1078-0432.CCR-10-0889.Full.Pdf

Author Manuscript Published OnlineFirst on June 15, 2010; DOI: 10.1158/1078-0432.CCR-10-0889 AuthorPublished manuscripts OnlineFirst have been peer on reviewedJune 15, and 2010 accepted as 10.1158/1078-0432.CCR-10-0889for publication but have not yet been edited.

Clinical implications of gene dosage and gene expression

patterns in diploid breast carcinoma

∗ Toshima Z. Parris,1, Anna Danielsson,1 Szilárd Nemes,1 Anikó Kovács,2 Ulla

Delle,1 Ghita Fallenius,1 Elin Möllerström,1 Per Karlsson,1 and Khalil Helou1

Running Title: Clinical relevance of integrative genomics in breast cancer

Keywords: Diploid breast carcinoma, array-CGH, copy number aberration,

gene expression microarray, aggressive phenotype

Authors’ Affiliations: 1Department of Oncology, Institute of Clinical Sciences, and 2Laboratory of Clinical Pathology and Cytology, Sahlgrenska Academy at University of Gothenburg, Gothenburg, Sweden

Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).

∗ Corresponding Author: Toshima Z. Parris, Department of Oncology, Sahlgrenska Academy at University of Gothenburg, Gula stråket 2, SE-41345 Gothenburg, Sweden. Phone: 46-31- 3427855; Fax: 46-31-820114; E-mail: [email protected].

Grant support: This work was supported by grants from the King Gustav V Jubilee Clinic Cancer Research Foundation (K. Helou) and the Wilhelm and Martina Lundgren Research Foundation (T. Parris). Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Copyright © 2010 American Association for Cancer Research

Downloaded from clincancerres.aacrjournals.org on September 30, 2021. © 2010 American Association for Cancer Research. Author Manuscript Published OnlineFirst on June 15, 2010; DOI: 10.1158/1078-0432.CCR-10-0889 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Translational Relevance

This study provides novel information on the specific DNA copy number aberrations and

underlying changes in gene expression associated with clinicopathological features in

diploid breast carcinoma (DBC). Using an integrative genomics approach, we identified

transcripts whose expression patterns were directly impacted by gene dosage. In

addition, the transcriptional levels of 12 genes were consistently associated with a more

malignant phenotype in DBC. Integrating data on DNA and mRNA dysregulation in

relation to breast cancer behavior is, therefore, a step toward understanding tumor

initiation and progression contributing to unfavorable prognosis. Furthermore, these

findings provide potential targets in the development of individualized patient therapy

for aggressive breast neoplasms.

Abstract

Purpose: Deregulation of key cellular pathways is fundamental for the survival and

expansion of neoplastic cells. In cancer, regulation of gene transcription can be mediated

in a variety of ways. The purpose of this study was to assess the impact of gene dosage

on gene expression patterns, the effect of other mechanisms on transcriptional levels,

and to associate these genomic changes to clinicopathological parameters.

Experimental Design: We screened 97 invasive diploid breast tumors for DNA copy

number alterations and changes in transcriptional levels using array comparative

genomic hybridization and expression microarrays, respectively.

Results: The integrative analysis identified an increase in the overall number of genetic

alterations during tumor progression and fifteen specific genomic regions with aberrant

DNA copy numbers in at least 25% of the patient population, i.e. 1q22, 1q22-q23.1,

1q25.3, 1q32.1, 1q32.1-q32.2, 8q21.2-q21.3, 8q22.3, 8q24.3, and 16p11.2 were

recurrently gained, while 11q25, 16q21, 16q23.3, and 17p12 were frequently lost

(P<0.01). An examination of the expression patterns of genes mapping within the

detected genetic aberrations identified 47 unique genes and 1 Unigene cluster

significantly correlated between the DNA and relative mRNA levels. In addition, more

malignant tumors with normal gene dosage levels displayed a recurrent over-expression

of UBE2C, S100A8 and CBX2, and down-regulation of LOC389033, STC2, DNALI1, SCUBE2,

NME5, SUSD3, SERPINA11, AZGP1 and PIP.

Conclusions: Taken together, our findings suggest that the dysregulated genes identified

here are critical for breast cancer initiation and progression, which could be used as novel

therapeutic targets for drug development to complement classical clinicopathological

features.

Introduction

Copy number alterations (CNA) are a fundamental feature of neoplastic cells that

influence crucial cancer-specific processes (1). It is estimated that more than 1% of

protein-coding genes in the human genome contribute to tumorigenesis, whereas only

3% of these have been implicated in sporadic breast carcinoma (2). In comparison with

many tumor types of non-epithelial origin, breast carcinomas arising from epithelial cells

have numerous genetic alterations (3). Many of the recurrent alterations occurring in

carcinomas of the breast have been identified, including copy number gains on 1q, 8p,

8q, 11q, 16p, 17q, and 20q and losses on 1p, 6q, 11q, 16q, 17p, and 22q (4).

Tumorigenesis is however a dynamic evolutionary process that promotes genetic

heterogeneity and thereby produces a complex combination of random and non-random

aberrations. To study complex phenotypes such as cancer, a straightforward approach is

to provide comprehensive information on the disease by integrating multiple platforms

on several biological levels (DNA-RNA-protein). Integrative analyses of data produced by

genome-wide profiling techniques such as array comparative genomic hybridization

(array-CGH; (5)) and expression microarrays can provide detailed information on the

genomic locations of recurrently altered DNA regions, the impact of these CNAs on gene

deregulation, as well as enhance our understanding of these genetic events coupled with

clinicopathological parameters (6, 7). However, there are limitations associated with the

array-CGH approach as it will not detect inversions, balanced translocations, polyploidy,

mutations, or epigenetic modulations. It is, therefore, necessary to also study changes in

gene expression patterns independent of copy number alterations.

In the present investigation, we chose to conduct genome-wide screening on a series of

diploid breast carcinomas (DBC). A previously suggested theory postulated that the

aneuploid state is not fundamental to malignancy as diploid and aneuploid breast tumors

share many of the acquired genetic aberrations characteristic of the disease, despite the

increase in number of events in aneuploid disease (8-14). Approximately 1 in 4 diagnosed

breast carcinoma cases have a diploid DNA content (15). Generally, diploid tumors are

composed of a nearly homogenous mix of slow-growing cells which may permit these

neoplasms to follow a more favorable clinical course than aneuploid tumors (16, 17). The

more aggressive phenotype displayed by aneuploid tumors can possibly be explained by

clonal heterogeneity represented by the presence of fewer cells expressing the steroid

hormone receptors, increased proliferative activity, and a higher percentage of poorly

differentiated tumors (18). Taken together, these observations suggest that despite

diversity there are similar molecular mechanisms active in the development of breast

carcinomas, regardless of ploidy status. It is, therefore, of particular interest to conduct

integrative genomics using diploid breast carcinomas to mitigate some of the obstacles

associated with the heterogeneity of genetic alterations in carcinomas.

Therefore, we investigated the copy number- and non-CNA induced gene expression

patterns of DBC from 97 patients using 38K BAC array-CGH and the Illumina HumanHT-12

platform. With the application of this approach on a series of DBC we were able to Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Copyright © 2010 American Association for Cancer Research

reduce biological noise and thereby identify key putative candidate genes associated with

an aggressive phenotype in breast carcinoma. Additionally, we were able to develop an

approach to a) test the relationship between gene dosage and mRNA levels, b) determine

the proportional or non-proportional gain/loss in CNA and gene expression patterns, and

c) determine if a significant difference in expression levels exist between neoplasms

possessing a given genetic aberration from those lacking these aberrations.

