Journal of Mammary Gland Biology and Neoplasia, Vol. 9, No. 3, July 2004 ( C 2004)
The Genetic Epidemiology of Breast Cancer Genes
Deborah Thompson1 and Douglas Easton1,2
Genetic susceptibility to breast cancer in women is conferred by a large number of genes,of which six have so far been identified. In the context of multiple-case families, BRCA1and BRCA2 are the most important. Mutations in these genes confer high lifetime risks ofbreast cancer and ovarian cancer, and more moderate risks of prostate cancer and some othercancer types. Mutations in theCHEK2 andATM genes, by contrast, cause much more modest(24 fold) risks of breast cancer. Genes so far identified explain approximately 20% of thefamilial aggregation of breast cancer. The remaining susceptibility genes have, so far, provedillusive, suggesting that they are numerous and confer moderate risks. A variety of techniquesincluding genome-wide association studies, use of quantitative intermediate endpoints, andresequencing of genes may be required to identify them. The identification of such genes canprovide a basis for targeted prevention of breast cancer.
KEYWORDS: breast cancer; BRCA1; BRCA2; ATM; CHEK2; TP53; PTEN.
A womans risk of breast cancer is determinedby both genetic and lifestyle factors. While thevariation in breast cancer between populations islargely explicable in terms of lifestyle factors suchas reproductive patterns and diet, there is sub-stantial variation between individuals that is ge-netically determined. In this paper we review thegenetic epidemiology of the known breast cancersusceptibility genes, concentrating specifically onBRCA1, BRCA2, ATM, and CHEK2. We also dis-cuss the prospects of the identification of furthergenes.
The overall genetic variation in disease riskcan be quantified by the familial aggregation ofthe disease. Overall, breast cancer is approximatelytwice as common in women with an affected first-degree relative; this risk increases with the number
1 Cancer Research U.K. Genetic Epidemiology Unit, Universityof Cambridge, Cambridge, U.K.
2 To whom correspondence should be addressed at Cancer Re-search U.K. Genetic Epidemiology Unit, Strangeways ResearchLaboratory, Worts Causeway, Cambridge, CB1 8RN, U.K.;e-mail: email@example.com.
of affected relatives and is greater for women withrelatives affected at a young age (1). Twin stud-ies have demonstrated a substantially higher riskto monozygotic twins of affected relatives than todizygotic twins, suggesting that most of the famil-ial aggregation is determined by genetic susceptibil-ity rather than lifestyle or environmental risk factors(2,3).
BRCA1 AND BRCA2
In addition to systematic epidemiological evi-dence, certain families display a very high degreeof clustering of early-onset breast cancer cases, con-sistent with the inheritance of a high-risk mutation(e.g., 4). Such multiple-case families provided theimpetus for identification of high-risk susceptibilitygenes, using linkage analysis to identify markers thatcosegregate with the disease. This approach led to
Abbreviations used: AT, aaxia telangiectasia; BCLC, breast can-cer linkage consortium; BIC, Breast Cancer Information Core;BMI, body mass index; ER, estrogen receptor; LFS, Li-FraumeniSyndrome; LOD, logarithm of the odds; OCCR, ovarian cancercluster region.
2211083-3021/04/0700-0221/0 C 2004 Springer Science+Business Media, Inc.
222 Thompson and Easton
the demonstration of linkage to chromosome 17q (5)and to chromosome 13q (6) in subsets of breast can-cer families. Subsequent positional cloning lead tothe identification of the BRCA1 and BRCA2 geneswith mutations in families linked to these regions(79). Many families linked to BRCA1 and BRCA2also contain multiple cases of ovarian cancer, consis-tent with epidemiological observations of significantcoaggregation of breast and ovarian cancer in fami-lies (10,11).
Although neither BRCA1 nor BRCA2 has anapparent close homologue in the human genome,they have several features in common. Both are rea-sonably large genes: BRCA1 has 22 exons, spans ap-proximately 100kb of genomic DNA, and encodes a1863 amino acid protein, while BRCA2 has 27 ex-ons, spans around 70kb, and encodes a protein of3418 amino acids (Fig. 1). Both are ubiquitously ex-pressed in humans, with the highest levels in thetestis, ovaries, and thymus, and both are relativelypoorly conserved between species. Breast and ovar-ian tumours in carriers show frequent loss of thewild-type chromosome 17q or 13q, consistent witha tumour suppressor gene (12,13). The functions ofBRCA1 and BRCA2 are discussed in detail else-where (e.g., 14).
