Polymorphisms cMyc-N11S and p27-V109G and breast cancer risk and prognosis.

Abstract

cMyc and p27 are key genes implicated in carcinogenesis. Whether polymorphisms in these genes affect breast cancer risk or prognosis is still unclear. In this study, we focus on a rare non-synonymous polymorphism in cMyc (N11S) and a common polymorphism in p27 (V109G) and determine their role in risk and prognosis using data collected from the Ontario Breast Cancer Family Registry.
Risk factor data was collected at baseline on a large group of women (cases = 1,115 and population-based controls = 710) and clinical data (including treatment and follow-up) were collected prospectively by periodic review of medical records for a subset of cases (N = 967) for nearly a decade. A centralized pathology review was conducted. Unconditional logistic regression was used to determine the association of polymorphisms with breast cancer risk and the Cox proportional hazards model was used to determine their association with survival.
Our results suggest that while cMyc-N11S can be considered a putatively functional polymorphism located in the N-terminal domain, it is not associated with risk, tumor characteristics or survival. The p27-G109 allele was associated with a modest protective effect in adjusted analyses and higher T stage. We found no evidence to suggest that p27-V109G alone or in combination with cMyc-N11S was associated with survival. Age at onset and first-degree family history of breast or ovarian cancer did not significantly modify the association of these polymorphisms with breast cancer risk.
Further work is recommended to understand the potential functional role of these specific non-synonymous amino acid changes and a larger, more comprehensive investigation of genetic variation in these genes (e.g., using a tagSNP approach) in combination with other relevant genes is needed as well as consideration for treatment effects when assessing their potential role in prognosis.

Full text

Page 1

BioMed Central
Page 1 of 8
(page number not for citation purposes)
BMC Cancer
Open Access
Research article
Polymorphisms cMyc-N11S and p27-V109G and breast cancer risk
and prognosis
Jane C Figueiredo1,4, Julia A Knight1,4, Stewart Cho1,5, Sevtap Savas1, U
Venus Onay1, Laurent Briollais1,4, Pamela J Goodwin1,4,
John R McLaughlin1,3,4, Irene L Andrulis1,5 and Hilmi Ozcelik*1,2,5
Address: 1Samuel Lunenfeld Research Institute, Mount Sinai Hospital, Toronto, Ontario, Canada, 2Department of Pathology and Laboratory
Medicine, Mount Sinai Hospital, Toronto, Ontario, Canada, 3Division of Preventive Oncology, Cancer Care Ontario, Toronto, Ontario, Canada,
4Department of Public Health Sciences, Faculty of Medicine, University of Toronto, Ontario, Canada and 5Department of Laboratory Medicine
