Antoniou 2006 (other genes).pdf

8/14/2019 Antoniou 2006 (other genes).pdf

http://slidepdf.com/reader/full/antoniou-2006-other-genespdf 1/8

REVIEW

Models of genetic susceptibility to breast cancer

AC Antoniou and DF Easton

Cancer Research UK Genetic Epidemiology Unit, Department of Public Health and Primary Care, Strangeways ResearchLaboratory, University of Cambridge, Cambridge, UK

One of the most important risk factors for breast cancer isfamily history of the disease, indicating that geneticfactors are important determinants of breast cancer risk.A number of breast cancer susceptibility genes have beenidentified, the most important being BRCA1 and BRCA2.However, it is estimated that all the currently known

breast cancer susceptibility genes accounts for less than25% of the familial aggregation of breast cancer. In thispaper, we review the evidence for other breast cancersusceptibility genes arising from twin studies, pedigreeanalysis and studies of phenotypes associated with breastcancer, and the progress towards finding other breastcancer susceptibility genes through linkage and associa-tion studies. Taken together, the available evidenceindicates that susceptibility to breast cancer is mediatedthrough variants in many genes, each conferring amoderate risk of the disease. Such a model of suscept-ibility has implications for both risk prediction and forfuture gene identification studies.Oncogene (2006) 25, 5898–5905. doi:10.1038/sj.onc.1209879

Keywords: polygenic inheritance; genetic model; twinstudies; mathematical model; association; linkage

Introduction

A family history of breast cancer is one of the mostimportant risk factors for the disease. Epidemiologicalstudies estimate that breast cancer is approximatelytwice as common among first-degree relatives of breast

cancer patients, suggesting strongly that genetic factorsare important determinants of disease risk (Collabora-tive Group in Hormonal Factors in Breast Cancer,2001). The two most important breast cancer suscept-ibility genes, BRCA1 and BRCA2, were identified bylinkage analysis and positional cloning in the 1990s(Miki et al ., 1994; Wooster et al ., 1995). Mutations inBRCA1 and BRCA2 are rare, but confer high risks of breast and ovarian cancer and smaller risks for other

cancers (The Breast Cancer Linkage Consortium, 1999;Thompson and Easton, 2002; Antoniou et al ., 2003).Well over 1000 different mutations have been identifiedin BRCA1 and BRCA2, and genetic testing for muta-tions in these genes in high-risk families is now wellestablished (Walsh et al ., 2006).

In addition to BRCA1 and BRCA2, five other genescan be considered well established breast cancersusceptibility genes. Germline mutations in the TP53gene have been implicated as the cause of theLi–Fraumeni syndrome (Malkin et al ., 1990). Breastcancer appears as a feature of this syndrome and carriersof TP53 mutations are at high risk of developing early-onset breast cancer (Garber et al ., 1991). However, theproportion of early-onset breast cancer in the generalpopulation explained by TP53 mutations is small asmutations are very rare (Borresen et al ., 1992; Sidranskyet al ., 1992; Lalloo et al ., 2003). Breast cancer is also afeature of the Cowden syndrome caused by mutations inthe PTEN gene (Nelen et al ., 1996), but such mutations

are also uncommon in the general population. Patientswith Petz–Jeghers syndrome have been found to be atincreased risk of breast cancer (Boardman et al ., 1998).This syndrome is caused by truncating mutations in theLKB1 gene, but its involvement in breast cancer appearsto be only in patients with the syndrome (Hemminkiet al ., 1998; Jenne et al ., 1998; de Jong et al ., 2002).

In addition to these cancer syndromes, two othergenes are associated with more moderate risks of breast cancer. Approximately 0.5% of the populationis estimated to carry a germline mutation in theATM gene (Swift et al ., 1986; FitzGerald et al ., 1997;Ahmed and Rahman, 2006). Studies based on relativesof ataxia-telangiectasia patients, and breast cancer case– control studies, have estimated that the relative riskof breast cancer in heterozygous carriers of ATM mut-ations is approximately 2 (Olsen et al ., 2001; Cavaciutiet al ., 2005; Thompson et al ., 2005b), with someevidence of higher relative risk under the age of 50years. Specific mutations, notably 7271T>G, mayconfer higher breast cancer risks (Chenevix-Trenchet al ., 2002), although the evidence is limited (Bernsteinet al ., 2003; Thompson et al ., 2005a). More recently, atruncating variant in CHEK2, 1100delC, has also beenshown to confer an approximately twofold risk of breastcancer (Meijers-Heijboer et al ., 2002). The variantoccurs with a frequency of 0.5–2% in the European

populations (The CHEK2-Breast Cancer Consortium,

Correspondence: Dr AC Antoniou, Cancer Research UK GeneticEpidemiology Unit, Department of Public Health and Primary Care,Strangeways Research Laboratory, Worts Causeway, Cambridge CB18RN, UK.