Materials and Methods

Patient and tumor specimens. Primary invasive tumors from 97 DBC patients were

selected from the fresh-frozen tissue tumor bank at the Sahlgrenska University Hospital

Oncology Lab (Gothenburg, Sweden). Patient stratification according to axillary lymph

node status (node-negative, pN0; node-positive, pN1) and overall survival (>8-year

survivors, henceforth termed long-term survivors; breast cancer-specific mortality within

8 years of diagnosis, henceforth termed short-term survivors) is shown in Supplementary

Table S1. All samples were assessed for DNA content at the time of diagnosis from 1991-

1999 (data not shown) by flow cytometry at the Laboratory for Clinical Chemistry,

Sahlgrenska University Hospital. Tumor specimens with a DNA index of 1.0 were classified

as diploid. The presence of malignant cells was assessed in all samples by evaluation of

touch preparation imprints stained with May-Grünwald Giemsa (Chemicon). Highly

representative specimens possessing greater than 70% neoplastic cell content were

included in the microarray and fluorescence in situ hybridization (FISH) analyses. All

procedures were performed in accordance with the Declaration of Helsinki and approved

by the Medical Faculty Research Ethics Committee (Gothenburg, Sweden). The

clinicopathological features of the 97 cases are shown in Table 1.

Array comparative genomic hybridization. Whole-genome tiling arrays with 38,043

reporters mapping to the UCSC May 2004 hg17: NCBI Build 35 were manufactured as

previously described (19) at the SCIBLU Genomics DNA Microarray Resource Center Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Copyright © 2010 American Association for Cancer Research

(SCIBLU), Department of Oncology, Lund University. Images and raw signal intensities

were acquired using an Agilent G2505B DNA microarray scanner (Agilent Technologies)

and GenePix Pro 6.0.1.22 (Axon Instruments) image analysis software.

Data preprocessing and normalization were performed using the web-based BioArray

Software Environment system (BASE) provided by SCIBLU (20).3 Further analysis to

segment the data into regions of gains and losses was performed using the Rank

Segmentation algorithm with Nexus Copy Number Professional 4.1 software

(BioDiscovery), as previously described (21). Further, log2ratio thresholds for low-level

gain (LLG), high-level gain/amplification (HLG), heterozygous loss (HL), and homozygous

deletion (HD) were set at +0.2, ≥+0.5, -0.2, and ≤-1.0, respectively, using a 0.01 p-value

cutoff. Genomic regions covered entirely by previously reported copy number variations

(CNV) in the human genome were removed (22). Unsupervised hierarchical clustering

was applied using complete linkage with Pearson correlation to group tumors based on

detected genetic alterations (LLG and HL). The association between clinicopathological

features and CNAs was analyzed using a 0.05 p-value cutoff with the Mann–Whitney U or

Kolmogorov-Smirnov test, as appropriate.

Verification of HER2/neu gene amplification using FISH. Dual-color FISH was performed

to confirm HER2/neu gene amplification in tumors with positive (log2ratio ≥+0.5) array-

CGH results (11%). Touch preparation imprints were prepared from fresh cuts of frozen

3 Internet address: http://base2.thep.lu.se/onk/. Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Copyright © 2010 American Association for Cancer Research

tumor samples on Superfrost Plus microscope slides (Erie Scientific Company). Probe

hybridization and post-hybridization washes were performed according to the

manufacturer’s guidelines using the ZytoLight SPEC HER2/CEN 17 Dual Color Probe

(ZytoVision GmbH). The slides were evaluated using a Leica DMRA2 fluorescence

microscope (Leica) equipped with an ORCA Hamamatsu charged-couple devices camera

(Hamamatsu Corporation). Digitalized black and white images were captured at 100X

magnification using the Leica CW4000 software package (Leica Microsystems Imaging

Solutions Ltd). HLG was classified in cells with ≥2.5-fold the centromere count.

Gene expression analysis. Total RNA from the same extraction was used for both

expression profiling and subsequent validation with quantitative real-time PCR. The RNA

samples were processed at SCIBLU using Illumina HumanHT-12 Whole-Genome

Expression BeadChips (Illumina), according to the manufacturer’s instructions. The

expression microarrays contained approximately 49,000 probes representing more than

25,400 RefSeq (Build 36.2, Release 22) and Unigene (Build 199) annotated genes. Images

and raw signal intensities were acquired using the Illumina BeadArray Reader scanner

and BeadScan 3.5.31.17122 (Illumina) image analysis software, respectively.

Data preprocessing and quantile normalization were applied to the raw signal intensities

using BASE. Further data processing was performed in Nexus Expression 2.0

(BioDiscovery) using log2-transformed, normalized expression values and a variance

filter. Normalized values from five normal breast samples profiled with Illumina Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Copyright © 2010 American Association for Cancer Research

HumanWG-6 Expression Beadchips (GEO, accession number GSE17072) were used as

reference (23). Differentially expressed genes were determined using the Benjamini-

Hochberg method (24) to control for the false discovery rate (FDR) with FDR-corrected p-

values <0.01. The diploid dataset was stratified into the molecular breast cancer subtypes

using the five centroids (Normal-like, Basal-like, Luminal subtype A, Luminal subtype B,

and HER2/ER-) and Genomic Grade Index (GGI; low, high) using ER-positive tumors as

previously described (25-27). All microarray data discussed in this publication are

accessible through NCBI’s Gene Expression Omnibus (GEO accession number GSE20486)4.

Quantitative real-time PCR (qPCR). Validation of the expression microarray data was

performed as previously described using qPCR with pre-designed TaqMan® Gene

Expression Assays (Applied Biosystems) (28). In brief, 82/97 total RNA samples were used

to validate the expression patterns of 16 transcripts and three endogenous controls, i.e.

PPIA, PUM1, and HPRT1. The endogenous controls were selected based on their

constitutive expression using the Illumina HumanHT-12 platform. The geometric mean of

the three endogenous controls was used to normalize the data and relative gene

expression levels were calculated with the relative standard curve method. The Student’s

t-test was calculated to determine the difference in expression between studied groups

and the Spearman correlation coefficients (two-tailed) to establish the relationship

between microarray and qPCR expression patterns.

4 Internet address: http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE20486. Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Copyright © 2010 American Association for Cancer Research

Integrated DNA copy number and expression analysis. Illumina HumanHT-12 probe

nucleotide sequences were mapped to genomic locations (NCBI Build 35) using

sequences downloaded from the UCSC Genome Browser (29)5. A pair-wise comparison

of the Illumina probe and BAC clone nucleotide sequences was then conducted to

generate Illumina-BAC probe pairs with 100% sequence similarity. Illumina-BAC probes

spanning the recurrent aberrations were selected from smoothed array-CGH data using

log2ratio ±0.2. CNA-induced genes were assessed by Pearson correlation (Q<0.05).

Further, the change in mRNA levels as a function of DNA copy numbers was estimated by

robust piecewise linear regression. Regression slopes significantly higher than 1, i.e.

positive allometric change, indicated higher gene expression levels in comparison with

CNA levels; the opposite for slopes significantly smaller than 1, i.e. negative allometric

change; a slope of 1 indicated a proportional gain/loss in CNA and gene expression, i.e.

isometric change. Lastly, the difference between the mean relative mRNA values for

tumors containing either gain/loss versus tumors without these aberrations was

estimated with t-test for Illumina-BAC probes showing significant association between

CNA and expression patterns. Statistical analyses were performed in R/Bioconductor.

