Document Outline

; Bakkenist and Kastan

; Kurz and Lees-

Miller

The ATM gene is located on chromosome 11q22-23, and

the entire gene spans almost 150 kb. The gene consists of
66 exons and is transcribed to a 13-kb mRNA with a coding

M. Schrauder · S. Frank · P. L. Strissel · M. P. Lux · M. R. Bani ·
C. Rauh · K. Heusinger · A. Hartmann · R. Schulz-Wendtland ·
R. Strick · M. W. Beckmann · P. A. Fasching (&)
University Breast Center for Franconia,
Erlangen University Hospital, Universitaetsstrasse 21–23,
91054 Erlangen, Bavaria, Germany
e-mail: peter.fasching@uk-erlangen.de

M. Schrauder
e-mail: michael.schrauder@uk-erlangen.de

C. C. Sieber
Institute for Biomedicine of Aging,
Department of Internal Medicine V (Geriatrics),
University of Erlangen-Nuremberg,
Prof. Ernst-Nathan-Strasse 1, 90419 Nuremberg,
Bavaria, Germany

874

J Cancer Res Clin Oncol (2008) 134:873–882

123

sequence of 9,168 bp. It encodes a 350-kDa protein belong-
ing to the PI3K-related protein kinases (PIKKs) (Uziel
et al.

1996

; Platzer et al.

1997

Hereditary homozygous or compound heterozygous

mutations in the ATM gene cause the rare autosomal reces-
sive disease ataxia telangiectasia (A-T). Most of the muta-
tions that cause A-T are base substitutions, insertions, or
deletions that generate premature termination codons or
splicing abnormalities (Shiloh

2003

). A-T is clinically char-

acterized by skin and ocular telangiectasia, extreme hyper-
sensitivity to ionizing radiation, progressive
immunode

Wciency, chromosomal instability, and an

increased risk of malignancies (primarily of lymphoid ori-
gin) (Eyfjord and Bodvarsdottir

Studies of female relatives of A-T patients have sug-

gested that heterozygous carriers of ATM mutations have a
signi

Wcant increase in the risk of breast cancer (Swift et al.

; Borresen et al.

; Olsen et al.

;

Thompson et al.

). These

Wndings are of epidemiologi-

cal importance, as it has been suggested that heterozygous
carriers account for 0.35–1% of the general population
(Swift et al.

1986

; Gatti et al.

1999

). However, the role of

ATM as a breast cancer susceptibility gene in non-A-T fam-
ilies has been controversial.

A case control study, which selected cases based on a

family history of breast cancer, found truncating, splicing
and missense mutations to be associated with an increased
breast cancer risk (Renwick et al.

Furthermore studies with selected breast cancer patients

(early age of onset, and/or bilateral breast cancer) suggested
an increased breast cancer risk for patients with heterozy-
gous ATM mutations (Broeks et al.

2000

; Teraoka et al.

2001

). These studies included selected cases with either a

family history, bilaterality of breast cancer or early onset of
disease. The selection of these kind of cases increases the
likelihood to

Wnd susceptibility alleles (Antoniou and Eas-

ton

2003

). There are however reports that ATM variations

are not found more frequently in early onset breast cancer
cases (FitzGerald et al.

1997

) or familial breast cancer

cases (Tommiska et al.

Case control studies, including sporadic breast cancer

cases, have presented inconsistent results regarding the risk
of breast cancer among heterozygous ATM mutation carri-
ers. Most of these studies have failed to show an elevated
breast cancer risk associated with ATM mutations or poly-
morphisms (Tommiska et al.

; Einarsdottir et al.

It has been suggested that the common polymorphism

ATMex39 5557G>A (p.D1853N; rs1801516) is associated
with bilateral breast cancer among familial breast cancer
patients when found in the cis position with ATMivs38-
8T>C (Heikkinen et al.

). Other groups have reported

no signi

Wcant association of these variants or any haplotype

containing them with the risk of breast cancer, bilateral
breast cancer, or multiple primary cancers (Ralhan et al.

2007

; Teraoka et al.

2001

; Tommiska et al.

; Heikki-

nen et al.

; Angele et al.

2003

). The carrier frequency

of the ivs38-8T>C variant (or the combined variant, as all
the carriers of this variant also carried 5557G>A) was mar-
ginally (but not signi

Wcantly) higher in familial breast can-

cer patients (8.1%) than in healthy controls (5.6%)
(Tommiska et al.

