Journal of Cancer 2013, Vol. 4
Jo ou ur rn na al l o of f C
2013; 4(2): 104-116. doi: 10.7150/jca.5002
Ca an nc ce er r
In Vitro Assessment of the Inflammatory Breast Cancer
Cell Line SUM 149: Discovery of 2 Single Nucleotide
Polymorphisms in the RNase L Gene
Brandon T. Nokes1, Heather E. Cunliffe2, Bonnie LaFleur3, David W. Mount1,4, Robert B. Livingston1,5,
Bernard W. Futscher1,6, Julie E. Lang7
1. University of Arizona Cancer Center;
2. Translational Genomics Research Institute, Phoenix, Arizona;
3. Division of Epidemiology and Biostatistics;
4. Bioinformatics Shared Services;
5. Department of Medicine, Division of Hematology and Oncology, University of Arizona College of Medicine, Tucson, Arizona;
6. Department of Pharmacology and Toxicology, University of Arizona College of Medicine, Tucson, Arizona;
7. Norris Comprehensive Cancer Center, University of Southern California, Los Angeles, California.
Corresponding author: Julie E. Lang, email: Julie.Lang@med.usc.edu. Address: 1510 San Pablo Street, Suite 412, Los Angeles, CA 90033
FAX: 323-865-3539 PHONE: 323-422-5772.
© Ivyspring International Publisher. This is an open-access article distributed under the terms of the Creative Commons License (http://creativecommons.org/
licenses/by-nc-nd/3.0/). Reproduction is permitted for personal, noncommercial use, provided that the article is in whole, unmodified, and properly cited.
Received: 2012.08.08; Accepted: 2012.11.12; Published: 2013.01.10
Background: Inflammatory breast cancer (IBC) is a rare, highly aggressive form of breast
cancer. The mechanism of IBC carcinogenesis remains unknown. We sought to evaluate
potential genetic risk factors for IBC and whether or not the IBC cell lines SUM149 and
SUM190 demonstrated evidence of viral infection.
Methods: We performed single nucleotide polymorphism (SNP) genotyping for 2 variants of
the ribonuclease (RNase) L gene that have been correlated with the risk of prostate cancer
due to a possible viral etiology. We evaluated dose-response to treatment with interfer-
on-alpha (IFN-α); and assayed for evidence of the putative human mammary tumor virus
(HMTV, which has been implicated in IBC) in SUM149 cells. A bioinformatic analysis was
performed to evaluate expression of RNase L in IBC and non-IBC.
Results: 2 of 2 IBC cell lines were homozygous for RNase L common missense variants 462
and 541; whereas 2 of 10 non-IBC cell lines were homozygous positive for the 462 variant (p=
0.09) and 0 of 10 non-IBC cell lines were homozygous positive for the 541 variant (p = 0.015).
Our real-time polymerase chain reaction (RT-PCR) and Southern blot analysis for sequences
of HMTV revealed no evidence of the putative viral genome.
Conclusion: We discovered 2 SNPs in the RNase L gene that were homozygously present in
IBC cell lines. The 462 variant was absent in non-IBC lines. Our discovery of these SNPs
present in IBC cell lines suggests a possible biomarker for risk of IBC. We found no evidence
of HMTV in SUM149 cells. A query of a panel of human IBC and non-IBC samples showed no
difference in RNase L expression. Further studies of the RNase L 462 and 541 variants in IBC
tissues are warranted to validate our in vitro findings.
Key words: inflammatory breast cancer, SUM149, HMTV, interferon-alpha, MMTV, RNase L.
Inflammatory breast cancer (IBC) is a rare form
of breast cancer, accounting for only 5% of breast
cancer cases annually in the United States. Survival
outcomes are improving for patients with
Journal of Cancer 2013, Vol. 4
non-inflammatory breast cancer (non-IBC), but re-
main poor for patients with IBC despite aggressive
multimodal treatment 1-3. IBC, the most lethal form of
breast carcinoma, is characterized by distinct clinico-
pathologic features, including rapid disease progres-
sion and onset of swelling, enlargement of the breast,
skin tenderness, induration, edema, warmth, and er-
ythema, which is commonly combined with peau
d’orange 4-8. The 5-year survival rate for patients pre-
senting with non-metastatic IBC is only about 40%,
even with modern multidisciplinary therapy 1,7,9-12.
Treatment factors associated with improved pa-
tient survival include use of multimodality therapeu-
tic strategies, including chemotherapy, modified rad-
ical mastectomy, and postmastectomy radiation 12-14.
Although excellent rates of locoregional control are
achievable 6, mortality is usually related to systemic
recurrence. Because of the rarity of IBC and the in-
herent difficulty in obtaining tumor tissue from IBC
patients, who may lack a tumor mass at presentation
and typically receive upfront systemic chemotherapy,
few studies have been performed to characterize its
molecular biology 15-17. Understanding the distinct
biologic and molecular behavior of IBC is likely to
provide insight into carcinogenic mechanism(s) and
aid discovery of novel targets for future treatment
Human mammary tumor virus (HMTV), a hu-
man homologue to the mouse mammary tumor virus
(MMTV) has been proposed by Pogo et al. to play a
role in IBC 18. MMTV predictably leads to tumor for-
mation in mice 18; however, HMTV may or may not be
associated with human cancers 18-26. Recently, Pogo et
al. reported that HMTV sequences were detected in
71% of IBC cases in American women and, in turn,
were associated with a more malignant breast cancer
phenotype than non-IBC 26. Importantly, these find-
ings have not been independently validated, and the
significance of these putative viral DNA sequences in
humans remains unclear 18-26.
