ATM-Heterozygous Germline Mutations Contribute to Breast Cancer–Susceptibility

Summary

Approximately 0.5%–1% of the general population has been estimated to be heterozygous for a germline mutation in the ATM gene. Mutations in the ATM gene are responsible for the autosomal recessive disorder ataxia-telangiectasia (A-T) (MIM 208900). The finding that ATM-heterozygotes have an increased relative risk for breast cancer was supported by some studies but not confirmed by others. In view of this discrepancy, we examined the frequency of ATM germline mutations in a selected group of Dutch patients with breast cancer. We have analyzed ATM germline mutations in normal blood lymphocytes, using the protein-truncation test followed by genomic-sequence analysis. A high percentage of ATM germline mutations was demonstrated among patients with sporadic breast cancer. The 82 patients included in this study had developed breast cancer at age <45 and had survived ⩾5 years (mean 15 years), and in 33 (40%) of the patients a contralateral breast tumor had been diagnosed. Among these patients we identified seven (8.5%) ATM germline mutations, of which five are distinct. One splice-site mutation (IVS10-6T→G) was detected three times in our series. Four heterozygous carriers were patients with bilateral breast cancer. Our results indicate that the mutations identified in this study are “A-T disease–causing” mutations that might be associated with an increased risk of breast cancer in heterozygotes. We conclude that ATM heterozygotes have an approximately ninefold-increased risk of developing a type of breast cancer characterized by frequent bilateral occurrence, early age at onset, and long-term survival. The specific characteristics of our population of patients may explain why such a high frequency was not found in other series.

Introduction

Individuals with the autosomal recessive disorder ataxia-telangiectasia (A-T) (MIM 208900) are homozygous or compound heterozygous for a mutation in the ATM gene (Savitsky et al. 1995; Lavin and Shiloh 1997). More than 300 different disease-causing mutations, located throughout this gene's open reading frame (ORF), have been identified thus far. The disease has an estimated incidence of 1/40,000–1/100,000 live births and is characterized by several clinical features—for example, cerebellar ataxia, dilated blood vessels in the eyes and skin (telangiectasia), immune deficiency, chromosomal instability, predisposition to cancer, and increased sensitivity to ionizing radiation (Gatti et al. 1991; Segdwick and Boder 1991; Shiloh 1995; Lavin and Shiloh 1996, 1997). The typical phenotype is caused by ATM null alleles that either truncate (frameshift or stop-codon mutations) or severely destabilize (mutations affecting mRNA splicing) the ATM protein (Gilad et al. 1998; Sandoval et al. 1999; Teraoka et al. 1999).

A number of epidemiological studies have indicated that female ATM-heterozygous carriers have an excess risk of breast cancer, with an estimated relative risk of 3.9 (95% confidence interval [CI] 2.0–6.0) (Swift et al. 1987, 1991; reviewed in Easton 1994; Inskip et al. 1999; Janin et al. 1999). Several recent studies have been conducted to verify this hypothesis. Athma et al.'s (1996) study of families with A-T showed that female ATM-mutation carriers have an increased relative risk for breast cancer: 3.8. In spite of that, studies of patients with breast cancer have so far failed to confirm the findings in families with A-T. Fitzgerald et al. (1997), analyzing a series of young patients with breast cancer, suggested that ATM-heterozygous germline mutations do not contribute to early-onset (age <40 years) sporadic breast cancer. Moreover, in a third recent report, Chen et al. (1998) conclude that the contribution of heterozygous ATM mutations to familial breast cancer is minimal.

In the present study we examined whether ATM germline mutations can be found in a selected group of Dutch patients with breast cancer. We determined the ATM mutation frequency in women who had developed breast cancer at a young age. Using the protein-truncation test (PTT) as the mutation-detection method, which was followed by genomic sequencing, we expected to have a mutation-detection sensitivity of ∼75% (Broeks et al. 1998). We had reported elsewhere that there was no common ATM founder mutation present in a small series of 12 Dutch patients with A-T and that the mutations had been found throughout the ORF of the ATM gene, with no apparent hotspots.

