methodological and analytical issues in assessing modifiers of cancer risk in mutation carriers
  5. Genetic modifiers of risk
  6. Genetic modifiers of ovarian cancer risk
  7. Environmental, hormonal and reproductive modifiers of risk
  8. Common alleles and cancer risks for mutation carriers
  9. Future challenges
  10. Conclusions
  11. Acknowledgements
  12. Conflict of interest statement
  13. References

The tumour suppressor genes BRCA1 [1] and BRCA2 [2, 3] have been associated with maintenance of genomic stability and are involved in the homologous recombination pathway for double-strand DNA repair [4, 5]. Germline mutations in BRCA1 and BRCA2 confer increased risks of developing breast and ovarian cancer. Population-based data suggest that approximately one in 800 individuals carry a BRCA1 and one in 500 individuals carry a BRCA2 mutation [6], although mutations are more frequent in certain populations such as the Icelandic and the Ashkenazi Jewish. BRCA1 and BRCA2 mutations are estimated to account for approximately 15% of the excess familial risk of breast cancer amongst first-degree relatives of patients with breast cancer [7–9]. Several studies have estimated the penetrance associated with BRCA1 and BRCA2 mutations. The risk of breast cancer for BRCA1 mutation carriers by age 70 years has been estimated in the range 40–87% and for ovarian cancer 16–68%. The corresponding risks for BRCA2 mutation carriers were estimated to be 40–84% for breast cancer and 11–27% for ovarian cancer [6, 10–18]. Although the observed differences in risk across studies may be, in part, because of random variation or differences in the analytical methods used, risks have been found to vary by several other factors. Risks have been found to increase with more recent birth [10, 19], suggesting that changing patterns in environmental, lifestyle, reproductive and hormonal factors may influence cancer risks in mutation carriers. Penetrance estimates have also been found to vary by the age at disease onset, the cancer site (e.g. breast or ovary or bilateral breast) of the individual that led to the family ascertainment and the method of ascertainment (i.e. through unselected cancer cases or families with multiple affected individuals) [10, 11, 15]. Such observations are consistent with the hypothesis that factors clustering in families, genetic and/or environmental, modify cancer risk in mutation carriers (Fig. 1). Another source of evidence for genetic modifiers comes from investigations into factors that are associated with disease risk in mutation carriers, but themselves are known to be strongly influenced by genetic factors. For example, familial aggregation studies have shown that mammographic density has a strong genetic component [20, 21], but has also been associated with breast cancer risk for BRCA1 and BRCA2 mutation carriers [22]. In addition, mutation position within BRCA1 and BRCA2 has been shown to be associated with different breast and/or ovarian cancer risks [17, 18]. Segregation analysis studies have also quantified the extent of genetic variation in breast cancer risk in mutation carriers [6, 11].

image

Figure 1. Several lines of evidence suggest that genetic, or other familial, factors modify breast and ovarian cancer risks for mutation carriers.

Download figure to PowerPoint

As an adjunct to predictive testing for a high-risk BRCA1 or BRCA2 mutation, more individualized risk estimates that take into account explicit information on genetic and environmental modifiers require identifying and characterizing such modifiers of risk and understanding of how these risk factors interact. These modifiers may then become useful in developing therapeutic interventions for mutation carriers and can provide better understanding of the disease biology in mutation carriers. Identifying modifiers of risk in mutation carriers may also lead to the better understanding of breast and ovarian cancer risk development in the general population and ultimately also benefit nonmutation carriers.

Over the past few years, considerable progress has been made in understanding how genetic and nongenetic factors contribute to the observed variation in cancer risk for BRCA1 and BRCA2 mutation carriers, in particular through the work of large international consortia. In this review, we discuss briefly the key methodological issues in the study of modifiers of cancer risk in mutation carriers and provide an overview of the latest evidence for genetic modifiers of breast and ovarian cancer risk for BRCA1 and BRCA2 mutation carriers. We also discuss the implications for risk prediction in mutation carriers.

Methodological and analytical issues in assessing modifiers of cancer risk in mutation carriers

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methodological and analytical issues in assessing modifiers of cancer risk in mutation carriers
  5. Genetic modifiers of risk
  6. Genetic modifiers of ovarian cancer risk
  7. Environmental, hormonal and reproductive modifiers of risk
  8. Common alleles and cancer risks for mutation carriers
  9. Future challenges
  10. Conclusions
  11. Acknowledgements
  12. Conflict of interest statement
  13. References

