Missense Variants in ATM in 26,101 Breast Cancer Cases and 29,842 Controls

This article is featured in Highlights of This Issue, p. 2113

Ataxia-telangiectasia (A-T) is an autosomal recessive disorder characterized by cerebellar ataxia, telangiectases, immune defects, radiosensitivity, and a predisposition to malignancy (MIM #208900). The gene that is mutated in A-T, ATM (MIM #607585), encodes a protein kinase that plays a key role in cellular responses to DNA damage. The large majority of A-T cases are known to harbor mutations in ATM leading to a truncated or absent protein. Epidemiologic studies of families of A-T patients have shown a 2- to 5-fold increased risk of breast cancer for female relatives who are obligate heterozygous carriers of an A-T mutation (1, 2).

The increased risk of breast cancer in ATM mutation carriers has been confirmed by direct analysis of ATM mutations in breast cancer cases compared with controls. In a study of British familial breast cancer cases and controls, Renwick and colleagues identified 9 mutations that result in premature termination or exon skipping among 443 strongly familial cases (2.0%) compared with 2 in 551 controls (0.4%, P = 0.028; ref. 3). They also found three cases and no controls who carried one of two missense variants for which there is strong a priori evidence of a pathogenic phenotype in individuals with A-T (V2424G or SV2855_2856RI). Bernstein and colleagues identified 7 heterozygotes for the V2424G missense variant among 3,743 population-based breast cancer cases (0.2%) unselected for family history and none among 1,268 controls (P = 0.1; ref. 4). Based on the breast cancer history of first- and second-degree relatives of carrier cases, the breast cancer risk to age 70 years for heterozygotes was estimated to be 52% [95% confidence interval (95% CI), 28-80%; P < 0.0001].

An association between other ATM variants, particularly amino acid substitutions that are not expected to be associated with A-T, and breast cancer has also been hypothesized (5), but to date, there has been little evidence to support this (6, 7). In a previous study, we genotyped nine missense variants in ATM in 473 bilateral breast cancer cases and 2,463 controls as part of a high-throughput screen of 1,037 nonsynonymous single-nucleotide polymorphisms (SNP) within candidate “cancer genes” (8). None of these variants was common, with minor allele frequencies (MAF) in controls ranging from <0.1% (0 of 4,924 chromosomes) to 2.4% (116 of 4,926 chromosomes). Although no single ATM missense variant was significantly associated with breast cancer risk, there was a significant trend in risk with increasing numbers of variant ATM SNPs [odds ratio (OR), 1.27; 95% CI, 1.04-1.56; P_trend = 0.02]. We selected four variants with MAF >1% [S707P (rs4986761), F858L (rs1800056), P1054R (rs1800057), and L1420F (rs1800058)] for further analysis in 26,101 invasive breast cancer cases and 29,842 controls in 23 studies within the Breast Cancer Association Consortium (BCAC). We also included a fifth variant [S49C (rs1800054)] with a MAF of 1.2%, which was not genotyped in our previous analysis (8) but for which there had been some prior evidence of an association with breast cancer risk (OR, 1.13; 95% CI, 0.99-1.30 P = 0.08) in an earlier BCAC analysis (9) that included a subset of the current studies.

Supplementary Table S1 summarizes the study details and genotyping platform for all studies that contributed data. Genotyping was done with 5′ nuclease assay (TaqMan), Sequenom iPLEX, or Illumina Golden Gate technology. TaqMan genotyping reagents were designed by Applied Biosystems as Assays-by-Design^SM and distributed by the University of Cambridge group to each of the centers that used this technology. Genotyping was done using the ABI PRISM 7900HT or 7500 Sequence Detection System according to the manufacturer's instructions. For five studies, SNPs were genotyped using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry for the determination of allele-specific primer extension products using the Sequenom MassARRAY system and iPLEX technology. The design of oligonucleotides was carried out according to the guidelines of Sequenom and performed using MassARRAY Assay Design software (version 1.0). In one study, SNPs were genotyped using customized Illumina Sentrix Bead Arrays according to the manufacturer's instructions.