These data might suggest a functional e

Vect of the

p.D1853N variant in BC, but it is still not clear whether it is
associated with a higher BC risk or even with a protective
e

Vect. We therefore analyzed the genotype of the ATM

polymorphism 5557G>A (p.D1853N) in a hospital-based
case–control study in Germany to further elucidate its
potentially modifying role on the risk of breast cancer, the
prognosis, and its association with histopathological and
clinical patient characteristics.

Patients and methods

Study population

We carried out a hospital-based case–control study con-
ducted in Germany—the Bavarian Breast Cancer and Con-
trols (BBCC) study. Starting in 2002, primary breast cancer
patients who were treated at the University Breast Center
for Franconia were asked to take part in the study. Both
newly diagnosed patients and breast cancer patients who
were attending the follow-up care unit at the Breast Center
were invited to participate in the study. The patients were
followed up for cancer recurrence and mortality. Mortality
data were obtained from the German death registry. Data on
recurrences were documented prospectively for patients
attending the Breast Center for follow-up care, in accor-
dance with the data management standards of the German
certi

Wcation board for Breast Centers (

http://www.onkoz-

ert.de

) and the European Society of Mastology (EUSOMA)

(Blamey and Cataliotti

http://www.appliedbiosystems.com

). Patients who were not

attending the Breast Center were contacted annually by
mail. Patients who did not respond to the letter were con-
tacted by phone. The median follow-up period for the
cohort was 4.1 years.

The study population consisted of 514 patients with his-

tologically con

Wrmed breast cancer and 511 cancer-free

control individuals selected from the same geographical
area and matched with the cases by age. All of the patients
provided informed consent to the recording of personal data
and DNA specimens. Approval for the study was obtained
from the Ethics Committee of the University of Erlangen.
Trained interviewers used a pretested questionnaire to sys-
tematically collect epidemiological data from the partici-

J Cancer Res Clin Oncol (2008) 134:873–882

875

123

pants with regard to demographic factors (e.g., age) and
known risk factors, including occupational history, smok-
ing status, and family history of cancer. Approximately
10 mL of whole blood was collected after the interview.
The median age at the onset of breast cancer was
55.8 § 11.6 years in this group of patients. The clinical
stage was classi

Wed in accordance with the American Joint

Committee on Cancer TNM staging system.

Blood DNA isolation

DNA was extracted from 8-mL blood samples using a
genomic DNA puri

Wcation kit (Puregene; Gentra Systems,

Minneapolis, MN, USA/Qiagen Ltd., Hilden, Germany)
with modi

Wcations. BrieXy, after initial centrifugation, the

white blood cell layer was removed and added to RBC lysis
bu

Ver, pH 7.3, containing 0.15 M NH

Cl, 0.01 M K

and 0.1 M Na-EDTA. After 10 min incubation, the cells
were centrifuged at 2,000 g and incubated with 3 mL cell
lysis bu

Ver containing 20 mM Tris pH 7.4, 15 mM Na-

EDTA, and 1% SDS, and treated with RNase A and pro-
teinase K (all from Sigma-Aldrich Chemie Ltd., Schnell-
dorf, Germany). The proteins were precipitated with 1 mL
of protein precipitation solution (Puregene), and the DNA
was precipitated by the addition of isopropanol, washed
with 70% ethanol, dried, solubilized with a Tris-EDTA
bu

Ver, pH 7.5, quantitated using a spectrophotometer, and

stored at ¡80°C. Using this methodology, an average of
70–100

g of DNA per patient was obtained.

Polymorphism genotyping

Real-time PCR for the following single nucleotide polymor-
phism (SNP) of the ATM gene at chromosome 11q22.3: ref-
SNP ID, rs1801516 (nucleotide exchange G5557A, amino
acid exchange D1853N) in accordance with the National
Center for Biotechnology Information (NCBI;

http://

www.ncbi.nlm.nih.gov

) was purchased from Applied Bio-

systems (ABI, Foster City, CA, USA/Applera Deutschland
Ltd, Darmstadt, Germany) and analyzed for all patients with
an ABI7000 in accordance with the manufacturer’s instruc-
tions. The used SNP assay rs1801516 was purchased as a
TaqMan

SNP pre-designed genotyping assay functionally

tested for accuracy, reliability and reproducibility by
Applied Biosystems. This function test consists of either 10
or 20 unique DNA samples comprised of a mixed ethnic
population and both male and female representation (refer-
ence guide under

). The

real-time PCR instrument ABI7000 was routinely calibrated
for region of interest (ROI), background and pure dye spec-
tra according to manufacturer’s instructions.