and Pathobiology, Faculty of Medicine, University of Toronto, Ontario, Canada
Email: Jane C Figueiredo - jane@mshri.on.ca; Julia A Knight - knight@mshri.on.ca; Stewart Cho - cho@mshri.on.ca;
Sevtap Savas - savas@mshri.on.ca; U Venus Onay - onay@mshri.on.ca; Laurent Briollais - laurent@mshri.on.ca;
Pamela J Goodwin - pgoodwin@mtsinai.on.ca; John R McLaughlin - john.mclaughlin@cancercare.on.ca;
Irene L Andrulis - andrulis@mshri.on.ca; Hilmi Ozcelik* - ozcelik@mshri.on.ca
* Corresponding author
Abstract
Background: cMyc and p27 are key genes implicated in carcinogenesis. Whether polymorphisms in these
genes affect breast cancer risk or prognosis is still unclear. In this study, we focus on a rare non-
synonymous polymorphism in cMyc (N11S) and a common polymorphism in p27 (V109G) and determine
their role in risk and prognosis using data collected from the Ontario Breast Cancer Family Registry.
Methods: Risk factor data was collected at baseline on a large group of women (cases = 1,115 and
population-based controls = 710) and clinical data (including treatment and follow-up) were collected
prospectively by periodic review of medical records for a subset of cases (N = 967) for nearly a decade.
A centralized pathology review was conducted. Unconditional logistic regression was used to determine
the association of polymorphisms with breast cancer risk and the Cox proportional hazards model was
used to determine their association with survival.
Results: Our results suggest that while cMyc-N11S can be considered a putatively functional
polymorphism located in the N-terminal domain, it is not associated with risk, tumor characteristics or
survival. The p27-G109 allele was associated with a modest protective effect in adjusted analyses and
higher T stage. We found no evidence to suggest that p27-V109G alone or in combination with cMyc-
N11S was associated with survival. Age at onset and first-degree family history of breast or ovarian cancer
did not significantly modify the association of these polymorphisms with breast cancer risk.
Conclusion: Further work is recommended to understand the potential functional role of these specific
non-synonymous amino acid changes and a larger, more comprehensive investigation of genetic variation
in these genes (e.g., using a tagSNP approach) in combination with other relevant genes is needed as well
as consideration for treatment effects when assessing their potential role in prognosis.
Published: 14 June 2007
BMC Cancer 2007, 7:99doi:10.1186/1471-2407-7-99
Received: 14 November 2006
Accepted: 14 June 2007
This article is available from: http://www.biomedcentral.com/1471-2407/7/99
© 2007 Figueiredo et al; licensee BioMed Central Ltd.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0),
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Page 2