The biology of IBC is distinct from that of
non-IBC, in that IBC progresses much more rapidly
(weeks to months, rather than months to years) and
has unique clinical features, such as skin erythema,
warmth, lack of a discrete mass, and often the pres-
ence of dermal lymphatic invasion. Given the mark-
edly different clinical presentation of IBC (as com-
pared with non-IBC) along with diverse viral etiolo-
gies implicated in other types of cancer, we hypothe-
sized that IBC may have a viral cause, possibly in-
volving putative HMTV infection. Given the scarcity
of available human IBC tissue samples available for
research at any single institution, we chose to leverage
the 2 commercially available immortalized IBC cell
lines SUM149 (the most widely used cell line model in
IBC studies 27-29) and SUM190 as appropriate models
for testing our hypothesis. Cell culture experiments
were performed using SUM149 due to the extremely
fastidious nature of manipulating SUM190 cells in
Considering HMTV is still a putative virus, we
assessed our in vitro IBC model for carrying a viral
infection by multiple modalities. First, we sought to
define single nucleotide polymorphisms (SNPs) for
the ribonuclease (RNase) L gene, the product of which
combats viral infection by degrading viral RNA and
inducing apoptosis of infected cells 30-33. Several
non-synonymous coding SNPs have already been
associated with high risk of prostate cancer 30,31 and
further reported to be associated with a putative on-
cogenic viral infection 32. We chose to investigate the
significance of two common missense variants R462Q
(rs486907) and E541D (rs627928) reported to associate
with the incidence rate of sporadic prostate cancer in
several studies. Given the utility of these RNase L
SNPs as high-risk biomarkers for susceptibility to
prostate cancer, we sought to determine whether IBC
cell lines contained the same high-risk genotypes,
potentially serving as an indicator of genetic suscep-
tibility to viral infection and IBC carcinogenesis.
Moreover, the RNase L gene is a downstream effector
of the type 1 interferon pathway, which is utilized in
mitigating viral infections (Figure 1), and altered
function of this gene product may render cells more
susceptible to cancer development, as the normal
function of type 1 IFN-induced RNase L expression is
to trigger destruction of viral RNA 30-32.
An allelic discrimination assay for the down-
stream interferon (IFN) effector RNase L was utilized
to detect 2 SNP variants associated with cancer risk,
impacting amino acids 462 and 541 34. We hypothe-
sized that variations in genotypic frequency of these
RNase L SNPs may indicate an association of in-
creased risk of a viral infection and potential etiology
within IBC vs. non-IBC cell lines. We also conducted a
search on the National Center for Biotechnology In-
formation (NCBI) Gene Expression Omnibus (GEO)
website to assess differences in expression of RNase L
between IBC and non-IBC tumor tissues 35.
Our second approach to determining the possi-
bility of viral infection of IBC cells, we assayed for a
selective decrease in proliferation of SUM149 cells in
response to interferon-alpha (IFN-α) treatment (a
naturally occurring antiviral cytokine). Interferons
(IFNs) are effective molecules for studying the biology
of viral cause: in vertebrates, they are produced nat-
urally by many nucleated cells in response to viral,
parasitic, and tumor-derived challenges. IFNs assist
Journal of Cancer 2013, Vol. 4
the immune response by inhibiting viral replication
within host cells, activating natural killer cells, and
increasing antigen presentation to lymphocytes 36,37.
Moreover, the IFN-mediated 2-5A pathway is a key
innate response to viral infection via RNase L-induced
viral RNA degradation, as well as a mediator of
Lastly, to investigate findings by Pogo et al 18, we
sought whether determine whether HMTV elements
are present in the IBC cell line SUM149 using RT-PCR
and Southern blot analysis.
Figure 1: Schematic representation of the innate IFN-RNase L pathway as influenced by viral infection. Viral infection
triggers the release of Type I interferons, which act on the infected cell in an autocrine fashion. IFN signaling initiated the Oligoadenylate
Synthase (OAS)/RNase L pathway for nucleic acid degradation (Adapted from Silverman 2007) .
MATERIALS AND METHODS
Allelic Discrimination Assay We tested for the
presence of RNase L variants R462Q and E541D as
potential indicators of a genetic risk factor for IBC,
using the TaqMan allelic discrimination assay (Life
Technologies, Carlsbad, CA). Primer probes were
identical to those designed by Shook et al. 31. The
primers and probes for R462Q were: forward primer
verse primer 5’-TGCA- GATCCTGGTGGGTGTA-3’,
and probes 5’-VIC-CAGGACATTTCGGG-
CAA-MGB and 5’-FAM-CAGGACATTTTGGGCAA-
MGB. The primers and probes for E541D were: for-
ward primer 5’-TCTATGTGGTAAAGAAGGGAAGC
A-3’, reverse primer 5’-TTGAACCACCTCTTCATT
ACTTTGAG-3’, and probes 5’-VIC-TTTCAGATCCT-
CAAAT-MGB and 5’-FAMTTTCAGCTCCTCAAAT-
We extracted genomic DNA from 12 cell lines
using the QIAamp DNA purification kit (Qiagen, Va-
lencia, CA). We used 20 ng of DNA per reaction in
96-well plate format. All reactions were conducted in
triplicate using an ABI 7500 Fast RT PCR System and
analyzed using SDS 2.0 software (Life Technologies).
P values for the cell line SNP comparisons were cal-
culated using the online contingency table from Vas-
Journal of Cancer 2013, Vol. 4
Bioinformatics Analysis A comparative search
performed on the
(http://www.ncbi.nlm.nih.gov/geo/) in order to
assess differences in expression of RNase L between
IBC and non-IBC tumor tissues. The GSE5847 entry
originally published by Boersma et al. provided a
suitable data set with 15 IBC and 35 non-IBC tumor
samples, with 2 normal breast tissue samples as a
control 35. The decision to use this data set was based
on lack of available entries for IBC samples, as well as
the consistent nature of the expression profile of
stromal tissue relative to tumor samples. Moreover,
because we were looking for a DNA-based marker of
genetic susceptibility, and tumor cells are highly het-
erogeneous, we selected the stromal data set for this
Statistical Considerations. To assess SNP prev-
alence between IBC and non-IBC breast stromal tissue
samples, a computational script was written in the
language R based on the sample size analysis rec-
ommendations made by Pfeiffer et al., 40. The script
was then independently verified using the Bioinfor-
matics Institute’s (BII) Online Sample Size Estimator
(http://osse.bii.a-star.edu.sg/). Using a case-control
design, based on the lowest minor allele frequency
(MAF) for SNP rs486907; we based these estimates on
the MAF for rs486907, because the lower allele fre-
quency will require a larger sample size. We calcu-
lated the required sample sizes to test a significant
risk ratio between normal samples and IBC or
non-IBC samples, as well as the risk ratio between IBC
and non-IBC samples.
