Top Banner
Integrated Systems and Technologies microRNA-Associated Progression Pathways and Potential Therapeutic Targets Identified by Integrated mRNA and microRNA Expression Profiling in Breast Cancer Francesca M. Buffa 1 , Carme Camps 2 , Laura Winchester 2 , Cameron E. Snell 1 , Harriet E. Gee 1 , Helen Sheldon 1 , Marian Taylor 1 , Adrian L. Harris 1 , and Jiannis Ragoussis 2 Abstract microRNA expression profiling plays an emerging role in cancer classification and identification of therapeutic strategies. In this study, we have evaluated the benefits of a joint microRNAmRNA analysis in breast cancer. Matched mRNA and microRNA global expression profiling was conducted in a well-annotated cohort of 207 cases with complete 10-year follow-up. Penalized Cox regression including microRNA expression, mRNA expression, and clinical covariates was used to identify microRNAs associated with distant relapse-free survival (DRFS) that provide independent prognostic information, and are not simply surrogates of previously identified prognostic covariates. Penalized regression was chosen to prevent overfitting. Furthermore, microRNAmRNA relationships were explored by global expression analysis, and exploited to validate results in several published cohorts (n ¼ 592 with DRFS, n ¼ 1,050 with recurrence-free survival). Four microRNAs were independently associated with DRFS in estrogen receptor (ER)-positive (3 novel and 1 known; miR-128a) and 6 in ER-negative (5 novel and 1 known; miR-210) cases. Of the latter, miR-342, -27b, and -150 were prognostic also in triple receptor-negative tumors. Coordinated expression of predicted target genes and prognostic microRNAs strengthened these results, most significantly for miR-210, -128a, and -27b, whose targets were prognostic in meta-analysis of several cohorts. In addition, miR-210 and -128a showed coordinated expression with their cognate pri-microRNAs, which were themselves prognostic in independent cohorts. Our integrated microRNAmRNA global profiling approach has identified microRNAs independently asso- ciated with prognosis in breast cancer. Furthermore, it has validated known and predicted microRNAtarget interactions, and elucidated their association with key pathways that could represent novel therapeutic targets. Cancer Res; 71(17); 563545. Ó2011 AACR. Introduction Breast cancer is a heterogeneous disease, with treatment resistance a consequence of pathologic and genomic char- acteristics. mRNA expression profiling in clinical cohorts has led to identification of functional pathways with roles in tumor progression and collections of genes (gene signatures) associated with disease outcome (1), some of which are now approved by Food and Drug Administration for clinical use, such as MammaPrint and OncotypeDX. microRNAs are small noncoding RNA molecules regulating cell function both at transcriptional and posttranscriptional levels which open up a new area of prognostic marker research complementary to established transcriptional gene signature or traditional marker studies (2). A study by Blenkiron and colleagues of 93 breast cancers identified several human microRNAs associated with intrinsic breast cancer subtype (3). Another recent study of 38 breast cancers (4) selected 12 microRNAs associated with clinicopathologic variables for analysis in a cohort of 261 breast cancers. Among these, 4 (miR-7, -128a, -210, and -516-3p) were prognostic, miR-210 having been previously identified (5). However, integrated analysis of microRNA and mRNA global expression profiles has yet to be explored in prognostic studies. Such analyses have the potential not only to identify microRNAs that are independent prognostic factors, but also to elucidate microRNA function in vivo, and identify interac- tions between microRNAs and targeted mRNAs for enhanced marker and therapeutic target discovery. Thus, we conducted comprehensive microRNA and mRNA expression profiling in a large, well-annotated cohort of 207 early-invasive breast cancers. To identify microRNAs that provide independent informa- tion, and are not simply surrogates of previously identified Authors' Affiliations: 1 Oncology Department, Weatherall Institute of Molecular Medicine, University of Oxford and 2 Genomics Research, Well- come Trust Centre for Human Genetics, Oxford, United Kingdom Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/). A.L. Harris and J. Ragoussis are the senior co-leads. Corresponding Authors: Francesca M. Buffa, The Weatherall Institute of Molecular Medicine, University of Oxford, OX3 9DS, United Kingdom. Phone: 44-01865-222443; Fax: 44-01865-222737; E-mail: francesca. [email protected] or Jiannis Ragoussis, The Wellcome Trust Centre for Human Genetics, Roosevelt Drive, Oxford, OX3 7BN, United Kingdom. Phone: 44-01865-287526; Fax: 44-01865287533; E-mail: [email protected] doi: 10.1158/0008-5472.CAN-11-0489 Ó2011 American Association for Cancer Research. Cancer Research www.aacrjournals.org 5635 on August 15, 2020. © 2011 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from Published OnlineFirst July 7, 2011; DOI: 10.1158/0008-5472.CAN-11-0489
12

microRNA-Associated Progression Pathways and Potential … · To identify microRNAs that provide independent informa-tion, and are not simply surrogates of previously identified Authors'

Jul 10, 2020

Download

Documents

dariahiddleston
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: microRNA-Associated Progression Pathways and Potential … · To identify microRNAs that provide independent informa-tion, and are not simply surrogates of previously identified Authors'

Integrated Systems and Technologies

microRNA-Associated Progression Pathways and PotentialTherapeutic Targets Identified by Integrated mRNA andmicroRNA Expression Profiling in Breast Cancer

Francesca M. Buffa1, Carme Camps2, Laura Winchester2, Cameron E. Snell1, Harriet E. Gee1,Helen Sheldon1, Marian Taylor1, Adrian L. Harris1, and Jiannis Ragoussis2

AbstractmicroRNA expression profiling plays an emerging role in cancer classification and identification of therapeutic

strategies. In this study, we have evaluated the benefits of a joint microRNA–mRNA analysis in breast cancer.Matched mRNA and microRNA global expression profiling was conducted in a well-annotated cohort of 207

cases with complete 10-year follow-up. Penalized Cox regression including microRNA expression, mRNAexpression, and clinical covariates was used to identify microRNAs associated with distant relapse-free survival(DRFS) that provide independent prognostic information, and are not simply surrogates of previously identifiedprognostic covariates. Penalized regression was chosen to prevent overfitting. Furthermore, microRNA–mRNArelationships were explored by global expression analysis, and exploited to validate results in several publishedcohorts (n ¼ 592 with DRFS, n ¼ 1,050 with recurrence-free survival).Four microRNAs were independently associated with DRFS in estrogen receptor (ER)-positive (3 novel and 1

known; miR-128a) and 6 in ER-negative (5 novel and 1 known; miR-210) cases. Of the latter, miR-342, -27b, and-150 were prognostic also in triple receptor-negative tumors. Coordinated expression of predicted target genesand prognostic microRNAs strengthened these results, most significantly for miR-210, -128a, and -27b, whosetargets were prognostic in meta-analysis of several cohorts. In addition, miR-210 and -128a showed coordinatedexpression with their cognate pri-microRNAs, which were themselves prognostic in independent cohorts.Our integrated microRNA–mRNA global profiling approach has identified microRNAs independently asso-

ciated with prognosis in breast cancer. Furthermore, it has validated known and predicted microRNA–targetinteractions, and elucidated their association with key pathways that could represent novel therapeutic targets.Cancer Res; 71(17); 5635–45. �2011 AACR.