Subjects and Methods

Study Population

The patients in this study were selected from a large cohort of patients with breast cancer who had been treated at either The Netherlands Cancer Institute (Amsterdam) or The Dr. Daniel den Hoed Cancer Center (Rotterdam), during 1970–90. Patients with contralateral breast cancer (n=33) were eligible if their first breast cancer had been diagnosed at age <45 and ⩾5 years before the diagnosis of contralateral disease. Patients with unilateral disease (n=49) also had to have been diagnosed with breast cancer at age <45 and had to have survived ⩾5 years without developing contralateral disease. On the basis of their birth names, all 82 patients were inferred to be of Dutch ethnic origin. Information regarding a family history of breast cancer was obtained by means of a questionnaire mailed to all patients. The extent of family history of breast cancer was divided into three categories (one case–i.e., the index case—had to be diagnosed at age <45, paternal inheritance was considered to be first degree): (1) hereditary—three family members with breast cancer in two generations, who were first-degree relatives of each other; (2) sporadic—either a single breast cancer case or only one second-degree relative with breast cancer; and (3) familial—all others. Each of these patients was asked to donate a 20-ml blood sample for DNA and RNA isolation, and informed consent was obtained. To determine the frequency of specific ATM mutations in the Dutch population, DNA samples from 268 healthy individuals of Dutch ethnic origin (179 males and 89 females of which 51 were age <45 years) were obtained from the Blood Bank at Leiden.

Generation of Genomic DNA and Total RNA and Preparation of cDNA

Peripheral blood lymphocytes were used to isolate genomic DNA and total RNA, with DNAzol (Gibco) and RNAzol B (Ampro) methods, respectively, according to the manufacturers' protocol. A 25–1.25-μg sample of total RNA was used in the cDNA reaction, which was conducted by means of a Superscript II kit, according to the manufacturer's (Gibco BRL) protocol.

PTT

The ATM ORF was divided into 10 overlapping fragments of ∼1–2 kb each, with an overlap of 400–500 bp to avoid the possibility that truncations close to the ends of the fragments could be missed. Forward primers all include a T7-promoter sequence for the transcriptional initiation of T7 RNA polymerase and an ATG codon for translational initiation. Nested PCR was necessary only for amplification of region I. Primer sequences are available on request.

The PCR of each region was performed in a total volume of 30 μl containing 1 μl of cDNA template, 3 μl PCR buffer 6 (Stratagene), 0.2 mM of each dNTP, 15 pmol of each (forward and reverse) primer, 1.5 μl of dimethyl sulfoxide, 1 unit of Taq DNA-polymerase (Gibco BRL), and 0.25 units of PWO DNA-polymerase (Boehringer). The PCR regions were amplified in a PTC200 apparatus (MJ Research) for 1 min at 94°C (for denaturing); 35 cycles of 10 s at 92°C, 1 s at 70°C (with a decrease of 0.5°C/s until 55°C was attained), 10 s at 55°C (with an increase of 0.7°C/s until 72°C was attained), 2.5 min at 72°C; and, finally, 5 min at 72°C. The outer PCR of region I was performed in 15 cycles in a total volume of 30 μl, under the same conditions as described above, and 1 μl of the product of the first PCR was used for the nested inner PCR. PCR products were analyzed by agarose gel-electrophoresis.