Polymerase chain reaction was performed with a total

volume of 10

L, which contained approximately 3.0 ng

DNA, 1£ TaqMan

Universal PCR Mastermix, each

primer at a concentration of 900 nM, and each probe at a
concentration of 200 nM. Quality controls (5 samples per
96 well plate) were included in the genotyping assays. The
PCR results were evaluated without knowledge of the
patients’ status. PCR call rates were greater than 97.1%.

Histopathological features such as tumor type, hormone

receptor status, proliferation status assessed using Ki-67
immunohistochemistry, and HER2/neu status were all ana-
lyzed at the Institute of Pathology at the University of
Erlangen.

Statistical analysis

Statistical analysis was carried out using the SPSS program
(version 14.0.1, SPSS Inc., Chicago, IL, USA). A total of
514 patients were included in the analysis. Data on survival
status, recurrence status, and genotype were available for
these patients. The objectives analyzed were overall survival
and disease-free survival, de

Wned as any local or distant

recurrence, and death, whichever occurred

Wrst. Any sur-

vival function was calculated from the time of the onset of
disease to the occurrence of an objective. Disease-free sur-
vival was censored at the time of the last contact, and overall
survival was censored on 1 March 2007, which was the date
on which the survival data were correlated with the German
death registry. Kaplan–Meier estimates are presented for the
survival function, and di

Verences in survival were analyzed

using the log-rank test. Cox proportional hazards regression
analysis was used to estimate hazards ratios (HR) and 95%
con

Wdence intervals (95% CI) were calculated for disease-

free and overall survival. Associations between the geno-
types and histopathological and clinical parameters were
analyzed using Chi-squared tests (Pearson and Mantel–
Haenszel tests). Logistic regression analysis was used to
estimate the odds ratios for variables contributing to the risk
of breast cancer and to adjust the analyses for confounders
contributing to breast cancer susceptibility.

Results

The results for the association between commonly known
risk factors and the risk of breast cancer are summarized in
Table

, and the genotype and allele frequencies are pre-

sented in Table

. Hardy–Weinberg equilibrium testing of

rs1801516 was consistent with the distribution (Chi-
squared test P = 0.59).

Association with breast cancer risk

The frequencies of homozygous ATM 5557G/G (wt), het-
erozygous 5557G/A, and homozygous 5557A/A were 79.0,

876

J Cancer Res Clin Oncol (2008) 134:873–882

123

19.3, and 1.8% for BC patients, and 72.2, 25.2 and 2.5% for
healthy controls, respectively (Table

). This di

Verence was

signi

Wcant with a P value of 0.04 for the Chi-squared test

(Pearson) and a P value of 0.013 for the Chi-squared test
(Mantel–Haenszel). In the logistic regression model, raw
odds ratios (OR) for the heterozygous genotype were 0.70
(95% CI, 0.52–0.94; P = 0.018) and 0.63 (95% CI, 0.27–
1.49; P = 0.292) for the homozygous variant genotype.
After adjustment for age, the number of

Wrst-degree rela-

tives a

Vected by breast cancer, the number of Wrst-degree

relatives a

Vected by other cancers, body mass index, age at

Wrst birth, number of pregnancies, number of live births,

and age at menarche, the genotype was still statistically sig-
ni

Wcant: the OR for the heterozygous genotype was 0.68

(95% CI, 0.47–0.97; P = 0.03) and 0.59 (95% CI, 0.22–
1.55; P = 0.28) for the homozygous variant genotype. The
incidences of breast cancer, ovarian cancer, and other carci-
nomas among

Wrst-degree relatives did not diVer signiW-

cantly between cases and controls (Pearson Chi-squared
test P = 0.85, P = 0.316, and P = 0.096). The number of
pregnancies was the only parameter for which a signi

Wcant

Verence between cases and cancer-free controls was

found (P = 0.003). Breast cancer risk factors are summa-
rized in Table

Table 1 Comparison of cases and controls by selected demographic characteristics and major risk factors for breast cancer

Values are presented as means § standard deviation among cases and controls, unless otherwise noted