Introduction

Breast cancer is a heterogeneous disease, with treatmentresistance a consequence of pathologic and genomic char-acteristics. mRNA expression profiling in clinical cohorts hasled to identification of functional pathways with roles intumor progression and collections of genes (gene signatures)associated with disease outcome (1), some of which are nowapproved by Food and Drug Administration for clinical use,such as MammaPrint and OncotypeDX.

microRNAs are small noncoding RNA molecules regulatingcell function both at transcriptional and posttranscriptionallevels which open up a new area of prognostic marker researchcomplementary to established transcriptional gene signatureor traditional marker studies (2). A study by Blenkiron andcolleagues of 93 breast cancers identified several humanmicroRNAs associated with intrinsic breast cancer subtype(3). Another recent study of 38 breast cancers (4) selected 12microRNAs associated with clinicopathologic variables foranalysis in a cohort of 261 breast cancers. Among these, 4(miR-7, -128a, -210, and -516-3p) were prognostic, miR-210having been previously identified (5).

However, integrated analysis of microRNA and mRNAglobal expression profiles has yet to be explored in prognosticstudies. Such analyses have the potential not only to identifymicroRNAs that are independent prognostic factors, but alsoto elucidate microRNA function in vivo, and identify interac-tions between microRNAs and targeted mRNAs for enhancedmarker and therapeutic target discovery. Thus, we conductedcomprehensive microRNA and mRNA expression profilingin a large, well-annotated cohort of 207 early-invasive breastcancers.

To identify microRNAs that provide independent informa-tion, and are not simply surrogates of previously identified

Authors' Affiliations: 1Oncology Department, Weatherall Institute ofMolecular Medicine, University of Oxford and 2Genomics Research, Well-come Trust Centre for Human Genetics, Oxford, United Kingdom

Note: Supplementary data for this article are available at Cancer ResearchOnline (http://cancerres.aacrjournals.org/).

A.L. Harris and J. Ragoussis are the senior co-leads.

Corresponding Authors: Francesca M. Buffa, The Weatherall Institute ofMolecular Medicine, University of Oxford, OX3 9DS, United Kingdom.Phone: 44-01865-222443; Fax: 44-01865-222737; E-mail: [email protected] or Jiannis Ragoussis, The Wellcome Trust Centrefor Human Genetics, Roosevelt Drive, Oxford, OX3 7BN, United Kingdom.Phone: 44-01865-287526; Fax: 44-01865287533; E-mail:[email protected]

doi: 10.1158/0008-5472.CAN-11-0489

�2011 American Association for Cancer Research.

CancerResearch

www.aacrjournals.org 5635

on August 15, 2020. © 2011 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst July 7, 2011; DOI: 10.1158/0008-5472.CAN-11-0489

Page 2: microRNA-Associated Progression Pathways and Potential … · To identify microRNAs that provide independent informa-tion, and are not simply surrogates of previously identified Authors'

prognostic covariates, a Cox regression for distant relapse-freesurvival (DRFS) was carried out, including all microRNAs,clinical covariates and gene signatures. Penalized least-squareminimization with variable selection and regularization wasused to prevent overfitting.

To investigate the functional role of prognostic microRNAs,relationships between host genes (pri-microRNAs), maturemicroRNAs, and cognate target mRNAs were examined byglobal expression analysis. Finally, findings were confirmed by2 further independent analyses. First, pri-microRNA (upstreamofmicroRNAexpression) and genes in themicroRNAprocessingmachinery were analyzed. In addition, expression of targetgenes, reflecting the functional effect downstream of microRNAexpression, was considered. Prognostic significance of bothanalyses was assessed in published cohorts (n > 1,000 cases).

Materials and Methods

Patient characteristicsA retrospective series of 219 patients with early primary

breast cancer was considered (5); extended demographics areprovided in Supplementary Table S1 and SupplementaryInformation. Informed consent was obtained for each subject.Clinical investigations were conducted after approval by thelocal research ethics committee and in accordance with theethical principles expressed in the Declaration of Helsinki.Main endpoint was DRFS.

mRNA and microRNA profilingMatched microRNA and mRNA profiling was successfully

obtained from 207 of 219 samples by using Illumina HumanRefSeq-8 and miRNAv1 arrays (see Supplementary Informa-tion). Data have been submitted to GEO (6), superSeriesGSE22220. In addition, Affymetrix U133A-B/plus2 array datafrom previously published breast cancer cohorts were ana-lyzed [Supplementary Table S2, n¼ 1,050 with recurrence-freesurvival (RFS), n ¼ 592 with RFS outcome].

Statistical methodsWorkflow with study design is provided in Supplementary

Figure S1. Analyses were implemented and carried out inR v2.9.0, and Sweave (7) was used for Automatic Generation ofReports.

Penalized regression allows identification of microRNAswhose expression is prognostic, or associated with a clinico-pathologic factor, independently fromothercovariates.Genomicdatasets are characterized by a greater number of variables thansamples and high structure. Therefore, we used L1- and L2-penalized regression to enable efficient variable selection andencourage a grouping effect (8). Penalization parameters wereoptimizedbycross-validation,and leave-one-outwasused totestthe models. Penalized Cox regression was carried out in 2 steps:

1. microRNAs associated with DRFS were selected by usingpenalized Cox regression including all microRNAs on thearray [Supplementary Information, Equation (SM1)].

2. Selected microRNAs were analyzed further to assesswhether they were prognostic independently from

known clinicopathologic factors. Covariates consideredwere microRNA expression, clinical factors, treatment,and gene signatures of biological processes [Supp-lementary Information, Equation (SM2)].

Only microRNAs selected by both steps were consideredindependently prognostic of DRFS.

Penalized linear regression analysis was used to studyassociation of microRNA expression with clinicopathologicfactors [Supplementary Information, Equation (SM3)].

Meta-analysis of published cohorts was carried out aspreviously described (9); datasets are summarized in Supple-mentary Table S2 and selection criteria are provided inSupplementary Information.

Global expression analysis of microRNA–targetrelationship

An extremely small number of microRNA targets have beenvalidated previously, thus analyses of microRNA–target rela-tionship relies on in silico predictions. Simultaneous use ofmultiple prediction algorithms has been suggested (10), and toavoid bias resulting from underlying correlations we consid-ered the union of all predicted targets from 6major algorithms(further details in Supplementary Information). We expectedthat if a microRNA were functional, a significant proportion ofits predicted targets would be downregulated. Because of thechallenges involved, the following 3 methods were compared:

1. Cumulative relative risk (RR) plots. All transcripts presenton the array were ranked on the basis of the correlation oftheir expression with that of the microRNA under study.The RRwas defined as the probability of observing, at eachcorrelation level, a given number of predicted targetsamong genes whose expression is inversely correlatedwith that of the microRNA, with respect to genes withpositive correlation [Supplementary Information, Equa-tion (SM4)]. If targets are regulated by themicroRNA, RR>1. These plots were also used to study association ofmicroRNA targets with clinical factors [SupplementaryInformation, Equations (SM5) and (SM6)].

2. Predicted target signature (PTSign) score. In each sample,the inverse of the weighted average expression of allpredicted targets was calculated. Weights were the insilico prediction scores [Supplementary Information,Equation (SM7)].

3. Regulatory effect (RE) score. For each sample, transcriptswere ranked by their expression level. The difference ofthe average rank between microRNA target and non-target transcripts was calculated as described in (11).High PTSign or RE scores indicate global target down-regulation.