From each sample, 2 μl of PCR product was used for the in vitro transcription-translation reaction using the TNT T7 Coupled Reticulocyte Lysate System (L4610; Promega). The reaction was performed according to protocol in a 6.25-μl total volume containing 2 μl of PCR product, 3.1 μl of rabbit reticulocyte lysate, 0.25 μl of TNT-buffer, 0.125 μl of amino acid mixture (minus methionine), 0.5 μl of [³⁵S]-methionine (10 mCi/ml), 0.125 μl (40 units/μl) of RNAse inhibitor, and 0.125 μl of T7 RNA polymerase. The reaction mixture was incubated for 60 min at 30°C. The translation products were separated on a 15% SDS-polyacrylamide gel, and the gels were fixed, incubated in fluorographic reagent (Amplify; Amersham), dried, and subjected to autoradiography. Each gel contained a ¹⁴C-labeled high-molecular-weight rainbow marker (Amersham) as a size standard.

Nucleotide-Sequence Analysis

All aberrant PTT patterns were analyzed by sequence analysis, first on cDNA and finally on genomic DNA, to identify candidate mutations. Sequence analysis was performed by means of the “DNA Sequencing Kit” big-dye primer cycle sequencing ready reaction with 21M13 and M13 reverse primers (PE Biosystems). Sequencing products were analyzed with the ABI 377 automated sequencing apparatus and corresponding software.

IVS10-6T→G Mutation–Specific Assay

A 100- ng sample of genomic DNA was amplified, in a total volume of 30 μl containing 3 μl of PCR buffer with 3 mM MgCl₂, 15 pmol of each primer, an intron 10 forward primer (IVS10-146: 5′-ctg agg cag gag aat cac ttg) and an exon 11 reverse primer (nucleotide 1213: 5′-tct gtg act tct gaa ggt gat c), 0.7 μl of AmpliTaq gold, and 3 μl (2 mM each) of each dNTP, in the PTC200 (MJ research) apparatus, for 6 min at 94°C; 20 cycles of 30 s at 94°C, 20 s at 65°C (with a decrease of 0.5°C/s/cycle until 55°C was attained), 2.5 min at 72°C, followed by 20 cycles annealing at 55°C; and, finally, 5 s at 72°C). PCR fragments were digested with 5 units of RsaI for 2 h at 37°C and were analysed on a 2% agarose gel.

IVS10-6T→G Splicing

PCR with forward primer FSQ1 (nucleotide −2: 5′-gtg tgt tct gaa att gtg aac) and reverse primer RSQ2 (nucleotide 1370: 5′-cgt aac aca tat ggt gta cg) resulted in two fragments—one including exon 11 and the other excluding it—of 1,373 and 1,203 bp, respectively. These products were separated on a 1.2% agarose gel and were isolated from gel. The products were individually sequenced, and a polymorphism at nucleotide position 37, 37C/T, was used as an allele-specific marker; nucleotide 37C is linked to the normal sequence IVS10-6T, whereas nucleotide 37T is linked to the mutation IVS10-6G.

Tumor-Tissue Histology and Immunohistochemistry

For each paraffin-embedded tumor, six 10-μm paraffin slides were used for genomic-DNA isolation (whereas three 10-μm paraffin slides are used in the standard protocol [Sambrook et al. 1989]), and an hematoxylin-eosin–stained slide was used for histopathological examination.

Loss of Heterozygosity (LOH)

LOH of the microsatellite marker D11S2179 (located in ATM) in tumor material was determined by comparison of peak-height ratios of normal tissue with those of tumor tissue. The peak height of the longer allele peak was divided by the peak height of the shorter allele peak. The ratio obtained in the tumor DNA was divided by the ratio obtained in normal-tissue DNA. LOH was determined by ratios >2 or <0.5, depending on which allele was lost. Microsatellite analysis was performed by PCR in a total volume of 30 μl containing 5–10 ng of genomic DNA, 3 μl PCR buffer 6 (Stratagene), 0.2 mM of each dNTP, 9 pmol of each primer (listed in The Genome Database), and 1.5 units of Taq polymerase, in a multicycler PTC200 apparatus (MJ Research). The PCR products were amplified in 30 cycles of 30 s for 94°C, 1 min at 62°C, and 1 min at 72°C; were electrophoresed on an ABI 373 or 377 automated sequencing apparatus; and were analysed by means of GENESCAN 2.1. Subsequently, sequence analysis of the tumor DNA was used to determine whether the wild-type or mutant allele was lost.

