Association of common ATMvariants with familial breast cancer in a South American population

The incidence of breast cancer in Chile is still increasing but the mortality rate in the last decade has remained constant. Early detection contributes to mortality reduction and genetic testing may identify high-risk individuals. Mutations in BRCA1 and BRCA2 genes (BRCA1/2) have been identified as high penetrance alleles. However, these alleles explain only a small fraction of familial breast cancers [1, 2]. We have previously screened for germ line mutations in BRCA1/2, 64 Chilean families with several cases of breast and/or ovarian cancer [3]. Only 15.6% of these presented mutations in one of these two genes. The most widely accepted model proposes that familial breast cancer susceptibility is a consequence of a small number of mutations in BRCA1/2 and a much larger variability in ethnic-specific genes of moderate and/or low penetrance [4]. The Ataxia-Telangiectasia Mutated gene (ATM) has been frequently involved in hereditary breast cancer as a low-penetrance susceptibility gene. The ATM kinase has an essential role in maintaining genomic integrity. It is a key activator of the cellular responses to DNA double-strand breaks [5]. Individuals heterozygous for ATM mutations have been reported to have an increased risk for female breast cancer. A number of studies have searched for germ line ATM mutations in breast cancer cases and/or compared the frequency of common ATM variants among breast cancer cases to population controls, but evidence regarding the role of ATM as a breast cancer susceptibility gene has been contradictory [6, 7]. Recently, large epidemiological and molecular studies have finally provided conclusive evidence that ATM mutations that cause ataxia-telangiectasia are breast cancer susceptibility alleles [6, 8]. There was no evidence that other classes of ATM variants confer a risk of breast cancer [8]. A common ATM variant IVS38-8T>C in cis with the 5557G>A ATM variant, has been suggested to be associated with bilateral breast cancer [9]. The 5557G>A variant has previously been reported in the homozygous state to associate with enhanced clinical radiosensitivity in breast cancer patients [10‐12].

In this study, we screened the entire coding region and exon-intron boundaries of the ATM gene from 137 Chilean familial breast cancer patients. We further evaluated the 5557G>A missense variant and the IVS38-8T>C and the IVS24-9delT polymorphisms, for breast cancer risk in a subgroup of 126 familial breast cancer cases without BRCA1/2 mutations and in 200 healthy controls.

Table 3 shows the sequence variants found in the ATM gene in the 137 patients. We detected two missense mutations and eight intronic polymorphisms. The most frequent variants were: IVS4+36insAA (46%); IVS48-69insATT (50.5%); IVS24-9delT (20.6%) and 5557G>A (20.6%). All the polymorphisms and the aminoacid substitutions detected in the present study have been previously described in different populations such as Caucasians, Africans and Amerindians [20‐22]. Of the detected variants, eight were intronic polymorphisms, the majority of which were located far of the donor or acceptor splice sites. Then, the probability that they represent risk variants is very low. Thus, we considered not necessary to study the frequency of the following variants in controls: IVS4+36insAA, IVS17-56G>A, IV25-12insA, IVS38-15G>C, IVS47-65G>C, and IVS48-69insATT.

In the current investigation, we performed a case-control study between the subgroup of 126 cases BRCA1/2 negative and 200 controls, for the 5557G>A missense variant and for the IVS38-8T>C and IVS24-9delT polymorphisms. Heikkinen et al. [9] proposed a cancer risk-modifying effect for the ATM 5557G>A – IVS38-8 T>C composite allele. The IVS24-9delT was studied because it is located in the acceptor splice site of intron 24 and for its higher frequency in cases. We analyzed the allelic, genotype and composite genotype frequencies for the variants mentioned above and compared them with controls. These variants were in Hardy-Weinberg equilibrium in the studied groups. The frequency of 5557A allele (allele analysis) was 0.10 in BRCA1/2 negative cases versus 0,06 in controls (p = 0,056, OR = 1.67 [95% CI 0.94–2.92]). Nevertheless, the frequency of 5557A allele carriers (genotype analysis) was 0.21 in cases versus 0.13 in controls (p = 0.048, OR = 1.74 [95%CI 0.96–3.16]) (Table 4). Similar results were observed for the allelic and genotype frequencies of the IVS24-9delT polymorphism (Table 4). With respect to IVS38-8T>C polymorphism, the IVS38-8C allele frequency was higher in BRCA1/2 negative cases (0.05) than in healthy controls (0.01) and the difference was significant (p = 0.025, OR = 3.00 [95%CI 1.09–8.21]) (Table 4). Moreover, we observed a higher frequency of IVS38-8C allele carriers in cases versus controls, being the difference statistically significant (p = 0.024, OR = 3.09 [95%CI 1.11–8.59]). Therefore, each of the genotypes may confer an increase in breast cancer risk. Nevertheless the higher significance was observed in the carriers of the IVS38-8T>C. With respect to the 11 BRCA1/2 positive index cases, none of them carried the analyzed variants.