Results

microRNAs independently prognostic of distant relapsein breast cancer

A 2-step Cox analysis accounting for clinical, pathologic,and molecular features was used (Fig. 1; SupplementaryFig. S2). Specifically, prognostic microRNAs were first selected

Buffa et al.

Cancer Res; 71(17) September 1, 2011 Cancer Research5636

on August 15, 2020. © 2011 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst July 7, 2011; DOI: 10.1158/0008-5472.CAN-11-0489

Page 3: microRNA-Associated Progression Pathways and Potential … · To identify microRNAs that provide independent informa-tion, and are not simply surrogates of previously identified Authors'

in a penalized Cox analysis where all microRNAs were testedsimultaneously within the same model. Selected microRNAswere then assessed to test prognostic capability independentfrom clinical covariates [patient age, tumor size and grade,nodal involvement, estrogen receptor (ER) status, Tamoxifenand Chemotherapy treatment] and key biological processes asmeasured by gene expression signatures derived from largecancer cohort studies. Specifically, these were proliferation,ESR1 and HER2 signaling (12, 13), hypoxia (9), stem cell (14),invasion, immune response, and apoptosis (13).This method identified 3 novel and 1 known (miR-128a)

prognostic microRNA in ERþ breast cancer, and 5 novel and1 known (miR-210) in ER� breast cancer (Fig. 1). Specifically,high miR-767-3p, -128a, and/or -769-3p expression wasassociated with poor prognosis, and high miR-135a withgood prognosis in ERþ cases; high miR-27b, -144, and/or -210was associated with poor prognosis, and high miR-342, -150,and/or -30c with good prognosis in ER� cases. Full resultsare reported in Supplementary Figure S2A–G. Three furthermicroRNAs, miR-29c, -642 (high values, good prognosis), and-548d (high values, poor prognosis) were identified in ananalysis including all samples (Supplementary Fig. S2C). Ananalysis including intrinsic subtype classification (Supple-mentary Table S1) produced similar results (data notshown); however, because the stability of this classificationhas been recently questioned (15) we have used the HER2/

ER/proliferation signature classification for the main ana-lysis (Supplementary Fig. S2C–E).

The above analysis confirmed 2 known prognostic micro-RNAs from previous studies (miR-210 and -128a); however,miR-7 and -516-3p, also prognostic in a previous study (4),could not be confirmed. In our study, miR-7 was associatedwith DRFS in univariate Cox analysis but not in a modelincluding all microRNAs, suggesting that this microRNA is notan independent prognostic factor, whereas miR-516-3p wasnot significantly associated with DRFS.

Also, real-time PCR was carried out to measure expressionsof 4 prognostic microRNAs, miR-210, miR-342, miR-144 andmiR-27b, relative to 3 small nucleolar RNAs (snoRNAs) controls(Supplementary Information). Strength of correlation betweenreal-time PCR and arrays results varied but was significant forall 4 microRNAs (P < 0.0001 in all cases); analysis with non-normalized CT values produced similar results (see Supple-mentary Information for discussion on PCR normalization).

microRNA signatures divided patients efficiently into goodand poor prognosis groups when either a simple median split(Fig. 1E and G) or clustering with Bayesian InformationCriterion (Supplementary Fig. S2F) were considered, andimportantly they performed well when tested on the left-out cases (Supplementary Fig. S2B and G). Among microRNAsprognostic in ER� samples, 5 (miR-150, -342, good prognosis;and miR-210, -144, -27b, poor prognosis) were also prognostic

Figure 1.microRNAs independently prognostic of DRFS in breast cancer (n¼ 207). Prognostic microRNAs were selected by 2-step penalized Cox regression(seeMaterials andMethods). A–C, first amodel including all microRNAs (x-axis) was considered (full results in Supplementary Fig. S2A). Heat maps display thecross-validated model at each leave-one-out iteration (y-axis) for analyses in all 207 (A), 82 ER� (B), and 90 ERþ Tamoxifen-treated (C) samples. Similar modelsare clustered (standard correlation). Color reflects the HR (logged and rescaled per column, see bar). microRNAs that were selected in at least 5% of iterationsare shown. Consistently selected microRNAs (>95% iterations) were further analyzed by a Cox model including clinicopathologic factors and gene expressionsignatures (D–G). microRNAs consistently significant in this model were considered prognostic and are indicated with an asterisk (red, not known; black,previously known to be prognostic). Summary results are shown for ERþ Tamoxifen-treated (D) and ER� (F) samples (full results in Supplementary Fig. S2C–E).miR-768-3p is in parenthesis as it was retired from miRBase. Kaplan–Meier plots of prognostic microRNA signature in ERþ (E) and ER� (G) cases are alsoshown. Expression of each prognostic microRNA (asterisks fromD and F, respectively) was ranked. In each case, the mean rank of poor prognosis microRNAswas subtracted from that of good prognosis microRNAs to provide a nonparametric summary score. Cases were categorized as high and low risk by medianvalue of this score. Box plots of the microRNA ranked expression in the 2 groups show median, quartiles, and range.

Integrated mRNA and microRNA Analysis in Breast Cancer

www.aacrjournals.org Cancer Res; 71(17) September 1, 2011 5637

on August 15, 2020. © 2011 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst July 7, 2011; DOI: 10.1158/0008-5472.CAN-11-0489

Page 4: microRNA-Associated Progression Pathways and Potential … · To identify microRNAs that provide independent informa-tion, and are not simply surrogates of previously identified Authors'

Tab

le1.

Ana

lysisof

microRNAho

st-gen

e(pri-microRNA)ex

pressionin

pub

lishe

dco

horts

pri-m

icroRNA

HR

forDRFS

inpub

lishe

dco

horts(n

=59

2)b

pri-m

icroRNA

HR

forRFS

inpub

lishe

dco

horts(n

=1,05

0)b

Oxf

GSE65

32GUY

GSE65

32KI

GSE65

32OXFc

GSE91

95Sum

mary

effect

dOxf

GSE14

56GSE20

34GSE65

32GUY

GSE65

32KI

GSE65

32OXFc

GSE91

95Sum

mary

effect

dProgno

stic

microRNAsa

pri-m

icroRNA

host

tran

script

(Illumina,

Affym

etrix,

and

Gen

Ban

kID

s)HR

(P)

HR

(P)

HR

(P)

HR

(P)

HR

(P)

HR

(95%

CI)

HR

(P)

HR

(P)

HR

(P)

HR

(P)

HR

(P)

HR

(P)

HR

(P)

HR

(95%

CI)

miR-128

aGI_31

5435

34,

2027

54_a

t,BC04

1093

2.53 (0.018

5)3.20

(0.082

5)4.03

(0.027

5)0.14

(0.054

9)9.63 (0.061

4)2.49 (1

.44–

4.32

)2.34 (0.024

3)2�0

4 (0�19

49)

2.79 (0.002

8)3.20

(0.082

5)3.08 (0.022

6)0.59

(0.477

8)4.88 (0.117

1)2.43 (1

.68–

3.52

)

miR-210

NA,23

0710

_at,

AK12

3483

NA

2.94

(0.102

9)3.76

(0.056

4)6.39

(0.040

1)0.74 (0.783

0)3.1 (1

.43–

6.74

)NA

3�06 (0�04

86)