*P value, Pearson Chi-squared test

Among parous women

Among postmenopausal women

P value, Mantel–Haenszel test

Patient characteristics

Total (n = 1,025)

Cases (n = 514)

Controls (n = 511)

Demographic factors

Age (year)

56.4 § 9.5

55.9 § 11.6

57.0 § 6.8

0.64*

Major risk factors

Breast cancer in

Wrst-degree relatives, n (%)

93 (9.4)

44 (9.2)

49 (9.6)

0.853*

Ovarian cancers in

Wrst-degree relatives, n (%)

14 (1.4)

5 (1.1)

9 (1.8)

0.316*

Other cancers in

Wrst-degree relatives, n (%)

311 (31.5)

138 (28.9)

173 (33.9)

0.096*

Body mass index (kg/m

)

25.6 § 4.6

25.8 § 4.6

25.5 § 4.7

0.286*

Age at

Wrst birth (year)

25.2 § 4.9

25.5 § 5.0

25.0 § 4.8

0.184*

Number of pregnancies, n (%)

104 (11.2)

55 (11.5)

49 (10.9)

0.003*

186 (20.1)

111 (23.3)

75 (16.7)

335 (36.2)

181 (37.9)

154 (34.4)

300 (32.4)

130 (27.3)

170 (37.9)

Age at menarche (year)

13.5 § 1.5

13.6 § 1.7

13.4 § 1.4

0.52*

Age at menopause (year)

49.4 § 4.9

49.7 § 4.6

49.1 § 5.2

0.135*

ATM polymorphism 5557G>A (rs1801516)

775 (75.6)

406 (79.0)

369 (72.2)

0.04*

228 (22.2)

99 (19.3)

129 (25.2)

0.013

22 (2.1)

9 (1.8)

13 (2.5)

775 (75.6)

406 (79.0)

369 (72.2)

0.012*

GA + AA

250 (24.4)

108 (21.0)

142 (27.8)

Table 2 Genotype and allele frequency of the single nucleotide polymorphism evaluated (rs1801516)

Genotype frequency

Allele frequency

Total n (%)

Wild-type n (%)

Heterozygous
n (%)

Homozygous
variant n (%)

Wild-type n (%)

Variant n (%)

Total alleles
n (%)

Total (cases + controls)

1,025 (100)

G/G 775 (75.6)

G/A228 (22.2)

A/A 22 (2.1)

G 1,778 (86.7)

A 272 (13.3)

2,050 (100)

Breast cancer cases

514 (100)

G/G 406 (79.0)

G/A9 9 (19.3)

A/A 9 (1.8)

G 911 (88.6)

A 117 (11.4)

1,028 (100)

Controls

511 (100)

G/G 369 (72.2)

G/A 129 (25.2)

A/A 13 (2.5)

G 867 (84.8)

A 272 (15.2)

1,022 (100)

J Cancer Res Clin Oncol (2008) 134:873–882

877

123

Table 3 Patient and tumor characteristics and association with disease-free and overall survival (P values obtained using the Kaplan–Meier esti-
mate)

pT tumor stage, pN nodal stage, ER estrogen receptor, PgR progesterone receptor

*Chi-squared test (Pearson)

Mantel–Haenszel test for linear association

Characteristic Total

n (%)

Genotype: G/G
Protein: DD

Genotype: G/A
Protein: DN

Genotype: A/A
Protein: NN

Total (breast cancer patients)

514 (100)

406 (79.0)

99 (19.3)

9 (1.8)

Age

514 (mean 55.9 § 11.6)

55.8 § 11.4

55.7 § 12.0

58.8 § 14.5

0.748

Body mass index (kg/m

)

472 (mean 25.8 § 4.6)

35.9 § 4.4

26.1 § 5.3

23.8 § 2.44

0.406

Tumor type

Invasive ductal

328 (63.8)

257 (63.3)

63 (63.6)

8 (88.9)

0.601*

Invasive lobular

82 (16)

66 (16.3)

16 (16.2)

0 (0)

0.625

Other

104 (20.2)

83 (20.4)

20 (20.2)

1 (11.1)

Total

514 (100)

406 (100)

99 (100)

9 (100)

Menopausal status

Premenopausal

176 (34.9)

137 (34.3)

37 (38.1)

2 (25)

0.654*

Postmenopausal

328 (65.1)

262 (65.7)

60 (61.9)

6 (75)

0.751

Total

504 (100)

399 (100)