The analysis of modifiers of cancer risk for mutation carriers is complicated by numerous factors. BRCA1 and BRCA2 mutations are rare in the population, and standard epidemiological study designs, such as case–control studies, would identify only a small number of mutation carriers. In addition, current cohort studies of mutation carriers have not yet accumulated enough years of follow-up to study associations with modifiers of risk in mutation carriers reliably. Another limiting factor is that a large proportion of BRCA1 and BRCA2 mutation carriers may opt for prophylactic surgery, such as bilateral mastectomy or bilateral prophylactic oophorectomy. Therefore, most studies to date have focused on sampling mutation carriers through genetic clinics. The advantage of this approach is that it allows a large number of affected and unaffected mutation carriers to be identified quickly in a relatively inexpensive manner. However, genetic testing is aimed at young affected individuals with strong family history of disease. As a consequence, ascertainment of mutation carriers is nonrandom with respect to disease phenotype. The nature of this nonrandom ascertainment requires more sophisticated analytical techniques. Whittemore has reviewed in detail possible study designs for obtaining estimates of risk modification in carriers of rare mutations, mainly in the context of environmental modifiers [23]. These include: (i) a case only design (carriers versus noncarriers), (ii) comparison of carrier cases and untyped controls, (iii) comparison of disease risk in exposed and unexposed carriers and (iv) comparison of carrier cases and carrier controls. Some of these designs are susceptible to the weaknesses with respect to available sample sizes described previously.

Alternative methods proposed to deal with the nonrandom sampling of affected and unaffected mutation carriers include a weighted cohort model that assigns sampling weights separately to cases and controls to mimic a true cohort [24] and a retrospective likelihood approach that models the conditional likelihood of the exposure, given the disease phenotype [25]. Simulations have shown that these methods provide estimates that are close to unbiased and have greater statistical power to detect associations compared with other standard methods of analysis (such as Cox regression [26]). Furthermore, for risk factors potentially associated with multiple diseases (e.g. breast and ovarian cancer), extensions of these methods to analyse associations with both diseases simultaneously, in a competing risks framework, have been shown to provide valid tests of association with respect to both diseases, and also provide unbiased estimates of associations for both diseases [26]. The majority of the studies investigating genetic modifiers of cancer risk have used the retrospective likelihood or the weighted cohort approaches to evaluate associations.

Genetic modifiers of risk

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methodological and analytical issues in assessing modifiers of cancer risk in mutation carriers
  5. Genetic modifiers of risk
  6. Genetic modifiers of ovarian cancer risk
  7. Environmental, hormonal and reproductive modifiers of risk
  8. Common alleles and cancer risks for mutation carriers
  9. Future challenges
  10. Conclusions
  11. Acknowledgements
  12. Conflict of interest statement
  13. References

Several small studies have reported associations between common polymorphisms in candidate gene studies and the risk of breast or ovarian cancer for mutation carriers, but the majority of these associations did not replicate in subsequent studies [27–32]. More recent findings are awaiting replication in larger studies [33–36]. The main evidence for genetic modifiers of risk to date comes from studies of the Consortium of Investigators of Modifiers of BRCA1/2 (CIMBA) [37]. With sample sizes of up to 15 000 BRCA1 and 10 000 BRCA2 mutation carriers from around 50 studies, CIMBA has been able to provide strong evidence that several single nucleotide polymorphisms (SNPs) are associated with breast and ovarian cancer risk for mutation carriers. CIMBA studies have focused on studying (i) polymorphisms in candidate genes for which there was evidence of association from smaller studies in mutation carriers or breast cancer studies in the general population; (ii) polymorphisms from genome-wide association studies (GWAS) of breast or ovarian cancer in the general population and (iii) through GWAS using BRCA1 and BRCA2 mutation carriers. Tables 1 and 2 summarize the common genetic variants found to be associated with breast or ovarian cancer risk for BRCA1 and/or BRCA2 mutation carriers using CIMBA data.

Table 1. Loci found to be associated with breast cancer risk for BRCA1 or BRCA2 mutation carriers in analyses by the CIMBA consortium
SNPGene/regionBRCA1BRCA2Reference
Unaffected/affectedMAF (%)HR (95% CI)PVE (%)Unaffected/affectedMAF (%)HR (95% CI)PVE (%)
  1. MAF, Minor Allele Frequency in unaffected mutation carriers; HR, hazard ratio: all HRs are per-allele unless otherwise stated; CI, confidence interval; VE, proportion of the polygenic modifying variance explained [104].

  2. aHR for genotype CC versus GG.

  3. b2 df test.

  4. cHR under dominant model.

  5. dStage 1 and 2 combined.