Table 5 shows the composite genotype analysis for IVS24-9delT, IVS38-8T>C and 5557G>A. In accordance with the findings of Heikkinen et al. [9] and Langholz et al. [23], we also observed that IVS38-8T>C carriers were also carriers for 5557G>A in BRCA1/2 negative cases and controls. The IVS24-9 T/(-T), IVS38-8 T/C, 5557 G/A composite genotype showed a higher frequency in cases (8,7%) compared to controls (3.0%) (p = 0.021, OR = 3.19 [95%CI 1.16–8.89]) (Table 5). Thus, this composite genotype confers a 3.19 fold increase in breast cancer risk. In addition, the frequency of T/(-T), T/T, G/A composite genotype did not differ between BRCA1/2 negative cases (11.9%) versus controls (10.0%) (p = 0.289, OR = 1.31 [95%CI 0.63–2.66]). Table 5 also shows that the analyzed variants are highly correlated in relation to the risk for breast cancer. All the IVS38-8T>C heterozygotes were also IVS24-9delT and 5557G>A heterozygotes, both in cases (11/126) and controls (6/200). The difference is given by IVS24-9delT heterozygotes that are also associated to IVS38-8 T/T (15/126 in cases and 20/200 in controls), but not to 5557 G/G (0/126 cases, 0/200 controls), and 5557G>A heterozygotes that are not associated to IVS24-9 T/T (0/126 cases, 0/200 controls), and are associated to IVS38-8 T/T (15/126 cases, 20/200 controls). This fact raises the possibility for these variants to be in strong linkage disequilibrium.

Although phase for alleles at these 3 variants could not be determined directly from the screening data, haplotypes where constructed from genotype data using UNPHASED software, which uses a maximum-likelihood aproach. The total number of haplotypes with the three markers are eight, but in the cases and controls we observed only three which correspond to wildtype (IVS24-9 T – IVS38-8 T – 5557 G); IVS24-9 (-T) – IVS38-8 T – 5557 A; and IVS24-9 (-T) – IVS38-8 C – 5557 A haplotypes. The haplotype estimation suggested a strong linkage disequilibrium between the three markers (coefficient of linkage disequilibrium, D' = 1), a result which had previously been suggested by composite genotype analysis.

Heterozygous individuals for germline ATM mutations have been reported to have an increased risk for malignancy, in particular, female breast cancer. In this study we detected a low ATM diversity in breast cancer women, and this variation was minor in the coding region with respect to noncoding regions. This result is in accordance to Thorterson et al. [22], which established that the nucleotide diversity of ATM in the coding and non coding region have a ratio of 1:7.5, probably due to selective pressure for maintaining the protein sequence. The ATM variations in this study were also minor with respect to those informed by Tapia et al. [24]in Chilean women with hereditary breast cancer. Probably, this may be due to the ethnic origin of the families included. In Tapia et al. study [24], 28% of the families were of European ancestry, probably biasing the results of this study. In our study, all of the families were of Chilean ancestry several generations ago.