Results

Seven ATM Germline Mutations in 82 Selected Patients with Breast Cancer

Among 82 patients with sporadic breast cancer, we have identified in 7 (8.5%) ATM germline mutations, of which five are distinct (tables 1 and 2); these mutations either already had been described as A-T disease causing mutations elsewhere (Ataxia-Telangiectasia Mutation Database; T. Dörk, personal communication), as causing the disease in patients with A-T or are truncating mutations (table 2; patients ATMb56 and ATMb73) and therefore most likely are pathological mutations.

Table 1.

Frequency of All ATM Germline Mutations and of the ATM IVS10-6T→G Splice-site Mutation in Patients and Controls

Group	All ATM Mutations	ATM IVS10-6T→G
Breast cancer at age <45 years:
Unilateral	3/49	2/49
Bilateral	4/33	1/33
Total	7/82 (8.5%)	3/82 (3.7%)
Controls	.5%–1.0%a	2/268 (.7%)

^aData are from the literature.

Table 2.

Seven ATM Germline Mutations in Patients With Breast Cancer at Age <45 Years

Patient	Genomic Mutationa	Predicted Alteration in ATM Proteinb
Bilateral breast cancer:
ATMb56	3114A→T (exon 23)	Stop codon, truncation at amino acid 1039
ATMb57	IVS10-6T→G (exon 11)	Out-of-frame deletion, truncation at amino acid 419
ATMb58	IVS14+2T→G (exon 14)	In-frame deletion codons 601-633
Unilateral breast cancer:
ATMb73	1660delA (exon 13)	Truncation at amino acid 555
ATMc211	1562delAG (exon 12)	Truncation at amino acid 564
ATMc214	IVS10-6T→G (exon 11)	Out-of-frame deletion, truncation at amino acid 419
ATMc232	IVS10-6T→G (exon 11)	Out-of-frame deletion, truncation at amino acid 419

^aNucleotide positions indicated are numbered in accordance with reference sequence ATG = 1 (Platzer et al. 1997).

^bLength is 3,056 amino acids.

All patients with breast cancer who were included in our study were diagnosed with breast cancer at age <45 and had survived for 5 years after diagnosis (mean duration of follow-up was 15 years), and in 33 (40%) of these patients a contralateral breast tumor was diagnosed ⩾5 years after the first diagnosis of breast cancer. The proportion of women in the ATM-mutated group who had hereditary or familial versus sporadic breast cancer were roughly similar to those in the group of women who did not have an ATM germline mutation (table 3A). Additional characteristics of these patients and their tumors are listed in table 3. Patients with an ATM germline mutation were slightly older at the first diagnosis of their first breast cancer and had a slightly longer follow-up period, and a higher proportion of them had bilateral disease (tables 1 and 3A). However, none of these differences was statistically significant, possibly because of the small number of patients. Furthermore, no specific clinicopathological characteristics (e.g., histology or tumor stage) of ATM-heterozygous patients were apparent (table 3B).

Table 3.

Patient and Histopathological Characteristics of All Patients with Breast Cancer and of ATM Heterozygotes

A. All Patients with Breast Cancer
Characteristic	Patients without ATM Mutation (n=75)	Patients with ATM Mutation (n=7)
Mean age [range] at diagnosis (years):
First breast cancer	38.0 (22.1–45.0)	40.5 (35.7–43.3)
Second breast cancer (n=33)	46.3 (27.3–55.7)	50.6 (43.7–59.3)
Interval [range] between diagnosis of first tumor and diagnosis of second tumor (years)	9.9 (4.6–19.9)	11.0 (6.8–19.6)
Mean follow-up period [range] since diagnosis (years):
First tumor	15.0 (7.5–25.7)	16.8 (13.9–24.1)
Second tumor (n=33])	5.3 (.7–16.8)	6.4 (4.4–8.4)
Type of breast cancer in family:a
Hereditary	10%	14%
Familial	16%	14%
Sporadic	74%	72%