Cell Culture SUM149 cells, BT474 cells, and
MDA-MB-231 cells were used in this study. All cell
lines were acquired from the American Type Culture
Collection (ATCC, Manassas, VA) except the SUM149
cell line sourced from Asterand (Detroit, MI). All cell
lines were subjected to genotyping with the ABI
Identifiler Assay (Life Technologies) for validation of
cell line identity. SUM149 cells were grown in Ham’s
F-12 medium supplemented with 5% heat-inactivated
fetal bovine serum (FBS)
ics/antimycotics (Invitrogen, Carlsbad, CA). BT474
cells were grown in Roswell Park Memorial Institute
(RPMI) medium supplemented with 10% FBS and 1%
antibiotics/antimycotics (Invitrogen). MDA-MB-231
cells were grown in Dulbecco’s modified Eagle’s me-
dium (DMEM) supplemented with 10% FBS and 1%
antibiotics/antimycotics (Invitrogen). All cell lines
were maintained in a humidified incubator with 5%
CO2 atmosphere at 37o C. Cells were plated in tripli-
cate at a density of 5,000 cells per well in 96-well tissue
culture plates and grown to greater than or equal to
30% confluence at the time of treatment. Purified
was NCBI GEO site
and 1% antibiot-
DNA from the SUM190 cell line was graciously pro-
vided by Dr. Cunliffe from the Translational Ge-
nomics Research Institute (TGen) in Phoenix, AZ.
IFN-α Treatment IFN-α (Imgenex, San Diego,
CA) was dissolved in phosphate-buffered saline (PBS)
with 5% fetal bovine serum to a final stock concentra-
tion of 100 µg/ml. Before treatment, the complete
medium was removed and the cell monolayers were
rinsed once with PBS. Cells were then treated with 0,
500, 1000, 2500, and 5000 U/ml of IFN-α for 24 hours
and 48 hours. At each time point, we removed the
IFN-α, rinsed the cells with PBS, and evaluated cell
proliferation. For IFN-α block experiments, we
pre-incubated the cells for 15 minutes with 1 µg/ml
IFN-α specific antibody (Sigma, St. Louis, MO).
Proliferation Assays Cell proliferation was
evaluated in triplicate for all treatments with the
CyQUANT cell proliferation assay kit (Invitrogen),
per the manufacturer’s instructions. Briefly, IFN-α
treatment was removed and the cells washed once
with PBS. 100 µl CyQUANT cell proliferation assay
working reagent was then added to the cells and in-
cubated for 1 hour at 37o C. Emission (directly pro-
portional to proliferation) was recorded on the BioTek
FLx800 plate reader (Winooski, VT). Proliferation data
was analyzed by one-way analysis of variance using
the Tukey multiple comparisons test and GraphPad
Prism software (GraphPad Prism Software, Inc., La
Jolla, CA). All assays were performed in at least trip-
licate and were analyzed together at the same time
Primer Design and PCR Putative HMTV se-
quence were obtained from the National Center for
cific primers were designed to the env/LTR and late
LRT regions. The primers used for PCR of the
env/LTR region were as follows: 5’TCT GCG TTA
CAC CAC TAC CG 3’ and 5’TGA ACT CGA CCT
TCC TCC TG 3’. The primers used for PCR of the late
LTR region were as follows: 5'ACC TTC CTC CTG
AGC CTA GC 3’ and 5’TTT ATT AGC CCA ACC TTG
CG 3’. For reverse-transcription polymerase chain
reaction (RT-PCR), total RNA was isolated from
SUM149, BT474 and MDA-MB-231 cells using the
RNeasy purification kit (Qiagen) and cDNA produced
using the first strand cDNA synthesis kit from MBI
Fermentas (Glen Burnie, MD). To conduct PCR, we
used MBI Fermentas reagents and evaluated products
on a 1% Tris-borate-EDTA (TBE)/agarose gel. Gel
images were acquired using a Gel Logic 200 Imaging
System with Kodak 1D 3.6 software (Carestream Mo-
lecular Imaging, Rochester, NY).
Cloning and Sequencing PCR products of interest
Journal of Cancer 2013, Vol. 4
were excised from agarose gels and the DNA purified
using the QIAquick gel extraction kit (Qiagen). Puri-
fied PCR products were then ligated into the pGEM-T
Easy Vector (Promega, Madison, WI) overnight at 4o C
and transformed into TOP10 Chemically Competent
E. coli (Invitrogen). Blue-white colony selection was
used to screen for recombinant plasmids containing
ligated PCR fragments (recombinants resulting in
disrupted β-galactosidase function, preventing me-
tabolism of X-gal substrate). Recombinant plasmids
were evaluated by Sanger sequencing, performed by
the Genomic Analysis and Technology Core Facility at
the BIO5 Institute at the University of Arizona (Tuc-
Western Blot Analysis Proteins were resolved
by sodium dodecyl sulfate-polyacrylamide gel elec-
trophoresis (SDS-PAGE) on a 4% to 20% gradient
minigel (Bio-Rad, Hercules,
Mini-PROTEAN 3 Cell, run at 100 V for 1.5 hours at
25° C. 15 μg of total protein from cytoplas-
mic/membrane extracts were resolved and trans-
ferred to nitrocellulose membranes using a Mini
Trans-Blot Electrophoretic Transfer Cell (Bio-Rad).
Efficient transfer of proteins was confirmed by
SYPRO Ruby protein blot stain (Bio-Rad) and Kalei-
doscope molecular weight markers (Bio-Rad). Nitro-
cellulose membranes were blocked for 2 hours in
blocking buffer (4.0% bovine serum albumin [BSA], 10
mM PBS, 0.05% Triton X-100, pH 7.4) at 25° C. Mem-
branes were next incubated with antihuman-specific
rabbit monoclonal IFN receptor alpha IFNAR1 anti-
body (ab45172, Abcam, Cambridge, MA), diluted
1:20,000 v/v in blocking buffers overnight at 4° C with
gentle agitation. Following incubation with primary
antibody, membranes were washed 3 times in 10 mM
PBS, 0.05% Triton X-100, pH 7.4 then incubated in
donkey anti-rabbit secondary antibody with alkaline
phosphatase (AP) conjugate (1:1000 v/v; Jackson
ImmunoResearch, West Grove, PA) at room temper-
ature for 1 hour. Nitrocellulose membranes were
washed 3 times in 10 mM PBS, 0.05% Triton X-100 pH
7.4; and protein products visualized after 1 to 5
minutes following addition of AP substrate (MBI
Fermentas). Quantification of a specific protein band
was established with GelQuant.NET software pro-
vided by biochemlabsolutions.com. In the densitom-
etry analysis, relative pixel density of the IFN receptor
wells was normalized to that of β-actin.