BMC Cancer 2007, 7:99http://www.biomedcentral.com/1471-2407/7/99
Page 2 of 8
(page number not for citation purposes)
Background
Several studies have implicated cMyc and p27 in breast
cancer [1,2]. cMyc is amplified in 20–30% of breast
tumors and amplification has been correlated with pre-
menopausal status, specific tumor features (i.e., high
tumor grade, lymph-node metastases, large tumor size
and negative progesterone receptor status) and worse
prognosis [1]. Loss of p27 expression is also a common
event in breast cancers, and has been strongly associated
with high tumor grade and poor prognosis [2].
Whether genetic variation in these two genes affects cancer
risk or prognosis is not yet known. To our knowledge,
only seven studies have examined polymorphisms in
cMyc and p27 and all except one, which looked at haplo-
types in p27, have focused on either cMyc-N11S or p27-
V109G [3-9]. cMyc-N11S was recently reported by Wirten-
berger et al. (2005) to be associated with non-BRCA famil-
ial breast cancer [3], but has not been investigated as a
potential prognostic factor. A few studies have investi-
gated the association of the p27-V109G polymorphism
with cancer risk and progression, but results have been
inconsistent [4-7]. A previous publication by our group
reported no association for p27-V109G and breast cancer
risk in a smaller sample of breast cancer cases (N = 398)
and controls (N = 372) [8], but were unable to explore the
association with tumor characteristics and survival. Ma et
al. (2006) also showed no association of this polymor-
phism with breast cancer risk among Chinese women [9].
Another breast cancer study observed that the p27-V109G
polymorphism was correlated with nodal involvement,
but not with p27 tumor expression [4]. In univariate anal-
ysis among the node-negative group, V109G was signifi-
cantly associated with shorter disease-free survival [4].
In this study, we explore whether these non-synonymous
single nucleotide polymorphisms (nsSNPs) in cMyc
(N11S) and p27 (V109G) are important risk and prognos-
tic factors in breast cancer using a large, population-based
cohort of incident breast cancer with systematically col-
lected clinical data from the Ontario Familial Breast Can-
cer Registry (OFBCR).
Methods
Study design and subjects
Our study sample consisted of incident histo-pathologi-
cally confirmed cases of primary breast cancer from the
population-based OFBCR [10,11]. Recruitment of cases
and controls has been described previously [12-14]. In
brief, all cases were identified from the Ontario Cancer
Registry which registers >97% of all cases in the province.
All women aged 20–54 years who met the OFBCR defini-
tion for high genetic risk (family history of specific cancers
particularly breast and ovarian, early onset disease,
Ashkenazi ethnicity or a diagnosis of multiple breast can-
cer) were asked to participate by completing risk factor
questionnaires and providing a blood sample. A 25% ran-
dom sample of individuals in this age category who did
not meet the OFBCR definition, 35% of those aged 55–69
at high risk and 8.75% aged 55–69 at low risk were also
asked to participate. This multi-step sampling scheme
enriched the population for genetically predisposed indi-
viduals, which was an objective of the Ontario Familial
Breast Cancer Registry [11]. Response rates were as fol-
lows: consent to contact patients was 92%, response to
initial family history questionnaire was 65%, response to
risk factor questionnaires was 73% of all eligible, and
donation of a blood sample was 63% of all eligible. Less
than 2% died before initial contact.
To conduct case-control studies, the OFBCR also collected
unrelated, unaffected population controls (N = 710).
They were recruited by calling randomly selected residen-
tial telephone numbers throughout the same geographi-
cal region. Eligible controls were women with no history
of breast cancer and characteristics of the control popula-
tion have been shown to be representative of the target
population [14]. Approximately, 65% of identified eligi-
ble women returned questionnaires, and 63% of these
donated a blood specimen.
For the prognostic study, those patients who provided a
blood sample, had no prior malignancy (except for breast
carcinoma-in-situ, non-melanoma skin cancer or cervix
carcinoma-in-situ) and consented to retrieval of medical
records were followed prospectively for clinical outcomes.
The study methodology has been reported elsewhere [15].
In brief, clinical factors including stage, surgical treat-
ment, radiation therapy, chemotherapy, and hormonal
therapy were extracted from patient medical records by
registered nurses at each clinic using validated data collec-
tion forms. Tumor pathological factors including tumor
size, grade, number of positive lymph nodes, histologic
subtype, status of margins, lymphatic and blood vessel
invasion and hormone receptor status (estrogen, proges-
terone) were extracted from pathology reports and also
obtained from a review of histologic slides by study
pathologists. Medical records were reviewed annually for
the occurrences of new primary cancer, local-regional and
distant recurrences, changes in treatment and vital status.
Data were reviewed, verified and coded centrally. The
expected relationships between established prognostic
and predictive factors and clinical outcomes in breast can-
cer were observed [16]. Individuals who were missing
clinical data or who were ineligible due to refusal to pro-
vide a blood sample or consent to access medical records
did not differ on main tumor pathological variables, sur-
vival and other patient characteristics from those with
complete clinical follow-up data [16]. In total, there are
967 individuals with clinical follow up data.

