SUBJECTS AND METHODS

Subjects

A total of 215 breast cancer patients from 162 families originating from central and northern Finland were chosen for ATM mutation screening. All families, except one in which two members tested positive for a newly discovered BRCA2 mutation (5797G>T, exon 11) after the ATM study had been performed, had previously been excluded for BRCA1, BRCA2, and TP53 mutations.^{3–5, 7} Inclusion criteria for the families with moderate to high genetic susceptibility to breast cancer were one or more of the following: (1) at least three (two in combination with other selection criteria) cases of breast cancer in first or second degree relatives, (2) early disease onset (≤35 years alone, or <45 in combination with other inclusion criteria), (3) bilateral breast cancer, or (4) multiple tumours including breast cancer in the same person. Additionally, 85 sporadic breast cancer cases from the Oulu area were included. All patients provided informed consent for obtaining pedigree data and blood specimens for a study on cancer susceptibility gene mutations. Reference blood samples from 200 healthy, geographically matched controls were used to validate the observations from the two test groups. Approval to perform the study was obtained from the Ethical Board of the Northern Ostrobotnia Health Care District and the Finnish Ministry of Social Affairs and Health.

Methods

DNA extraction from blood lymphocyte specimens was performed using the standard phenol-chloroform method. The screening of eight distinct ATM germline mutations (table 1) previously detected in Finnish AT families (partly unpublished results)¹³ was mainly done by conformation sensitive gel electrophoresis (CSGE).⁵ Simultaneously, the complete exons and exon-intron boundary regions at the sites of the known mutations were evaluated for other possible aberrations. Samples with a band shift were reamplified and purified with the QIAquick PCR purification Kit (Qiagen). Sequencing analysis was performed with the Li-Cor IR² 4200-S DNA Analysis system (Li-Cor Inc, Lincoln, USA) and using the SequiTherm EXCEL^TM II DNA Sequencing Kit-LC (Epicentre Technologies). Oligonucleotides for exons 14, 37, 62, and 65 were as reported by Shayegi et al.²⁰ For exons 4, 5, 48, 49, and 53, the oligonucleotides were designed by using the Primer3 software. Primer sequences and PCR conditions for CSGE and sequencing are available upon request.

RESULTS

Identification of pathogenic ATM variants

All detected germline alterations are summarised in tables 2 and 3. Two of the eight ATM mutations previously identified in Finnish AT families were also detected in the breast cancer patients studied. 7522G>C (exon 53, Ala2524Pro) was originally observed in two AT families, FIAT6 and FIAT7, and was now identified in two women with breast cancer, belonging to separate cancer families (table 2, fig 1, families 003 and 005). The woman in family 003 had breast cancer at the age of 50. Her sister had breast cancer at the age of 59, but owing to lack of DNA we were unable to determine her carrier status. However, of her four children, the two sons carried the 7522G>C aberration, both cancer free at ages 49 and 54. Her two daughters, on the other hand, tested negative for the mutation, even though they had breast and thyroid cancer at the ages of 51 and 47, respectively. Interestingly, there were also multiple cases of cancer on the other side of the family. In family 005, an identified mutation carrier was diagnosed with both breast and stomach cancer at 57 and 41, respectively. In her relatives, other observed cancers included a basal cell carcinoma in the proband's uncle, several cases of stomach cancer (in the maternal grandmother, paternal uncle, and cousin), as well as an unknown cancer in the paternal aunt and thyroid cancer in a cousin.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Families with breast cancer exhibiting (A) 7522G>C and (B) 6903insA ATM germline mutations. Filled/open symbols indicate cancer/non-cancer status. Age at diagnosis, when known, is shown in brackets after the cancer type (Br, breast; Bs, basal cell; Bt, brain; Csu, cancer site unknown; Pan, pancreas; Pro, prostate; Pul, pulmonary; Sar, sarcoma; Sto, stomach; Th, thyroid). Subjects tested for a specific mutation are marked + if positive and − if negative. In addition, in a subsequent BRCA2 study of a previously unscreened branch of family 004, two subjects marked with an asterisk tested positive for the 5797G>T mutation. Therefore, other available members of this family were also evaluated but were all found to be negative for this mutation.