Southern Blot Analysis 10 µg purified SUM149
genomic DNA was digested with 5U FastDigest
BamHI restriction enzyme (MBI Fermentas) at 37o C
for 30 minutes, then heat inactivated at 80o C for 5
minutes. DNA was subsequently extracted with an
equal volume of isopropyl alcohol and resuspended
CA), using the
in 10 µl nuclease-free water at room temperature for
15 minutes. The entire volume was electrophoresed
through a 1% agarose gel in 1X TBE buffer for 6 hours
at 3 V/cm (until the bromphenol blue marker reached
the bottom of the gel). Halfway through the gel elec-
trophoresis, we loaded a positive control synthetic
fragment (IDT Technologies, San Diego CA) encoding
for a 172 base-pair region of the HMTV env region 5’
TAT GAT TTT ATC TGC GTT ACA CCA CTA CCG
TAT AAT GCT TCT GAG AGC TGG GAA AGA ACC
AAG GCT CAT TTA CTG GGC ATT TAA AAT AAC
AAT GAG ATT TCA TAT AAC ATA CAA AAA TTA
ACC AAC CTA ATT AGT GAT ATG AGC AAA CAA
CAT ATT GAC GCA GTG GAC CTT A 3’. Before
blotting, the gel was rinsed in deionized water, incu-
bated in denaturing solution for 30 minutes at room
temperature with shaking, rinsed again in deionized
water and incubated in neutralization for 15 minutes
at room temperature with shaking. We repeated this
procedure and then transferred the DNA by tradi-
tional upward capillary action for 18 hours at room
temperature. After transfer, the membrane was
washed in 2X SSC solution to remove any residual
agarose, dried at room temperature, and fixed by UV
crosslinking for 2 minutes.
Probe Synthesis HMTV env DNA (500 ng) was
labeled using the Biotin DecaLabel DNA Labeling Kit
(MBI Fermentas). The HMTV env DNA template was
combined with 5x decanucleotide reaction buffer and
nuclease-free water. The
pulse-spun for 5 seconds, incubated in a boiling water
bath for 10 minutes, and quickly cooled on ice. Biotin
Labeling Mix and 5U of Klenow fragment were added
and the reaction incubated for 1 hour at 37° C. The
reaction was stopped by adding 1μl 0.5M EDTA, pH
8.0. The labeled DNA was directly used for hybridi-
Hybridization and Detection The membrane
was incubated in a pre-hybridization solution con-
taining 5X SSC/5X Denhardt’s, 0.5% SDS, 100 µg/ml
nonspecific DNA (Sigma) at 42° C for 4 hours with
agitation in a ProBlot 12 hybridization oven (Labnet
International, Woodbridge, NJ). During this time, the
biotin-labeled probe was denatured at 100° C for 5
minutes and chilled on ice. The denatured probe was
added to the pre-hybridization solution to obtain a
final probe concentration of 100 ng/ml and incubated
it overnight at 42° C with shaking. After hybridiza-
tion, the membrane was washed twice with 2X SSC,
0.1% SDS for 10 minutes at room temperature, then
twice with 0.1X SSC, 0.1% SDS for 20 minutes at 65º C.
Excess liquid was removed from the membrane by
briefly placing it on filter paper. The biotin-labeled
DNA was detected using the Biotin Chromogenic
tube was vortexed,
Journal of Cancer 2013, Vol. 4
Detection Kit (MBI Fermentas), according to the
manufacturer’s directions. Color development started
to be visible after 1 hour. We acquired both gel and
membrane images using the Gel Logic 200 Imaging
System with Kodak 1D 3.6 software (Carestream Mo-
lecular Imaging, Rochester, NY).
Analysis of R462 and E541 SNPs in human IBC
and non-IBC cell lines
SNP genotyping was first performed for RNase
L variants R462Q and E541D in 2 IBC and 10 non-IBC
cell lines with results shown in Figure 2 and Table 1.
The SUM149 and SUM190 IBC cell lines were homo-
zygous G at rs486907 (homozygous Arginine at resi-
due 462) and homozygous G at rs627928 (homozy-
gous Glutamic acid at residue 541). The 541 GG and
462 AA genotype is the same as those previously re-
ported to be associated with increased risk for spo-
radic prostate cancer 30,31. Of note, both IBC cell lines
displayed 462 GG and 541 GG homozygous geno-
types, of which, 462 GG is not associated with prostate
cancer development, whereas 541 GG is. However,
these genotypes did differ significantly from non-IBC
genotypes at these residues suggesting the possibility
of a novel risk allele for IBC. All but two of the
non-IBC cell lines were either heterozygous or ho-
mozygous for the A allele at rs486907 (residue 462),
and all 10 non-IBC cell lines were heterozygous or
homozygous for the T allele at rs627928 (residue 541).
Analysis by the Fisher’s exact probability test yielded
a two-tailed p value of 0.09 for the 462 variant and
0.015 for the 541 variant, as calculated using Vas-
sarStats (Vassarstats.net) contingency tables (Table 1).
Figure 2: Single nucleotide polymorphism (SNP) allele
discrimination scatter plot forribonuclease (RNase) L
variants (A) 462 (rs486907) and (B) 541 (rs627928). The
Y-axis indicates detection of the 462Q(A) and 541D (B) allele by a
FAM reporter (diamonds), and the X-axis indicates detection of
the 462R (A) and 541E (B) allele by a VIC reporter (circles).
Heterozygous genotypes displayed amplified copies of both alleles
and were found in an intermediate position between homozygous
mutants and homozygous wildtype cell lines (triangles). The No
Template Control (NTC) wells are denoted as squares. *Non-IBC
cell lines, **Corresponds to IBC cell lines.
Table 1. Human ancestral alleles for RNase L are “G” at SNP 462 (rs486907) and “G” at SNP 541 (rs627928). Amino acid residue
abbreviations: R, Arginine; Q, Glycine; E, Glutamic Acid; D, Aspartic acid. Data from dbSNP Reference SNP build 137. The “Normal Cell
Line” and “BRCA+” cell lines are both primary cell lines established by the Lang laboratory from prophylactic mastectomy specimens from
individuals without breast cancer; these cell lines include both stromal and breast parenchymal components.