97 (100)

8 (100)

268 (52.1)

207 (51)

56 (56.6)

5 (55.6)

0.687*

189 (36.8)

150 (36.9)

36 (36.3)

3 (33.3)

0.132

26 (5.1)

21 (5.2)

4 (4)

1 (11.1)

31 (6)

28 (6.9)

3 (3)

0 (0)

Total

514 (100)

406 (100)

99 (100)

9 (100)

330 (64.2)

260 (64)

67 (67.7)

3 (33.3)

0.119*

184 (35.8)

146 (36)

32 (32.3)

6 (66.6)

0.672

Total

514 (100)

406 (100)

99 (100)

9 (100)

Grading

42 (9)

31 (8.4)

10 (11.1)

1 (11.1)

0.741*

318 (68.2)

248 (67.6)

64 (71.1)

6 (66.7)

0.214

106 (22.7)

88 (24)

16 (17.8)

2 (22.2)

Total

466 (100)

367 (100)

90 (100)

9 (100)

HER2/neu status

Negative

282 (80.1)

224 (80.6)

53 (79.1)

5 (71.4)

0.814*

Positive

70 (19.9)

54 (19.4)

14 (20.9)

2 (28.6)

0.588

Total

352 (100)

278 (100)

67 (100)

7 (100)

ER status

Negative

130 (27.7)

105 (28.5)

21 (22.6)

4 (50)

0.191*

Positive

340 (72.3)

264 (71.5)

72 (77.4)

4 (50)

0.707

Total

470 (100)

369 (100)

93 (100)

8 (100)

PgR status

Negative

174 (36.9)

136 (36.7)

34 (36.6)

4 (50)

0.739*

Positive

298 (63.1)

235 (63.3)

59 (63.4)

4 (50)

0.706

Total

472 (100)

371 (100)

93 (100)

8 (100)

Proliferation (Ki-67)

>15%

186 (45.9)

142 (44.4)

42 (53.2)

2 (33.3)

0.307*

15%

219 (54.1)

178 (55.6)

37 (46.8)

4 (66.7)

0.353

Total

405 (100)

320 (100)

79 (100)

6 (100)

878

J Cancer Res Clin Oncol (2008) 134:873–882

123

Association with breast cancer tumor characteristics

Selected patient characteristics as well as tumor character-
istics and their association with the genotypes are summa-
rized in Table

. When associations between the genotype

and patient and tumor characteristics were analyzed, no
associations were found between age, body mass index,
menopausal status, tumor size, nodal status, grading, estro-
gen or progesterone receptor status, Her2/neu status, or pro-
liferation (as measured by KI-67) and the di

Verent

genotypes.

Association with disease-free and overall survival

The 5-year disease-free survival (DFS) rate for the group of
breast cancer patients studied was 85.7%. No signi

Wcant

Verence in the DFS was found between subgroups with

Verent ATM 5557G>A (D1853N) variants (Fig.

). The

5-year overall survival (OS) was 93%, and again no corre-
lation was found between the ATM polymorphism studied
and the OS (Fig.

). However, commonly known prognos-

tic factors did correlate with the DFS and OS (Table

Discussion

The analysis of the single nucleotide polymorphism
ATMex39 5557G>A (D1853N) that was conducted
revealed a slight but signi

Wcant diVerence in the carrier fre-

quency between unselected breast cancer patients and can-
cer-free controls. The breast cancer patients had a lower
percentage of the heterozygous and homozygous ATM
5557G>A (D1853N) missense variant (Table

). Compar-

ing the homozygous variant group with the homozygous

wildtype group however alone, the di

Verence was statisti-

cally not signi

Wcant. No associations between the histopa-

thological

Wndings, patient characteristics, or disease-free

and overall survival were observed.

Although most of the ATM mutations that cause A-T that

have been described so far are truncating mutations result-
ing in little or no detectable ATM protein, the initial studies
that examined the association between truncating ATM
mutations and BC risk failed to reveal signi

Wcant correla-

tions (FitzGerald et al.

1997

; Bebb et al.

1999

). On the

other hand, female relatives of A-T patients were found to
have a signi

Wcantly higher incidence of breast cancer, sug-

gesting that heterozygous carriers of ATM mutations might
be at higher risk (Swift et al.

1991

1987

; Borresen et al.