In breast cancer family 004 (table 2, fig 1), an alteration in exon 49 was observed. The insertion of adenine (6903insA) results in a loss of 685 amino acids from the carboxy terminus of the ATM protein. The 6903insA mutation had previously been detected in two AT families, FIAT4 and FIAT5. Interestingly, this alteration was now found in three sisters, who all had breast cancer at ages 40, 47, and 50. Further investigation of family 004 showed ambiguity in the genotype-phenotype association (fig 1), as the probands' cousin who had been diagnosed with breast cancer aged 54 did not carry the mutation. Her sister, however, who was unaffected at the age of 49, was a carrier. Also, two additional breast cancer patients (diagnosed at ages 37 and 45) from another branch of this cancer family tested negative for 6903insA. Interestingly, however, after completion of our ATM mutation screening, these two cancer cases tested positive for a newly discovered BRCA2 mutation (5797G>T, exon 11), which probably explains their hereditary susceptibility to breast cancer. We then tested all the other available family members for this particular BRCA2 mutation, but no additional carriers were found. At present, it appears unlikely that the BRCA2 mutation positive branch of family 004 would segregate the 6903insA ATM mutation.

Neither 6903insA nor 7522G>C was seen in the sporadic breast cancer cases studied or in the healthy controls, indicating that in addition to being instrumental to AT, these two ATM mutations could also be related to hereditary predisposition to breast cancer.

Polymorphic variants

We also found three other germline sequence variants, two in exon 5 (133C>T and 146C>G) and one in the intronic region between exons 62 and 63. All these changes were regarded as polymorphisms (table 3). 133C>T leads to an amino acid substitution Arg45Trp and was observed in a woman with bilateral breast cancer at the age of 45. The second exon 5 variant, 146C>G, results in Ser49Cys and was detected in a woman diagnosed with breast cancer at 60. She had two sisters, both with breast cancer at ages 48 and 58, but neither of them carried the alteration. We did not observe any of these alterations in the sporadic breast cancer group or healthy controls. Furthermore, neither Arg45Trp nor Ser49Cys resides in a known functional ATM domain, and are thus less likely to interfere with ATM kinase function. The third sequence alteration, IVS62+8A>C, was detected in 5/215 (2.3%) of the familial breast cancer patients, 2/85 (2.4%) of the sporadic breast cancer cases, and 4/200 (2.0%) of the control samples.

DISCUSSION

We have examined the prevalence of eight different ATM germline mutations, originally found in Finnish AT families, in a cohort of 215 breast cancer cases from 162 cancer families, as well as in 85 sporadic breast cancer patients. Altogether, we detected five different ATM sequence alterations (tables 2 and 3), two of which (7522G>C and 6903insA) potentially relate to breast cancer susceptibility. The 7522G>C mutation was previously found in two AT families also displaying signs of genetic predisposition to breast cancer (see below). The 6903insA mutation present in two known AT families (see below) may also be associated with susceptibility to cancer, as this mutation leads to a truncated ATM protein with loss of 685 amino acids, including the region containing the PI-3 kinase motif.²¹ The remaining three variants (133C>T, 146C>G, and IVS62+8A>C) were classified as polymorphisms. This assumption was based both on their location away from essential functional domains, and for 146C>G also on previous studies by Vorechovsky et al²² and Izatt et al,²³ who did not observe loss of heterozygosity (LOH) of the relevant chromosome region in tumours carrying this alteration. Both groups also identified 146C>G among healthy controls.