NA

2.94

(0.102

9)3.48 (0.022

2)2.89

(0.116

2)2.50 (0.360

4)3.07 (1

.74–

5.41

)miR-27b

GI_24

4320

57,

2335

99_a

t,AK02

5151

1.75 (0.155

8)1.13

(0.853

4)0.87

(0.821

5)2.40

(0.313

4)5.64 (0.137

4)1.56 (0.91–

2.67

)1.64 (0.199

5)0�3

3 (0�05

16)

NA

1.13

(0.853

4)0.82 (0.675

9)1.15

(0.829

7)3.19 (0.234

8)0.83 (0

.48–

0.42

)

miR-29c

NA,22

8528

_at,

AK12

3264

NA

1.18

(0.801

0)0.35

(0.084

6)1.69

(0.555

7)0.01 (0.004

0)0.58 (0.27–

1.24

)NA

0�15 (0�00

16)

NA

1.18

(0.801

0)0.37 (0.040

3)0.68

(0.574

4)0.04 (0.006

1)0.37 (0

.21–

0.64

)miR-30c

NA,21

1797

_s_a

t,AF1

9174

4NA

0.96

(0.954

9)0.59

(0.396

6)0.69

(0.658

6)2.26 (0.465

7)0.83 (0.40–

1.74

)NA

2�09 (0�17

65)

0.54 (0.066

4)0.96

(0.954

9)0.82 (0.678

1)0.84

(0.795

0)1.96 (0.489

6)1.16 (0.67–

1.99

)miR-30e

-3p

GI_11

4969

77,

2022

15_s

_at,

AF1

9174

4

0.72 (0.417

4)0.38

(0.168

0)1.49

(0.524

6)0.53

(0.467

8)0.43 (0.442

3)0.71 (0.41–

1.22

)0.9 (0.785

2)0�8

7 (0�80

57)

0.52 (0.056

3)0.38

(0.168

0)1.14 (0.791

9)0.47

(0.268

9)0.58 (0.576

7)0.96 (0.48–

1.00

)

miR-342

GI_77

0668

6,22

7232

_at,

AL1

3364

2

0.42 (0.036

9)0.87

(0.825

6)0.45

(0.189

)0.12

(0.028

9)0.43 (0.439

7)0.44 (0

.26–

0.76

)0.51 (0.090

3)0�2

0 (0�00

50)

NA

0.87

(0.825

6)0.46 (0.097

2)0.09

(0.001

4)0.79 (0.802

0)0.36 (0

.21–

0.61

)

miR-548

dGI_24

4976

17,

2227

40_a

t,NM_0

1410

9

3.58 (0.001

4)4.04

(0.031

7)2.15

(0.218

8)5.47

(0.073

5)6.22 (0.120

3)3.54 (2

.06–

6.09

)2.57 (0.012

9)3�7

0 (0�02

04)

NA

4.04

(0.031

7)2.24 (0.097

8)5.97

(0.013

9)2.90 (0.275

6)3.36 (1

.93–

5.82

)

Abbreviation:

NA,no

tav

ailable.

apri-microRNAsaresh

ownthat

couldbemap

ped

tothearrays

andsh

owed

sign

ifica

ntco

rrelationwith

microRNAex

pressionintheOxf

datase

tora

subse

tof6

9sa

mpleswhe

reAffym

etrix

arrays

wereav

ailable

(Spea

rman

'srank

correlationtest

P<0.05

).bUniva

riate

Cox

analysis

HR

and2-sided

sign

ifica

nce(P)provided

.Exp

ress

ionwas

cons

idered

asco

ntinuo

usva

riable

(rank

ed,no

rmalized

betw

een0an

d1).

cGSE65

32ov

erlaps

partia

lly(n

¼69

)with

theprese

ntda

tase

t.Th

iswas

exploite

dto

establishtheco

rrelationbe

twee

nmicroRNAan

dpri-microRNAex

pressionas

mea

suredbyAffym

etrix

arrays

;ho

wev

er,theco

mmon

case

sweretake

nou

tfrom

GSE65

32OXFin

allp

rogn

ostic

analysis

whe

rethis

datas

etis

used

(herean

dalso

inFig.

4).

dSum

maryHR

and95

%CIforthemeta-an

alysis.Signific

antresu

ltsarein

bold.

Buffa et al.

Cancer Res; 71(17) September 1, 2011 Cancer Research5638

on August 15, 2020. © 2011 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst July 7, 2011; DOI: 10.1158/0008-5472.CAN-11-0489

Page 5: microRNA-Associated Progression Pathways and Potential … · To identify microRNAs that provide independent informa-tion, and are not simply surrogates of previously identified Authors'

in univariate analysis of triple negative (TRN) breast cancers(n¼ 37, see legend of Supplementary Table S1). miR-342, -27b,and -150 were also significant in a Cox analysis of this groupincluding clinicopathologic factors and gene signatures (Sup-plementary Fig. S2H).Otherindependentprognosticfactorswerenumberofpositive

nodes(SupplementaryFig.S2C–E)andtumorgrade(Supplemen-tary Fig. S2C). In agreement with previous studies (9, 13), pro-liferation and hypoxia signatures were prognostic in ERþ cases(Supplementary Fig. S2E); hypoxia, invasion, and immuneresponse signatures in ER� cases (Supplementary Fig. S2D).Crucially, microRNAs identified by our analysis were prognosti-callyindependentofthesesignatures(SupplementaryFig.S2C–E).

Prognostic microRNA clustersmiR-767-3p (prognostic in ERþ cases), andmiR-27b andmiR-

144 (prognostic in ER� cases) are part of microRNA clusters.Analysis of clustered microRNAs revealed that miR-451, clus-tered with miR-144, was significantly associated with goodprognosis in ERþ cases (Supplementary Table S3), suggestingan independent role for these 2 clusteredmicroRNAs inERþ andER� tumors. Conversely, the miR-24/27/23 cluster was consis-tentlyassociatedwithpoorprognosis,withall butonemicroRNAof this family (miR-27a/b, -23a/b, -24; but not -189) significantlyassociated with DRFS in univariate analysis of all cases, and allmicroRNAs but one (miR-23b) significant in analysis of ER�

cases (Supplementary Table S3). However, none of them wasassociatedwith prognosis in ERþ cases, providing evidence thatthis family plays a specific role in ER� breast cancer.

Transcriptional regulation of microRNAs and analysisof pri-microRNA in independent cohorts

We investigated whether prognostic microRNAs and theirpri-microRNAs showed coordinated expression, and studiedthe prognostic potential of pri-microRNAs in independentcohorts (n ¼ 1,050 with RFS, n ¼ 592 with DRFS information;for further details on these cohorts see Supplementary Infor-mation and Supplementary Table S2). Initially, transcriptscontaining pri-microRNA for 6 prognostic microRNAs couldbe mapped to the arrays and their expression found to besignificantly correlated with that of the mature microRNAin our series (Spearman's rank correlation test P < 0.05,microRNAs are listed in Table 1). This suggests transcriptionalregulation of these microRNAs in breast cancer, rather thanmodulation of the mature microRNA due to changes inprocessing. Among these, pri-miR-128a, 210, -29c, -342, and-548d showed significant association with prognosis in meta-analysis of the independent cohorts and a concordant effectwith respect to the microRNA prognostic analysis in Figure 1;specifically, pri-miR-128a, -210, -342, and -548d were signifi-cant for DRFS, and -128a, -210, -27b, -342, and -548d weresignificant for RFS (Table 1).