We applied BCAC standard quality control (QC) guidelines (http://www.srl.cam.ac.uk/consortia/bcac/). In addition, we imposed a threshold of 99% for the call rate (compared with the standard threshold of 95%) and we excluded SNPs from studies where cluster plots, scored from 1 (poor) to 4 (good), scored by a single reader blinded to identifiers, scored 2 or less. These more stringent thresholds were imposed because the minor alleles of these SNPs are rare and therefore more susceptible to differential calling between cases and controls. S49C was not genotyped by 3 studies and data were excluded from analyses for QC criteria for 3 studies. S707P was not genotyped by 2 studies and data were excluded from analyses for QC criteria for 8 studies. F858L was genotyped by all studies; data were excluded from analyses for QC criteria for 1 study. P1054R was genotyped by all studies and data were excluded from analyses for QC criteria for 3 studies. L1420F was not genotyped by 2 studies and data were excluded from analyses for QC criteria for 8 studies. The full details of the studies that contributed data for each SNP, the numbers of cases and controls genotyped by each study, and the genotypes of cases and controls for each SNP are given in Supplementary Tables S1 and S2.

The OR for each SNP and for being a carrier or rare homozygote for any SNP was tested using logistic regression with “study” as a stratifying covariate. To maximize the amount of data included in the analysis, SNPs that were not genotyped by a study or were excluded for QC criteria were coded as 0 for all subjects for the analysis of being a carrier or rare homozygote for any SNP. The effect of this will be to bias our OR estimate, marginally, toward the null.

Linkage disequilibrium metrics between SNPs (r² and D′; Supplementary Table S3) were computed separately for each study using the Tagzilla module as implemented in GLU version 1.0a6. rs1800056 (F858L) and rs1800057 (P1054R) are correlated (r² = 0.38-0.71; Supplementary Table S3), otherwise these rare SNPs are independent of each other (r² < 0.001). Maximum likelihood estimates of haplotype frequencies for the four alleles defined by F858L and P1054R (i.e. F858+P1054, F858+1054R, 858L+1054R, and 858L+P1054) were estimated in cases and controls separately and in each of the studies separately using HaploStats (http://mayoresearch.mayo.edu/mayo/research/schaid_lab/software.cfm); Supplementary Table S3). ORs for F858+1054R, 858L+1054R, and 858L+P1054 versus the common allele F858+P1054 were estimated using unconditional logistic regression weighted for the phase assignment probability and with study as a stratifying covariate.

Statistical analyses were done using STATA version 10 (State College). All P values reported are two-sided. Meta-analyses (Fig. 1) were carried out using the Metan routine within STATA, using inverse variance weighting of the study specific estimates. Cochran's Q statistic and the I² statistic (10) to quantify the proportion of the total variation due to heterogeneity between studies were calculated.

The distribution of genotypes in cases and controls in each study for each ATM SNP is shown in Supplementary Table S2. Subjects reporting ethnicities other than Caucasian were excluded (see footnote in Supplementary Table S2). The MAFs for each of the five SNPs genotyped in this analysis differed significantly (P < 0.007; see footnote in Supplementary Table S1) among the 22 studies of Caucasian subjects; medians (ranges) were S49C, 1.2% (0.2-1.7%); S707P, 0.9% (0.6-1.6%); F858L, 1.5% (0.2-2.4%); P1054R, 2.6% (0.6-3.7%); and L1420F, 1.6% (0.2-2.7%). In the study in which the majority of subjects were of Asian ethnicity (SEBCS), three SNPs were monomorphic (S49C, S707P, and F858L), and for the other two SNPs (P1054R and L1420F), there was only one carrier among 872 control subjects.