Journal of Cancer 2013, Vol. 4
An evaluation of genotype frequency for these
RNase L SNPs in IBC tumors is required in order to
define a statistically significant correlation with po-
tential risk of IBC in patient samples. Due to the rarity
of IBC, and lack of large biorepositories for this dis-
ease, we calculated a statistical estimate of how many
cases and controls would be required to perform a
comprehensive analysis of this nature. The dbSNP
database lists the minor allele frequency (MAF) for
rs486907 to be 24% and the MAF for SNP rs627928 to
be 48%. Using a case-control design, based on the
MAF for SNP rs486907 (the lower allele frequency will
require a larger sample size), we calculated the re-
quired sample sizes to test a significant risk ratio (es-
timated for a 2-fold increased risk) between healthy
patient samples (without cancer) and IBC and
non-IBC patient samples, as well another 2-fold risk
ratio between IBC and non-IBC samples. The mini-
mum sample size required to have 80% power to de-
tect at least a 2-fold increase in risk between IBC and
non-IBC cancer is 356 samples of each cancer; further,
in order to test a difference of at least 2-fold level of
risk between either IBC or non-IBC cancers to healthy
individuals will necessitate 160 healthy (control)
samples. This resulted in a total of 872 defined cancer
type/control samples being required to evaluate the
significance of these SNPs in IBC. This analysis could
be accomplished should major IBC investigators
choose to pool case/control resources to do so.
Our bioinformatic query using NCBI GEO data
set GSE5847 regarding the differences in gene expres-
sion of RNaseL between human IBC and non-IBC
samples found no significant difference in gene ex-
pression between the two groups within the stroma of
human breast tissue samples (Figure 3).
Investigating the possibility of viral infection-
mediated response to IFN-α α treatment in
We next sought to determine whether SUM149
cells showed an altered proliferative potential fol-
lowing IFN-α stimulation compared to non-IBC cell
lines (Figure 4). We observed a dose- and
time-dependent decrease in cell proliferation of
SUM149 cells treated with IFN-α. At 24 hours, cell
proliferation decreased by 32% relative to controls at
the highest dose of 5000 U/ml IFN-α (p < 0.001). At 48
hours, cell proliferation continued to decrease at this
dose, by 41% relative to controls (p < 0.01). We did not
observe a corresponding decrease in SUM149 prolif-
eration after pre-incubation of IFN-α with a specific
neutralizing antibody (Figure 5, proliferation de-
creased by only 5% at this dose relative to controls).
The results of the same INF-α response proliferation
assay performed with 2 non-IBC cell lines (BT474 and
MDA-MB-231) are shown in the lower panel of Figure
4. In contrast to the dose- and time-dependent de-
crease in cell proliferation of SUM149 cells, we did not
see any decrease in cell proliferation in the
MDA-MB-231 cells, and saw only a weak decrease in
the BT474 cells at 24 hours that was not evident at 48
hours. Moreover, only the SUM149 cell line demon-
strated a direct and specific response to IFN-α treat-
ment (Figure 5). Note that each of the 3 cell lines ex-
pressed the IFN-α receptor IFNAR1 RNA (Figure 6A)
and protein (Figure 6B), indicating that the absence of
an IFN- α treatment response in the BT474 and
MDA-MB-231 cell lines was not due to the lack of the
receptor. The results from Western blot showed that
all three cell lines expressed IR, however the expres-
sion level is much higher in SUM149 cell than the
other two cell lines. Correspondingly, the relative
IFNAR1 expression ratios for SUM149, MDA-MB-231
and BT474 cell lines were fold changes of 2.05
(59,601/29,037), 0.51 (107,956/212,049), and 0.17
(26,893/149,868), respectively, based on quantification
of the Western blot shown in Figure 6B. Notably, there
were subtle differences detected in the amount of
protein loaded into each well, but this was accounted
for in the densitometry analysis by normalizing each
well to the amount of β-actin.
Figure 3: Plot of expression of gene RNASEL on Affymetrix expression microarrays of IBC and non-IBC tumor sam-
ples. This plot was generated on the NCBI GEO website from data set GSE5847 originally published by Boersma et al. 2008 . Log
expression value (red) and rank order of expression value (blue) are shown for each sample.
Journal of Cancer 2013, Vol. 4
Figure 4: Cell proliferation as a function of Interferon (IFN- α) dosing. Increasing doses of IFN-α decreased cell proliferation
significantly in a dose and time-dependent manner in SUM149 cells. Cells were treated with 0, 500, 1000, 2500, and 5000 U/ml of IFN-α
for 24 and 48 hours. At each time point, the IFN-α was removed, the cells rinsed with PBS, and cell proliferation was evaluated (* P < 0.05,
** P < 0.01, (*** P < 0.001). The reduction in cell proliferation induced by treatment with IFN-α is an IBC-specific response. The 2 non-IBC
cell lines (A) MDA-MB-231 and (B) BT474 did not demonstrate a significant decrease in cell proliferation with IFN-α treatment (* P < 0.05,
** P < 0.01, *** P < 0.001).
Journal of Cancer 2013, Vol. 4
Figure 5: IFN- α response abrogated with antibody block. (A) SUM149 cells were treated for 24 hours with increased doses of
IFN-α; cell proliferation decreased in a dose-dependent manner. (B) IFN-α complexed to IFN-α specific antibody abrogated the
dose-dependent decrease in cell proliferation. For IFN-α block experiments, each treatment was pre-incubated for 15 minutes with 1
μg/ml IFN-α specific antibody (* P < 0.05, ** P < 0.01, *** P < 0.001).
Figure 6: (A) Reverse transcription polymerase chain reaction (RT-PCR) and (B) Western blot for human IFN-α Re-
ceptor. (A) RT-PCR showed that the IFN- α receptor was basally expressed in untreated cultured SUM149, MDA-MB-231, and BT474
cells. L: DNA ladder, IR: IFN-α receptor, (+): GAPDH controls, (-): no template negative controls. (B) Western blot analysis indicating
human IFN-α receptor protein expressed in the membrane fraction of SUM149, MDA-MB-231, and BT474 cell lines. M: Molecular weight
marker, IR: Interferon-alpha receptor type I. β-actin served as controls for SDS-PAGE protein loading and for complete immunoblot
Journal of Cancer 2013, Vol. 4
PCR-based analysis of HMTV sequences in
The results of our PCR-based analysis to detect
the presence of HMTV sequences are shown in Figure
7. PCR analysis of genomic DNA (Figure 7A) and
RT-PCR analysis (Figure 7B) of 4 primer sets for the
HMTV env/LTR and late LTR regions showed that
each of these viral elements were not detected in the
SUM149 cell line. In addition, our Southern blot
analysis using a specific probe to the env region of
HMTV did not detect the presence of HMTV inte-
grated in the SUM149 genome (Figure 8), consistent
with our PCR findings. Although we were able clone
and sequence PCR amplicons generated by our anal-
yses, none of these fragments revealed homology to
HMTV (data not shown). Our findings do not indicate
the presence of putative HMTV in SUM149 cells.