Page 3

BMC Cancer 2007, 7:99 http://www.biomedcentral.com/1471-2407/7/99
Page 3 of 8
(page number not for citation purposes)
Approval for this study was obtained from the Research
Ethics Board of Mount Sinai Hospital and the University
of Toronto.
Polymorphism selection and genotyping
We identified non-synonymous polymorphisms in cMyc
and p27 that may be biologically relevant using the NCBI
dbSNP database by considering whether they were delete-
rious changes using two bioinformatics tools: SIFT [17]
and PolyPhen [18], which have been recently advocated
to be useful tools in identifying potentially causal variants
[19]. We identified cMyc-N11S (rs4645959 A>G dbSNP
Build 123). We also decided to genotype p27-V109G
(rs2066827 G>T dbSNP Build 125) which was not
deemed to be deleterious using either tool, but has been
previously studied in the published literature. The
5'nuclease Taqman assay was used and technical details
will be provided upon request from the authors. Water
control, internal controls and previously genotyped sam-
ples were included in each plate to ensure accuracy of gen-
otyping. Positive and negative controls were used in each
genotyping assay, and 10% of the samples were randomly
selected to be duplicated with 100% concordance. In
addition, a total of 236 cases were genotyped twice for
both polymorphisms in two different laboratories (The
Centre for Applied Genomics at the Hospital for Sick Chil-
dren, Toronto and H.O.'s lab at Mount Sinai Hospital,
Toronto) using the same technique, and concordance was
100%.
Statistical analysis
Pearson's chi-squared test or Fisher's exact test was used to
determine the association between genotypes and
patient/tumor characteristics. Variables were defined
according to standard convention in order to facilitate
comparison with other published studies. Age was dichot-
omized at 50 years of age to represent the approximate age
of menopause for stratified analysis. An individual was
considered to have a family history of cancer if she had a
first-degree female relative with breast or ovarian cancer at
the time of diagnosis (or date of entry for controls) since
first-degree family history has been shown to be valid and
reproducible by self-report for breast cancer [20]. Stage
was defined according to the American Joint Committee
on Cancer Staging System (1988) and T stage was catego-
rized as low (pT1, <2 cm) or high (pT2, pT3, pT4 or >= 2
cm), which is prognostically relevant in breast cancer.
Nodal status was categorized as no regional lymph nodes
affected (pN0) or at least one nodal metastasis. His-
topathological grade was defined according to the Scarff,
Bloom and Richardson definition (I: well differentiated,
II: moderately differentiated and III: undifferentiated).
Estrogen and progesterone receptor status (ER or PgR)
were classified as negative, equivocal or positive; equivo-
cal tumors were combined with positive tumors.
Genotype frequencies among the controls were tested for
Hardy-Weinberg equilibrium (HWE) using Pearson's chi-
square test with 1 df. We report association results for a co-
dominant model unless there were few variant homozy-
gotes. The associations between SNPs and breast cancer
risk were estimated as odds ratios (OR) and 95% confi-
dence intervals (95% CI) by unconditional logistic regres-
sion adjusting for age (years) and ethnicity (White,
Other). We stratified by age and family history to compare
risk estimates in each category. Tests for interaction by
inclusion of the corresponding product terms in logistic
models were non-significant (data not shown).
Contingency table analyses were used to examine the
associations between selected tumor characteristics and
genotypes among cases with complete clinical follow-up.
The primary clinical outcomes were time to distant recur-
rence and death. Survival time was calculated from date of
surgery to these endpoints, censoring at the date of last
contact or date of non-breast primaries. The Cox propor-
tional hazard model was used to evaluate the crude and
covariate-adjusted associations of factors with survival.
The final multivariable model included established prog-
nostic factors regardless of level of significance or con-
founding. Family history and ethnicity were also included
because of their potential association with genotype.
Graphical evaluation by Schoenfeld's residual plot indi-
cated that the proportional hazard assumption of the Cox
model could not be rejected for any of the covariates.
Stratified analyses by age or family history were not con-
ducted due to limited sample size.
To account for the sampling design in this study we also
conducted a weighted analysis using the inverse of the
sampling fractions as weights, and found no material dif-
ferences. Therefore, we report results only from the
unweighted analysis. Furthermore, our associations
remain unchanged if limited to non-Ashkenazi or non-
BRCA cases (data not shown).
For analyses of combined cMyc and p27 alleles we report
all possible combinations using the most common geno-
types as the reference category in order to obtain stable
risk estimates, and avoid testing all possible combina-
tions.
A priori power calculations using Quanto [21] showed
that our study (1,115 cases and 710 controls) had 75%
and 99% statistical power to detect an odds ratio of 1.5 for
polymorphisms with an allele frequency of 0.05 (approx-
imate for cMyc-N11S) and 0.25 (p27-V109G) assuming
the dominant model. In our prognostic study (967 cases),
with a follow-up of 6 years, recruitment of 3 years and
hypothesized failure rate of 15% we had 70% power for
the cMyc polymorphism to detect a relative risk of 2.0 and