In the combined analysis across studies, the point estimates for each of the heterozygote ORs were above 1.0 and the estimates of the homozygote ORs were higher (Table 1). The only significantly elevated OR was for L1420F homozygotes (5.31; 95% CI, 1.35-20.87). Two SNPs, F858L (rs1800056) and P1054R (rs1800057), are correlated (r² = 0.38-0.71 across studies; Supplementary Table S3). The G allele of rs1800057 (1054R) is more common than the C allele of rs1800056 (858L; Supplementary Table S2); thus, the rare C allele of rs1800056 (858L) is almost completely contained on the rare G allele of rs1800057 (1054R) such that there are three main haplotypes for these two allelic variants (F858_P1054, F858_1054R, and 858L_1054R) and one extremely rare haplotype (858L_P1054; Supplementary Table S3). The trend OR estimates for each of the two haplotypes that carried the rare (C) allele of rs1800056 (858L_1054R and 858L_P1054) compared with the most common haplotype (F858_P1054) were 1.05 (95% CI, 0.95-1.16; P = 0.47) and 1.12 (95% CI, 0.61-2.05; P = 0.71), respectively. The OR estimate for the haplotype that carried the rare (G) allele of rs1800057 with the common T allele of rs1800056 (F858_1054R) was 0.97 (95% CI; 0.86-1.10, P = 0.65).

Combining the data from all five SNPs, the OR was 1.05 for being a heterozygote for any of the SNPs and 1.51 for being a rare homozygote for any of the SNPs (P_trend = 0.04), with an overall OR_trend of 1.06 (Table 1) and no evidence of heterogeneity between studies (Fig. 1A; Cochrane Q = 21.5 21df, P = 0.43, I² = 2.4%). Restricting the analysis to bilateral cases and those with a family history of breast cancer, the overall OR_trend was stronger (OR_trend, 1.12; P_trend = 0.02; Table 1) with no evidence of heterogeneity between studies (Fig. 1B; Cochrane Q = 15.5 18df, P = 0.62, I² = 0%).

Based on Swift's demonstration that carrier status for recessively inherited A-T is associated with a 3-fold increase in risk of female breast cancer (1) and on the more recent molecular validation of this observation (3), it is arguable that there is a high prior likelihood that a subset of polymorphic (MAF >1%) missense ATM variants will be associated with a modest increase in breast cancer risk. In our previous analysis of the combined effects of nine missense ATM variants (MAF <0.1-2.4%), we showed that, on average, each missense ATM SNP was associated with an OR of 1.27 (95% CI, 1.04-1.56) in bilateral breast cancer cases, implying an OR of 1.13 (95% CI, 1.02-1.25) for cases with a single primary breast cancer (11, 12).

We selected five SNPs for further investigation. Despite restricting our follow-up analysis to SNPs with MAFs estimated to be ≥1%, we did not have power to estimate the individual effects for these SNPs or the effects of individual haplotypes. The aim of this present analysis was, therefore, to test the composite hypothesis that rare polymorphic ATM variants are, on average, associated with an increased risk of breast cancer. The five SNPs we genotyped in this analysis had a combined carrier frequency of ∼12.5%; by genotyping 20,000 cases and 20,000 controls, we had 90% power at 1% significance to detect an OR of 1.10.

Our OR estimate of 1.06 (95% CI, 1.00-1.12) provides independent evidence that polymorphic missense variants in ATM are associated with a very modest increase in breast cancer risk, albeit at a nominal level of statistical significance (P = 0.04). The stronger OR estimate for bilateral cases and cases with a family history of breast cancer (OR, 1.12; 95% CI, 1.02-1.23; P = 0.02) provides additional support.