Figure 7: PCR products from SUM149 genomic DNA. Agarose gel electrophoresis of PCR amplicons from genomic DNA (A) and
RT-PCR amplicons from RNA (B) using 4 primer sets (see materials and methods for sequences) for the HMTV env/LTR and late LTR
regions. Expected product sizes if HMTV sequences were present are shown at the bottom of each gel. A. PCR analysis resulted in
nonspecific amplicons from chromosomes 7 and 17. The white box highlights bands of interest that were cloned and sequenced. B.
RT-PCR using several primer pairs over an annealing temperature gradient produced nonspecific amplicons, as compared with the positive
controls (+) GAPDH. Molecular weight marker was 47kDa.
Figure 8: Southern blot analysis for HMTV sequences in SUM149 genome. Southern blot analysis indicated the env region of
HMTV was not present in the SUM149 genome. (A) Ethidium bromide visualization of BamHI-digested and gel electrophoresed SUM149
genomic DNA and a synthetic 172 base-pair HMTV fragment. (B) Southern blot probed with biotinylated HMTV env probe. L: DNA
ladder, S: 10 μg digested SUM149 genomic DNA, (+): HMTV controls.
Journal of Cancer 2013, Vol. 4
Our experimental analysis of IBC cell lines
SUM149 and SUM190, revealed 2 SNPs in the RNase
L gene with known association to prostate carcino-
genesis, particularly those of possible viral etiology.
The consistent IBC homozygous variants 462R and
541E were infrequent or absent in non-IBC cell lines
respectively suggesting these variants may represent
novel risk alleles for IBC onset. This finding warrants
investigation in patient samples. In an effort to de-
termine if these SNPs could portend a genetic pre-
disposition to IBC, we first performed an analysis of
RNase L transcript levels between IBC and non-IBC
breast tissue samples from publicly available data.
Within the stroma of breast tissue, there was no sig-
nificant difference in the gene expression of these
RNase L SNPs in IBC and non-IBC tissue samples
(Figure 3). However, the expected differences in
prevalence, if there are any, should be in the somatic
tissues, such as breast parenchyma. This data suggests
that altered activity of RNase L is not likely due to
In 2012, Jin et al. 33 reported using two sequence
homology-based computational tools [Sort Intolerant
from Tolerant (SIFT) and Polymorphism Phenotype
(PolyPhen)] to predict the functional contributions of
several non-synonymous RNase L variants. Their
analysis suggested the R462Q and E541D variants are
predicted to be ‘tolerated’ changes, however func-
tional studies on these SNPs are yet to be conducted,
particularly in the context of prior viral infection.
Further, despite the prediction of a tolerated change,
variants at RNase L amino acid residues 462 and 541
have been definitively associated with altered cancer
Notably, Next Generation Sequencing would
yield the most appropriate dataset for this compari-
son, but such datasets are not currently available. A
significant confounding factor for a study of this na-
ture is the paucity of archived IBC samples, owing to
its rare clinical presentation and particular difficulties
with biospecimen collection of this tumor type, which
typically lacks a mass lesion and is generally treated
with systemic therapy prior to surgery 13. Further
validation with patient samples is required to evalu-
ate the frequency of RNase L variants in IBC, and
whether they are indeed a biomarker of genetic risk
for IBC. Of note, the sample detection could be
through patient cheek swaps, since the polymorphism
would likely affect all somatic tissues. In turn, tumor
isolation may not be necessary for SNP detection in
IBC patients. Our findings suggest that the 541G var-
iant may serve as a susceptibility factor in the devel-
opment of IBC, but larger sample sizes are needed to
better assess the role of 462G, which may potentially
serve as an IBC biomarker.
In contrast to our negative findings for HMTV in
IBC DNA, our finding of a selective response to IFN-α
treatment may indicate the potential to respond to
viral infection in IBC, as the 2-5A pathway is a key
mediator in the innate response to viral infection 37.
Additionally, RNase L 462R and 541 have normal
enzymatic functionality (both 462 and 541 variants are
in the RNase L protein kinase domain), and may, in
turn, decrease the likelihood of viral infection 31.
It is not without precedent that cell lines have
been found to have a constitutive viral load. Pattillo et
al. have shown that the cervical cancer cell line CaSKI
has an average viral load of 600 particles per cell 43.
However, a constitutive viral load has yet to be
demonstrated within IBC cell lines.
Of note, Reuben et al. recently suggested that
Epstein-Barr virus (EBV) may also play a potential
role in the pathogenesis of IBC 44. They demonstrated
that 20% of IBC patients have been exposed to EBV, as
determined by the detection of EBV-specific immu-
noglobulin G (IgG) antibody from peripheral blood 44.
They also demonstrated that a large portion of IBC
patients have reactivated EBV infection. However, no
causal relationship between EBV and IBC has been
proven, so the role of EBV in IBC is yet to be deter-
mined, despite the more clearly defined role of EBV in
carcinogenesis of diseases such as nasopharyngeal
carcinoma and Burkitt’s lymphoma 45.
Concomitantly, in our endeavor to detect the
presence of the HMTV virus, we did not detect HMTV
env sequences, by either RT-PCR or Southern blot
analysis, in SUM149 cells. In contrast, Pogo et al. re-
cently reported an increased detection of HMTV in
IBC samples and increased expression of HMTV en-
velope (Env) and capsid (Ca) proteins in 10 primary
cultures of human breast cancer containing HMTV
sequences (MSSM) 46. These cells were derived from
discarded ascitic fluids or pleural effusions obtained
from patients with metastatic breast cancer 46.
Pogo et al. also reported that, by using nested
priming PCR and Southern blot techniques, they de-
tected the presence of MMTV-like env sequences in
71.5% of the 67 human tissue samples evaluated 46-48.
Although those studies by Pogo et al. are
thought-provoking, a causal relationship has yet to be
definitively established, given the conflicting nature
of the reports currently available on the prevalence of
MMTV-like sequences within IBC patient specimens
49,50. In our study, both our PCR and Southern blot
analysis failed to detect HMTV-like env sequences in
the SUM149 cell line—despite the lower limits of de-
Journal of Cancer 2013, Vol. 4
tection in our Southern blot assay being in the femto-
gram level 51.