E-mail: [email protected]

Oncogene (2006) 25, 5898–5905

& 2006 Nature Publishing Group All rights reserved 0950-9232/06 $30.00

www.nature.com/onc



2004). Whereas no other deleterious variants occur ata significant frequency in the UK, other deleteriousmutations including S428F, CHEK2del5567 and per-haps Il57T have been identified in other populations(Walsh et al ., 2006).

Evidence for other breast cancer susceptibility genes

The clearest evidence for other susceptibility genescomes from studies of BRCA1 and BRCA2 mutationscreening in population-based series of breast cancerpatients. Notwithstanding the fact that BRCA1/2mutations are common in women with a strong familyhistory of the disease, these studies demonstrate thatonly a minority of cases without a family history of thedisease harbour a mutation in either of these genes. Twostudies have estimated that the proportion of thefamilial risk of breast cancer that is accounted for by

BRCA1 and BRCA2 is approximately 15% (Peto et al .,1999; Anglian Breast Cancer Study Group, 2000). Evenallowing for the more minor effects of the other knowngenes, it has been estimated that the known genescan account for no more than 25% of the familialaggregation of breast cancer (Thompson and Easton,2004), indicating that the majority of the familialclustering is still unexplained. Environmental factorsthat cluster in families are unlikely to explain all theresidual familial clustering. Simulation studies haveshown that, even with complete correlation amongrelatives in the exposure to the environmental factor,such risk factors need to confer at least a 10-fold

increase in risk to lead to even modest increases to thefamilial relative risk (Hopper and Carlin, 1992). Amongthe known risk factors for breast cancer, none conferssuch high risks.

Twin studiesTwin studies provide a powerful means for evaluatingwhether the familial clustering of disease is due toenvironmental or hereditary factors. As monozygotic(MZ) twins are genetically identical, whereas dizygotic(DZ) twins share only half their genetic material, geneticsusceptibility should be reflected in a greater diseaseconcordance among MZ than DZ twins (Risch, 2001).The largest such study was conducted by Lichtensteinet al . (2000), who combined data from the Nordic twinand cancer registries. They estimated that the relativerisk of breast cancer in MZ twins of breast cancerpatients is 5.2, compared to 2.8 in DZ twins of breastcancer patients. Similar evidence of higher relative risksamong MZ twins compared to DZ twins was found inother studies (Swerdlow et al ., 1997; Mack et al ., 2002).Swerdlow et al . (1997) found that the differences in therelative risks between MZ and DZ twins of affectedprobands were highest when the first affected twin wasdiagnosed under the age of 35 years. For this age group,the relative risk among MZ twins was estimated to be34.7, although this was based on small numbers. Based

on the known effects of BRCA1 and BRCA2, the

authors estimated that mutations in these genes alonecannot explain this high relative risk among MZ twinsand other high-penetrant genes or a combination of genes must explain the residual risk. Using a multi-factorial model, Lichtenstein et al . (2000) estimated thatgenetic factors account for approximately 27% of breast

cancer phenotypic variance. However, this estimate is of limited value as it is highly dependent on the assumedmodel.

Another twin study that has provided interestinginsights into possible models of susceptibility was con-ducted by Peto and Mack (2000). Their study was aprospective follow-up of MZ twins of breast cancerpatients. In this study, they estimated the breast cancerincidence to be constant and approximately 1.3% peryear, independent of age and duration of follow-up.More recently, Hemminki and Li (2002) have found thisincidence rate to decrease with time since the diagnosisof the first twin, but the numbers in this study were small