We identified four previous studies (13-16) in which at least 100 Caucasian breast cancer cases and 100 Caucasian controls were genotyped and for which individual effect sizes for S49C (rs1800054), S707P (rs4986761), F858L (rs1800056), P1054R (rs1800057), or L1420F (rs1800058) were reported (Table 2); we also obtained data for all five variants from the Wellcome Trust Case Control Consortium analysis (Table 2, ref. 17). For three of these (13, 14, 16), the case control series overlap with the current analysis; the other two (15, 17) do not support an association but are entirely consistent with a per-SNP OR of 1.06. A recent analysis of rare (MAF <1%), evolutionarily unlikely missense substitutions in ATM (18) reported a per-SNP OR estimate of 1.14 (95% CI, 0.90-1.44; P = 0.39) for the combined effects of 121 variants in 1,948 cases and 1,852 controls. We also identified two studies that compared the frequency of ATM variants in bilateral breast cancer cases versus unilateral breast cancer cases. One (19) reported no difference in the frequency of missense variants between bilateral cases and unilateral cases overall but a longer median time to developing a second cancer in carriers of a missense variant who also received radiotherapy. In the other (20), a study of gene-environment interactions (WECARE study) in which bilateral cases were counter-matched to unilateral “controls” on the basis of exposure to radiotherapy, rare (MAF <1%) A-T associated variants and those that were classified as deleterious according to the prediction algorithm SIFT (21) were associated with a nonsignificantly increased risk of a second breast cancer, whereas those that were classified as tolerated and several of the more common missense variants were associated with a protective effect. For the linked variants F858L and P1054R, this was statistically significant [OR, 0.5 (95% CI, 0.3-1.0) and 0.5 (95% CI, 0.3-0.9) for F858L and P1054R, respectively], raising the possibility of an interaction between radiotherapy and a subset of ATM variants.

It is not yet clear whether polymorphic (MAF >1%) missense variants in ATM and other validated breast cancer genes could make a contribution to explaining the excess familial risk of breast cancer. With a combined carrier frequency of 12.6% in Caucasian controls and an estimated average OR of 1.06, these five ATM variants explain 0.03% of excess familial risk of breast cancer, compared with between 0.07% and 1.7% explained by each of the common variants identified in recent genome-wide association studies (7, 22-27). Rare SNPs (MAF ≤5%), however, account for a relatively large proportion of genetic variation (28); there are 83 rare missense SNPs in ATM listed in dbSNP (including the five genotyped in this study) and large numbers in other breast cancer genes (29-32).

Testing the combined effects of rare missense variants in known breast cancer genes in large collaborative studies should, eventually, clarify their overall contribution to breast cancer susceptibility. Gutierrez-Enriquez et al. (33) compared the radiosensitivity of lymphoblastoid cell lines (LCL) from breast cancer cases who were carriers of one or more rare allele(s) of S707P, F858L, P1054R, and L1420F to that of LCLs from healthy controls. They showed increased radiosensitivity in the LCLs from the breast cancer cases compared with the LCLs from controls generally, and specifically for the six LCLs from patients with at least one copy of the 858L + 1054R haplotype. Incorporating information from such functional assays and from next-generation in silico prediction algorithms may help to identify a subset that are most likely to be predictive of risk (34-36).

No potential conflicts of interest were disclosed.