Page 4

BMC Cancer 2007, 7:99http://www.biomedcentral.com/1471-2407/7/99
Page 4 of 8
(page number not for citation purposes)
80% for the p27 polymorphisms to detect a relative risk of
1.6 [22].
Missing data for given variables were reported in tables if
>10% and all tests were two-sided. We did not adjust for
multiple testing since this study focused on a few a priori
defined hypotheses.
Results
The majority of breast cancer cases were White pre-meno-
pausal women and slightly more than a third had a first-
degree family history of breast or ovarian cancer (Table 1).
There were 75 (7.8%) confirmed BRCA1 or 2 carriers
among the cases. There were no significant differences
between all cases and those with clinical follow-up data.
Controls were more likely to have children compared to
cases. Genotype frequencies for each polymorphism
showed no deviation from HWE among controls.
cMyc-N11S genotypes were not associated overall with
breast cancer risk (Table 2). The p27-G109 allele was asso-
ciated with a significant, but modest protective effect in
adjusted analyses [GT vs. TT: OR, 0.70 (0.52–0.93) and
GG vs. TT: OR, 0.83 (0.42–1.65)]. Age at onset (under or
over 50 years) and first-degree family history of breast or
ovarian cancer did not significantly modify the associa-
tion between these polymorphisms and breast cancer risk
(data not shown). The combined effect of the two poly-
morphisms, using the most common genotype as the ref-
erence, did not show any relationship with risk, and a test
for interaction between cMyc and p27 was non-signifi-
cant. Estimates of risk were similar in the entire sample as
when restricted to Caucasians.
Table 3 shows the relationships between selected tumor
characteristics with the polymorphisms. There were no
differences by genotypes for cMyc-N11S. The p27-G109
allele was associated with high T stage (p = 0.01) and pos-
sibly with nodal involvement (p = 0.07). Similar results as
shown for ER status were observed if the data were ana-
lyzed for PgR or combined ER and PgR status (data not
shown).
There was no association between any of the polymor-
phisms or their combined alleles and either distant-recur-
rence free survival or overall survival (Table 4). Tests for
interaction between cMyc and p27 allele in these models
were non-significant. There was no association between
polymorphisms and survival within treatment groups:
radiation, chemotherapy or hormonal therapy (data not
shown). There were also no material differences if we
examined these associations stratified by nodal status or
ER/PgR status (data not shown).
Discussion and conclusion
There has been great interest in understanding the role of
cMyc and p27 amplification/expression in breast cancer
risk and prognosis, but surprisingly little on the role of
polymorphisms. Our study fills this gap by presenting
findings on the role of two specific polymorphisms in
these genes. Our data suggests that the p27-G109 allele
may confer a protective effect against breast cancer. This
observation needs to be confirmed by other breast cancer
studies, since there is disagreement in the published liter-
ature about its potential role. A previous publication by
our group showed no association with a smaller sample
size [8], as did a study by Ma et al. (2006) among Chinese
women (cases = 368, controls = 467) [9]. Furthermore,
one case-control study of prostate cancer (cases = 92, con-
trols = 106) found a positive association with this poly-
morphism especially in cases under 66 at the time of
diagnosis [7], while a family-based study of hereditary
prostate cancer (N = 188 families) that resequenced p27
did not confirm this association, but identified the -32T
polymorphism in the promoter site as a risk factor espe-
cially among cases diagnosed under age 65 [6]. Another
study of oral squamous cell carcinoma did not find an
association between p27-V109G and risk (cases = 713,
controls = 1,224), but did show an association with over-
all tumor stage [5]. In this study, we also show that p27-
V109G is associated with T stage and possibly nodal sta-
tus, which was previously reported by Schondorf et al. [4].
Furthermore, our data do not suggest that the p27-G109
allele is associated with breast cancer survival, which con-
firms the overall null association with breast cancer sur-
vival as reported previously [4]. The latter study; however,
did show a significant association with distant recurrence
free survival among the node-negative tumors (N = 46).
We did not confirm this finding with a larger, but still lim-
ited sample of node-negative cases.
Previous studies have shown that reduced p27 expression
correlates with poor clinical outcomes, invasiveness, poor
prognosis, high tumor grade and progression in breast
cancer [2]. The missense V109G change may alter the
interaction between p27 and its negative regulator p38jab1
because it is located in the interaction surface [23]. Since
p27 is rarely mutated and decreased protein levels are
found in tumors, it can be hypothesized that this decrease
may be the result of changes in degradation of p27. There-
fore, the V109 allele may alter p27 affinity for p38jab1 and
thereby modify p27 degradation. However, since data
from a homology-based bioinformatic tool suggests that
this amino acid substitution is not deleterious [24], this is
a hypothesis that needs to be confirmed in functional
studies.
Our results do not confirm the recent findings by Wirten-
berger et al. (2005) [3], who found that an increased risk