and therefore not necessarily inconsistent. Peto andMack also note that this incidence rate is about twice theincidence rate for contralateral breast cancer (which arealso roughly constant with time), consistent with thehypothesis that the high rate of contralateral breastcancer chiefly reflects genetic susceptibility (but withonly one breast being at risk). Such a high incidence inMZ twins is incompatible with a Mendelian modelof inheritance with only two susceptibility categories.This has led the authors to postulate that their resultsare more consistent with the hypothesis that geneticsusceptibility to breast cancer may be the resultof multiple low-penetrance alleles which may coexist inhigh penetrant combinations, that is, a type of polygenic

model (Peto and Mack, 2000; Mack et al ., 2002). Theyalso propose that the high constant risk could reflect amodel in which women reach a high risk of breast cancerat a genetically determined age. Such a model would beradically different from the usual models of carcinogen-esis, in which the risk of disease is driven by a series of rate-limiting events and age itself is relatively unim-portant. However, the constancy of the incidence rate isalso supported by the absence of inflexion in contra-lateral breast cancer risks around menopause, which isusually observed in the general population.

Mathematical modelsPossible genetic models can be assessed by segregationanalysis, in which different statistical models are fitted topedigree data. Several such studies have been per-formed, but most pre-date the identification of BRCA1and BRCA2 (e.g., see Bishop et al ., 1988; Newman et al .,1988; Amos et al ., 1991; Claus et al ., 1991). Many,although not all, found evidence for dominant suscept-ibility to breast cancer. These models provided someimpetus for linkage studies that led to the identificationof the high-risk susceptibility genes BRCA1 andBRCA2. However, these models are now of limitedvalue as they do not consider the effects of the knownsusceptibility genes. More recently, Cui et al . (2001) used

data on families ascertained through a population-based



5899

Oncogene



series of breast cancer patients from Australia toinvestigate the models that best explain the residual,non-BRCA1/2 familial clustering of breast cancer. Theytested all probands for mutations in BRCA1 andBRCA2, and when they excluded the families of mutation carriers from the analysis, the most parsimo-

nious model for the familial clustering of breast cancerwas a mixed model of inheritance, including both arecessive and a polygenic component. The disease alleleat this recessive locus was estimated to have a frequencyof 7%, and homozygote carriers were at a very high riskof developing breast cancer. However, this analysis maystill have included some families with BRCA1 andBRCA2 mutations because of the incomplete sensitivityof the mutation-screening techniques used.

An alternative approach was followed by Antoniouet al . (2002), who used combined data from familiesascertained through population-based studies of breastcancer patients and data from families with multiple-

affected individuals to model the simultaneous effectsof BRCA1, BRCA2 and other genes, while allowingfor the reduced sensitivity of the mutation testing. Thissegregation analysis found that familial aggregationof breast cancer is best explained by a model thatincludes the effects of BRCA1, BRCA2 and a polygeniccomponent. This polygenic model represents the effectsof a large number of genes, each conferring a smalleffect on risk and combining multiplicatively. There wasno significant evidence for an additional major gene.In a recent reanalysis, using additional data frompopulation based studies of breast cancer (Peto et al .,1999; Lalloo et al ., 2003) and mutation positive families(Antoniou et al ., 2003), there was evidence that the

polygenic variance decreased with age (Antoniou et al .,2005; Antoniou and Easton, 2006; manuscript inpreparation). This model would imply that at leastsome of the component polygenes confer higher relativerisks at young ages. This model seems to provide a goodfit to the observed age-specific familial relative risks of breast cancer (Table 1). The polygenic model implies

a log-normal distribution of risk in the general popula-tion (Pharoah et al ., 2002) and would also support thesuggestion of Peto and Mack (2000) that most breastcancers occur among a susceptible minority of women(see ‘Twin studies’ above). This model is the basis for theBOADICEA model of genetic susceptibility to breast

cancer, which can be used for genetic counsellingpurposes (Antoniou et al ., 2004; Antoniou and Easton,2006).

The reasons for the difference between the Cui et al .(2001) and Antoniou et al . (2002) models is unclear, butmight be due to the early age at the onset of the cases(below at 40 years) in the Cui et al . study, perhapsreflecting some specific genetic effects at young age. It isalso important to note that, although the presenceof a polygenic component suggests that the number of susceptibility loci is large, the analyses cannot predictthe precise number, or the risks they confer.