B. ATM Heterozygotes
Patient (Mutation) and Diagnosis of Breast Cancer	Age(years)	Histology	TNM Stage	Interval(years)	Follow-Up Period since First Diagnosis of Breast Cancer
ATMb56 (3114A→T):
First diagnosis	40	Ductal carcinoma in situ	T0N0M0		14.3
Second diagnosis	46	Infiltrating ductal carcinoma/ductal carcinoma in situ	T2N1M0	6.8
ATMb57 (IVS10-6T→G):
First diagnosis	39	Ductal/lobular carcinoma in situ	T0N0M0		24.1
Second diagnosis	59	Ductal carcinoma in situ	T0N0M0	19.6
ATMb58 (IVS14+2T→G):
First diagnosis	35	Infiltrating ductal carcinoma	T1N0M0		16.4
Second diagnosis	43	Infiltrating ductal carcinoma	T2N0M0	8
ATMb73 (1660delA):
First diagnosis	42	Infiltrating ductal carcinoma	T2N1M0		14.8
Second diagnosis	52	Infiltrating ductal carcinoma	T1N0M0	9.7
ATMc211 (1562delAG):
First diagnosis	39	Infiltrating lobular carcinoma	T1N0M0		17.8
ATMc214 (IVS10-6T→G):
First diagnosis	43	Infiltrating ductal carcinoma	T1N1M0		13.9
ATMc232 (IVS10-6T→G):
First diagnosis	42	Infiltrating ductal carcinoma	T2NxM0		16.2

^aFor category descriptions, see the Subjects and Methods section. Data were available for 7 mutation carriers and for 70 non– mutation carriers.

ATM Germline Mutation IVS10-6T→G

The splice-site mutation IVS10-6T→G, localized in the highly conserved pyrimidine tract of the 3′ splice-acceptor site of intron 10, was detected in three unrelated patients with breast cancer in this study (3.7% of the sample; see table 1). We assessed the frequency of this particular mutation in the general population, using a mutation-specific assay (see the Subjects and Methods section). In the general population, we detected this mutation in 2 (both males) of 268 healthy controls, which results in an estimated frequency of ∼0.7% (95% CI 0%–1.7%) for this mutation alone. Haplotype analysis of the five individuals with this mutation, using seven microsatellite markers located in the ATM region (Laake et al. 1999), revealed that the ATM IVS10-6T→G mutation is present on at least two different alleles, each sharing a different set of at least three of these seven markers (authors' unpublished observations).

The IVS10-6T→G mutation leads to incorrect splicing of exon 11 and to exon skipping, resulting in a frameshift and subsequent truncation of the protein at amino acid residue 419. Assaying for a mutation-linked polymorphism at nucleotide position 37, we determined that a small proportion (<10%) of the mutant allele is spliced correctly, resulting in some full-length mRNA (data not shown). The wild-type allele did not display any incorrect splicing.

LOH: Not a Prerequisite for Initiation of Breast Cancer in ATM Heterozygotes

To obtain further insight into the precise role of ATM in tumorigenesis in ATM-heterozygous patients, we investigated whether the loss of the wild-type ATM allele in these tumors was a prerequisite for tumor development. We analyzed LOH of the ATM locus—that is, chromosomal region 11q22-23—in four breast tumors in three ATM heterozygotes, using the microsatellite marker D11S2179, located within the ATM gene. One patient (ATMb56) with bilateral breast cancer showed LOH in the first tumor (data not shown); however, sequence analysis revealed that it was the mutant allele that was lost in the tumor—in, remarkably, only a proportion (∼50%) of the tumor cells. Two other informative tumors, one each in patients ATMb57 and ATMb58, showed no LOH.