Page 5

BMC Cancer 2007, 7:99http://www.biomedcentral.com/1471-2407/7/99
Page 5 of 8
(page number not for citation purposes)
for breast cancer associated with the S11 allele (OR = 1.54,
95% CI, 1.05–2.26) and a stronger effect among women
> 50 years (OR = 2.24, 95% CI, 1.20–4.21) [3]. This lack
of replication could be due to differences in study design
including the selection of the study population. Wirten-
berger et al. focused on non-BRCA1/2 familial cases
selected from two countries (n = 349 Polish; n = 356 Ger-
man) and non-BRCA1/2 healthy controls (n = 441 Polish;
n = 655 German) collected from clinics from 1997–2003.
The current investigation is this population-based study of
Table 1: Characteristics of Breast Cancer Cases and Population Controls in the OFBCR
Characteristic All Cases (N = 1,115)Cases with Clinical Data (N = 967)All Controls (N = 710)
Age1
Mean ± SD
Menopausal Status2
Pre-
Post-
Ethnicity
White
Non-White
Parity
Nulliparous
1–2 child
> 3 children
First-degree Family History3
No
Yes
OFBCR Defined Genetic Risk Case4
No
Yes
Histology
Infiltrating Ductal No Special Type
Lobular
Other Special Type5
T Stage
pT1 (<2 cm)
pT2 (2–5 cm)
pT3/pT4 (>5 cm)
pTx (not accessible)
# Positive Lymph Nodes
None
1–3
≥4
Nx (not accessible)
Grade
I (well differentiated)
II (moderately differentiated)
III (poorly differentiated)
Lymphatic Vessel Invasion
Negative
Positive
Estrogen Receptor Status
Negative
Equivocal
Positive
Progesterone Receptor Status
Negative
Equivocal
Positive
48.6 ± 9.0 48.9 ± 9.2 48.5 ± 9.1
828 (75.0%)
276 (25.0%)
710 (73.7%)
253 (26.3%)
497 (70.2%)
211 (29.8%)
1062 (95.3%)
53 (4.7%)
899 (93.0%)
68 (7.0%)
653 (95.2%)
33 (4.8%)
201 (19.5%)
620 (60.2%)
209 (20.3%)
204 (21.5%)
562 (59.2%)
183 (19.3%)
106 (15.0%)
434 (61.2%)
168 (23.8%)
731 (65.6%)
384 (34.4%)
618 (63.9%)
349 (36.1%)
641 (90.3%)
69 (9.7%)
311 (27.9%)
804 (72.1%)
281 (29.1%)
686 (70.9%)
NA
--824 (90.2%)
65 (7.1%)
25 (2.7%)
NA
--607 (64.0%)
281 (29.6%)
34 (3.6%)
26 (2.7%)
NA
--538 (56.6%)
243 (25.6%)
113 (11.9%)
56 (5.9%)
NA
--194 (21.5%)
351 (38.8%)
359 (39.7%)
NA
-- 582 (66.1%)
299 (33.9%)
NA
--230 (24.5%)
49 (5.2%)
659 (70.3%)
NA
--259 (27.8%)
55 (5.9%)
618 (66.3%)
NA
1 age at diagnosis for cases and age at interview for controls; 2 peri-menopausal women are grouped with pre-menopausal women; 3 self-reported
cancer histories of breast or ovarian cancer; 4 criteria used by the OFBCR to oversample more informative cases in order to enrich registry for
genetically predisposed individuals (see methods); 5 includes medullary, tubular, cribriform, micropapillary, mucinous, metaplastic; NA, not
applicable