1990

). Various models have been developed to explain this

controversy. It has been proposed by Gatti et al. that hetero-
zygotes with the truncating type of mutation and those with
missense mutations have a di

Verent cancer risk. The

increased BC risk may be con

Wned to dominant-negative

mutations—missense changes in particular—due to inter-
ference of the mutated protein with the product of the wild-
type allele during the transphosphorylation process (Bakke-
nist and Kastan

; Gatti et al.

1999

). By contrast, large

case–control studies analyzing di

Verent ATM missense and

truncation variants were not able to reproduce the associa-
tion with BC or other cancer types and failed to detect any
di

Verences in the BC risk among diVerent types of ATM

mutation (Cavaciuti et al.

; Bernstein et al.

It has also been proposed that the increased BC risk

could be restricted to women under the age of 55 years and
may be due mainly to the very high risk in the group of
mothers of individuals with A-T (Olsen et al.

Screening of 443 familial breast cancer patients for ATM

sequence variants identi

Wed 12 mutations in aVected indi-

Fig. 1 Disease-free survival relative to the single nucleotide polymor-
phism (SNP) genotype rs1801516; ATM variant 5557G>A (D1853N)

Time (years)

yti

1,0

0,8

0,6

0,4

0,2

0,0

ATM rs1801516

G/G 5y

DFS:

81.4%

G/A 5y

DFS:

79.1%

A/A 5y

DFS:

75.0%

Log-rank

value: 0.793

Fig. 2 Overall survival relative to the single nucleotide polymorphism
(SNP) genotype rs1801516; ATM variant 5557G>A (D1853N)

Time (years)

yti

1,0

0,8

0,6

0,4

0,2

0,0

ATM rs1801516

G/G 5y

DFS:

91.6%

G/A

5y DFS: 91.1%

A/A

5y DFS: 88.9%

Log-rank

value: 0.811

J Cancer Res Clin Oncol (2008) 134:873–882

879

123

Table 4 Patient and tumor characteristics and association with disease-free and overall survival (P values by log-rank test)

Characteristic

Total n (events)

5-year DFS

Total n (cases)

5-year OS

Total

512 (73)

81.0

514 (36)

91.4

Age

<45

112 (14)

83.1

0.122

67 (2)

97.0

0.159

45–54

147 (19)

83.5

148 (8)

93.0

55–64

148 (20)

84.2

149 (15)

87.9

>65

105 (20)

66.5

105 (10)

0.874

Total

512 (73)

469 (35)

Body mass index (kg/m

)

<20

31 (4)

84.7

0.996

22 (2)

85.5

0.868

20–25

192 (24)

82.5

171 (11)

92.9

25–30

174 (23)

82.1

166 (10)

92.1

>30

74 (10)

82.5

71 (6)

88.7

Total

401 (61)

430 (29)

Tumor type

Invasive ductal

326 (51)

77.6

0.252

294 (26)

89.0

0.218

Invasive lobular

82 (11)

84.2

81 (5)

92.8

Other

104 (11)

87.2

94 (4)

94.0

Total

512 (73)

469 (35)

Menopausal status

Premenopausal

176 (19)

85.1

0.082

132 (4)

96.4

0.022

Postmenopausal

326 (51)

79.3

327 (30)

88.4

Total

502 (70)

459 (34)

268 (20)

89.6

<0.001

246 (7)

96.3

<0.001

188 (34)

74.5

171 (15)

88.8

26 (7)

70.7

21 (4)

77.9

30 (12)

48.9

31 (9)

66.0

Total

512 (73)

469 (35)

330 (33)

85.9

0.001

309 (9)

96.9

<0.001

182 (40)

73.0

160 (26)

80.1

Total

512 (73)

469 (35)

Grading

42 (2)

94.7

0.066

41 (1)

97.2

<0.001

318 (41)

82.9

294 (16)

93.9

104 (19)

74.1

88 (15)

76.1

Total

464 (62)

423 (32)

HER2/neu status

Negative

280 (34)

81.3

0.093

261 (19)

90.4

0.101

Positive

70 (14)

75.4

63 (9)

82.5

Total

350 (48)

324 (28)

ER status

Negative

129 (24)

77.1

0.055

109 (19)

78.5

<0.001

Positive

339 (39)

83.7

318 (14)

94.4

Total

468 (63)

427 (33)

880

J Cancer Res Clin Oncol (2008) 134:873–882

123

viduals, in comparison with two in controls, and demon-
strated that the ATM mutations that cause A-T in biallelic
carriers are breast cancer susceptibility alleles in monoall-
elic carriers (Renwick et al.