Although the findings of Pogo et al. conflicted
with ours in that they found HMTV in tumors from
IBC patients, some marked differences between the 2
models could explain this discrepancy 25. In our study,
we used authenticated SUM149 cells that were de-
rived from a primary inflammatory ductal carcinoma
of the breast and established as an immortalized cell
line that is well characterized 27-29. To our knowledge,
no other group has investigated the presence of puta-
tive viral HMTV elements in this cell line, so our re-
sults provide novel insights, as SUM149 is the most
widely used in vitro model for IBC.
Even though SUM149 cells serve as an excellent
IBC model, the dynamics of disease progression may
differ in this established cell line, as compared with
the primary tumors evaluated by Pogo et al. Given the
rarity of IBC clinical specimens, we were not able to
examine tumors for viral sequences in our in vitro
study. Our power calculation of the minimum sample
size required to have 80% power to detect at least a
2-fold increase in risk between IBC and non-IBC can-
cer is 356 samples of each cancer (IBC and non-IBC)
with 160 healthy (control) samples. Given that few
institutions would be likely to have 356 IBC banked
specimens or access to 356 IBC patients currently be-
ing followed in the clinic who might participate in the
evaluation of the prevalence of these RNase L SNPs,
validation of these SNPs becomes problematic.
Moreover, clinical biomarker validation studies with
fewer than the calculated number of patients may be
misleading or even futile 52.
However, given the large number of IBC speci-
mens required to establish whether these SNPs cor-
relate with IBC, this study would be best performed
using either multi-institutional datasets or in the con-
text of a clinical trial’s biospecimen collection. One
such approach would be genome wide association
studies (GWAS) of relevant datasets to query the
prevalence of RNase L SNPs in IBC; unfortunately, no
such dataset exists at this time.
However, since SUM149 is the most widely
studied IBC cell line model, our finding that it may be
discordant with regard to the presence of the HMTV
genome in a cohort of IBC primary tumors calls into
question its appropriateness as a model for IBC, pro-
vided that the findings of Pogo et al. are validated.
Nonetheless, in the absence of independent confirma-
tion of Pogo et al.’s findings that HMTV is prevalent in
IBC, SUM149 remains the commercially available in
vitro model of choice for IBC.
Alternatively, our in vitro study has identified 2
SNPs, specific genotypic variants of which identify
important possible genetic risk determinants for IBC.
Future studies investigating the genotypic frequency
of these SNPs within human IBC tumors are war-
ranted to validate our in vitro findings.
Financial support: seed funds from the Arizona
Cancer Center and the Department of Surgery, Uni-
versity of Arizona. NIH P30CA014089 supports the
USC Norris Comprehensive Cancer Center. NIH
grant P30CA023074 supports the Arizona Cancer
This study was presented in part at the 2011
American Society of Clinical Oncology Breast Cancer
Symposium, San Francisco, CA.
The authors have declared that no competing
1. Gonzalez-Angulo AM, Hennessy BT, Broglio K, et al. Trends for
inflammatory breast cancer: is survival improving? Oncologist
2. Bristol IJ, Woodward WA, Strom EA, et al. Locoregional treatment
outcomes after multimodality management of inflammatory breast
cancer. Int J Radiat Oncol Biol Phys 2008;72:474-84.
3. Charafe-Jauffret E, Tarpin C, Viens P, Bertucci F. Defining the molecular
biology of inflammatory breast cancer. Semin Oncol 2008;35:41-50.
4. Hance KW, Anderson WF, Devesa SS, Young HA, Levine PH. Trends in
inflammatory breast carcinoma incidence and survival: the surveillance,
epidemiology, and end results program at the National Cancer Institute.
J Natl Cancer Inst 2005;97:966-75.
5. Dirix LY, Van Dam P, Prove A, Vermeulen PB. Inflammatory breast
cancer: current understanding. Curr Opin Oncol 2006;18:563-71.
6. Singletary SE, Cristofanilli M. Defining the clinical diagnosis of
inflammatory breast cancer. Semin Oncol 2008;35:7-10.
7. Dawood S, Merajver SD, Viens P, et al. International expert panel on
inflammatory breast cancer: consensus statement for standardized
diagnosis and treatment. Ann Oncol 2011;22:515-23.
8. Lerebours F, Bieche I, Lidereau R. Update on inflammatory breast
cancer. Breast cancer research : BCR 2005;7:52-8.
9. Tai P, Yu E, Shiels R, et al. Short- and long-term cause-specific survival of
patients with inflammatory breast cancer. BMC Cancer 2005;5:137.
10. Cristofanilli M. Novel targeted therapies in inflammatory breast cancer.
11. Dushkin H, Cristofanilli M. Inflammatory breast cancer. J Natl Compr
Canc Netw 2011; 9:233-40.
12. Ellis GK, Barlow WE, Gralow JR, et al. Phase III comparison of standard
doxorubicin and cyclophosphamide versus weekly doxorubicin and
daily oral cyclophosphamide plus granulocyte colony-stimulating factor
as neoadjuvant therapy for inflammatory and locally advanced breast
cancer: SWOG 0012. Journal of clinical oncology : official journal of the
American Society of Clinical Oncology 2011;29:1014-21.
13. Smoot RL, Koch CA, Degnim AC, et al. A single-center experience with
inflammatory breast cancer, 1985-2003. Arch Surg 2006;141:567-72.
14. Taghian A, El-Ghamry MN, Merajver SD. Inflammatory breast cancer:
Clinical features and treatment. Waltham, MA: Uptodate; 2011.
15. Robertson FM, Bondy M, Yang W, et al. Inflammatory breast cancer: the
disease, the biology, the treatment. CA Cancer J Clin 2010;60:351-75.
16. Gonzalez-Angulo AM, Sneige N, Buzdar AU, et al. p53 expression as a
prognostic marker in inflammatory breast cancer. Clin Cancer Res
17. Houchens NW, Merajver SD. Molecular determinants of the
inflammatory breast cancer phenotype. Oncology (Williston Park)
18. Pogo BG, Melana SM, Holland JF, et al. Sequences homologous to the
mouse mammary tumor virus env gene in human breast carcinoma
Journal of Cancer 2013, Vol. 4 Download full-text
correlate with overexpression of laminin receptor. Clin Cancer Res
19. Fantozzi A, Christofori G. Mouse models of breast cancer metastasis.
Breast cancer research : BCR 2006;8:212.