We thank the thousands of women who participated in this research. The HEBCS thanks Dr. Kirsimari Aaltonen and RN Hanna Jäntti for their help with the patient data and gratefully acknowledges the Finnish Cancer Registry for the cancer data. The GC-HBOC thanks Sandrine Tchatchou for participating in genotyping. The SBCS thanks Sabapathy Balasubramanian, Simon Cross, Helen Cramp, and Dan Connley for their contribution to the study. The ABCFS thanks Maggie Angelakos, Judi Maskiell, and Gillian Dite. The HABCS and HMBCS gratefully acknowledge their German colleague Johann H. Karstens for his support of the breast cancer studies at Hannover Medical School. The ORIGO study thanks P.E.A. Huijts, E. Krol-Warmerdam, and J. Blom for patient accrual, administering questionnaires, and managing clinical information. The SEARCH study thanks the SEARCH and EPIC teams for recruitment of case patients and control subjects. kConFab thanks Heather Thorne, Eveline Niedermayr, all the kConFab research nurses and staff, the heads and staff of the Family Cancer Clinics, the Clinical Follow-Up Study for its contributions to the resource, and the many families who contribute to kConFab. The AOCS Management Group (D. Bowtell, G. Chenevix-Trench, A. de Fazio, D. Gertig, A. Green, and P. Webb) gratefully acknowledges the contribution of all the clinical and scientific collaborators, the AOCS, and the ACS Management Group (A. Green, P. Parsons, N. Hayward, P. Webb, D. Whiteman), as well as all of the project staff, collaborating institutions, and study participants. The GENICA study acknowledges Christian Baisch for the collection of clinical and histopathologic data; Beate Pesch, Volker Harth, and Thomas Brüning for their involvement in the recruitment of study subjects and the collection of epidemiologic data; and Christina Justenhoven for genotyping and data management. The CNIO-BCS thanks Primitiva Menendez from the Hospital Central Universitario de Asturias (HUCA-Oviedo), Pilar Zamora from the La Paz University Hospital in Madrid, and Anna González-Neira, Charo Alonso, and Tais Moreno from the CNIO. The KBCP is thankful to Helena Kemiläinen and Aija Parkkinen for their contribution. The PBCS thanks Drs. Neonila Szeszenia-Dabrowska and Beata Peplonska of the Nofer Institute of Occupational Medicine (Lodz, Poland), Witold Zatonski of the Department of Cancer Epidemiology and Prevention, The M. Sklodowska-Curie Cancer Center and Institute of Oncology (Warsaw, Poland), Mark Sherman from the Division of Cancer Epidemiology and Genetics of the National Cancer Institute (USA), Jeff P Struewing from the National Human Genetics Research Institute (USA), and Pei Chao from Information Management Services (Sliver Spring, MD) for their valuable contributions to the study. The GESBC thanks Ursula Eilber for competent data coordination and management and Tanja Koehler for excellent technical assistance. The ABCS acknowledges L. Braaf, R. van Hien, R. Tollenaar, and the other contributors to the “BOSOM” study and H.B. Bueno-de-Mesquita for support in organizing the release of control DNA.

