Abstract
Deficiencies in tasks of detecting and repairing DNA damage lead to mutations and chromosomal abnormalities, a hallmark of cancer. The gene mutated in ataxia-telangiectasia (A-T), ATM, is a proximal component in performing such tasks. Studies of A-T families have suggested an increased risk of breast cancer among obligate female heterozygous carriers of ATM mutations. Paradoxically, studies of sporadic and familial breast cancer have failed to demonstrate an elevated prevalence of mutations among breast cancer cases. We characterized the prevalence and distribution of 20 ATM missense mutations/polymorphisms in a population-based case-control study of 854 African-American, Latina, Japanese, and Caucasian women aged ≥45 years participating in the Multiethnic Cohort Study. The study population included 428 incident breast cancer cases and 426 controls. The prevalence of variants ranged from 0% to 13.6% among controls and varied by ethnicity (0–32.5%). Overall, these data provide little support for an association of ATM missense mutations with breast cancer among older women. We observed only one sequence variation (L546V), common among African-American women, to be overrepresented among all high-stage breast cancer cases (odds ratio, 3.35; 95% confidence interval, 1.27–8.84). After correction for multiple comparisons, this observed risk modification did not attain statistical significance. The distribution of ATM missense mutations and polymorphisms varied widely across the four ethnic groups studied. Although a single missense variant (L546V) appeared to act as a modest predictor of risk, the remaining variants were no more common in breast cancer cases as compared with controls.
Introduction
A-T3 is a pleiotropic inherited disease characterized by neurodegeneration, oculocutaneous telangiectases, an increased incidence of cancer, immunodeficiencies, radiation sensitivity and genetic instability (1 , 2) . The gene mutated in A-T, ATM, spans more than 150 kb, is composed of 66 exons (62 coding), and produces a 13-kb transcript encoding a protein of 3056 amino acids (3) . ATM is a member of the PI3K family as a consequence of its sequence similarity to the PI3K catalytic domain near the COOH terminus of the protein (4) . Heterozygous carriers of germ-line ATM variants are estimated to constitute 0.35–1% of the general population; the majority of these variants (>70%) are predicted to lead to either truncation or altered splicing of the protein (2 , 5) .
ATM plays a key and proximal role in monitoring and responding to DNA damage; initial evidence came from case reports of A-T patients who had fatal reactions to radiation therapy (6, 7, 8) . It has also been shown that cultured fibroblasts from A-T patients are multiplicatively more sensitive to the cytotoxic effects of ionizing radiation as compared with control cells (9 , 10) . Similarly, cell lines derived from A-T patients exhibit defects in several ionizing radiation-inducible cell cycle checkpoints, the most critical of which is arrest in the G1 phase of the cell cycle (11 , 12) .
Recent evidence has identified ATM as an essential and proximal component of cell cycle restriction point control. Its scope of interaction includes the phosphorylation and activation of p53 (13) , c-Abl (14 , 15) , and Chk2 proteins (16 , 17) as well as the inactivation of Cdc25 (18) , all of which promote either apoptosis or cell cycle arrest. More germane to the oncogenesis of breast cancer, however, is the recently demonstrated biochemical connection between ATM and the inherited breast cancer susceptibility gene BRCA1, wherein ATM phosphorylates BRCA1 in a cluster of residues at the COOH terminus in response to γ-radiation (19) .
Studies of A-T families have suggested an increased risk of breast cancer among obligate female heterozygous carriers of A-T variants (20, 21, 22, 23) ; a meta-analysis estimated the relative risk to be 3.9 (24) . Paradoxically, studies of sporadic and familial breast cancer have failed to consistently demonstrate an elevated prevalence of germ-line ATM gene variants among breast cancer cases (25 , 26) . To resolve these apparently disparate findings, Gatti et al. (27) proposed a model for the role of ATM heterozygosity in breast and other cancers, positing two classes of ATM mutations: null or truncating mutations that lead to A-T; and missense mutations that cause cancers. Whereas truncating mutations would block expression of ATM protein, missense mutations could code for stable ATM proteins that are present at normal intracellular concentrations but function abnormally.