). This study found no sta-

tistically signi

Wcant diVerences in carrier frequencies

between cases and controls of 35 nonsynonymous missense
variants that do not cause A-T, including ATM 5557G>A
(D1853N). The heterozygous and homozygous carrier fre-
quency of the variant was found to be even higher in con-
trols (25.1 and 3.6%) in comparison with familial BC cases
(22.1 and 1.3%), but without reaching statistical signi

W-

cance. Moreover, there was no evidence of a di

Verence in

the relative risk between carriers under or over the age of
50 (Renwick et al.

Various other studies have found associations between

ATM polymorphisms that do not cause disease and breast
cancer. An association with a substantially elevated BC risk
has been indicated for the rare ATM c.7271T>G variant, as
well as the missense mutation p.S49C, in large population-
based studies (Bernstein et al.

; Stredrick et al.

It has previously been shown, among familial breast can-

cer patients in Finland, that the ATMivs38-8T>C polymor-
phism occurring in the cis position with 5557G>A
(D1853N) appears to be associated with bilateral breast
cancer (Heikkinen et al.

In other studies by Langholz et al. (

) and Tommiska

et al. (

) neither 5557G>A (D1853N) nor ins38-8T>C,

nor any haplotype containing these variants, was associated
with an increased risk of breast cancer or bilateral breast
cancer. The 5557G>A (D1853N) carrier frequency among
cancer-free controls was again found to be higher (37% for
heterozygotes and 5.4% for homozygous variants) in com-
parison with unselected breast cancer patients (35.5 and

4.1%; OR 0.89; P = 0.27). A German study showed a
higher proportion of node-positive BC patients in
p.D1853N homozygotes, and rare missense substitutions—
including p.D1853N—have been shown to be more preva-
lent in breast cancer patients than in healthy individuals
(7.9 versus 5.3% of alleles, P > 0.01) (Dork et al.

2001

When these di

Verent results are viewed together, the

5557G>A (D1853N) ATM variant, although not a classical
A-T mutation, might modulate the course, prognosis, and
survival of patients with BC. In the present study, the rele-
vance of the 5557G>A (D1853N) ATM polymorphism for
the occurrence and progression of breast cancer was there-
fore assessed.

The G to A variant at position 5557 (D1853N) of the

DNA double-strand break repair gene ATM was found with
a signi

Wcantly higher frequency among cancer-free controls

in comparison with unselected breast cancer patients, but
there was no statistically signi

Wcant association with

reduced disease-free survival or overall survival. Whether
this speci

Wc polymorphism, or another variation that could

be in linkage equilibrium with the change, is responsible
for this e

Vect is unclear. Theoretically the negatively

charged Asp at amino acid position 1853 is highly con-
served from clawed frogs (X. laevis) to humans [ATM
amino acid position 1843–1863: PYl-
iHdiLLqDtnesWRnlLS (according to protein sequence
NCBI: NP_000042) with the capital letters conferring to
highly conserved amino acids]. Data about the functional
impact of amino acid changes regarding this position is
presently unknown. It is possible that with an exchange to
an uncharged Asn this alteration may not be deleterious but
favourable for the structure and function of the ATM pro-
tein and for genomic integrity.

Table 4 continued

DFS disease-free survival, OS overall survival, pT tumor stage, pN nodal stage, ER estrogen receptor, PgR progesterone receptor

Characteristic

Total n (events)

5-year DFS

Total n (cases)

5-year OS

PgR status

Negative

173 (35)

74.1

0.001

156 (23)

81.5

<0.001

Positive

297 (28)

86.7

273 (10)

95.4

Total

470 (63)

429 (33)

Proliferation (Ki-67)

>15%

186 (13)

88.6

<0.001

178 (5)

95.9

<0.001

15%

217 (44)

72.3

192 (28)

80.9

Total

403 (57)

370 (33)

Rs1801516

404 (57)

81.4

0.793

406 (27)

91.6

0.811

99 (14)

79.1

99 (8)

91.1

9 (2)

75.0

9 (1)

88.9

Total

512 (73)

514 (36)

J Cancer Res Clin Oncol (2008) 134:873–882

881

123

Furthermore, the sample size of the cohort in the present

study is rather small, and there is a risk of false-positive
reporting (Wacholder et al.