BOADICEA somewhat underpredicts the risks to MZ

twins of cases reported by Peto and Mack (2000) and therisk of contralateral breast cancer (Antoniou et al .,2004). This suggests either some additional geneticcomponent not captured by the Antoniou et al . modelor perhaps that the Peto and Mack twin risks areoverestimates and the high risk of contralateral breastcancer (and perhaps also the MZ twin risk) is partly dueto individual non-genetic factors. However, the predic-tions are closer to the estimates based on data fromthe Swedish Family Cancer Registry (Vaittinen andHemminki, 2000; Antoniou et al ., 2004).

An alternative model to BOADICEA was developedby Tyrer et al . (2004), which assumes the effects of BRCA1, BRCA2 and of a third, dominantly inherited

hypothetical gene. However, this was developed bymodelling the published recurrence risks among first-degree relatives as opposed to family data.

Associated phenotypesA number of risk factors other than family history havebeen implicated in the development of breast cancer.Some of these risk factors will themselves be influencedby genetic factors, and studying these factors mayprovide clues to underlying models of susceptibility andlead to the identification of further susceptibility genes(Hopper and Carlin, 1992). Mammographic density isperhaps the most important such risk factor. Mammo-graphic density is strongly predictive of breast cancerrisk in the general population (Boyd et al ., 1995; Byrneet al ., 1995; Harvey and Bovbjerg, 2004), with relativerisks of four- to six-fold associated with the highestcategory of breast density compared with the lowest.Twin studies have demonstrated that mammographicdensity has a strong genetic component (Boyd et al .,2002; Stone et al ., 2006), and it has also been reportedthat first-degree relatives of women with increasedmammographic density are at higher risk of developingbreast cancer (Ziv et al ., 2003). Such findings provideevidence that breast cancer and mammographicdensity are likely to have a common genetic basis. It

has been estimated that genetic factors that influence

Table 1 Observed and predicted age-specific FRR associated withfamily history in a first-degree relative

Age (years) FRR

BOADICEAa Observed b

25 6.030 5.8 5.735 3.540 2.6

2.045 2.250 1.9

1.655 1.760 1.565 1.4 1.470 1.3

Abbreviations: FRR, familial relative risk. aPredicted FRR assumingan affected mother. Based on the BOADICEA model of geneticsusceptibility with an age-specific polygenic variance (see text).b

Collaborative Group in Hormonal Factors in Breast Cancer (2001).



5900

Oncogene



mammographic density may explain between 5 and 8%of the observed excess risk of breast cancer among first-degree relatives (Boyd et al ., 2005). More recently, it hasalso been demonstrated that breast cancer risk is asso-ciated with density in BRCA1 and BRCA2 carriers(Mitchell et al ., 2006), in line with the findings that high

mammographic density is associated with the risk of developing both oestrogen receptor (ER)-positive andER-negative breast cancer (Ziv et al ., 2004). Mammo-graphic density is not, however, higher in BRCA1 orBRCA2 carriers, indicating that genes influencingdensity act independently of BRCA1/2 mutation status.A number of candidate gene studies for mammographicdensity have been performed, but no genes have beenimplicated with certainty (e.g. Hong et al ., 2004; Mulhallet al ., 2005; van Duijnhoven et al ., 2005). Segregationanalyses of mammographic density have provided someevidence for a major gene component, but have not beenable to distinguish between various models of suscept-

ibility (Pankow et al ., 1997).Another well-established predictor of breast cancerrisk is age at menopause. An earlier age at menopause(both natural and surgical) is associated with a lowerbreast cancer risk (Collaborative Group in HormonalFactors in Breast Cancer, 2001). Estimates for theheritability of age at menopause are in the range31–87% (Snieder et al ., 1998; Treloar et al ., 1998; deBruin et al ., 2001; van Asselt et al ., 2004; Murabitoet al ., 2005). A few candidate gene studies and linkageanalyses have been performed for age at menopause(see, for example, te Velde and Pearson, 2002; Koket al ., 2005), but none have provided concrete evidenceof any susceptibility genes. Other phenotypes with a

genetic component that have also shown associationswith breast cancer risk include body mass index, bonemineral density (Hunter et al ., 2001), chromosomalradiosensitivity (Roberts et al ., 1999), levels of insulin-like growth factor-I and their binding proteins (Canzianet al ., 2005; Fletcher et al ., 2005) and sex hormone levels(Jaquish et al ., 1997; Key et al ., 2002; Dunning et al .,2004).