Some previous studies have proposed a phenotypic effect for the common ATM missense variation 5557G>A in exon 39 [10‐12]. Heikkinen et al. [9], in their mutation analysis of the ATM gene in patients from 121 Northern Finnish breast or breast-ovarian cancer families reported that the haplotype composed of the alleles 5557A and IVS38-8C was significantly associated with bilateral breast cancer. In a later report, Langholz et al. [23] in the WECARE Study population did not find that the haplotype of 5557G>A and IVS38-8T>C confered an increased risk to women with bilateral breast cancer when compared with women with unilateral breast cancer. Nevertheless, the WECARE Study is population-based and unselected for family history. Tommiska et al. [25] evaluated the 5557G>A and IVS38-8T>C variants in an extensive analysis that included familial breast cancer cases and healthy controls of Southern Finland. In their study neither 5557G>A nor IVS38-8T>C or any haplotype containing these variants, was associated with breast cancer risk or bilateral cancer, but in familial breast cancer cases there was a higher frequency of IVS38-8T>C carriers than in controls (8.1% and 5.6%, respectively). Tapia et al. [24] in a study in Chilean women with hereditary breast cancer found a positive association of 5557G>A (OR = 2.52, p = 0.008), when analyzed alone and in combination with the IVS24-9delT intronic variant (OR = 3.97, p = 0.0003). In our case-control study of the IVS38-8T>C, 5557G>A and IVS24-9delT variants, the frequency of IVS38-8C allele was higher in BRCA1/2 negative cases (0.05) than in controls (0.01) and the difference was significant (p = 0.025, OR = 3.00 [95%CI 1.09–8.21]) (Table 4). The database (dbSNP rs3092829) reported for IVS38-8C allele frequencies of 0.023 in the Hispanic population, 0.016 in Caucasians and 0.01 in Europeans which are in accordance with those of our control group. The frequency of the 5557A allele in controls is also similar to those published in the database (dbSNP rs1801516). Our results with respect to IVS38-8T>C and 5557G>A allele control frequencies differ from those reported by Tapia et al. [24], which selected the control sample from a blood bank of Santiago, Chile. In our study controls were interviewed, age matched with cases, and correspond to individuals with no personal or familial history of breast and other cancer. Cases and controls were from the same ethnic group, and from the same geographic area. The major problem in case-control studies is ensuring a good match between their genetic background, so that any genetic difference between them is related to the disease under study and not to biased sampling [26]. This may be the reason why in the Tapia et al. study [24] controls showed no significant differences with cases in the allele frequencies of these markers.

In our study the carriers of the variant IVS24-9delT, or IVS38-8T>A, or 5557G>A (genotype frequencies) showed an increase in breast cancer risk, although the statistical significance was higher for the variant IVS38-8T>C (OR = 3.09 [95%CI 1.11–8.59] p = 0.024) (Table 4).

We also analyzed the effect of the composite genotypes in breast cancer risk. In accordance with the findings of Heikkinen et al. [9] and Tommiska et al. [25] in the Finnish population, and Langholz et al. [23] in the WECARE sample, we observed that all the carriers of IVS38-8C allele were also carriers of 5557A allele. Then, it is possible that IVS38-8T>C occurs in combination with 5557G>A. We observed, both in cases and controls, only three of the 27 possible composite genotypes, and of these the triheterozygote composite genotype showed a higher frequency in cases (8.7%) than controls (3.0%), being the difference statistically significant (OR = 3.19 [95%CI 1.16–8.89], p = 0.021) conferring an increased risk among women with familial breast cancer. The composite genotype T/(-T), T/T, G/A presented a similar frequency in cases (11.9%) and in controls (10.0%) (Table 5). In agreement with Heikkinen's report [9], our results could indicate that the heterozygous condition for IVS38-8T>C and 5557G>A may be responsible of the association between T/(-T), T/C, G/A composite genotype and an increased risk of familial breast cancer. Nevertheless, we cannot rule out the possibility that T/(-T), T/C, G/A constitutes a risk composite genotype. Another possibility is that this composite genotype in combination with certain genetic background and/or environmental factors, could modify the cancer risk by increasing genetic inestability or by altering the effect of the normal DNA damage response. The present observations need to be confirmed by additional case-control and functional studies.

We observed a strong linkage disequilibrium between the three analyzed markers. Therefore, of the three haplotypes detected in the cases and control samples some of theese may be founder haplotypes, given that the contemporary Chilean population stems from the admixture of Amerindian peoples with the Spanish settlers (European Caucasian). The relationship between ethnicity, Amerindian admixture, genetic markers, and socioeconomic strata has been extensively studied in Chile [27‐29]. Moreover, it has been reported a low prevalence of breast cancer in those regions of the country where the Amerindian population is concentrated and represents demographically a substancial proportion of the people of such regions.

With the exception of the studies performed in Northern and Southern Finnish populations, there are no reports for other European populations were the association between familial breast cancer with the IVS24-9delT, IVS38-8T>C and 5557G>A variants has been analyzed. Therefore, probably the association reported in our study may also exist in other populations with a different ethnic background, such as other European, Asian or Amerindian populations.

With respect to the Amerindian populations a consensus among anthropologists was reached in the sense that Amerindians derived from Asians which, moving from Asia, crossed the Bering Land Bridge [30]. Then, we also raise the hypothesis that the haplotype IVS24-9(-T) – IVS38-8C – 5557A may be a risk haplotype in some contemporary South American populations that stem from the admixture of Amerindian peoples with the Spanish settlers. Up to date, there are no reports of these variants in Asian and Spanish populations.

PRIMER3 software: http://​frodo.​wi.​mit.​edu