20. Holland JF, Pogo BG. Mouse mammary tumor virus-like viral infection
and human breast cancer. Clin Cancer Res 2004;10:5647-9.
21. Fernandez-Cobo M, Melana SM, Holland JF, Pogo BG. Transcription
profile of a human breast cancer cell line expressing MMTV-like
sequences. Infect Agent Cancer 2006;1:7.
22. Bindra A, Muradrasoli S, Kisekka R, Nordgren H, Warnberg F, Blomberg
J. Search for DNA of exogenous mouse mammary tumor virus-related
virus in human breast cancer samples. The Journal of general virology
23. Mok MT, Lawson JS, Iacopetta BJ, Whitaker NJ. Mouse mammary tumor
virus-like env sequences in human breast cancer. Int J Cancer
24. Etkind PR, Stewart AF, Wiernik PH. Mouse mammary tumor virus
(MMTV)-like DNA sequences in the breast tumors of father, mother, and
daughter. Infect Agent Cancer 2008;3:2.
25. Pogo BG, Melana SM, Moran H, Holland JF. Presence of MMTV-like env
gene sequences in human breast cancer. Breast Cancer Res Treat
26. Pogo BG, Holland JF, Levine PH. Human mammary tumor virus in
inflammatory breast cancer. Cancer 2010;116:2741-4.
27. Ethier SP, Mahacek ML, Gullick WJ, Frank TS, Weber BL. Differential
isolation of normal luminal mammary epithelial cells and breast cancer
cells from primary and metastatic sites using selective media. Cancer Res
28. Forozan F, Veldman R, Ammerman CA, et al. Molecular cytogenetic
analysis of 11 new breast cancer cell lines. Br J Cancer 1999;81:1328-34.
29. Hoffmeyer MR, Wall KM, Dharmawardhane SF. In vitro analysis of the
invasive phenotype of SUM 149, an inflammatory breast cancer cell line.
Cancer cell international 2005;5:11.
30. Rokman A, Ikonen T, Seppala EH, et al. Germline alterations of the
RNASEL gene, a candidate HPC1 gene at 1q25, in patients and families
with prostate cancer. Am J Hum Genet 2002;70:1299-304.
31. Shook SJ, Beuten J, Torkko KC, et al. Association of RNASEL variants
with prostate cancer risk in Hispanic Caucasians and African Americans.
Clin Cancer Res 2007;13:5959-64.
32. Urisman A, Molinaro RJ, Fischer N, et al. Identification of a novel
Gammaretrovirus in prostate tumors of patients homozygous for R462Q
RNASEL variant. PLoS Pathog 2006;2:e25.
33. Jin W, Wu DD, Zhang X, Irwin DM, Zhang YP. Positive Selection on the
Gene RNASEL: Correlation between Patterns of Evolution and Function.
Molecular biology and evolution 2012.
34. De la Vega FM, Lazaruk KD, Rhodes MD, Wenz MH. Assessment of two
flexible and compatible SNP genotyping platforms: TaqMan SNP
Genotyping Assays and the SNPlex Genotyping System. Mutation
35. Boersma BJ, Reimers M, Yi M, et al. A stromal gene signature associated
with inflammatory breast cancer. Int J Cancer 2008;122:1324-32.
36. Sen GC. Viruses and interferons. Annu Rev Microbiol 2001;55:255-81.
37. Samuel CE. Antiviral actions of interferons. Clin Microbiol Rev 2001; 14:
38. Silverman RH. A scientific journey through the 2-5A/RNase L system.
Cytokine & growth factor reviews 2007;18:381-8.
39. Silverman RH. Viral encounters with 2',5'-oligoadenylate synthetase and
RNase L during the interferon antiviral response. J Virol 2007;81:12720-9.
40. Pfeiffer RM, Gail MH. Sample size calculations for population- and
family-based case-control association studies on marker genotypes.
Genet Epidemiol 2003;25:136-48.
41. Kaiser J. Chronic fatigue syndrome. Studies point to possible
contamination in XMRV findings. Science 2011;331:17.
42. Smith RA. Contamination of clinical specimens with MLV-encoding
nucleic acids: implications for XMRV and other candidate human
retroviruses. Retrovirology 2010;7:112.
43. Pattillo RA, Hussa RO, Story MT, Ruckert AC, Shalaby MR, Mattingly
RF. Tumor antigen and human chorionic gonadotropin in CaSki cells: a
new epidermoid cervical cancer cell line. Science 1977;196:1456-8.
44. Reuben J. Epstein-Barr Virus (EBV) in Patients with Inflammatory Breast
Cancer: A Potential Role in Pathogenesis of the Disease. In: First
International Inflammatory Breast Cancer Conference. Houston, Texas:
45. Kaneda A, Matsusaka K, Aburatani H, Fukayama M. Epstein-barr virus
infection as an epigenetic driver of tumorigenesis. Cancer Res
46. Pogo BG. Human Mammary Tumor Virus and IBC. In: First
International Inflammatory Breast Cancer Conference; 2008; Houston,
Texas: Blackwell-Wiley; 2008.
47. Melana SM, Nepomnaschy I, Hasa J, Djougarian A, Holland JF, Pogo BG.
Detection of human mammary tumor virus proteins in human breast
cancer cells. J Virol Methods 2010;163:157-61.
48. Levine PH, Young H. A, Mark L., Rojowsky H., Holland J. F, Pogo B. G. .
Increased detection of breast cancer virus sequences in inflammatory
breast cancer. Advances in Tumor Virology 2009; 1:3-7.
49. Stewart A. MMTV-related env sequences in human breast tumor. Int J
50. Park DJ, Southey MC, Giles GG, Hopper JL. No evidence of MMTV-like
env sequences in specimens from the Australian Breast Cancer Family
Study. Breast Cancer Res Treat 2011;125:229-35.
51. Murakami S, Mizobuchi M, Nakashiro Y, Doi T, Hato N, Yanagihara N.
Bell palsy and herpes simplex virus: identification of viral DNA in
endoneurial fluid and muscle. Ann Intern Med 1996;124:27-30.
52. Gail MH, Pfeiffer RM, Wheeler W, Pee D. Probability of detecting
disease-associated single nucleotide polymorphisms in case-control
genome-wide association studies. Biostatistics 2008;9:201-15.