The methods are described in more detail in Supplementary Materials and Methods.

5 Internet address: http://hgdownload.cse.ucsc.edu/goldenPath/hg17/chromosomes/.

Results

Overview of chromosomal aberrations detected by array-CGH in the diploid patient

population. Ninety-seven diploid breast tumors were analyzed for DNA copy number

alterations using 38K BAC CGH arrays. Chromosomal aberrations were detected in all

tumors, including all human autosomes and the X chromosome. These segments varied

in size from small aberrations to gain and/or loss of whole chromosomes (range, 103 kb

to 109 Mb; mean, 6.6 Mb). The mean number of DNA copy number alterations (±SEM)

corresponding to clinicopathological features are presented in Supplementary Table S2.

On average, 40.1±3.9 chromosomal aberrations (range, 2-204) were detected per diploid

tumor sample, including 20.1±2.0 LLG (range, 0-117), 4.9±0.6 HLG (range, 0-31), 15.1±1.7

HL (range, 0-104), and 0.03±0.02 HD (range, 0-1). Recurrent regions of gains and losses

observed with a frequency of ≥25% (P<0.01) were identified using the Rank

Segmentation algorithm in Nexus Copy Number 4.1 (Fig. 1a). LLG was observed on nine

chromosomal sub-regions on 1q22 (53%), 1q22-q23.1 (53%), 1q25.3 (53%), 1q32.1 (53%),

1q32.1-q32.2 (53%), 8q21.2-q21.3 (33%), 8q22.3 (26%), 8q24.3 (26%), and 16p11.2

(26%). Loss was observed on six sub-regions on 11q25 (26%), 16q21 (38%), 16q23.3

(38%), and 17p12 (26%). Recurrent regions of HLG in at least 10% (P<0.01) of the cases

were identified on three sub-regions on 1q32.1 (20%), 1q32.2 (20%), and 11q13.3-q13.4

(10%). In addition, several amplicons were detected in less than 10% of the patient

population, i.e. 3q26.2 (5%), 8p11-p12 (6%), 8q (5-9%), 17q (5-9%), and 20q13 (6%; Fig.

1b). These findings are consistent with published array-CGH data on breast carcinoma (9,

30-35) and are listed in more detail in Table 2.

Unsupervised clustering of the dataset according to the detected LLG and HL in each

tumor produced two distinct groups. Cluster 1 (n=21) contained tumors displaying very

few aberrations in select portions of the genome including exclusion of 1q gain which is

common in most breast neoplasms, while Cluster 2 (n=76) contained tumors with an

abundance of genetic aberrations covering the entire genome (Fig. 1c). In addition, the

former group lacked genetic alterations on chromosomes 3, 18, and 20, and loss of

chromosome 6 was more characteristic of this group. In general, Cluster 1 consisted

predominantly of low-grade tumors (P=0.006), low GGI (P=0.021), and long-term

survivors (P=0.034), whereas Cluster 2 consisted predominantly of tumors from short-

term survivors (P=0.034), high GGI (P=0.035), short-term survivors with pN1 disease

(P=0.044), and tumors with gain of the 8p11-p12 locus (P=0.044).

Correlation of DNA copy number alterations to clinicopathological features. To

determine if clinicopathological features are characterized by specific genetic alterations,

recurrently altered genomic regions were compiled and are shown in Supplementary

Table S3. A high number of genetic alterations were observed in tumors with high GGI

(P=0.0004), tumors from short-term survivors (P=0.001), Cluster 2 (P=0.001), and high-

grade tumors (P=0.007), in comparison with their corresponding counterparts.

Specifically, a higher number of LLG (P=0.001), HLG (P=0.005), and HL (P=0.005) were

detected in short-term survivors compared to long-term survivors, suggesting that the Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Copyright © 2010 American Association for Cancer Research

number of genetic alterations may have a detrimental effect on clinical outcome. Using

the Rank Segmentation algorithm and a 25% difference cutoff, HL on 3p14.2, 17p13.1-

p12, and 18q23 were more prevalent in short-term survivors than long-term survivors.

Further, LLG and HLG were generally more frequent in high-grade than in low-grade

tumors (P=0.002 and P=0.01, respectively); all forms of genetic aberrations, excluding HD

were more prevalent in tumors with high than low GGI; LLG (P=0.04) was more prevalent

in larger tumors (pT3/pT4) than in smaller tumors (pT1/pT2); LLG and HLG were generally

more frequent in Cluster 2 than in Cluster 1 (P=0.00003 and P=0.02, respectively). LLG of

the 17q23.3-q24.1 region were detected solely in high-grade and never in low-grade

tumors. With respect to GGI status, LLG of 1q, 20q and HL of 3p14.2, 8p, 11q, 17p12-

p11.2, and 18q23 were more prevalent in high GGI tumors. Lastly, LLG on chromosomes

2q, 7p, 12q, 17q, and 20q were observed predominantly in larger tumors.

In our dataset, there was no distinct difference in the number of or type of observed

CNAs in tumors with regard to their axillary lymph node status. These findings suggest

that pN0 and pN1 tumors are biologically similar entities. Biological differences became

apparent when these tumors were further stratified by either overall survival status

and/or by the number of positive axillary lymph nodes (PALN). LLG of 6p21.32-p21.31

and HL of 10q26.12 were more prevalent in short-term survivors with pN0 tumors than in

short-term survivors with pN1 tumors. Furthermore, heterozygous loss of 8p23.2 was

predominantly detected in tumors from patients with ≥4 PALN than pN0 tumors. LLG of

8p12 was more prevalent in long-term survivors with 1-3 PALN than long-term survivors

with pN0 tumors; LLG of the 6p25.3 and 17q12-q21.2 regions were more prevalent in

long-term survivors with ≥4 PALN than long-term survivors with pN0 tumors; lastly, LLG

of 17q12-q21.2 was more prevalent in long-term survivors with ≥4 PALN than long-term

survivors with 1-3 PALN. To validate the array-CGH results, dual-color interphase FISH

was performed on tumors displaying HLG of the HER2/neu gene using the ZytoLight SPEC

HER2/CEN 17 Dual Color Probe. We observed a 100% concordance between array-CGH

and FISH for the assessment of HER2/neu amplification (Fig. 2).

Gene expression profiling. The number of deregulated transcripts in each group

comparison identified by transcriptome profiling of the DBC samples are shown in Table

3, as well as, Gene Ontology biological processes associated with these genes. A large

number of transcripts were deregulated 1.5-fold based on short-term overall survival,

endocrine insensitivity, high GGI, and triple negative status. Similar to results on the DNA

level, there were few differences between tumors stratified according to axillary lymph

node status, as only the CNTNAP2 gene was differentially expressed. Similarly, few

transcripts differed statistically between tumors of varying SBR grade, implying that there

may be some overlap between SBR grade II and low- and high-grade tumors. One gene

(SHISA2) was up-regulated in SBR grade I compared to grade II tumors, which was also

among transcripts differentially regulated between SBR grade I and grade III tumors. Six

genes were on both lists for SBR grade I vs III and SBR grade II vs III (S100A8, S100A9,

CBX2, MAPT, AZGP1, PIP). Lastly, a comparison of the gene expression levels was

performed on the two clusters determined using genetic alterations. In total, five

transcripts were differentially regulated (2 up-regulated and 3 down-regulated). These

results show the recurrent up-regulation of transcripts associated with a more malignant

phenotype (short-term overall survival, endocrine insensitivity, triple negative status,

poor tumor differentiation), i.e. UBE2C, S100A8 and CBX2, while LOC389033, STC2,

DNALI1, SCUBE2, NME5, SUSD3, SERPINA11, AZGP1 and PIP are down-regulated

(Supplementary Table S4). Five of these genes were also differentially regulated between

tumors of high and low GGI, i.e., UBE2C, S100A8, STC2, SCUBE2, SERPINA11. To validate

these results, sixteen genes showing more than 1.5-fold differential expression levels in

either the receptor status or overall survival groups were assessed with qPCR. A linear

relationship was shown between the Illumina and qPCR results (rS=0.97; P<0.01). The

results of the validation experiments are listed in Supplementary Table S5.