SNPs identified by candidate gene studies
 rs1801320RAD512902/28768.01.59a (0.96–2.63)0.067 1174/15746.03.18a (1.39–7.27)0.00040.96[25]
 D302HCASP82241/260312.30.85 (0.76–0.97)0.028b0.381061/144811.31.06 (0.88–1.27)0.79b [38]
SNPs identified by population-based GWAS
 rs2981582FGFR23822/444639.01.03 (0.97–1.09)0.31 2160/271638.71.30 (1.20–1.40)6.8 × 10−111.96[41]
 rs3803662TOX3/TNRC93911/449228.01.09 (1.03–1.16)0.00490.242135/267928.01.17 (1.07–1.27)0.000290.61[41]
 rs889312MAP3K14152/440429.00.99 (0.93–1.05)0.63 2282/284029.01.10 (1.01–1.19)0.0220.22[41]
 rs3817198LSP14480/538332.01.05 (0.99–1.11)0.11 2636/326632.81.14 (1.06–1.23)0.000790.46[41]
 rs132816158q244730/549842.51.00 (0.95–1.05)0.93 2723/333841.21.06 (0.98–1.13)0.13 [41]
 rs133870422q354554/538352.21.11c (1.01–1.21)0.0260.142646/330051.21.15c (1.02–1.29)0.0210.19[41]
 rs4973768SLC4A7/NEK104844/543949.01.03 (0.98–1.08)0.26 2783/337049.21.10 (1.03–1.18)0.00640.26[41]
 rs6504950STXBP4/COX114885/551726.91.02 (0.96–1.08)0.59 2813/340126.01.03 (0.95–1.11)0.47 [41]
 rs109416795p124420/527125.30.96 (0.90–1.02)0.16 2591/326323.41.09 (1.01–1.19)0.0320.16[41]
 rs20462106q25.15302/551534.71.17 (1.11–1.23)4.5 × 10−90.882807/338135.91.06 (0.99–1.14)0.09 [39]
 rs93974356q25.16201/63747.11.28 (1.18–1.40)1.3 × 10−80.753313/38048.21.14 (1.01–1.28)0.0310.17[39]
 rs112494331p11.25328/558341.30.97 (0.92–1.02)0.2 2827/342339.71.09 (1.02–1.17)0.0150.21[39]
 rs999737RAD51L14372/448321.00.96 (0.90–1.03)0.27 2632/313622.40.96 (0.88–1.04)0.30 [39]
SNPs identified by BRCA1 or BRCA2 GWAS
 rs817019p134203/416017.01.26 (1.17–1.35)2.3 × 10−91.32N/AN/AN/AN/A [40]
 rs236395619p134199/416052.00.84 (0.80–0.89)5.5 × 10−91.15N/AN/AN/AN/A [40]
 rs16917302ZNF365N/AN/AN/AN/A 2026/216211.00.75 (0.66–0.86)3.8 × 10−5d0.75[42]
Table 2. Loci found to be associated with ovarian cancer risk for BRCA1 or BRCA2 mutation carriers in analyses by the CIMBA consortium
SNPGene/regionUnaffected/affectedMAF (%)HR95% CIPReference
  1. MAF, Minor Allele Frequency in unaffected mutation carriers; HR, hazard ratio. All HRs are per-allele unless otherwise stated; CI, confidence interval.

  2. a2 df test.

rs3814113
 BRCA19p22.28142/188734.20.780.72–0.854.8 × 10−9[43]
 BRCA2 5314/52332.60.780.67–0.905.5 × 10−4 
D302H
 BRCA1CASP84268/57611.90.690.53–0.890.008a[38]
 BRCA2 2360/14912.31.270.73–2.230.718a 

Genetic variants associated with breast cancer risk for BRCA1 mutations carriers

In total, five loci have been found so far to be associated with breast cancer risk for BRCA1 mutation carriers using CIMBA data. A candidate gene study in the general population found that the minor allele of CASP8-D302H was associated with a reduced risk of breast cancer [44, 45]. CASP8 is involved in receptor-induced programmed cell death [46]. The same polymorphism was later studied by CIMBA who also found that the minor allele of this SNP is associated with a 15% reduction in risk per copy of the minor allele [38], a similar magnitude of risk reduction found in the general population.

Four SNPs, at three separate loci, have been found to modify breast cancer risk for BRCA1 mutation carriers by investigating breast cancer susceptibility SNPs identified through GWAS of breast cancer risk in the general population. The strongest evidence of association was for two SNPs (rs2046210 and rs9397435) at 6q25.1 near the ESR1 gene. rs2046210 was identified through a breast cancer GWAS in Chinese women [47] and rs9397435 through fine mapping of the region in women of European ancestry [48], although the latter study suggested that the association in the region could be explained by only rs9397435. CIMBA found that the two SNPs were independently associated with breast cancer for BRCA1 mutation carriers risk [39]. The minor allele of both SNPs was associated with increased breast cancer risk. The SNPs were weakly correlated (r2 = 0.14), and in a model that included both SNPs, none could be rejected in favour of the other, suggesting that either the observed associations are driven by another causative variant that is partially associated with both SNPs or that more than one causative variant is located in this region. Of the remaining 11 known breast cancer susceptibility variants that have been investigated by CIMBA so far, only a TOX3/TNRC9 SNP (rs3803662) and SNP rs13387042 at 2q35 were found to be associated with BRCA1 breast cancer risk [25, 39–41, 49, 50]. There was no evidence that any of the other common breast cancer susceptibility alleles modify BRCA1 breast cancer risk.