Discussion

We have identified a high percentage of ATM germline mutations among selected patients with breast cancer. The mutations that are described in this study are most likely ATM null alleles, since most of them either truncate the ATM protein or have been described, elsewhere (Ataxia-Telangiectasia Mutation Database; T. Dörk, personal communication), as occurring in patients who are homozygous or compound heterozygous for A-T. Our results do not show an overrepresentation of hereditary breast cancer among patients with breast cancer and heterozygous ATM germline mutations, thereby excluding the likelihood of the presence of a breast cancer–susceptibility gene, such as BRCA1 or BRCA2, that would account for predisposition to breast cancer.

One mutation, the exon 11 splice-site mutation IVS10-6T→G, was found frequently among the patients with breast cancer in this study (frequency 3.7%, vs. 0.7% in the general population). Despite this high frequency in the Dutch population, the IVS10-6T→G mutation has not been detected thus far in a small series of Dutch patients with A-T (Broeks et al. 1998). However, a patient with A-T who was homozygous for the exon 11 splice-site mutation has been identified in Germany (T. Dörk, personal communication). Our data suggest that this particular ATM mutation might be associated with an increased risk of breast cancer in ATM heterozygotes.

None of the tumors of the ATM heterozygotes tested for LOH showed LOH of the wild-type allele. Thus, it seems that a mutant ATM germline allele plays a role in tumor initiation but that complete loss of the normal protein is not a prerequisite for tumor initiation. Loss of the mutant allele, as was found in patient ATMb56, could be explained by LOH directed by a locus in close proximity to ATM, which would be important at a later stage during tumor development (Laake et al. 1997, 1999).

Our results support the hypothesis that ATM heterozygotes do have an increased risk of development of breast cancer. The specific characteristics of our the patients whom we studied (i.e., early age at onset, frequent occurrence of contralateral disease, and long-term survival) may explain why such a high frequency of ATM germline mutations was not found in other studies of patients with breast cancer (Fitzgerald et al. 1997; Chen et al. 1998). For a precise assessment of their risk, data are needed on the prevalence of ATM-heterozygous mutations in a large group of women without breast cancer. In the absence of such data, we have used the commonly accepted upper estimate of 1% heterozygotes in the Dutch population (based on an upper estimate of 1/40,000 A-T live births in The Netherlands; Jaspers, personal communication) to estimate that ATM heterozygotes have an approximately ninefold-increased risk, compared with nonheterozygotes, of developing a breast tumor with specific characteristics (early age at onset, increased risk of contralateral disease, and long-term survival).

It is well known that A-T is associated with high sensitivity to ionizing radiation and that fibroblast strains or cultures from ATM heterozygotes display moderate levels of radiation sensitivity; on average, they are more sensitive than are normal controls (Paterson et al. 1979; West et al. 1995). All patients with breast cancer who were included in the present study received low- dosage diagnostic radiation at a young age, as part of a screening program for tuberculosis. Moreover, all these patients had received radiation treatment for their first breast tumor, and a nonsignificant trend toward a higher incidence of ATM mutations was noted in women with bilateral breast cancer. Thus, radiation might indeed be an induction trigger for development of breast cancer in ATM heterozygotes, and these data strongly suggest that the role of ATM heterozygosity in the pathogenesis of radiogenic cancers needs further investigation. Although our findings indicate an approximately ninefold-increased risk of heterozygous ATM mutation carriers for breast cancer, we did not observe an extended family history of breast cancer in these patients. However, since an induction trigger might be a prerequisite for tumor development, their relatives might not have been exposed to such an environmental factor, as has been suggested as well by a study of the incidence of breast cancer in female relatives with A-T (Swift et al. 1991 Inskip et al. 1999). We hope that in the near future we can provide more insight into the quantitative contribution of ATM heterozygosity to the risk of radiation-related cancers.