Stratification of the diploid dataset into molecular gene expression subtypes and GGI.

The diploid dataset was stratified into the five molecular breast cancer subtypes by

assigning each tumor to one of the five centroids using the Pearson correlation

coefficient (25). Subsequently, no tumors were classified as Normal-like (0%), seven as

Basal-like (7%), one as Luminal subtype A (1%), seventy-eight as Luminal subtype B (80%),

and eleven as HER2/ER- (11%). As expected, all tumors classified as both HER2+ and ER-

(n=4) were correctly classified in the HER2/ER- subtype. The remaining HER2+ tumors

were either classified in the HER2/ER- (n=1) or Luminal subtype B groups (n=6). In

addition, Basal-like (86%) and HER2/ER- (64%) tumors were predominantly associated

with short-term overall survival; Basal-like (67%) and HER2/ER- (73%) tumors were more

prevalent in Cluster 2 than in Cluster 1. Eighty-one percent of tumors grouped in Cluster 1

were of Luminal B subtype.

Stratification of the 81 ER-positive tumors in our dataset into the GGI groups produced a

slight overlap between the low and high GGI groups whereas 2 SBR grade I and 4 grade III

tumors were classified with high and low GGI, respectively. As expected, SBR grade II

tumors were a mix of both GGI groups. In general, low GGI was predominantly associated

with long-term overall survival (P=0.008), long-term survivors with 1-3 PALN (P=0.009),

SBR grade I (P=0.009), pT1 (P=0.012), Cluster 1 (P=0.02), and 1-3 PALN (P=0.025), while

high GGI was associated with ≥4 PALN (P=0.001), short-term survivors with ≥4 PALN

(P=0.005), short-term survivors with pN1 disease (P=0.006), short-term overall survival

(P=0.008), pT2 (P=0.018), and Cluster 2 (P=0.02).

Integrated DNA copy number and gene expression analysis. To determine the effect of

CNAs on gene deregulation in DBC, a correlation analysis was performed in three steps

using identified CNAs. These analyses showed that the regulation of several breast-cancer

related candidate genes and genes not previously associated with breast cancer were

located within recurrent CNAs listed in Table 2. In total, 1161 Illumina-BAC probe pairs

contained nucleotide sequences within the eighteen observed CNAs, of which 149 probe

pairs (52 unique genes and 3 Unigene clusters) were identified as having a significant

correlation between DNA and relative mRNA levels (Supplementary Table S6). For thirty-

eight genes the expression levels were as expected given the CNA log2ratio; three genes

showed higher expression levels and fourteen genes showed lower expression levels than Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Copyright © 2010 American Association for Cancer Research

expected considering the CNA log2ratio. Comparison of relative mRNA levels between

samples showing no CNAs and gain/loss of specific genomic regions identified 128/149

probe pairs (47 genes and 1 Unigene cluster) with significantly different expression levels

between the two groups. Figure 3 illustrates an example of the integration analysis for

the HER2/neu gene in amplified and non-amplified tumors.

Discussion

Our integrative genomics approach demonstrated that diploid breast neoplasms exploit

different strategies to promote tumorigenesis via gene deregulation. These results

suggest that despite diversity DBCs contain common genetic anomalies that may mediate

malignant tumor behavior. Furthermore, these observations support the theory that

aberrant gene expression levels displayed in breast carcinoma are, in part, induced by

copy number alterations and other mechanisms of transcriptional control (3).

Copy number alterations are a common feature of genetic instability in breast

carcinomas (3). The finding that a progressive increase in genetic aberrations is seemingly

predictive of unfavorable clinical outcome has been shown in several studies (12, 14). In

the present study, this pattern was associated with tumors exhibiting aggressive

behavior, i.e. tumors originating from short-term survivors, larger tumors, poorly

differentiated tumors, tumors with high GGI, and tumors grouped in Cluster 2. The

presence of recurrent genetic alterations at specific loci was also reflective of an adverse

effect on patient clinical outcome. Among the recurrently altered regions identified here,

transcripts on chromosome 1 are of particular interest, as we have shown that patients

with tumors lacking genetic alterations on chromosomes 1, 3, 18, and 20 have better

prognoses. Although CNAs are a fundamental feature of tumorigenesis, we have shown

that changes in gene dosage do not necessarily relate to the amount of mRNA gene

To assess the clinical significance of disease-associated gene deregulation, we integrated

copy number data with expression patterns. We were, therefore, able to show that only

a fraction of the genes altered at the DNA level were actually transcribed at abnormal

levels. Relative mRNA levels of 48/360 (13%) transcripts located within the eighteen

common regions of genetic alteration were directly impacted by CNAs, thus further

emphasizing the notion that many genetic alterations are bystander mutations which

may not reflect biological effect. These findings are consistent with previous reports,

which showed that the percentage of over-expressed genes can range from 6-44%

depending on the level of gene dosage (6, 7, 32). However, the co-regulation and

expression of adjacent genes may play an important role in tumorigenesis (7). To further

characterize these transcripts, we demonstrated a variation in the overall rate of gene

transcription given a specific copy number. This may be a reflection of the robust

induction of gene transcription, as a single gene must bypass several checkpoints before

it is expressed. We hypothesize that transcripts with expression levels deviating from

expected values (higher and lower) given gene dosage may be controlled by multiple

molecular mechanisms including smaller CNAs not detected using BAC arrays.

Recurrent chromosomal regions of gain and amplification are common on 1q in several

types of human neoplasms, ranging from carcinomas of the breast to sarcomas (36).

Here, we show that 40/48 (83%) of the transcripts correlating with gene dosage were

located on chromosome 1q consisting of 5 clusters of adjacent genes. We find Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Copyright © 2010 American Association for Cancer Research

ARHGEF11, CLK2, DHX9, EFNA1, RAB7L1, RABIF, RAG1AP1, RIPK5, RNASEL, RNPEP, and

SNRPE of particular interest as the elevated mRNA levels of these genes combined with

increased copy number may promote oncogenic activity. Several of these genes regulate

similar cellular processes predominantly involved in cell motility, protein amino acid

phosphorylation, RNA splicing, and cell-cell signaling.

Alternative cellular processes by which breast neoplastic cells mediate abnormal relative

mRNA levels in this study are unclear. Interestingly, our data suggest the consistent

deregulation of twelve genes, i.e. UBE2C, S100A8, CBX2, LOC389033, STC2, DNALI1,

SCUBE2, NME5, SUSD3, SERPINA11, AZGP1 and PIP, in breast carcinomas with a

malignant phenotype. Functional studies have implicated several of these genes in a role

in tumor cell growth, motility, and progression in multiple cancer types (37-44). Here, we

could associate all twelve genes with progesterone expression and eleven with estrogen

expression (UBE2C excluded). Similarly, Gruvberger et al. (45) identified S100A8 and STC2

as estrogen receptor antagonist and responsive genes. Furthermore, Berlingieri et al. (46)

revealed that suppression of ERBB2 inhibits UBE2C activity and thereby cell growth.