A two-stage GWAS in BRCA1 mutation carriers [40] has so far reported on only the approximately 100 most significant SNPs from the first stage of the experiment. Genotyping of these SNPs in a total sample of 8369 BRCA1 mutation carriers (including both stage 1 and replication stage) revealed five SNPs at 19p13 that were associated with breast cancer risk for BRCA1 mutation carriers (P-values: 2.3 × 10−9–3.9 × 10−7) [40]. However, only two SNPs (rs8170 and rs2363956, r2 < 0.23) were independently associated with risk when all SNPs were considered jointly. The minor allele of rs8170 was associated with an increased risk of breast cancer by 25%, whereas the minor allele of rs2363956 amongst affected individuals was estimated to confer a HR of 0.84. The same SNPs were also found to be associated with ER-negative and triple-negative (ER-, PR- and HER2-negative) breast cancer in the general population, which are the predominant tumour subtypes in BRCA1 mutation carriers [51]. These findings confirm that studying modifiers of risk in mutation carriers may identify risk associations with specific disease subtypes in the general population, which otherwise may have been missed in the investigation of associations with overall breast cancer risk. Taken together, the five loci are estimated to account for 3% of the genetic variability of breast cancer risk in BRCA1 mutation carriers.

Genetic variants associated with breast cancer risk for BRCA2 mutations carriers

In total, eleven loci have been found to modify breast cancer risk for BRCA2 mutation carriers. RAD51 is the homolog of bacterial RecA, which is required for recombinational repair of double-strand DNA breaks [52, 53]. BRCA1 and BRCA2 interact with RAD51 [54, 55], providing a plausible candidate modifier of breast cancer risk in BRCA1 and/or BRCA2 mutation carriers. Several small candidate gene studies provided evidence of association between the RAD51 SNP 135G>C (rs1801320) and breast cancer risk for BRCA2 mutation carriers [56–58]. A subsequent CIMBA study found that BRCA2 mutation carriers who also carry two copies of the C allele were at a threefold increased risk of developing the disease [25].

Of the 12 breast cancer susceptibility loci identified through population-based GWAS, which have been investigated by CIMBA, nine provided evidence of association with breast cancer risk for BRCA2 mutation carriers. The strongest evidence of association has been for SNP rs2981582 in FGFR2. Each copy of the minor allele was estimated to confer a 30% increased risk [41]. SNPs in TOX3/TNRC9 and at 2q35, which were associated with BRCA1 breast cancer risk, were also found to be associated with breast cancer risk for BRCA2 mutation carriers [41]. There was some evidence that one of the SNPs at 6q25.1 (rs9397435) also modifies risk for BRCA2 mutation carriers, but the association was somewhat weaker amongst BRCA2 mutation carriers (estimated HR = 1.14 compared with 1.28 for BRCA1), and there was no evidence of association with SNP rs2046210 [39]. In addition to the FGFR2 SNP, another five SNPs have been found to be associated with breast cancer risk for BRCA2 mutation carriers, but not BRCA1 mutation carriers. The minor alleles of SNPs at or near LSP1, MAP3K1, SLC4A7/NEK10, 5p12 and 1p11.2 have been estimated to confer HRs between 1.09 and 1.14 [39, 41]. Although there was no significant evidence of association for SNPs rs13281615 at 8q24 and rs999737 in RAD51L1, the estimated HRs were similar to those reported in the population-based studies [59, 60].

An ongoing GWAS in BRCA2 mutation carriers [42] identified one novel breast cancer susceptibility locus. SNP rs16917302, in ZNF365, was estimated to confer a HR of 0.75 in sample of 4188 BRCA2 mutation carriers. The same locus, albeit a different SNP, was found through a separate GWAS to be associated with breast cancer risk in the general population [61] and subsequently to be associated with mammographic density [62]. As with the BRCA1, the BRCA2 GWAS has so far reported on only the approximately 100 most significant SNPs from the first stage of the experiment. A large-scale replication is expected to report in the near future. The 11 loci are estimated to account for 6% of the genetic variability in breast cancer risk for BRCA2 mutation carriers.

Patterns of association and tumour characteristics

In addition to the SNPs described previously and investigated by CIMBA so far, another five breast cancer susceptibility loci have been identified through GWAS in the general population [61, 63]. These are currently under investigation by CIMBA. Based on the available results, several patterns have emerged when comparing the associations of SNPs between BRCA1 and BRCA2 mutation carriers and the associations in the general population. The associations between these SNPs and breast cancer risk differ substantially between BRCA1 and BRCA2 mutation carriers. The SNPs with the strongest evidence of association with breast cancer risk for BRCA1 mutation carriers (19p13 SNPs, rs2046210 at 6q25.1), as well as CASP8, are not associated with breast cancer risk for BRCA2 mutation carriers. On the other hand, with the exception of the variants in TOX3 and at 2q35, which are associated with both BRCA1 and BRCA2 breast cancer risk, none of the other BRCA2-associated loci (FGFR2, LSP1, MAP3K1, SLC4A7/NEK10, 5p12, 1p11.2) provided significant evidence of association with BRCA1 breast cancer risk, despite the larger sample sizes of BRCA1 carriers studied. Several of these associations were found to be significantly different between BRCA1 and BRCA2 mutation carriers.