Modifiers of breast cancer risk in BRCA1/2 carriersAnother important question is whether the breast cancerrisks in BRCA1 and BRCA2 mutation carriers arethemselves influenced by genetic factors. Penetranceestimates based on mutation-positive families ascer-tained through population-based series of breast cancerpatients have generally been lower than estimates basedon families with multiple-affected individuals (Fordet al ., 1998; Antoniou et al ., 2003). Moreover, the riskshave been found to vary by age at diagnosis and the typeof cancer of the index patient. Such observations areconsistent with the hypothesis that the breast cancer riskin carriers is modified by genetic factors. Several breastcancer risk factors, including parity, early oophorect-omy and mammographic density, also influence therisk of breast cancer in carriers (Rebbeck et al ., 2002;Andrieu et al ., 2006; Mitchell et al ., 2006), and a

plausible model is that breast cancer susceptibility alleles

in the general population also confer similar relative riskin carriers. Segregation analysis has demonstrated thatsuch a model fits the observed data well, and suchmodifying effects are built into the BOADICEA model(Antoniou et al ., 2004). However, the possibilityof BRCA1/2-specific modifiers cannot be ruled out.

Conversely, one known risk allele, CHEK2*1100delC,does not appear to confer a breast cancer risk in BRCA1or BRCA2 carriers (or, more precisely, does not confera similar relative risk to that in non-carriers), perhapsreflecting the fact that these genes act in a commonpathway (Meijers-Heijboer et al ., 2002).

Progress in finding other genes

Evidence from linkage studiesTwo main strategies have been used to identify furthersusceptibility genes. The first approach has been to

conduct genetic linkage studies in multiple case breastcancer families that do not segregate BRCA1 or BRCA2mutations. Two genome-wide linkage screens have beenpublished. The first included 14 multiple case familiesfrom Finland. This study found evidence for linkage tomarkers on 2q32. However, a second much larger study,using 149 families negative for BRCA1 and BRCA2mutations analysed by the Breast Cancer LinkageConsortium, failed to confirm this finding (Smithet al ., 2006). The highest LOD score in the latter studywas 1.80 on 4q. Although this peak is suggestive, furtherstudies will be required to confirm or refute whether theregion harbours a susceptibility locus. Overall, the studyfound no excess in the number of LOD scores over 1,over that expected by chance, suggesting that all thepeaks could be spurious. Other studies have suggestedlinkage to chromosomes 8p and 13q12–13 regionsidentified through loss of heterozygosity studies (Ker-angueven et al ., 1995; Seitz et al ., 1997). Neither of theseregions was, however, confirmed by either the BCLCstudy or the Finnish study. Thus, to date, linkageanalyses have failed to find any compelling evidence of linkage to any region. The results suggest that no onegene is responsible for a significant fraction of breastcancer susceptibility, providing further support for thepolygenic model.

Association studiesThe other main approach to identifying susceptibilitygenes has been through case–control association studies.These studies test directly the frequency of putativesusceptibilility variants in breast cancer cases andmatched controls. Such studies do not require high-riskfamilies and have much greater power to detect alleles of moderate effect. Association studies have generallyconcentrated on candidate genes, chosen by virtue of their potential involvement in carcinogenesis – examplesinclude genes involved in carcinogen metabolism, genesin the oestrogen pathway, DNA double-strand breakrepair genes and genes involved in cell apoptosis. Studies

have either examined possible functional variants



5901

Oncogene



(usually single-nucleotide polymorphisms (SNPs)) orhave attempted to study all common variation in thegene using an ‘SNP tagging’ approach (Pharoah et al .,2004). Several positive associations have been reportedbut, to date, none of these has been convincinglyreplicated (Pharoah et al ., 2004). The likely reason for

this is that many association studies are being conductedbut only a small proportion of SNPs are truly associatedwith breast cancer, so that most reported associationsare likely to be false positives. Very stringent significantlevels (Po0.0001 or better) are required to avoid a highfalse-positive rate, but if the true effect sizes are small,most studies will lack the power to detect the associa-tions reliably. In a recent attempt to rectify this problem,the Breast Cancer Association Consortium (BCAC) hasbeen established. This is a collaborative effort toconduct more powerful association studies by combin-ing data from over 20 groups. BCAC has recentlypublished the results on 16 SNPs, which were genotyped

by at least three of the participating groups with samplesizes in excess of 10 000 cases and controls, representingthe majority of all available data on those SNPs (TheBreast Cancer Association Consortium, 2006). Therewas no significant evidence of association with breastcancer for 11 of those SNPs. Five polymorphismsshowed some evidence of association with P-valuesranging between 0.009 and 0.06: rare alleles of CASP8D302H, IGFBP3-202c>a and SOD2 V16A polymor-

phisms were associated with a reduced risk for breastcancer, whereas the rare alleles of PGR V660L andTGFB1 L10P were associated with an increased risk of breast cancer. These are currently being followed up in alarger study.