Another intriguing finding was the identification of a single transcript (CNTNAP2)

differently deregulated between the axillary lymph node groups. This transcript spans 2.3

Mb of chromosome 7 which makes it the largest known gene in the human genome and

is also located at a region containing a common fragile site (FRA7I) (47, 48). We did not,

however, observe loss of the 7q35-q36.1 chromosomal region spanning the CNTNAP2 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Copyright © 2010 American Association for Cancer Research

gene that could explain down-regulation in pN0 tumors. Expression of the CNTNAP2

protein has been primarily implicated in disorders of the vertebrate nervous system for

its functions as cell adhesion molecules and receptors (49). In cancer, the CNTNAP2

promoter has been shown to be hypermethylated in 82% of pancreatic cancers in

comparison with 3% in normal pancreas (50). Hypermethylation of the CNTNAP2

promoter may be one mechanism by which the transcriptional levels of this gene may be

silenced. In order to determine the methylation status of this gene and to establish its

role in breast tumor development and progression, further methylation studies using

breast neoplasms are warranted.

In summary, we have shown the feasibility of using microarray technology as a tool to

identify common biological features among breast carcinomas which can be used as

potential therapeutic targets. We identified eighteen recurrent regions of genetic

alteration in our dataset, of which 48 transcripts were abnormally expressed. In addition,

the expression levels of 12 genes displaying normal copy numbers were associated with

malignant phenotypes in breast carcinoma. This approach identified good candidates for

further investigations using an independent series of multiple cancer types to assess the

biological relevance of elevated DNA and/or mRNA levels on protein gene product. Taken

together, this information can potentially be used to establish cost-effective targeted and

individualized treatment regimens that will directly benefit a specific patient based on

the genetic/transcriptomic profile of their associated tumor to target specific cellular

pathways perturbed in the tumor without affecting the activity of non-neoplastic cells. Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Copyright © 2010 American Association for Cancer Research

Acknowledgments

We wish to thank Marcela Davila for invaluable bioinformatics support.

References

1. Vogelstein B, Kinzler KW. The multistep nature of cancer. Trends Genet 1993;9: 138-41.

2.Futreal PA, Coin L, Marshall M, et al. A census of human cancer genes. Nat Rev Cancer

2004;4: 177-83.

3.Albertson DG, Collins C, McCormick F, Gray JW. Chromosome aberrations in solid

tumors. Nat Genet 2003;34: 369-76.

4.Tsuda H. Gene and chromosomal alterations in sporadic breast cancer: correlation with

histopathological features and implications for genesis and progression. Breast Cancer

2009;16: 186-201.

5.Pinkel D, Segraves R, Sudar D, et al. High resolution analysis of DNA copy number

variation using comparative genomic hybridization to microarrays. Nat Genet 1998;20:

207-11.

6.Hyman E, Kauraniemi P, Hautaniemi S, et al. Impact of DNA amplification on gene

expression patterns in breast cancer. Cancer Res 2002;62: 6240-5.

7.Reyal F, Stransky N, Bernard-Pierrot I, et al. Visualizing chromosomes as transcriptome

correlation maps: evidence of chromosomal domains containing co-expressed genes--a

study of 130 invasive ductal breast carcinomas. Cancer Res 2005;65: 1376-83.

8.Farabegoli F, Santini D, Ceccarelli C, Taffurelli M, Marrano D, Baldini N. Clone

heterogeneity in diploid and aneuploid breast carcinomas as detected by FISH. Cytometry

2001;46: 50-6.

9.Hicks J, Krasnitz A, Lakshmi B, et al. Novel patterns of genome rearrangement and their

association with survival in breast cancer. Genome Res 2006;16: 1465-79.

10.Hicks J, Muthuswamy L, Krasnitz A, et al. High-resolution ROMA CGH and FISH analysis

of aneuploid and diploid breast tumors. Cold Spring Harb Symp Quant Biol 2005;70: 51-

63.

11.Rennstam K, Baldetorp B, Kytola S, Tanner M, Isola J. Chromosomal rearrangements

and oncogene amplification precede aneuploidization in the genetic evolution of breast

cancer. Cancer Res 2001;61: 1214-9.

12.Ried T, Heselmeyer-Haddad K, Blegen H, Schrock E, Auer G. Genomic changes defining

the genesis, progression, and malignancy potential in solid human tumors: a

phenotype/genotype correlation. Genes Chromosomes Cancer 1999;25: 195-204.

13.Tanner MM, Karhu RA, Nupponen NN, et al. Genetic aberrations in hypodiploid breast

cancer: frequent loss of chromosome 4 and amplification of cyclin D1 oncogene. Am J

Pathol 1998;153: 191-9.

14.Tirkkonen M, Tanner M, Karhu R, Kallioniemi A, Isola J, Kallioniemi OP. Molecular

cytogenetics of primary breast cancer by CGH. Genes Chromosomes Cancer 1998;21:

177-84.

15.Truong K, Vielh P, Guilly MN, et al. Quantitative FISH analysis on interphase nuclei may

improve diagnosis of DNA diploid breast cancers. Diagn Cytopathol 2002;26: 213-6.

16.Habermann JK, Doering J, Hautaniemi S, et al. The gene expression signature of

genomic instability in breast cancer is an independent predictor of clinical outcome. Int J

17.Ried T, Just KE, Holtgreve-Grez H, et al. Comparative genomic hybridization of

formalin-fixed, paraffin-embedded breast tumors reveals different patterns of

chromosomal gains and losses in fibroadenomas and diploid and aneuploid carcinomas.

Cancer Res 1995;55: 5415-23.

18.Leers MP, Nap M. Steroid receptor heterogeneity in relation to DNA index in breast

cancer: a multiparameter flow cytometric approach on paraffin-embedded tumor

samples. Breast J 2001;7: 249-59.

19.Jonsson G, Staaf J, Olsson E, et al. High-resolution genomic profiles of breast cancer

cell lines assessed by tiling BAC array comparative genomic hybridization. Genes

Chromosomes Cancer 2007;46: 543-58.

20.Saal LH, Troein C, Vallon-Christersson J, Gruvberger S, Borg A, Peterson C. BioArray

Software Environment (BASE): a platform for comprehensive management and analysis

of microarray data. Genome Biol 2002;3: SOFTWARE0003.

21.Moore SR, Persons DL, Sosman JA, et al. Detection of copy number alterations in

metastatic melanoma by a DNA fluorescence in situ hybridization probe panel and array

comparative genomic hybridization: a southwest oncology group study (S9431). Clin

Cancer Res 2008;14: 2927-35.

22.Iafrate AJ, Feuk L, Rivera MN, et al. Detection of large-scale variation in the human

genome. Nat Genet 2004;36: 949-51.

23.Lim E, Vaillant F, Wu D, et al. Aberrant luminal progenitors as the candidate target

population for basal tumor development in BRCA1 mutation carriers. Nat Med 2009;15:

24.Benjamini Y, Hochberg Y. Controlling the False Discovery Rate: A Practical and

Powerful Approach to Multiple Testing. J R Stat Soc 1995;57: 289-300.

25.Hu Z, Fan C, Oh DS, et al. The molecular portraits of breast tumors are conserved

across microarray platforms. BMC Genomics 2006;7: 96.