Figure 2. Biological processes and clinicopathologic factors associated with prognostic microRNAs. Expression signatures derived from large-scale analysisof cancer datasets were used as surrogate markers of biological processes; samples were ranked from lowest to highest as measured by summary scores ofthese signatures. microRNAs prognostic in ERþ (A) and ER� (B) cases are shown. Association between expression of each microRNA (y-axis) withclinicopathologic variables and gene signatures (x-axis) was obtained by using penalized linear regression. Heat maps illustrate themicroRNA association withsignature/clinical variable in the cross-validated model (see bar). Red, covariate levels high, microRNA expression high; blue, opposite. C, scatter plots withexamples of significant associations. MicroRNA expression (x-axis) and covariate (y-axis) are plotted with a second covariate shown as color stratification.Linear fits and Spearman's rank correlation are shown in the 2 strata. HS up, Hypoxia Signature Score above median; HS down, below median; Nþ, nodalinvolvement; N0, no involvement.

Integrated mRNA and microRNA Analysis in Breast Cancer

www.aacrjournals.org Cancer Res; 71(17) September 1, 2011 5639

on August 15, 2020. © 2011 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst July 7, 2011; DOI: 10.1158/0008-5472.CAN-11-0489

Page 6: microRNA-Associated Progression Pathways and Potential … · To identify microRNAs that provide independent informa-tion, and are not simply surrogates of previously identified Authors'

Association of prognostic microRNAs with processescritical for breast cancer biology

microRNA association with clinicopathologic factors wasstudied by multiple penalized linear regression (Fig. 2A and B).This confirmed, for example, the previously reported hypoxiaregulation of miR-210 (Fig. 2C), but suggested that miR-210expression is also associated with proliferation, ER positivity,grade, and nodal invasion (Fig. 2B and C). Interestingly,associations with hypoxia and proliferation were significantafter accounting for grade and ER status (Fig. 2B and C). Thishighlights the usefulness of testing associations simulta-neously by including clinical and molecular covariates inthe same regression model. miR-128a was found to be asso-ciated with ER positivity, both as measured by IHC and ESR1expression signature (Fig. 2A). Among the newly discoveredprognostic microRNAs, a reported role for miR-150 in immu-nity (16) agrees with an association with immune responsesignature in our dataset (Fig. 2C). miR-135a was inverselycorrelated with the proliferation signature, in agreement withreports of negative regulation of cell growth in Hodgkin'slymphoma (17). Finally, miR-27b was associated with theinvasion signature (Fig. 2B), in agreement with its reportedability to promote invasion and angiogenesis (18).

A recent study in 20 inflammatory breast cancers (IBC; ref.19), one of the most aggressive forms of locally advancedbreast cancer, identified microRNAs belonging to familiesmiR-29 (-29a), -30 (-30b), and -342 as downregulated in IBC.This agrees with our findings of expression of microRNAs inthese families (miR-29c, -30c, -342) being associated with goodprognosis. In the same study, miR-548a-5p was found upre-gulated in IBC, in agreement with high expression of the miR-548 family being associated with poor prognosis (-548d).

Target expression is consistent with regulatory effectof prognostic microRNAs

Downregulation of predicted target transcript was observedby using cumulative RR plots for all prognostic microRNAs,with miR-128a, -144, and -210 providing the most consistentand significant results (Fig. 3A; Supplementary Fig. S3).Furthermore, microRNA association with clinicopathologicfactors (Fig. 2) was reflected by coordinated downregulationof cognate target mRNAs (Fig. 3B; Supplementary Fig. S3). Themost significant were miR-210 and -144 targets, stronglyunderexpressed in highly proliferative tumors (Fig. 3). In 3cases (miR-144, -210, and -769-3p), results could be confirmedconsistently by both PTsign- and RE-score methods (Supple-mentary Table S4); however, PTsign and RE scores werestrongly correlated for all microRNAs [P << 0.00001 for allmicroRNAs, correlation coefficient range: (0.7–0.95)]. Contraryto RR plots, these measure inhibition of target expressionirrespective of amount of downregulation.

A global analysis of microRNA targets reveals pathwaysdysregulated in cancer

A pathway analysis revealed that a number of down-regulated predicted targets, greater than that expected bychance, were implicated in pathways which have importantroles in direct tumor growth and metastasis (Supplementary

Fig. S4); and also that microRNA target downregulationwould lead to induction of these pathways (SupplementaryFig. S4A). For example, downregulated targets includedMAPK8 and MAPK14 ((miR-144 and miR-27b targets, respec-tively), which can be proapoptotic under stress conditions(20); SPRY2 (miR-128a target) in the FGF receptor signalingpathway, which has a major role in inhibiting tyrosinekinases (21); and tumor suppressor genes PTEN and FOXO1(miR-144 and -128a targets). miR-27b predicted targets inthe metabotropic glutamate receptor pathway, namelyGRIN3A, GRM6, and GRIA4, and voltage-dependent L-typecalcium channel subunit, were consistently downregulated.Their association with neuronal cell death under hypoxicstress (22) suggests a new potential mechanism by whichcancer cells escape death under stress. Recently, mutationsin this pathway were reported as the commonest mutationsin melanoma (23). Wnt antagonists such as miR-144 pre-dicted target SRFP1, and targets involved in angiogenesissuch as Frizzled4 were also among downregulated predicted

Figure 3. Inhibition of cognate target expression by prognosticmicroRNAs. Cumulative RR plots are shown for the most significanteffects; further results in Supplementary Figure S3. A, plots showenrichment in the number of predicted targets among transcripts whosedownregulation is associated with microRNA overexpression(Supplementary Fig. S3 for workflow of this analysis). At each correlationlevel r, the RR of observing a given number of predicted targets whoseexpression is anticorrelated with microRNA expression, with correlation of�r or less, is plotted. Number of downregulated targets is shown on theplot. Dotted lines represent the mean RR for randomly sampled (�1,000)set of transcripts, with 5% and 95% CI. B, RR plots of enrichment in thenumber of predicted targets among transcripts whose expression isanticorrelated with gene expression signatures of biological processes.Proliferation was considered as biological process with strongestassociation with microRNA expression (Fig. 2A). RR plot statistics asdescribed in A.

Buffa et al.

Cancer Res; 71(17) September 1, 2011 Cancer Research5640

on August 15, 2020. © 2011 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst July 7, 2011; DOI: 10.1158/0008-5472.CAN-11-0489

Page 7: microRNA-Associated Progression Pathways and Potential … · To identify microRNAs that provide independent informa-tion, and are not simply surrogates of previously identified Authors'

targets, where excessive Frizzled4 has been associated withdisrupted embryonic vasculature (24).In TRN cancers, several signaling pathways related with

miR-150 predicted targets were found upregulated in poorprognosis cases including Akt2, insulin receptor, ErbB3, S6kinase, MAP kinase pathways, and stress response kinase JNK,as well as downstream protease MMP-13 (SupplementaryFig. S4B). miR-342 predicted target RRM2was also upregulatedin this group (Supplementary Fig. S4B). RRM2 is a target forseveral inhibitors of proliferation that affect cells in S-phase,and would be compatible with the high proliferation rate ofTRN cancer (6). Similar pattern was observed for miR-342target glucose 1,6-bisphosphate synthase, a critical componentof glycolysis (25) to which inhibitors are currently beingdeveloped; and TLE-1, a downstream component of notchsignaling (26), that is already recognized to be activated inTRN cancers.