Mutations in BRCA1 and BRCA2
BRCA1 and BRCA2 are the most importantbreast cancer susceptibility genes in high-risk fam-ilies, and identification of mutations in these genesforms an important component of the managementof high-risk women. The Breast Cancer InformationCore (BIC) database had recorded (as of February2004) 1220 distinct germline BRCA1 mutations and1384 BRCA2 mutations. Of these, 697 (57%) and870 (63%) have been reported just once. Mutationsappear to be reasonably evenly distributed acrossthe coding sequences, with no obvious mutationhot-spots. Most mutations found in breast and/orovarian cancer families are predicted to truncate theprotein product. The most common types of muta-tion are small frameshift insertions or deletions, non-sense mutations, or mutations affecting splice sitesresulting in deletion of complete or partial exons orinsertion of intronic sequence. The Breast CancerLinkage Consortium (BCLC) has estimated that ap-proximately 70% of BRCA1 mutations and 90% ofBRCA2 mutations in linked families are of this type(D. Easton, personal communication).
Large-scale rearrangements, including inser-tions, deletions, or duplications of more than 500kb
Fig. 1. The BRCA1 and BRCA2 genes, showing some functional domains, founder mutations, and other features described inthe text.
The Genetic Epidemiology of Breast Cancer Genes 223
of DNA, have also been identified, but as these arenot identifiable by exonic sequencing or other con-ventional screening techniques they are likely to beunderreported. To date there have been reports ofat least 19 distinct large genomic rearrangements inBRCA1 and two in BRCA2, identified using protein-truncation analyses or Southern blots. The majorityare deletions of one or more exons (reviewed in 15).The higher density of Alu repetitive sequences in theBRCA1 gene (42% vs. 20%) (16) is thought to con-tribute to the larger number of large deletions andduplications observed in this gene.
In addition to protein-truncating mutations,large numbers of amino-acid substitutions have beenidentified in both BRCA1 and BRCA2. A small num-ber of these, principally involving cysteine residuesin the BRCA1 RING domain, have occurred con-sistently in high-risk families and are regarded asdisease-associated (missense mutations), but the sta-tus of the majority (termed unclassified variants) isuncertain. Given their frequency and the fact thatmany occur in patients with another, deleterious, mu-tation, it is clear that the large majority of these vari-ants cannot be strongly associated with disease. Atpresent no reliable functional assay exists to deter-mine whether such a variant is likely to be delete-rious, and only the epidemiological evidence on thefrequency of the variant in breast cancer cases andcontrols, and on the cosegregation of the variant withdisease in families, can be regarded as definitive. Un-fortunately this evidence is lacking for most variants.Only two variants outside known functional domainsof BRCA1 are classified as missense mutations byBIC, and for some of these the evidence that they arepathogenic is not totally convincing. No clearly dele-terious missense BRCA2 mutations have yet beendefined.
It has been suggested that common polymor-phisms in BRCA1 and BRCA2 may be associatedwith moderately increased risks of breast or ovar-ian cancer. This hypothesis has been tested by com-paring polymorphism frequencies in cases and con-trols, but there is no consistent evidence that anyof the BRCA1 polymorphisms tested so far con-fers an increased risk of breast cancer (17,18). Onecommon BRCA2 variant, N372H, has been shownto be associated with a moderately increased riskof breast cancer (19,20). Intriguingly, among femalecontrols including newborns, the frequency of ho-mozygotes was significantly lower than that expectedunder Hardy-Weinberg equilibrium, whereas amongnewborn males a deficit of heterozygotes was identi-
fied, suggesting that BRCA2 has different roles in thefoetal development of males and females, leading todifferential selection (19).
Whilst the majority of BRCA1 and BRCA2 mu-tations are infrequently observed, certain mutationsin BRCA1 and BRCA2 have been observed multi-ple times. Haplotype analysis using markers flank-ing the genes has demonstrated that, in most cases,these recurrent mutations are descended from a sin-gle founder. Such mutations tend to be common inspecific populations, and absent elsewhere, indicatingthat most mutations observed now have arisen overthe past few hundred years.
Although BRCA1 and BRCA2 mutations arerare in most populations, mutations can be morefrequent if the population has risen from a relativelyrecent small founder population. The best character-ized examples occur in the Icelandic and the Ashke-nazi Jewish population.
Ashkenazi Jewish Founder Mutations
Three mutations are commonly found inthe Ashkenazi Jewish population: 185delAG and5382insC in BRCA1 (21) and 6174delT in BRCA2(9). Carriers of each mutation share a commonhaplotype. These three mutations account for almostall the BRCA1 and BRCA2 mutations found inthis population, facilitating qui