26.Loi S, Haibe-Kains B, Desmedt C, et al. Definition of clinically distinct molecular

subtypes in estrogen receptor-positive breast carcinomas through genomic grade. J Clin

Oncol 2007;25: 1239-46.

27.Sotiriou C, Wirapati P, Loi S, et al. Gene expression profiling in breast cancer:

understanding the molecular basis of histologic grade to improve prognosis. J Natl Cancer

Inst 2006;98: 262-72.

28.Karlsson E, Delle U, Danielsson A, et al. Gene expression variation to predict 10-year

survival in lymph-node-negative breast cancer. BMC Cancer 2008;8: 254.

29.Kent WJ, Sugnet CW, Furey TS, et al. The human genome browser at UCSC. Genome

Res 2002;12: 996-1006.

30.Andre F, Job B, Dessen P, et al. Molecular characterization of breast cancer with high-

resolution oligonucleotide comparative genomic hybridization array. Clin Cancer Res

2009;15: 441-51.

31.Fridlyand J, Snijders AM, Ylstra B, et al. Breast tumor copy number aberration

phenotypes and genomic instability. BMC Cancer 2006;6: 96.

32.Haverty PM, Fridlyand J, Li L, et al. High-resolution genomic and expression analyses of

copy number alterations in breast tumors. Genes Chromosomes Cancer 2008;47: 530-42.

33.Loo LW, Grove DI, Williams EM, et al. Array comparative genomic hybridization

analysis of genomic alterations in breast cancer subtypes. Cancer Res 2004;64: 8541-9.

34.Naylor TL, Greshock J, Wang Y, et al. High resolution genomic analysis of sporadic

breast cancer using array-based comparative genomic hybridization. Breast Cancer Res

2005;7: R1186-98.

35.Nessling M, Richter K, Schwaenen C, et al. Candidate genes in breast cancer revealed

by microarray-based comparative genomic hybridization of archived tissue. Cancer Res

2005;65: 439-47.

36.Knuutila S, Bjorkqvist AM, Autio K, et al. DNA copy number amplifications in human

neoplasms: review of comparative genomic hybridization studies. Am J Pathol 1998;152:

1107-23.

37.Arai K, Takano S, Teratani T, Ito Y, Yamada T, Nozawa R. S100A8 and S100A9

overexpression is associated with poor pathological parameters in invasive ductal

carcinoma of the breast. Curr Cancer Drug Targets 2008;8: 243-52.

38.Chapman EJ, Kelly G, Knowles MA. Genes involved in differentiation, stem cell

renewal, and tumorigenesis are modulated in telomerase-immortalized human urothelial

cells. Mol Cancer Res 2008;6: 1154-68.

39.Cheng CJ, Lin YC, Tsai MT, et al. SCUBE2 suppresses breast tumor cell proliferation and

confers a favorable prognosis in invasive breast cancer. Cancer Res 2009;69: 3634-41.

40.Cunha IW, Carvalho KC, Martins WK, et al. Identification of genes associated with local

aggressiveness and metastatic behavior in soft tissue tumors. Transl Oncol 2010;3: 23-32.

41.Fujita T, Ikeda H, Taira N, Hatoh S, Naito M, Doihara H. Overexpression of UbcH10

alternates the cell cycle profile and accelerate the tumor proliferation in colon cancer.

BMC Cancer 2009;9: 87.

42.Hale LP, Price DT, Sanchez LM, Demark-Wahnefried W, Madden JF. Zinc alpha-2-

glycoprotein is expressed by malignant prostatic epithelium and may serve as a potential

serum marker for prostate cancer. Clin Cancer Res 2001;7: 846-53.

43.Hassan MI, Waheed A, Yadav S, Singh TP, Ahmad F. Prolactin inducible protein in

cancer, fertility and immunoregulation: structure, function and its clinical implications.

Cell Mol Life Sci 2009;66: 447-59.

44.Okamoto Y, Ozaki T, Miyazaki K, Aoyama M, Miyazaki M, Nakagawara A. UbcH10 is the

cancer-related E2 ubiquitin-conjugating enzyme. Cancer Res 2003;63: 4167-73.

45.Gruvberger S, Ringner M, Chen Y, et al. Estrogen receptor status in breast cancer is

associated with remarkably distinct gene expression patterns. Cancer Res 2001;61: 5979-

84.

46.Berlingieri MT, Pallante P, Sboner A, et al. UbcH10 is overexpressed in malignant

breast carcinomas. Eur J Cancer 2007;43: 2729-35.

47.McAvoy S, Ganapathiraju SC, Ducharme-Smith AL, et al. Non-random inactivation of

large common fragile site genes in different cancers. Cytogenet Genome Res 2007;118:

260-9.

48.Smith DI, Zhu Y, McAvoy S, Kuhn R. Common fragile sites, extremely large genes,

neural development and cancer. Cancer Lett 2006;232: 48-57.

49.O'Dushlaine C, Kenny E, Heron E, et al. Molecular pathways involved in neuronal cell

adhesion and membrane scaffolding contribute to schizophrenia and bipolar disorder

susceptibility. Mol Psychiatry 2010.

50.Omura N, Li CP, Li A, et al. Genome-wide profiling of methylated promoters in

pancreatic adenocarcinoma. Cancer Biol Ther 2008;7: 1146-56.

Figures

Figure 1. Genome-wide frequency plots of gains and losses in 97 DBC samples. The x-axis

corresponds to the genomic region from chromosome 1 to X and the y-axis to

the percentage of gains and losses. Green, dark green, and red bars represent

the percentage of LLG, HLG, and HL in the given chromosomal region,

respectively. A, the frequency of LLG, HLG, and HL in DBC. B, schematic

overview of HLG in DBC. Statistically significant regions (P<0.01) present in

≥10% of DBC samples are indicated with black arrows showing candidate genes

located in the region. Previously reported genomic regions in breast cancer,

present in <10% of DBC samples, are indicated with red arrows. C, statistically

significant regions (≥25% difference) are shown in green and red in the upper

panel. The frequency of CNAs for the two clusters generated by unsupervised

hierarchical clustering using complete linkage with Pearson correlation are

shown in the two lower panels.

Figure 2. Schematic overview of HER2/neu gene amplification. A, depicts the genomic

profile covering chromosome 17 for tumors 8295 (black dots) and 8491 (gray

dots). BAC probes containing the HER2/neu gene are colored in green. The x- Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Copyright © 2010 American Association for Cancer Research

axis corresponds to the array-CGH log2ratio and the y-axis to the genomic

position along chromosome 17. B, a zoom-in of chromosome 17 spanning from

the centromere to the HER2/neu locus. C and D, dual-color FISH hybridization of

the ZytoLight SPEC HER2/CEN 17 Dual Color Probe with HER2 (green-labeled

fluorochrome) and CEN 17 (orange-labeled fluorochrome) for tumors 8295 and

8491, respectively. Tumor 8295 is shown with amplification of both the

centromere and HER2 locus while 8491 is amplified at the HER2 locus only.

Figure 3. Robust piecewise linear regression analysis showing the relationship between

gene dosage and transcriptional levels for the HER2/neu gene in 97 DBC

samples. The Illumina log2 (ILMN_2352131) and BAC log2ratio (RP11-909L6)

were compared using Pearson correlation. Shaded areas indicate the 95%

confidence intervals of the regression line.

Tables

Table 1. Clinicopathological characteristics of 97 patients with DBC.

Table 2. Comparison of recurrent CNAs in the current study with previous reports.

Table 3. Total deregulated transcripts and their involvement in GO biological processes.