To date, a confluence of evidence from epidemiological as well as cell and animal systems has provided initial support for this model. Indirect evidence comes from reports from cohorts of breast cancer cases carrying missense mutations, whereas truncating mutations are not common (28, 29, 30) . In particular, Stankovic et al. (31) identified in two A-T families a missense mutation in the PI3K region (T7271G or V2425G) that was associated with a 13-fold increased risk of breast cancer. Furthermore, the T7271G missense mutation appeared to be highly penetrant for breast cancer; expression and activity studies indicated that the mutation yielded a dominant-negative inhibitor of ATM (32) . In addition, an inducible expression system for ATM has been developed showing that several missense alleles outside of the kinase domain induce a partial A-T phenotype when introduced into normal cells also in a dominant-negative fashion, perhaps through a mechanism involving ATM-ATM interaction (33) . Lastly, a knock-in mouse model of a known A-T-causing in-frame deletion results in mice with a significant number of solid tumors (34) . This in-frame deletion results in the production of a functionally distrupted, nearly full-length ATM and hence has important implications for missense mutations. Taken together, these observations provide support for cancer predisposition among human A-T missense carriers.
It is important to note that not all missense mutations are equally important when considering breast cancer susceptibility. For instance, one might postulate that those mutations disrupting conserved domains of the ATM protein (e.g., the PI3K domain) could exert a greater effect on cancer risk as compared with those outside such regions. To date, the precise estimates of the risk of breast and other cancers associated with ATM missense variants are not clearly defined, but any elevated risk would carry with it significant clinical implications. The prevalence of ATM missense mutations has not been comprehensively evaluated in a multiethnic population, although striking differences in ATM sequence diversity between African and non-African populations have been reported (35) . In this study, we evaluated the relationship between 20 missense variants/polymorphisms in the ATM gene and breast cancer risk in a case-control study among African-America, Latina, Japanese, and Caucasian women participating in the Multiethnic Cohort Study.
Materials and Methods
Multiethnic Cohort Study Population.
This nested case-control study is part of a large, ongoing, multiethnic cohort study in Hawaii and Los Angeles, California with an emphasis on diet and other lifestyle characteristics in the etiology of cancer. Aspects of this large cohort as well as details of its design and implementation are described more fully elsewhere (36) . Briefly, participants were recruited between 1993 and 1996 from driver’s license files in Hawaii and California; the age range at baseline was between 45 and 75 years. The focus was on four main ethnic groups: African Americans; Japanese Americans; Latinos/Latinas; and Caucasians. The total number of male and female subjects who comprised the cohort was 215,251. Among women only, baseline data were collected on 22,251 African Americans, 29,957 Japanese, 26,502 Caucasians, and 24,620 Latinas. Participants have completed a baseline questionnaire designed for self-administration that included five sections: (a) background, including medical history and family cancer history; (b) diet history; (c) medication use; (d) physical activity; and (e) female reproductive history, including menstruation history, parity, age at first full-term pregnancy, oral contraceptive use, age at menopause, and the use of hormones.
Eligible cases were women enrolled in the cohort and diagnosed between 1993 and 1998 with a new primary, incident and histologically confirmed breast cancer (International Classification of Diseases-Oncology, codes C50.0 to C50.9) identified by linkage of the cohort to population-based cancer Surveillence, Epidemiology and End Results registries in Hawaii and California. Cases were contacted by letter and phone call and agreed to provide a blood specimen. The participation rate for providing a blood sample on request was 74% for cancer cases. Women with carcinoma in situ (non-infiltrating pathology) and neoplasms of the skin of the breast (International Classification of Diseases-Oncology code 44.5) were not included as breast cancer cases. Information on stage of disease was ascertained from tumor registries and used in subgroup analyses. Stage of disease was characterized as “localized” or “high stage,” which included regional (by direct extension and/or lymph node involvement) or systemic disease.
For this particular effort, a nested case-control study was designed with the intention of comprehensively analyzing the role of rare ATM missense variants in a multiethnic, population-based sample. Given that ATM missense mutation frequencies were hypothesized to be overrepresented among breast cancer cases, approximately 100 cases from each ethnic group were initially selected (n = 428), and women diagnosed with high-stage disease were oversampled (n = 222) as compared with cases who initially presented with localized disease (n = 206).
Blood samples had also been collected from an approximately 3% random sample of healthy cohort members at baseline (37 , 38) . In this effort, we selected approximately 100 controls for each of the four ethnic groups (n = 426); the participation rate for cohort controls was 66%. Only controls with no previous diagnosis of breast cancer were included. The study was approved by the Institutional Review Board of the Keck School of Medicine of the University of Southern California.
ATM Missense Variant Discovery.