Validation of prognostic microRNAs using targetexpression in the present and independent cohorts

We focused on prognostic microRNAs showing consis-tent and significant association with downregulation ofcognate target transcripts. Among these, miR-210, -128,and -27b predicted targets were prognostic for DRFSand RFS, both when using cumulative RR (SupplementaryFig. S5) and summary scores in meta-analysis of publishedcohort studies (Fig. 4A–F). Furthermore, tumors with con-comitant microRNA overexpression and target underex-pression had significantly worse prognosis, whereas theopposite was true for cases with low microRNA levels andhigh target expression (Fig. 4G–I).

We also conducted a single target analysis for experimen-tally derived miR-210 targets (Supplementary Fig. S4E). Threeof 5 targets showing most significant downregulation, namelyISCU, CBX7, and IGF1R, were significantly associated with

Figure 4. Validation of prognosticmicroRNAs in independent breastcancer datasets by analysis ofcognate targets. A–F, forest plotsare shown for miR-210 (A, D), -27b(B, E), and -128a (C, F) predictedtargets (PTSign summary score,see Materials and Methods) in thepresent (Oxf) and published BCdatasets (GEO IDs provided,further details in SupplementaryTable S2). Dots represent HR;dimensions are proportional todataset size. Gray bars are 95%CI. A–C, DRFS; D–F, RFS. G–I,Kaplan–Meier curves forexpression of mature miR-210 (G),-27b (H), and -128a (I), andrespective PTSigns, in the Oxfordbreast cancer dataset (n ¼ 207).microRNA expression levels andPTSign score were split by medianvalue (�, below; þ, above). HR fordeviation contrasts in a Coxregression comparing eachcategory with the overall effectare shown for significantcomparisons.

Integrated mRNA and microRNA Analysis in Breast Cancer

www.aacrjournals.org Cancer Res; 71(17) September 1, 2011 5641

on August 15, 2020. © 2011 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst July 7, 2011; DOI: 10.1158/0008-5472.CAN-11-0489

Page 8: microRNA-Associated Progression Pathways and Potential … · To identify microRNAs that provide independent informa-tion, and are not simply surrogates of previously identified Authors'

DRFS; specifically, low levels were associated with poor prog-nosis (Supplementary Fig. S4E). This suggests that novel targetscan be confirmed with this approach. To further assess thispotential we assessed protein expression by using immuno-histochemistry (IHC) for a predicted target with suitablecommercial antibody available. NFE2L2 (alias NRF2) waschosen as top ranking downregulated predicted target ofmiR-144, novel prognostic microRNAs showing evidence ofcoordinated downregulation of target mRNAs (Fig. 1; Supple-mentary Table S4). IHCwas done on 137 cases (SupplementaryInformation); staining was predominantly cytoplasmic withminimal stromal positivity (Supplementary Fig. S6). Tumorswith weak protein and mRNA expression showed significantlyhigher miR-144 expression than tumors with consistently highNFE2L2 expression (Supplementary Fig. S6). This was true onlyin ER� cases (Supplementary Fig. S6D and E), miR-144 beingprognostic only in ER� tumors (Fig. 1) in agreement. LowNFE2L2 IHC product score (IPS) was associated with worseDRFS (univariate Cox HR ¼ 0.64, P ¼ 0.18; IPS ranked andnormalized between 0 and 1); in agreementwithNFE2L2 actingas a tumor suppressor (23). Furthermore, tumors with con-comitant miR-144 overexpression and NFE2L2 underexpres-sion had significantly worse prognosis, whereas the oppositewas true for cases with low miR-144 levels and high NFE2L2expression (Supplementary Fig. S6F).

microRNAs are prognostic independent of expressionof microRNA-processing genes

We found that the expression of some, but not all, pri-microRNAs was correlated with that of the mature micro-

RNAs. Thus, we investigated whether microRNA-processinggenes might have a role in regulation of mature microRNAlevels. We tested the prognostic significance of severalprocessing genes. High expression of transport gene expor-tin 5 (XPO5) and RISC complex genes EIF2C2 and EIF2C3were significantly associated with poor prognosis, whereashigh expression of DICER genes DICER1 and TARBP1, andRISC complex gene EIF2C4 were associated with goodprognosis (Table 2). Of these, expression of XPO5 showeda significant correlation with mature microRNA expressionboth for the ER� and ERþ microRNA prognostic signatures(Fig. 5A and B), whereas EIF2C2 and EIF2C3 showed corre-lation only with the ER� microRNA prognostic signature.However, in all cases fold expression changes were verysmall (Fig. 5A and B). When samples were stratified by lowand high risk, based on microRNA signature and expressionof processing genes, the effect of the latter was neversignificant (Fig. 5C–F). A Cox analysis including singleprognostic microRNAs and processing genes showed thatin all cases, mature microRNA expression was significantlyindependent of the expression of the processing genes(Supplementary Table S5). Overall, these results suggestthat the expression profile and prognostic significance ofthese microRNAs is due to regulation of their transcriptionrather than differential processing. For miR-128a, -210, -342,and -548d, this is also confirmed by the prognostic sig-nificance of their pri-microRNA expression in independentcohorts (Table 1).

Interestingly, EIF2C2 (AGO2) expression in our samples,although mildly differential, was always very high (Fig. 5),

Table 2. Prognostic significance of microRNA processing genes

Gene symbol Function Probeset used HRa P

TARBP1 DICER GI_19743835-S 0.32 0.0041DICER1 DICER GI_29294648-I 0.44 0.0388EIF2C4 (AGO4) RISC GI_29029592-S 0.45 0.0420ESR1 p68 and p72 interaction GI_4503602-S 0.49 0.0657ARS2 Regulation of microprocessor GI_33383230-A 0.74 0.4313PRKRA DICER GI_32261293-S 0.86 0.7036DGCR8 Microprocessor GI_38488719-S 0.89 0.7642EIF2C1 (AGO1) RISC GI_29171732-S 0.90 0.7667RNASEN Microprocessor GI_21359821-S 1.06 0.8757TARBP2 DICER GI_19743837-A 1.24 0.5725ADARB1 Editing of specific microRNAs GI_7669476-I 1.27 0.5206RAN Transport GI_6042206-S 1.34 0.4535ESR2 p68 and p72 interaction GI_10835012-S 1.51 0.2767TRIM32 Binding to miRISC, enhancing microRNA activity GI_15208649-S 1.78 0.1326EIF2C3 (AGO3) RISC GI_29337285-A 2.46 0.0244EIF2C2 (AGO2) RISC GI_29171733-S 2.63 0.0128XPO5 Transport GI_22748936-S 3.11 0.0035

aCox univariate analysis. Gene expression considered as continuous variable, ranked, and normalized between 0 and 1. Whenmore than one probeset mapped to the same gene, the probeset with the greater effect is shown. Probesets with P < 0.05are highlighted in bold.

Buffa et al.

Cancer Res; 71(17) September 1, 2011 Cancer Research5642

on August 15, 2020. © 2011 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst July 7, 2011; DOI: 10.1158/0008-5472.CAN-11-0489

Page 9: microRNA-Associated Progression Pathways and Potential … · To identify microRNAs that provide independent informa-tion, and are not simply surrogates of previously identified Authors'

thus allowing mature microRNAs to modulate target geneexpression. This is particularly important in view of recentwork (11) suggesting that AGO expression is mandatory foran effect of microRNAs on the expression profiles of tumorsamples.