Supplementary tables

S1. Clinicopathological and experimental characteristics of 97 patients with DBC.

S2. Correlation of the number of detected CNAs and clinicopathological characteristics of

97 patients with DBC.

S3. Recurrent gains and losses identified in the sample population using BioDiscovery’s

Rank Segmentation algorithm.

S4. Differentially regulated transcripts identified with 1.5-fold change using Nexus

Expression.

S5. Correlation of gene expression analyses performed using Illumina HumanHT-12

probes and TaqMan® assays.

S6. Correlation of recurrent CNAs and gene expression levels.

Supplementary figures

F1. DNA ploidy analysis of 5 DBC patients assessed by flow cytometry.

Supplementary Materials and Methods

A B

17p13.3 2352.22.9 5512.4 7722.9 9833.4 2043. 41535.251 34.175 33.1 32.024 30.949 29.873 28.798 27.722 26.647 25.571 24.496 23.42 22.345

CEN 17 (17p11.1-q11.1) 17p13.1 C

17p11.2

17q11.1

17q12

HER2/neu

17q21.2

17q21.32 D 17q22

17q23.2

17q24.1

17q24.3

17q25.2 HER2/neu (17q12) (Mb)

−1 −0.5 −0.2 0 0.2 0.5 1 1.5 2 2.5 −1 −0.5 −0.2 0 0.2 0.5 1 1.5 2 2.5 Log2ratio Log2ratio Figure 2 / Parris TZ (CCR-10-0889)

HER2/neu (ILMN_2352131/RP11−909L6)

• • • • • • • 2 •

) • 2 •

••• Status •• ••• • HER2/neu ampliﬁcation 0 • ••••• • • • ••••••••••• • No ampliﬁcation • •• ••••• ••• • • • ••••• • • ••• • −1 •••• •• • •• • ••••• • •••• • Relative Gene Expression (Log • •• • •• • • • −2 •

• r = 0.51 0.0 0.5 1.0 1.5 2.0 Downloaded from clincancerres.aacrjournals.orgArray−CGH Log2Ratio on September 30, 2021. © 2010 American Association for Cancer Research. Figure 3 / Parris TZ (CCR-10-0889) Table 1. Clinicopathological characteristics of 97 patients with DBC.

Characteristic pN0 long-term survivors pN1 long-term survivors pN0 short-term survivors pN1 short-term survivors (N=25) (N=25) (N=22) (N=25) no of patients (%) Mean age (range) 58 (40-78) 57 (33-77) 62 (36-79) 51 (30-78) Histologic type Ductal 18 (72) 22 (88) 18 (82) 13 (52) Lobular Author Manuscript Published4 OnlineFirst (16) on June 15, 2010; DOI: 10.1158/1078-0432.CCR-10-08891 (4) 1 (5) 5 (20) Other Author manuscripts have been 3peer (12) reviewed and accepted for publication2 (8) but have not yet been edited. 3 (1) 7 (28)

Pathologic tumor size pT1 12 (48) 5 (20) 5 (9) 4 (16) pT2 9 (36) 15 (60) 15 (68) 12 (48) pT3 4 (16) 4 (16) 2 (9) 5 (20) pT4 0 (0) 1 (4) 0 (0) 4 (16)

SBR grade I 7 (28) 4 (16) 1 (5) 1 (4) II 15 (60) 12 (48) 9 (41) 12 (48) III 0 (0) 5 (20) 6 (27) 6 (24) Non-available 3 (12) 4 (16) 6 (27) 6 (24)

GGI Low 14 (56) 14 (56) 7 (32) 4 (16) High 9 (36) 9 (36) 9 (41) 15 (60) Non-available 2 (8) 2 (8) 6 (27) 6 (24)

No. of positive axillary lymph nodes 0 25 (100) 0 (0) 22 (100) 0 (0) 1 - 3 0 (0) 18 (72) 0 (0) 5 (20) ≥4 0 (0) 7 (28) 0 (0) 20 (80)

Neoadjuvant therapy Yes 0 (0) 0 (0) 1 (5) 1 (4) No 5 (20) 6 (24) 3 (14) 5 (20) Non-available 20 (80) 19 (76) 18 (82) 19 (76)

Surgery Lumpectomy 12 (48) 8 (32) 11 (50) 4 (16) Mastectomy 11 (44) 16 (64) 9 (41) 19 (76) Non-available 2 (8) 1 (4) 2 (9) 2 (8)

Chemotherapy Yes 1 (4) 13 (52) 1 (5) 15 (60) No 20 (80) 8 (32) 18 (82) 8 (32) Non-available 4 (16) 4 (16) 3 (14) 2 (8)

Endocrine therapy Yes 6 (24) 17 (68) 9 (41) 12 (48) No 16 (64) 3 (12) 10 (45) 10 (40) Non-available 3 (12) 5 (20) 3 (14) 3 (12)

Radiotherapy Yes 10 (40) 6 (24) 7 (32) 10 (40) No 13 (52) 15 (60) 12 (55) 12 (48) Non-available 2 (8) 4 (16) 3 (14) 3 (12)

Estrogen receptor Negative 2 (8) 2 (8) 6 (27) 6 (24) Positive 23 (92) 23 (92) 16 (73) 19 (76)

Progesterone receptor Negative 9 (36) 4 (16) 14 (64) 10 (40) Positive 16 (64) 21 (84) 8 (36) 15 (60)

HER2 status Negative 23 (92) 22 (88) 18 (82) 23 (92) Positive 2 (8) 3 (12) 4 (18) 2 (8)

Triple negative status Negative 23 (92) 25 (100) 16 (73) 22 (88) Positive 2 (8) 0 (0) 6 (27) 3 (12)

Current study (n = 97) Previous studies Cytoband Number of % of CNV Cytoband Region Size (Mb) Location Breast cancer-related candidate genes genes Frequency % overlap P-value Location Frequency % Reference Low-level gain chr1:151803323-152187362 0.4 1q22 DCST1, ADAM15, EFNA1, MUC1, C1orf2, FDPS 26 53 73 0.009 1q11-q23 25-48 [32, 35] chr1:153006542-154155681 1.1 1q22-q23.1 PMF1, CRABP2, C1orf66, NTRK1, INSRR 34 53 0 0.009 1q11-q23 48 [35] chr1:178971831-179547540 0.6 1q25.3 RGSL1, RNASEL, RGS16, DHX9 8 53 2 0.009 1q25-q31 24-45 [32, 35] chr1:197639011-202792328 5.2 1q32.1 LAD1, ELF3, UBE2T, JARID1B, ADIPOR1, ADORA1, MYOG, BTG2, KISS1, MDM4 82 53 13 0.009 1q32.1-q32.3 27-66 [32, 34] chr1:202921401-205404357 2.5 1q32.1-q32.2 SRGAP2, IKBKE, RASSF5, MAPKAPK2, IL10, IL24, CD55, CD34 26 53 34 0.009 1q32.1-q32.2 22-66 [32, 34] chr8:86674253-87135834 0.5 8q21.2-q21.3 5 33 98 0.000 8q21.12-q22.3 25 [32] chr8:102067884-102376617 0.3 8q22.3 1 26 0 0.000 chr8:145853807-146082345 0.2 8q24.3 RPL8 7 26 17 0.000 8q24 27-65 [30, 32, 35] chr16:31371886-31439871 0.1 16p11.2 TGFB1I1 4 26 0 0.000 16p11.2 12 [32] Author Manuscript Published OnlineFirst on June 15, 2010; DOI: 10.1158/1078-0432.CCR-10-0889 AuthorHeterozygous manuscripts loss have been peer reviewed and accepted for publication but have not yet been edited. chr11:131205533-131405980 0.2 11q25 1 26 24 0.000 11q22.3-q25 14-40 [32, 33] chr16:59474046-61647365 2.2 16q21 CDH8 1 38 9 0.004 16q21-q25 20-40 [9, 30, 32, 33] chr16:61922081-63590892 1.7 16q21 CDH11 1 38 10 0.004 16q21-q25 20-40 [30, 32, 33, 34] chr16:80651087-82423783 1.8 16q23.3 HSD17B2, CDH13, HSBP1 4 38 15 0.004 16q21-q25 18-40 [30, 32, 33] chr17:11053362-12022428 1.0 17p12 MAP2K4 4 26 7 0.006 17p13-p11 16-40 [9, 32, 33, 34, 35] chr17:12335195-15480166 3.1 17p12 ELAC2, PMP22, TRIM16 14 26 56 0.006 17p13-p11 16-40 [9, 32, 33, 35]