A separate sequencing effort had previously been undertaken to discover missense variants spanning the full-length sequence transcript of the ATM gene. Full sequence analysis of ATM was performed on cDNA from peripheral lymphocytes. Briefly, a nested reverse transcription-PCR approach was used to generate overlapping, internally labeled PCR products. These cover the entire sequence of ATM and were analyzed by sequencing of reverse transcription-PCR products. All base changes were reconfirmed. This study included a total of 274 individuals, mostly of European descent, comprised of 94 primary breast cancer patients, 70 bilateral cases, and 63 individuals without disease selected within a hospital-based group of breast cancer cases from the United States. From this effort, 20 missense variants of interest were identified; a number of the variants discovered (but not all) had been described previously (25 , 28 , 35 , 39 , 40) .
Genotyping.
Genomic DNA was purified from the buffy coats of peripheral blood samples for all cases and controls using the Puregene DNA Isolation protocol and kit (Gentra Systems, Minneapolis, MN). Single nucleotide polymorphism genotyping was performed using the fluorogenic 5′ nuclease assay (TaqMan Assay; Ref. 41 ). The TaqMan assay was performed using a TaqMan PCR Core Reagent Kit (Applied Biosystems) according to manufacturer’s instructions in a final volume of 20 μl. Using a fluorescent dye-labeled probe specific for each allele, the profile of each well was measured in a Sequence Detection System (model 7700 or model 7900HT; Applied Biosystems), and the results were analyzed with Sequence Detection Software (Applied Biosystems).
Statistical Analysis.
Data management and descriptive and univariate analyses were performed using SAS statistical software version 8.01 (SAS Institute, Cary, NC). The EpiLog software system (EpiCenter Software, Pasadena, CA) was used to estimate ORs and 95% CIs by unconditional logistic regression while adjusting for ethnicity. The Bonferroni correction for multiple comparisons was used to define the α level of significance to avoid spurious positive results. This α critical value for these analyses is ≤0.0025 (0.05/20). Given this level of significance, this study, as designed, has 80% power to detect a relative risk of 1.8 for a 25% minor allele, 2.2 for a 10% minor allele, 2.8 for a 5% minor allele, and 6.5 for a 1% minor allele.
Results
We characterized the prevalence and distribution of 20 ATM missense mutations/polymorphisms in a case-control study of 854 African-American, Latina, Japanese, and Caucasian women aged ≥45 years participating in the Multiethnic Cohort Study. Associations between established reproductive breast cancer risk factors and breast cancer risk were generally consistent with expectation in all ethnic groups among cases and controls (Table 1)⇓ . For instance, cases more often reported a family history of breast cancer (18.0% of cases versus 9.9% of controls; P = 0.01) and tended to have a later first full-term pregnancy (after age 30 years, 12.2% versus 6.7%; Pheterogeneity = 0.04).
Descriptive statistics of subjects stratified by case or control status (total observations with percentage in parentheses)
The prevalence of variants ranged in frequency from 0% to 13.6% among controls for all ethnicities combined and varied widely by ethnicity (0–33.0%; Table 2⇓ ). Two of the missense variants (D126E and D1853N) are previously described common polymorphisms (35) and are present in equal frequencies among cases and controls (Table 2)⇓ . Most of the other missense variants were uncommon and did not appear to be overrepresented among breast cancer cases as compared with controls (data not shown).
Ethnic-specific distribution of missense variants in the ATM gene among breast cancer cases and controls
We did observe an exon 13 missense variant (L546V) to be modestly overrepresented among all breast cancer cases as compared with controls (ORcrude, 2.44; 95% CI, 0.91–6.54). This association, however, was limited to African-American women as the L546V variant was relatively common within this group [7.7% overall; 10.3% among all cases (12.1% among those with high-stage disease) and 5.1% among controls]. The L546V missense mutation was also seen in two Latina cases but was not seen among any of the Japanese or Caucasian study participants.
Discussion
Studies of A-T families have documented an increased risk of breast cancer among both presumptive and obligate heterozygous carriers of ATM gene mutations (21 , 22) . Whereas this observation has been corroborated by a Dutch study (42) , to date, most other case-control studies have failed to support the hypothesis that ATM variant carriers are at an increased risk of breast cancer (25 , 26 , 43, 44, 45, 46, 47) . Initial surveys, guided by the suggestion that most “at risk” A-T alleles were truncating or null mutations (5) , relied on methods that identify aberrant pre-mRNA splice variants, namely, protein truncation test and single-strand conformational polymorphism methods. The protein truncation test method would necessarily overlook rare missense variants, and although the single-strand conformational polymorphism method is sensitive to missense mutations, early studies often only considered truncating mutations. Hence, these studies likely underestimated the prevalence of ATM variants in breast cancer cases and controls. As a result, few missense variants have been described or, alternatively, may have been overlooked. Nevertheless, the T7271G missense mutation in the PI3K region has been shown to be highly penetrant for breast cancer, associated with an estimated 13-fold increased risk, and yields a dominant-negative inhibitor of ATM (32) .