Discussion

In this study, microRNAs that are independently prognosticfor DRFS were identified and validated by integrated analysisof microRNA and mRNA profiling. Nine microRNAs werefound prognostic in ERþ and ER� breast cancer, 3 of whichremained significant in the clinically challenging TRN group.Our findings regarding mature miR-210 and miR-128a expres-sion confirm previous findings of a smaller study (4). Multiplepenalized linear regression revealed that prognostic micro-RNAs are associated with key biological processes inbreast cancer, such as proliferation (miR-135a), hypoxia

(miR-210, -342), invasion (miR-27b), and immune response(miR-150). However, these microRNAs are prognostic inde-pendent of gene expression signatures of these processes. Thishighlights the importance of considering both microRNA andtranscript expression data in prognostic studies.

Several microRNAs previously linked with breast cancersubtype or progression in experimental models were notidentified as independently prognostic. This agrees with pre-vious results from prognostic studies (4, 27); reasons couldinclude discrepancy of assays and/or the importance ofmicroRNA expression in specific cell populations which wouldnot appear in whole tumor sample analyses. However, it alsohighlights the need to differentiate between biomarkers oftumor presence and prognostic biomarkers, and identifyfactors carrying information independently rather than sur-rogates of known prognostic covariates.

To examine whether microRNAs might be prognostic dueto regulatory effects on cognate targets, associations between

Figure 5. Expression ofmicroRNA-processing genesdoes not affect microRNAprognostic significance. A and B,box plots of expression ofmicroRNA-processing genes thatwere prognostic in Cox univariateanalysis (Table 2) are shown forthe low-risk profile (LRP) and high-risk profile (HRP) in ERþ (A) andER� (B) samples as defined inFig. 1E and G legend. Foldchanges in mean expression (FC)and nonparametric Mann–Whitney test are shown forsignificant cases. C–F, Kaplan–Meier plots of DRFS. Samples arestratified into the LRP and HRP,and also by median value ofprocessing genes’ expression; �,below; þ, above median. Thisvalue is calculated by using all thesamples, thus the expression cut-point is equal in both LRP andHRP group. The HR of LRP andHRP groups was alwayssignificant, whereas the HR of [�]and [þ] groups was neversignificant (Cox model contrasts,threshold P < 0.05).

Integrated mRNA and microRNA Analysis in Breast Cancer

www.aacrjournals.org Cancer Res; 71(17) September 1, 2011 5643

on August 15, 2020. © 2011 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst July 7, 2011; DOI: 10.1158/0008-5472.CAN-11-0489

Page 10: microRNA-Associated Progression Pathways and Potential … · To identify microRNAs that provide independent informa-tion, and are not simply surrogates of previously identified Authors'

microRNAs and predicted targets were studied by globalexpression analysis. This is challenging, as the microRNA–mRNA interaction network is complex and the effect on eachindividual mRNA is often small. Nevertheless, prognostic miR-128a, -27b, and -210 showed evidence of cognate target down-regulation, and expression of their targets was prognostic inmeta-analysis of several cohorts (Fig. 4). Furthermore, com-bined use of microRNA and target expression identified caseswith the poorest prognosis, suggesting that a combined scorecould reflect not only expression but also functionality of themicroRNA.

Prognostic microRNAs could be used not only to selectpatients for specific interventions, but also to define ther-apeutic approaches. To this end, functional validation ofmicroRNA targets is needed to elucidate the role of micro-RNAs in cancer. In this respect, analysis of microRNA–mRNA relationships in clinical datasets could assist in targetprioritization, and cohorts such as the present one will be auseful resource for future validation studies. PrognosticmicroRNAs were found to regulate a wide range of pre-viously poorly investigated pathways that may be related tocancer progression and potential candidates for therapy,such as RRM2 and TLE-1 in TRN breast cancers. A validatedexample was miR-210 target ISCU (iron-sulfur cluster scaf-fold homolog), which was downregulated in samples withhigh miR-210 levels, and whose underexpression was asso-ciated with both hypoxia and poor prognosis. ISCU plays animportant role in mitochondrial respiration and DNA repair(28, 29). Tumors with high miR-210 might be for examplemore susceptible to DNA damaging agents combinedwith DNA repair inhibitors. Importantly, ISCU could notbe identified in traditional supervised gene expression prog-nostic analyses.

An additional benefit of merging microRNA andmRNA datawas that coordinate expression of microRNAs and precursor-containing pri-microRNAs, and expression of microRNA-pro-cessing genes could be explored. As this analysis suggestedthat several prognostic microRNAs are regulated at the tran-

scriptional level rather than through changes in RNA proces-sing mechanisms (Fig. 5), pri-microRNA data (Table 1) couldbe used for the validation of our findings in cohorts for whichonly mRNA profiling is available. The most striking resultswere for miR-210 and -128a, where coordinated expression ofpri-microRNA (upstream of microRNA expression), maturemicroRNA and predicted targets was observed. Expression ofeach one of these components was found to be prognostic.

In conclusion, this is the first large study integrating micro-RNA and mRNA global profiles in human breast cancer. Prog-nostic microRNAs in ERþ and ER� breast cancer wereidentified, and regulatory action on target transcripts shown.These results elucidated potential novel therapeutic targets,and could also be exploited to validate findings in independentcohorts. Our approach consistently validated known micro-RNA–target interactions, and may therefore be broadly applic-able to other biologically and clinically heterogeneous diseases.

Disclosure of Potential Conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank Helen Turley for assistance with immunohistochemistry,Dr. Russell Leek (tissue banking, Oxford), and the Computational BiologyResearch Group (hardware and services).

Grant Support

This work was supported by Oxford NIHR Comprehensive BiomedicalResearch and Experimental Cancer Medicine Centres; Cancer Research UK(A.L. Harris, M. Taylor, H. Sheldon, and C.E. Snell); EU 6th and 7th FrameworkPrograms (F.M. Buffa and A.L. Harris); Breast Cancer Research Foundation andFriends of Kennington Cancer Fund (A.L. Harris); Rhodes Trust Scholarship(H.E. Gee); and Wellcome Trust grant 075491/Z/04 (J. Ragoussis, C. Camps, andL. Winchester).

The costs of publication of this article were defrayed in part by the paymentof page charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received February 15, 2011; revised June 23, 2011; accepted June 27,2011; published OnlineFirst July 7, 2011.

References1. van't Veer LJ, Paik S, Hayes DF. Gene expression profiling of breast

cancer: a new tumor marker. J Clin Oncol 2005;23:1631–5.2. Nicoloso MS, Spizzo R, Shimizu M, Rossi S, Calin GA. MicroRNAs–

the micro steering wheel of tumour metastases. Nat Rev Cancer2009;9:293–302.

3. Blenkiron C, Goldstein LD, Thorne NP, Spiteri I, Chin SF, Dunning MJ,et al. MicroRNA expression profiling of human breast cancer identifiesnew markers of tumor subtype. Genome Biol 2007;8:R214.

4. Foekens JA, Sieuwerts AM, Smid M, Look MP, de Weerd V, BoersmaAW, et al. Four miRNAs associated with aggressiveness of lymphnode-negative, estrogen receptor-positive human breast cancer.Proc Natl Acad Sci U S A 2008;105:13021–6.

5. Camps C, Buffa FM, Colella S, Moore J, Sotiriou C, Sheldon H,et al. hsa-miR-210 Is induced by hypoxia and is an independentprognostic factor in breast cancer. Clin Cancer Res 2008;14:1340–8.