Grant Support: The BBCS is funded by Cancer Research UK (CR-UK) and Breakthrough Breast Cancer and acknowledges NHS funding to the NIHR Biomedical Research Centre and the National Cancer Research Network. The HEBCS study has been financially supported by the Helsinki University Central Hospital Research Fund, Academy of Finland grant 110663, the Finnish Cancer Society, and the Sigrid Juselius Foundation. The GC-HBOC study was supported by Deutsche Krebshilfe grant 107054, the Center of Molecular Medicine, Cologne, the Helmholtz Society, and the Dietmar-Hopp Foundation. The SBCS was supported by Yorkshire Cancer Research and the Breast Cancer Campaign. The ABCFS was supported by National Health and Medical Research Council of Australia (NHMRC) grant 145604, NIH grant CA102740-01A2, and National Cancer Institute, NIH grant CA-95-011 through cooperative agreements with members of the Breast Cancer Family Registry and principal investigators, Cancer Care Ontario grant CA69467, Columbia University grant CA69398, Fox Chase Cancer Center grant CA69631, Huntsman Cancer Institute grant CA69446, Northern California Cancer Center grant CA69417, and University of Melbourne grant CA69638. The content of this manuscript does not necessarily reflect the views or policies of the National Cancer Institute or any of collaborating centers in the Breast CFR, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government or the Breast CFR. The ABCFS was initially supported by the NHMRC, the New South Wales Cancer Council, and the Victorian Health Promotion Foundation. J.L. Hopper is an Australia Fellow of the NHMRC and Victorian Breast Cancer Research Consortium Group Leader. M.C. Southey and A.B. Spurdle are Senior Research Fellows of the NHMRC. Genotyping was in part supported by the Prostate Cancer Foundation of Australia. The MCBCS was supported by NIH grants CA122340 and CA128978, an NIH Breast Cancer Specialized Program of Research Excellence award to the Mayo Clinic (CA116201), and a Susan G. Komen Breast Cancer Foundation award. The HABCS was supported by an intramural grant from Hannover Medical School and German Research Foundation (DFG) grant Do761/2-1. The HMBCS was supported by short-term fellowships from the German Academic Exchange Program (N. Bogdanova) and the Friends of Hannover Medical School (N. Bogdanova). The ORIGO study was supported by the Dutch Cancer Society. The SASBAC study was supported by the Agency for Science, Technology and Research of Singapore (A*STAR), the NIH, and the Susan G. Komen Breast Cancer Foundation. SEARCH is funded by CR-UK programme grant C490/A11021. A.M. Dunning is supported by CR-UK grant C8197/A10865 and P.D.P. Pharoah is a Senior Clinical Research Fellow of CR-UK. kConFab is supported by grants from the National Breast Cancer Foundation, the NHMRC, the Queensland Cancer Fund, the Cancer Councils of New South Wales, Victoria, Tasmania, and South Australia, and the Cancer Foundation of Western Australia. The kConFab Clinical Follow-Up Study was funded by NHMRC grants 145684, 288704, and 454508. Financial support for the AOCS was provided by U.S. Army Medical Research and Materiel Command grant DAMD17-01-1-0729, the Cancer Council of Tasmania, Cancer Foundation of Western Australia, and NHMRC grant 199600. G. Chenevix-Trench is supported by the NHMRC. The UCIBCS is supported by NIH, National Cancer Institute grant CA-58860 and Lon V Smith Foundation grant LVS-18840. The GENICA study was supported by the German Human Genome Project and German Federal Ministry of Education and Research (BMBF) grants 01KW9975/5, 01KW9976/8, 01KW9977/0, and 01KW0114. Genotyping analysis was supported by the Robert Bosch Foundation of Medical Research, Stuttgart, Germany, and the Deutsches Krebsforschungszentrum, Heidelberg, Germany. The work of the BBCC was partly funded by ELAN-Fond of the University Hospital of Erlangen. Infrastructure support for the MCCS recruitment and follow-up is provided by The Cancer Council Victoria, whereas cohort recruitment was partly funded by VicHealth. This work using the MCCS was supported by NHMRC grants 209057, 251533, and 396414 and genotyping was in part supported by the Prostate Cancer Foundation of Australia. The CNIO-BCS was supported by the Genome Spain Foundation, the Red Temática de Investigación Cooperativa en Cáncer, the Asociación Española Contra Cáncer, and Fondo de Investigación Sanitario grants PI081120 (J. Beesley) and PI081583 (R.L. Milne.). The CGPS was supported by the Chief Physician Johan Boserup and Lise Boserup Fund, the Danish Medical Research Council, and Copenhagen University Hospital, Herlev Hospital. The SEBCS was supported National Research and Development (R&D) Program for Cancer Control grant 0620410-1 and Korea Health 21 R&D Project grant AO30001, Ministry of Health and Welfare, Republic of Korea. The KBCP is supported by the EVO funds of Kuopio University Hospital and the Finnish Cancer Foundation. The PBCS was funded by Intramural Research Funds of the National Cancer Institute, Department of Health and Human Services. The GESBC was supported by Deutsche Krebshilfe e.V. grant 70492. Funding for the ABCS was provided by Dutch Cancer Society grants NKI 2001-2423 and 2007-3839 and the Dutch National Genomics Initiative. KARBAC acknowledges funding from the Swedish Cancer Society and the Gustav V Julilee Foundation. The BCAC is funded by CR-UK grants C1287/A10118 and C1287/A7497. Meetings of the BCAC have been funded by the European Union COST Programme (BM0606). D.F. Easton is a Principal Research Fellow of CR-UK.

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.