Future directions

The evidence to date strongly suggests that the majorityof the familial clustering of breast cancer is unexplainedand, therefore, that other breast cancer susceptibilitygenes still remain to be identified (Figure 1). The failureof linkage analyses to find other breast cancer suscept-ibility genes, together with the segregation analyses,suggests that genes with effects similar to BRCA1 andBRCA2 are unlikely to exist, or at least that mutationsin such genes will be very rare. A more plausible modelis that residual genetic susceptibility is driven by variants

at many loci, each conferring a moderate risk of thedisease. The effects of mammographic density andmenopause on risk also appear to be consistent withsuch a polygenic model. Linkage studies, even in largenumbers of families, have limited power to detect suchgenes. However, recent advances in technology haveprovided the opportunity for genome-wide association(GWA) studies. In these studies, several hundredthousand SNPs are genotyped, sufficient to report on

TWIN STUDIES

KNOWN LOW

PENETRANCE GENES

LINKAGE STUDIES

EVIDENCE FOR

POLYGENIC

SUSCEPTIBILITY

ASSOCIATED

SEGREGATION

ANALYSIS

PHENOTYPES

FRR (MZ) > FRR (DZ)

High constant incidence in MZ

Breast cancer risk factors with a

Best fitting models include a

polygenic component

heritable component:

Mammographic density

Age at menopause

Sex hormone levels, etc

No substantial evidence of

linkage to any region in non-

BRCA1/2 families.

Variation in breat cancer risk

by family history.

Risk factors eg mammographic

density in common with the

general population.

CHEK2

BRCA1/2 RISK

MODIFIERS

ATM

twins

.

.

.

.

.. .

.

.

Figure 1 Other breast cancer susceptibility genes still remain to be identified. Current evidence suggests that the residual geneticsusceptibility is likely to be due to variants at many loci, each conferring a moderate risk of the disease. FRR ¼ familial relative risk,

MZ¼

monozygotic, DZ¼

dizygotic.



5902

Oncogene



the majority of common variation in the genome (Pe’eret al ., 2006). Thus, these studies offer the opportunity toidentify most common variants (allele frequenciesX5%) that are substantially associated with breastcancer. Several such studies are now ongoing and resultsare likely in the near future. GWA studies will not,

however, be able to identify rare variants conferringmoderate risks, as these are not well tagged by thecommon tagging SNPs. Instead, resequencing of suita-ble individuals followed by individual testing of variantsin case–control and/or family studies is required. Withcurrent technologies, this is only feasible on a candidategene basis.

The identification of new genes could make a majorimpact in risk prediction. For example, Pharoah et al .(2002) have estimated, on the basis of the polygenicmodel, that half of all breast cancer cases occur in the12% of women at the highest genetic risk. Although it isunrealistic to believe that all the susceptibility alleles will

be found in the near future, even identifying a morelimited number of susceptibility alleles could have

important consequences. This may be particularly truefor women with a family history of the disease, wheredetermining their genetic status could substantially altertheir risk and consequently their management.

A major challenge will be to incorporate the effects of new genes into mathematical models. The polygenic

model, as implemented in BOADICEA, provides aframework for this (Antoniou et al ., 2004). However, todo this it will be necessary to determine not only therisks associated with different variants but also theircombined effects and the interactions with lifestylefactors. Evaluating the combined effects of many genesis a problem that will require both statistical analysis of large association studies and a better biological under-standing of the pathways involved.

Acknowledgements

This work was supported by Cancer Research – UK. DFE is aPrincipal Research Fellow of Cancer Research UK.

References

Ahmed M, Rahman N. (2006). ATM and breast cancersusceptibility. Oncogene 25: 5906–5911.

Amos CI, Goldstein AM, Harris EL. (1991). Familiality of breast cancer and socioeconomic status in blacks. CancerRes 51: 1793–1797.