6. Heidel JD, Liu JY, Yen Y, Zhou B, Heale BS, Rossi JJ, et al. PotentsiRNA inhibitors of ribonucleotide reductase subunit RRM2 reduce cellproliferation in vitro and in vivo. Clin Cancer Res 2007;13:2207–15.

7. Leisch F. Sweave: dynamic generation of statistical reports usingliterate data analysis. In: H€ardle W, R€onz B, editors. Compstat; 2002.Heidelberg, Germany: Physica Verlag; 2002. p. 575–80.

8. Park MY, Hastie T. L1-regularization path algorithm for generalizedlinear models. J R Statist Soc B 2007;69:659–77.

9. Buffa FM, Harris AL, West CM, Miller CJ. Large meta-analysis ofmultiple cancers reveals a common, compact and highly prognostichypoxia metagene. Br J Cancer 2010;102:428–35.

10. Sethupathy P, MegrawM, Hatzigeorgiou AG. A guide through presentcomputational approaches for the identification of mammalian micro-RNA targets. Nat Methods 2006;3:881–6.

11. Cheng C, Fu X, Alves P, Gerstein M. mRNA expression profiles showdifferential regulatory effects of microRNAs between estrogen recep-tor-positive and estrogen receptor-negative breast cancer. GenomeBiol 2009;10:R90.

12. Wirapati P, Sotiriou C, Kunkel S, Farmer P, Pradervand S, Haibe-KainsB, et al. Meta-analysis of gene expression profiles in breast cancer:toward a unified understanding of breast cancer subtyping andprognosis signatures. Breast Cancer Res 2008;10:R65.

Buffa et al.

Cancer Res; 71(17) September 1, 2011 Cancer Research5644

on August 15, 2020. © 2011 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst July 7, 2011; DOI: 10.1158/0008-5472.CAN-11-0489

Page 11: microRNA-Associated Progression Pathways and Potential … · To identify microRNAs that provide independent informa-tion, and are not simply surrogates of previously identified Authors'

13. Desmedt C, Haibe-Kains B, Wirapati P, Buyse M, Larsimont D,Bontempi G, et al. Biological processes associated with breast cancerclinical outcome depend on the molecular subtypes. Clin Cancer Res2008;14:5158–65.

14. Ben-Porath I, ThomsonMW, Carey VJ, Ge R, Bell GW, Regev A, et al.An embryonic stem cell-like gene expression signature in poorlydifferentiated aggressive human tumors. Nat Genet 2008;40:499–507.

15. Weigelt B, Mackay A, A'hern R, Natrajan R, Tan DS, Dowsett M, et al.Breast cancer molecular profiling with single sample predictors: aretrospective analysis. Lancet Oncol; 11:339–49.

16. Vasilescu C, Rossi S, Shimizu M, Tudor S, Veronese A, Ferracin M,et al. MicroRNA fingerprints identify miR-150 as a plasma prognosticmarker in patients with sepsis. PLoS One 2009;4:e7405.

17. Navarro A, Diaz T, Martinez A, Gaya A, Pons A, Gel B, et al. Regulationof JAK2 by miR-135a: prognostic impact in classic Hodgkin lym-phoma. Blood 2009;114:2945–51.

18. Kuehbacher A, Urbich C, Zeiher AM, Dimmeler S. Role of Dicer andDrosha for endothelial microRNA expression and angiogenesis. CircRes 2007;101:59–68.

19. Van der Auwera I, Limame R, van Dam P, Vermeulen PB, Dirix LY,Van Laere SJ. Integrated miRNA and mRNA expression profiling ofthe inflammatory breast cancer subtype. Br J Cancer 2011;103:532–41.

20. Wagner EF, Nebreda AR. Signal integration by JNK and p38 MAPKpathways in cancer development. Nat Rev Cancer 2009;9:537–49.

21. Casci T, Vinos J, Freeman M. Sprouty, an intracellular inhibitor of Rassignaling. Cell 1999;96:655–65.

22. Won SJ, Kim DY, Gwag BJ. Cellular and molecular pathways ofischemic neuronal death. J Biochem Mol Biol 2002;35:67–86.

23. Martin-Montalvo A, Villalba JM, Navas P, de Cabo R. NRF2, cancerand calorie restriction. Oncogene 2011:30:505–20.

24. Ye X, Wang Y, Cahill H, Yu M, Badea TC, Smallwood PM, et al. Norrin,frizzled-4, and Lrp5 signaling in endothelial cells controls a geneticprogram for retinal vascularization. Cell 2009;139:285–98.

25. Zhao Y, Liu H, Liu Z, Ding Y, Ledoux SP, Wilson GL, et al. Overcomingtrastuzumab resistance in breast cancer by targeting dysregulatedglucose metabolism. Cancer Res 2011;71:4585–97.

26. Sonoshita M, Aoki M, Fuwa H, Aoki K, Hosogi H, Sakai Y, et al.Suppression of colon cancer metastasis by Aes through inhibition ofNotch signaling. Cancer Cell 2011 ;19:125–37.

27. Gee HE, Camps C, Buffa FM, Colella S, Sheldon H, Gleadle JM, et al.MicroRNA-10b and breast cancer metastasis. Nature 2008;455:E8–9;author reply E9.

28. Chan SY, Zhang YY, Hemann C, Mahoney CE, Zweier JL, Loscalzo J.MicroRNA-210 controls mitochondrial metabolism during hypoxia byrepressing the iron-sulfur cluster assembly proteins ISCU1/2. CellMetab 2009;10:273–84.

29. Favaro E, Ramachandran A, McCormick R, Gee H, Blancher C,Crosby M, et al. MicroRNA-210 regulates mitochondrial free radicalresponse to hypoxia and krebs cycle in cancer cells by targeting ironsulfur cluster protein ISCU. PLoS One 2010;5:e10345.

Integrated mRNA and microRNA Analysis in Breast Cancer

www.aacrjournals.org Cancer Res; 71(17) September 1, 2011 5645

on August 15, 2020. © 2011 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst July 7, 2011; DOI: 10.1158/0008-5472.CAN-11-0489

Page 12: microRNA-Associated Progression Pathways and Potential … · To identify microRNAs that provide independent informa-tion, and are not simply surrogates of previously identified Authors'

2011;71:5635-5645. Published OnlineFirst July 7, 2011.Cancer Res   Francesca M. Buffa, Carme Camps, Laura Winchester, et al.   Expression Profiling in Breast CancerTherapeutic Targets Identified by Integrated mRNA and microRNA microRNA-Associated Progression Pathways and Potential

  Updated version

  10.1158/0008-5472.CAN-11-0489doi:

Access the most recent version of this article at:

  Material

Supplementary

  http://cancerres.aacrjournals.org/content/suppl/2011/07/07/0008-5472.CAN-11-0489.DC1

Access the most recent supplemental material at:

   

   

  Cited articles

  http://cancerres.aacrjournals.org/content/71/17/5635.full#ref-list-1

This article cites 26 articles, 8 of which you can access for free at:

  Citing articles

  http://cancerres.aacrjournals.org/content/71/17/5635.full#related-urls

This article has been cited by 23 HighWire-hosted articles. Access the articles at:

   

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

  Subscriptions

Reprints and

  [email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/71/17/5635To request permission to re-use all or part of this article, use this link

on August 15, 2020. © 2011 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst July 7, 2011; DOI: 10.1158/0008-5472.CAN-11-0489