BRCA1 and BRCA2 breast cancer tumours have been previously reported to vary in terms of several characteristics. Breast cancers in BRCA1 mutation carriers are more often of the ‘basal’ phenotype [51, 64] and are predominantly ER-negative. It has been estimated that 90% of breast cancers in BRCA1 mutation carriers are ER-negative, compared with 35% in BRCA2 mutation carriers. Studies by the Breast Cancer Association Consortium (BCAC) and others, have demonstrated differences in the associations between these loci and tumour characteristics in the general population [65–68]. The majority of the SNPs indentified through GWAS of breast cancer in the general population to date have been found to have stronger associations with ER-positive tumours than ER-negative tumours. In particular, the estimated ORs for SNPs in FGFR2, TOX3, MAP3K1, 8q24, LSP1, 2q35, 5p12, NEK10/SLC4A7, RAD51L1, 1p11 and ZNF365 were larger for ER-positive breast cancer, than ER-negative breast cancer [60, 65, 69, 70]. A notable exception is 6q25.1 which was found to be associated with larger ORs for ER-negative breast cancer risk than ER-positive [47, 71]. The TOX3 and 2q35 SNPs were also found to be associated with ER-negative breast cancer in the general population, albeit with smaller estimated ORs than those for ER-positive breast cancer [65, 67]. Also, the 19p13 SNPs are associated with only ER-negative breast cancer in the general population [40] and the CASP8 polymorphism has been found to be associated with predominantly PR-negative breast cancer [66]. Therefore, the associations of these variants with breast cancer risk for BRCA2 mutation carriers appear to be similar to those for ER-positive breast cancer in the general population and the associations in BRCA1 mutation carriers more similar to those for ER-negative breast cancer. These patterns suggest that the differences in the associations between the SNPs and breast cancer risk for BRCA1 and BRCA2 mutations could potentially be explained by the differential effects of these SNPs on ER-positive and ER-negative breast cancer and differences in the prevalence of different tumour subtypes in BRCA1 and BRCA2 mutation carriers. Ongoing studies by CIMBA are investigating the associations of these SNPs with different disease subtypes in BRCA1 and BRCA2 and will clarify this hypothesis further.

Genetic modifiers of ovarian cancer risk

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methodological and analytical issues in assessing modifiers of cancer risk in mutation carriers
  5. Genetic modifiers of risk
  6. Genetic modifiers of ovarian cancer risk
  7. Environmental, hormonal and reproductive modifiers of risk
  8. Common alleles and cancer risks for mutation carriers
  9. Future challenges
  10. Conclusions
  11. Acknowledgements
  12. Conflict of interest statement
  13. References

The BRCA1 and BRCA2 GWAS have not yet reported on the associations between common alleles and ovarian cancer risk for BRCA1 and BRCA2 mutation carriers. The focus of studies for ovarian cancer risk modifiers has, therefore, been on SNPs identified through GWAS of ovarian cancer in the general population and through candidate gene studies. The Ovarian Cancer Association Consortium (OCAC) identified five loci through GWAS of ovarian cancer in the general population [72, 73]. The minor allele of SNP rs3814113 at 9p22.2 was associated with a decreased risk of ovarian cancer in the general population (OR = 0.82, 95% CI: 0.79–0.86, P = 5.1 × 10−19) [73]. In a subsequent study by Goode et al. [72], four additional ovarian cancer susceptibility loci were identified in more than 10 000 cases and 17 000 controls: rs2072590 at 2q31 (OR = 1.16, 95% CI: 1.12–1.21), rs2665390 at 3q25 (OR = 1.19, 95% CI: 1.11–1.27), rs10088218 at 8q24 (OR = 0.84; 95% CI: 0.80–0.89) and rs9303542 at 17q21 (OR = 1.11; 95% CI: 1.06–1.16). All associations were stronger for serous ovarian cancer, which is the predominant tumour histological in BRCA1 and BRCA2 mutation carriers [74]. Of these five loci, only SNP rs3814113 at 9p22.2 was investigated for its association with ovarian cancer risk for BRCA1 and BRCA2 mutation carriers. The minor allele of this SNP was found to be associated with a reduced risk of ovarian cancer for both BRCA1 and BRCA2 mutation carriers (HR = 0.78, 95% CI: 0.72–0.85 for BRCA1; HR = 0.78, 95% CI: 0.67–0.90 for BRCA2). The remaining four loci are good candidate modifiers of ovarian cancer risk for BRCA1 and BRCA2 mutation carriers and are currently under investigation by CIMBA.