We characterized the prevalence and distribution of 20 ATM missense mutations/polymorphisms in a multiethnic study population consisting of African-American, Latina, Japanese, and Caucasian women. In the aggregate, the variants characterized were rare, consistent with other ATM studies (35 , 48) . Furthermore, the ethnic distribution of specific variants was comparable with those reported in previous studies that observed striking differences between African and non-African populations (35) . We also observed the D1856N variant to be frequent among Caucasians (21.1% of controls), whereas the D126E variant was very common among African Americans (32.5%) but was less often observed among the other ethnic groups. Thorstenson et al. (35) observed a similar ethnic-specific distribution of these same two polymorphic markers and suggest that such variation may be the result of random genetic drift or in fact due to selective pressure.
With the exception of the L546V missense mutation, we did not note a specific increase in the frequency of ATM missense mutations in breast carcinoma cases as compared with controls. However, as a consequence of testing 20 variants, we did expect one to attain statistical significance as a consequence of multiple hypothesis testing. As such, a Bonferroni correction was used; no individual variant attained the critical level of significance as determined by this procedure.
There is increasing evidence that missense variants in ATM encode stable, functionally abnormal proteins. Overexpression of a mutant ATM polypeptide has previously been shown to increase genetic instability in normal cells, thus displaying a dominant-negative cellular phenotype (49) ; such dominant interference has been demonstrated using an in vitro mutagenesis approach (33) . Furthermore, two ATM mutations cosegregating with breast cancer in multiple-case families have been shown to yield a dominant-negative inhibitor of ATM (32) . Unlike truncating variants, which act effectively as null variants, missense variants exert distinct effects on ATM function and cancer risk (22) . Furthermore, the ATM protein exists as a component of a multiprotein complex (50) ; expression of a mutated protein from even a single missense allele might interfere with this complex.
Considerable molecular evidence places ATM as a key and proximal component in DNA damage response, maintenance of genomic integrity, and regulation of cell cycle checkpoints (51) . Additionally, the demonstrated functional interaction of ATM with BRCA1 (19) , along with an inferred relationship with BRCA2, defines a molecular pathway that may be disrupted in some fraction of breast cancer patients. Based on these observations, gene-gene interactions between ATM missense variants and variants/polymorphisms in BRCA1 and BRCA2 represent a promising avenue of further study.
This study lacked sufficient power to effectively evaluate some ethnic-specific risks (most particularly among Japanese and Caucasians) due to the low prevalence of the variant alleles. However, its multiethnic design will allow us to continue to examine these and other ATM variants in different ethnic groups in the future. Furthermore, population stratification, although a potential concern (52) , is an unlikely explanation for the demonstrated association in African Americans because the observed D126E (nucleotide 378) prevalence among cases and controls (∼30%) is identical to that reported among Africans in a comprehensive survey of ATM diversity (35) .
In this study, we evaluated 20 variant sites in four ethnic groups and their association with breast carcinoma. The L546V variant appeared to act as a modest but not statistically significant predictor of risk, although its effect was almost exclusively observed among African-American women. Additional evaluation of missense variants, particularly among younger women reporting a family history of breast cancer, is required to better characterize the effective contribution of this and other ATM missense variants. The degree to which ATM heterozygosity is associated with an increased risk in breast cancer remains an open debate. Not all missense mutations will have the same effect; hence, more research regarding the molecular structure and function of variant ATM is required.
Acknowledgments
We thank the members of the Multiethnic Cohort Study for their participation and cooperation and the cohort investigators in Los Angeles and Hawaii.
Footnotes
-
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
-
↵1 Supported by National Cancer Institute Grants CA 63464 and CA 54281.
-
↵2 To whom requests for reprints should be addressed, at Department of Preventive Medicine, University of Southern California/Norris Comprehensive Cancer Center, 1441 Eastlake Avenue, MS #44, P. O. Box 33800, Los Angeles, CA 90033. Phone: (323) 865-3997; Fax: (323) 865-0127; E-mail: bretsky{at}usc.edu
-
↵3 The abbreviations used are: A-T, ataxia-telangiectasia; PI3K, phosphatidylinositol 3′-kinase; OR, odds ratio; CI, confidence interval.
- Received November 25, 2002.
- Revision received April 28, 2003.
- Accepted May 6, 2003.
References
Volume 12, Issue 8