High-level gain/amplification chr1:197688254-201299020 3.6 1q32.1 LAD1, ELF3, UBE2T, JARID1B, ADIPOR1, ADORA1, MYOG, BTG2, KISS1, MDM4 60 20 13 0.001 chr1:204387556-205011692 0.6 1q32.2 CD34 2 20 36 0.001 chr11:69286642-70504982 1.2 11q13.3-q13.4 FGF4, FGF3, TMEM16A, FADD, PPFIA1, CTTN 7 10 11 0.000 11q13.3-q13.4 22-77 [9, 31, 32, 35]

Table 3. Total deregulated transcripts and their involvement in GO biological processes.

Gene expression changes (number of transcripts) 2-fold change 1.5-fold change Group comparisons Up Down Up Down Short-term survivors vs. long-term survivors 8 9 47 51

Estrogen receptor positive vs. receptor negative 122 77 353 287

Progesterone receptor positive vs. receptor negative 35 6 133 37 Receptor positive vs. negative 131 77 364 289 pN1, short-term survivors vs. long-term survivors 4 6 4 8 pN0, short-term survivors vs. long-term survivors 6 21 17 36

1-3 positive axillary lymph nodes, short-term survivors vs. long-term survivors 12 1 12 1

≥4 positive axillary lymph nodes, short-term survivors vs. long-term survivors 4 11 4 13 Short-term survivors, pN0 vs. 1-3 positive axillary lymph nodes 1 15 1 15

Short-term survivors, ≥4 vs. 1-3 positive axillary lymph nodes 4 14 4 14

Short-term survivors, ≥4 vs. pN0 positive axillary lymph nodes 2 1 2 1 HER2 negative vs. positive 3 13 3 15

pN0 vs. pN1 0 1 0 1 pN0 vs. ≥4 positive axillary lymph nodes 0 1 0 1 1-3 vs. ≥4 positive axillary lymph nodes 0 3 0 3 Long-term survivors, pN0 vs. ≥4 positive axillary lymph nodes 0 9 0 9 Long-term survivors, 1-3 vs. ≥4 positive axillary lymph nodes 0 10 0 10 Cluster 2 vs. Cluster 1 1 0 2 3

SBR grade 1 vs. 2 1 0 1 0 SBR grade 1 vs. 3 26 7 28 10

SBR grade 2 vs. 3 6 5 6 7

High vs. low GGI 45 14 181 105

Triple-negative vs. Non triple-negative 106 149 294 347

Gene Ontology biological process associated with differentially regulated genes (1.5-fold change) mitosis, cell division, cell cycle, positive regulation of transcription; DNA-dependent, DNA repair, negative regulation of signal transduction response to estradiol stimulus, anti-apoptosis, regulation of inflammatory response, regulation of cell proliferation, EGFR signaling pathway, negative regulation of mitotic cell cycle, negative regulation of cell proliferation potassium ion and chloride transport, cell aging, response to estrogen stimulus, regulation of cell proliferation, anti-apoptosis response to estradiol stimulus, regulation of inflammatory response, chloride transport, negative regulation of cell proliferation, negative regulation of mitotic cell cycle cellular metabolic processes, cyclin catabolic process, positive regulation of exit from mitosis, tyrosine catabolic process, endothelial cell migration response to estradiol stimulus, cholesterol efflux and homeostasis, negative regulation of TGFB3 production, regulation of cell-cell adhesion, regulation of blood vessel size by renin-angiotensin, regulation of gene-specific transcription, immune response-regulating cell surface receptor signaling pathway, apoptotic chromosome condensation, regulation of cytokine-mediated signaling pathway, negative regulation of blood vessel endothelial cell migration, regulation of epithelial cell proliferation, regulation of MAPKKK cascade nucleosome positioning and assembly, response to estradiol stimulus, blood vessel morphogenesis, endothelial cell migration, calcium-dependent cell-cell adhesion, cell migration vitamin D metabolic process, tyrosine catabolic process, cellular metabolic process, nucleosome assembly, response to estradiol stimulus, estrogen receptor signaling pathway, regulation of transcription; DNA-dependent, Notch signaling pathway regulation of blood vessel size by renin-angiotensin, C21-steroid hormone metabolic process, regulation of vasodilation, positive regulation of inflammatory response, response to toxin transmission of nerve impulse, neuron recognition, cellular alcohol metabolic process, collagen catabolic process induction of apoptosis by oxidative stress, negative regulation of immature T cell proliferation in the thymus, FGFR signaling pathway, positive regulation of MAP kinase activity, positive regulation of cell adhesion, EGFR signaling pathway, positive regulation of epithelial cell proliferation, cell-cell signaling, mammary gland development transmission of nerve impulse, neuron recognition, cell adhesion transmission of nerve impulse, neuron recognition, cell adhesion induction of apoptosis tyrosine catabolic process, cellular metabolic process, biosynthetic process, transport response to estradiol stimulus, TGFBR signaling pathway, transport regulation of hormone secretion, peptide hormone processing, antigen processing and presentation, neuropeptide signaling pathway, intracellular protein transport multicellular organismal development cyclin catabolic process, regulation of integrin biosynthetic process, positive regulation of exit from mitosis, regulation of mitosis, cell-cell signaling, tyrosine catabolic process regulation of integrin biosynthetic process, positive regulation of microtubule polymerization, response to estradiol stimulus, inflammatory response cell cycle, mitosis, cell division, DNA replication, phosphoinositide-mediated signaling, DNA repair, cell proliferation, regulation of cyclin-dependent protein kinase activity, protein amino acid phosphorylation, response to DNA damage stimulus, cell-cell signaling, cell differentiation, FGFR signaling pathway response to estradiol stimulus, regulation of cytokine biosynthetic process, regulation of inflammatory response, DNA replication checkpoint, positive regulation of interleukin-4 and 8 production, positive regulation of exit from mitosis, anti-apoptosis, cell-cell signaling, mammary gland development, regulation of cell proliferation, negative regulation of tumor necrosis factor production

Clinical Implications of Gene Dosage and Gene Expression Patterns in Diploid Breast Carcinoma

Toshima Z Parris, Anna Danielsson, Szilárd Nemes, et al.

Clin Cancer Res Published OnlineFirst June 15, 2010.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-10-0889

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2010/06/16/1078-0432.CCR-10-0889.DC1

Author Author manuscripts have been peer reviewed and accepted for publication but have not yet been Manuscript edited.

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/early/2010/06/15/1078-0432.CCR-10-0889. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.