A parallel GWAS of ovarian cancer prognosis by OCAC found that 2 SNPs at locus 19p13 were also associated with ovarian cancer risk in the general population, with stronger evidence of association for serous ovarian cancer (rs8170, OR = 1.12, 95% CI: 1.07–1.17, serous OR = 1.18, 95% CI: 1.12–1.25 and rs2363956, OR = 1.1, 95% CI: 1.06–1.15, serous OR = 1.16, 95% CI: 1.11–1.21) [75]. These are the same SNPs that provided independent evidence of association with breast cancer risk for BRCA1 mutation carriers, identified through the BRCA1 GWAS [40]. An analysis of the associations of these SNPs with breast and ovarian cancer risk simultaneously for BRCA1 mutation carriers found no significant evidence of association with ovarian cancer risk in BRCA1 mutation carriers [40]. However, the number of ovarian cancer cases included in the analysis was relatively small, and based on the reported confidence intervals, it is not possible at present to rule out an association with ovarian cancer risk that is of similar magnitude to those observed in the general population (e.g. for rs8170, HR for ovarian cancer amongst BRCA1 carriers = 1.07, 95% CI: 0.03–1.24). Ongoing CIMBA studies with a larger number of ovarian cancer cases aim to clarify this further and to assess the associations with ovarian cancer risk for BRCA2 mutation carriers. If locus 19p13 proves to also be associated with ovarian cancer risk for BRCA1 mutation carriers, then it would be the first locus of its kind that modifies both the risk for breast and ovarian cancer risk for BRCA1 mutation carriers.

The CASP8-D302H polymorphism identified through a candidate gene approach also provided some evidence of association with ovarian cancer risk in a sample of 4844 BRCA1 mutation carriers from CIMBA. The minor allele was associated with a 30% decreased ovarian cancer risk for BRCA1 mutation carriers [38]. No significant associations were found between this SNP and ovarian cancer risk for BRCA2 mutation carriers.

Environmental, hormonal and reproductive modifiers of risk

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methodological and analytical issues in assessing modifiers of cancer risk in mutation carriers
  5. Genetic modifiers of risk
  6. Genetic modifiers of ovarian cancer risk
  7. Environmental, hormonal and reproductive modifiers of risk
  8. Common alleles and cancer risks for mutation carriers
  9. Future challenges
  10. Conclusions
  11. Acknowledgements
  12. Conflict of interest statement
  13. References

Several studies assessed environmental/lifestyle, hormonal and reproductive factors as modifiers of risk in BRCA1 and BRCA2 mutation carriers. The large majority of the studies have been retrospective in design, with much smaller sample sizes than those of CIMBA, and reported associations jointly for BRCA1 and BRCA2 mutation carriers because sample sizes were small individually. Therefore, the evidence for the associations of some risk factors with breast or ovarian cancer risk for mutation carriers is still conflicting. A detailed review of the nongenetic risk factors is beyond the scope of this review, but it is worth reviewing briefly some of the risk factors that provided consistent evidence of association with breast cancer risk for mutation carriers across several studies. Briefly, these include parity and/or increasing number of live births, which were found to be associated with decreased breast cancer risk [76–78], oral contraceptive use, where there have been reports for both increased and decreased breast cancer risks associated with oral contraceptive use [79–81], bilateral prophylactic oophorectomy, which has been shown to be consistently protective for breast cancer risk in mutation carriers [82–84], and use of the chemoprevention drug tamoxifen, which has been shown to be protective for contralateral breast cancer in both BRCA1 and BRCA2 mutation carriers [86, 87]. In addition, exposure to chest X-rays has been associated with increased breast cancer risks for BRCA1 and BRCA2 mutation carriers [88]. One study also found that BRCA1 and BRCA2 mutation carriers with high mammographic density (≥50% density) had approximately two-fold increased risk of developing breast cancer compared with those with <50% breast density [22].

Parity and/or number of full-term pregnancies, oral contraceptive use and tubal ligation have been found to be associated with a reduced ovarian cancer risk for BRCA1 and/or BRCA2 mutation carriers in a number of studies [76, 89–91].

Ongoing prospective studies of BRCA1 and BRCA2 mutation carriers [92–95], which address some inherent biases in retrospective study designs such as recall bias and selection bias, aim to clarify the associations with these risk factors and are likely to report in the near future. However, many years of prospective follow-up will be required to obtain precise estimates of the associations.