Both of the two Finnish AT families (FIAT6 and FIAT7) showing 7522G>C originated from northern Finland. In family FIAT7, the AT patient carried two different mutant ATM alleles, 7522G>C and a 2 bp deletion in exon 48. The patient's mother, who carried the 7522G>C mutation, had been diagnosed with childhood leukaemia at the age of 4, followed by bilateral breast cancer at the age of 37. Tumour cells from her left breast showed LOH of the wild type allele, as determined by using the intragenic ATM marker D11S2179 (data not shown). Also, the AT patient's grandmother had died from breast cancer at the age of 52. In family FIAT6, the AT patient was a missense 7522G>C homozygote. It is known that at least the proband's maternal grandmother (aged 61) and also one of her sisters had breast cancer. So far, only a few missense homozygotes have been identified^{9, 13} and it has been suggested that these patients would express a milder AT phenotype. However, no clear differences were observed in the phenotypes of these two AT patients from families FIAT6 and FIAT7. The appearance of 7522G>C in two AT families from northern Finland suggested that this mutation could also be more frequent in breast cancer patients originating from the larger Oulu region. Indeed, the mutation was found in two families with breast cancer (table 2, families 003 and 005).

Besides 7522G>C, ATM 6903insA was the other putative breast cancer related alteration seen in this study. The geographical origin of the 6903insA allele, first identified from AT families FIAT4 and FIAT5, appeared to be a rural area slightly west of the city of Tampere. Both AT families showed excessive cases of breast cancer in the branches segregating the 6903insA allele, and at relatively young ages (FIAT4, the proband's paternal grandmother (aged 53); FIAT5, the proband's mother (aged 50) and maternal grandmother's sister (aged 51)). Interestingly, the branch of breast cancer family 004 displaying the 6903insA mutation also came from west of Tampere. Neither 7522G>C nor 6903insA was detected among 200 controls originating from the same geographical region, suggesting that these two mutations could have some effects on the carrier phenotype, even when heterozygous.

In families 003 and 004, where we were able to study additional family members, some discrepancies regarding cancer phenotype and carrier status were seen (fig 1). Even though this observation raises questions about the contribution of these mutations to breast cancer susceptibility, incomplete penetrance and occurrence of sporadic cancer cases must also be taken into account. Furthermore, the dominant inheritance model of cancer predisposition cannot automatically be assumed to apply to these mutations.

The large size of the ATM gene sets certain limitations to mutation screening. Most studies have used the protein truncation test, which only detects alterations resulting in a premature stop codon and therefore a shortened protein product. Missense mutations, however, contributing to the other of the two putative phenotypically distinct populations of AT carriers, are not detected with this method.^{24, 25} Also, if DNA is the only starting template available, as was the case in this study, it is quite impossible to cost effectively screen large sample sets for missense mutations. Therefore, and also because of the anticipation of seeing breast cancer related geographically recurrent mutations, we concentrated on studying only the ATM germline mutations previously recognised from all known Finnish AT patients (partly unpublished data)¹³ and used CSGE as the primary screening method because of its simplicity and high accuracy followed by DNA sequencing, if required.

ATM is one of the central components of the DNA damage response, and it interacts with several key proteins such as p53, Mdm2, BRCA1, Chk2, Nbs1, and Rad17.^{26, 27} Viewed against this background and considering the existing epidemiological evidence,¹⁰ the fact remains that ATM is a strong candidate contributing to cancer susceptibility. The low frequency of mutations within the protein encoding region in the breast cancer material analysed, however, raises the question of whether epigenetic mechanisms (for example, promoter region hypermethylation) could play an additional role in ATM silencing. These studies are currently under way in our laboratory.

We have analysed both familial and sporadic breast cancer patients for eight ATM germline mutations previously identified from Finnish AT families (partly unpublished results)¹³ and found a pathogenic mutation present in 3/162 (1.9%) of them displaying various indications of hereditary breast cancer. These mutations were not observed among sporadic breast cancer patients or healthy controls. Both of the detected mutations were also found in AT families originating from the same geographical region as the breast cancer families displaying the corresponding mutation, implying the possibility of a founder effect concerning the distribution of 7522G>C and 6903insA. However, to follow this preliminary lead, a more extensive screening for ATM mutations among breast cancer families is indicated. In addition, it would be interesting to investigate whether similar trends are seen with mutations in other AT families from different parts of the country.