Andrieu N, Goldgar D, Easton D, Rookus MA, Brohet R,Antoniou AC et al. (2006). Pregnancies, breastfeeding andbreast cancer risk in the International BRCA1/2 carrier

Cohort Study (IBCCS). J Natl Cancer Inst 98: 535–544.Anglian Breast Cancer Study Group (2000). Prevalence andpenetrance of BRCA1 and BRCA2 mutations in a popula-tion-based series of breast cancer cases. Br J Cancer 83:1301–1308.

Antoniou A, Pharoah PD, Narod S, Risch HA, Eyfjord JE,Hopper JL et al. (2003). Average risks of breast and ovariancancer associated with BRCA1 or BRCA2 mutationsdetected in case series unselected for family history:a combined analysis of 22 studies. Am J Hum Genet 72:1117–1130.

Antoniou AC, Durocher F, Smith P, Simard J, Easton DF.(2005). BRCA1 and BRCA2 mutation predictions using theBOADICEA and BRCAPRO models and penetranceestimation in high-risk French-Canadian families. Breast

Cancer Res 8: R3.Antoniou AC, Easton DF. (2006). Risk prediction models for

familial breast cancer. Future Oncol 2: 257–274.Antoniou AC, Pharoah PD, McMullan G, Day NE, Stratton

MR, Peto J et al. (2002). A comprehensive model forfamilial breast cancer incorporating BRCA1, BRCA2 andother genes. Br J Cancer 86: 76–83.

Antoniou AC, Pharoah PP, Smith P, Easton DF. (2004). TheBOADICEA model of genetic susceptibility to breast andovarian cancer. Br J Cancer 91: 1580–1590.

Bernstein JL, Bernstein L, Thompson WD, Lynch CF, MaloneKE, Teitelbaum SL et al. (2003). ATM variants 7271T>Gand IVS10-6T>G among women with unilateral andbilateral breast cancer. Br J Cancer 89: 1513–1516.

Bishop DT, Cannon-Albright L, McLellan T, Gardner EJ,

Skolnick MH. (1988). Segregation and linkage analysis

of nine Utah breast cancer pedigrees. Genet Epidemiol 5:151–169.

Boardman LA, Thibodeau SN, Schaid DJ, Lindor NM,McDonnell SK, Burgart LJ et al. (1998). Increased riskfor cancer in patients with the Peutz–Jeghers syndrome.Ann Intern Med 128: 896–899.

Borresen AL, Andersen TI, Garber J, Barbier-Piraux N,Thorlacius S, Eyfjord J et al. (1992). Screening for germ line

TP53 mutations in breast cancer patients. Cancer Res 52:3234–3236.Boyd NF, Byng JW, Jong RA, Fishell EK, Little LE, Miller

AB et al. (1995). Quantitative classification of mammo-graphic densities and breast cancer risk: results from theCanadian National Breast Screening Study. J Natl CancerInst 87: 670–675.

Boyd NF, Dite GS, Stone J, Gunasekara A, English DR,McCredie MR et al. (2002). Heritability of mammographicdensity, a risk factor for breast cancer. N Engl J Med 347:886–894.

Boyd NF, Rommens JM, Vogt K, Lee V, Hopper JL, YaffeMJ et al. (2005). Mammographic breast density as anintermediate phenotype for breast cancer. Lancet Oncol 6:798–808.

Byrne C, Schairer C, Wolfe J, Parekh N, Salane M, BrintonLA et al. (1995). Mammographic features and breast cancerrisk: effects with time, age, and menopause status. J Natl Cancer Inst 87: 1622–1629.

Canzian F, McKay JD, Cleveland RJ, Dossus L, Biessy C,Boillot C et al. (2005). Genetic variation in the growthhormone synthesis pathway in relation to circulatinginsulin-like growth factor-I, insulin-like growth factorbinding protein-3, and breast cancer risk: results fromthe European prospective investigation into cancer andnutrition study. Cancer Epidemiol Biomarkers Prev 14:2316–2325.

Cavaciuti E, Lauge A, Janin N, Ossian K, Hall J, Stoppa-Lyonnet D et al. (2005). Cancer risk according to type andlocation of ATM mutation in ataxia-telangiectasia families.

Genes Chromosomes Cancer 42: 1–9.



5903

Oncogene

Antoniou 2006 (other genes).pdf

Documents

Transcript of Antoniou 2006 (other genes).pdf