Common alleles and cancer risks for mutation carriers

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methodological and analytical issues in assessing modifiers of cancer risk in mutation carriers
  5. Genetic modifiers of risk
  6. Genetic modifiers of ovarian cancer risk
  7. Environmental, hormonal and reproductive modifiers of risk
  8. Common alleles and cancer risks for mutation carriers
  9. Future challenges
  10. Conclusions
  11. Acknowledgements
  12. Conflict of interest statement
  13. References

There has been considerable debate on the utility of common alleles identified through GWAS in the prediction of individual genetic risk [96–100]. Most studies to date have focused on the implications for risk prediction in the general population and found that the limited set of SNPs investigated have limited power to discriminate between affected and unaffected individuals [100], mainly because of the small proportion of the familial risks of breast cancer they account for [96]. However, it has been suggested that even a limited set of SNPs may have a role in stratifying individuals into high- and low-risk groups in the context of population screening programmes [98, 101].

The per-allele hazard ratio estimates associated with each modifying polymorphism are modest and vary between 0.84 to 1.28 for BRCA1 and 0.75 to 1.30 for BRCA2 mutation carriers (Table 1). Analysis of the interactions between pairs of nine of the modifying loci identified so far indicated that the combined effects on the breast cancer risk for BRCA1 and BRCA2 mutation carriers were consistent with a multiplicative model [41, 49, 50]. Under this model, and including only the FGFR2, TOX3, MAP3K1, LSP1, 2q35, SLC4A7/NEK10 and 5p12 loci, it was estimated that the hazard ratios for the combined effect of the SNPs on BRCA2 breast cancer risk can vary from one for those who were homozygous for the protective allele at all loci to 5.75 for those who were homozygous for the risk allele at all loci. The hazard ratio at the 5th percentile was 1.3 compared with 3.0 at the 95th percentile of the combined distribution at all loci (with a median hazard ratio of 1.9). However, as BRCA2 mutation carriers are already at elevated risk of developing breast cancer, the combined hazard ratios translate to large differences in the absolute risk of developing breast cancer between genotypes. Based on the seven SNPs above, it was estimated that the 5% of BRCA2 mutation carriers at lowest risk will have a lifetime risk between 42% and 50% of developing breast cancer, compared with 80–96% for the BRCA2 mutation carriers at highest risk [41]. These differences are illustrated in Fig. 2. Similar arguments apply for the risk variation by SNP combined profile in BRCA1 mutation carries. Such differences in risk may potentially be informative for genetic counselling purposes, for classifying mutation carriers into different risk categories. As more risk alleles are identified, which modify breast cancer, the differences in absolute risk are expected to become larger [102]. The absolute risk differences by SNP profile in mutation carriers are much greater than those for women in the general population [102]. In addition, SNP profiles in combination with other risk factors known to modify cancer risk for mutation carries, such as mammographic density for breast cancer [22], oral contraceptive use in the case of ovarian cancer [90, 91] or BRCA2 mutation position in the case of breast and ovarian cancer [17], may lead to even bigger differences in absolute cancer risk. These suggest that mutation carriers may be one of the first population groups for whom clinical applications could be developed for the SNPs identified through GWAS. However, additional research is required to evaluate how the various risk factors interact with the SNPs.

image

Figure 2. Predicted cumulative risk of breast cancer for BRCA2 mutation carriers at the 5th, 50th (median) and 95th percentiles of the combined genotype distribution of SNPs: rs2981582 in FGFR2, rs3803662 in TOX3/TNRC9, rs889312 in MAP3K1, rs3817198 in LSP1, rs13387042 in the 2q35 region, rs4973768 in SLC4A7/NEK10 and rs10941679 in the 5p12 region. Results based on Antoniou et al. [41]. Curves also show the risks at minimum and maximum of the combined genotype distribution.

Download figure to PowerPoint

Future challenges

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methodological and analytical issues in assessing modifiers of cancer risk in mutation carriers
  5. Genetic modifiers of risk
  6. Genetic modifiers of ovarian cancer risk
  7. Environmental, hormonal and reproductive modifiers of risk
  8. Common alleles and cancer risks for mutation carriers
  9. Future challenges
  10. Conclusions
  11. Acknowledgements
  12. Conflict of interest statement
  13. References

Over the past 5 years, there has been substantial progress in our understanding of genetic factors that modify breast and ovarian cancer risk for BRCA1 and BRCA2 mutation carriers. This was made possible to a great extent because of the availability of large numbers of mutation carriers from the CIMBA consortium and GWAS data. However, the five loci described in this review that are associated with breast cancer risk for BRCA1 mutation carriers are estimated to explain only approximately 3% of the genetic variability in breast cancer risk for BRCA1 mutation carriers. Similarly, the 11 SNPs associated with breast cancer risk for BRCA2 mutation carriers are estimated to account for approximately 6% of the genetic variability in breast cancer risk for BRCA2 mutation carriers. Therefore, the majority of the genetic variability in breast cancer risk for mutation carriers still remains unexplained. Several more breast and ovarian cancer susceptibility alleles have been identified through GWAS in the general population, but have not yet been investigated in mutation carriers [61, 63, 72, 75]. Given the observed association patterns in mutation carriers with previously identified loci, it is expected that at least a subset of these will also be associated with breast or ovarian cancer risk for mutation carriers. Additional genetic modifiers of risk may also be identified through not only the ongoing GWAS in BRCA1 and BRCA2 mutation carriers but also other GWAS from the general population or by GWAS focusing on specific cancer subtypes such as oestrogen-receptor-negative or triple-negative breast cancers, or serous ovarian cancer. However, it is likely that several of the alleles identified through population-based GWAS may be associated with modest relative risks in the range of 1.05–1.10. Despite sample sizes of approximately 15 000 BRCA1 and 10 000 BRCA2 mutation carriers, CIMBA would still be underpowered to detect modifying polymorphisms conferring such modest relative risks. Given the rarity of BRCA1 and BRCA2 mutations, increasing sample sizes is currently only possible through increased collaboration between studies and through continued recruitment of mutation carriers.

Population-based studies have demonstrated differences in the associations of common breast cancer susceptibility alleles with ER-positive and ER-negative breast cancer [65, 66]. The association patterns in BRCA1 and BRCA2 mutation carriers follow closely the associations in the general population once stratified by tumour ER status [41]. However, it is yet unclear how the SNPs are associated with ER-positive or ER-negative breast cancer for BRCA1 and BRCA2 mutation carriers. It is plausible, for example, that SNPs that are not associated with the overall breast cancer risk for BRCA1 mutation carriers are still associated with the risk of ER-positive breast cancer for BRCA1 mutation carriers. If proven, such associations could potentially have further implications for risk prediction in mutation carriers. For instance, if a BRCA1 mutation carrier is primarily at risk of developing ER-positive breast cancer (given the SNP profile), as opposed to ER-negative breast cancer, this may influence the choice of screening methods used, chemoprevention or prophylactic surgery.

Studies so far have demonstrated that common risk alleles for breast or ovarian cancer increase cancer risks in mutation carriers to the same relative extent and suggest that common susceptibility loci and BRCA1 and BRCA2 mutations interact multiplicatively on the risks of developing breast or ovarian cancer [103]. It still remains unclear whether there are genetic variants that are associated specifically with breast cancer risk for BRCA1 or BRCA2 mutation carriers and not with risk of breast cancer in the general population. The ongoing GWAS in BRCA1 and BRCA2 mutation carriers should be able to clarify this issue.

Segregation analysis models have estimated the extent of genetic variability in breast cancer risk for BRCA1 and BRCA2 mutation carriers [6]. Taken together with the already observed differences in the risk of developing breast cancer by the known SNP profiles [41], they suggest that as more modifiers of risk are identified, much greater improvements in profiling of mutation carriers can be achieved. However, in the context of genetic counselling, it is important to consider how the associations of the already identified SNPs vary with family history of breast or ovarian cancer. None of the studies have addressed this yet, because of lack of family history information in the currently available data sets. Future studies should aim to address this by taking into account explicit family history information in the estimation of the associated relative risks. These would require the development of novel analytical methods for modelling all factors simultaneously in this context, such as pedigree analyses.

Conclusions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methodological and analytical issues in assessing modifiers of cancer risk in mutation carriers
  5. Genetic modifiers of risk
  6. Genetic modifiers of ovarian cancer risk
  7. Environmental, hormonal and reproductive modifiers of risk
  8. Common alleles and cancer risks for mutation carriers
  9. Future challenges
  10. Conclusions
  11. Acknowledgements
  12. Conflict of interest statement
  13. References

As more cost-effective mutation screening techniques become available, the number of identified BRCA1 and BRCA2 mutation carriers in the population is likely to increase. Therefore, it will be important that all mutation carriers are provided with accurate information on their risk of developing breast and ovarian cancer, so that informed decisions on clinical management are made. Our understanding of factors influencing cancer risk variability in mutation carriers has increased over the last few years and is likely to improve further in the near future. Therefore, we are getting closer to the goal of being able to provide more individualized clinical management. Understanding how cancer risks are modified in BRCA1 and BRCA2 mutation carriers will also provide further insights for studying the biological mechanisms of cancer development in mutation carriers. These may lead to the development of novel therapies and more accurate prediction of breast and ovarian cancer progression in mutation carriers.

Studying genetic modifiers of breast and ovarian cancer risk for BRCA1 and BRCA2 mutation carriers has provided useful insights in study design, analytical methodology and applications, which could be used for studying modifiers of disease in carriers of other high-risk mutations such as the mismatch repair genes MSH2, MLH1, MSH6, PMS2 in colorectal cancer and CDKN2A in melanoma but also other noncancer-related diseases.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methodological and analytical issues in assessing modifiers of cancer risk in mutation carriers
  5. Genetic modifiers of risk
  6. Genetic modifiers of ovarian cancer risk
  7. Environmental, hormonal and reproductive modifiers of risk
  8. Common alleles and cancer risks for mutation carriers
  9. Future challenges
  10. Conclusions
  11. Acknowledgements
  12. Conflict of interest statement
  13. References