Relationship of Mammographic Density and Gene Expression ......able, raising the possibility of developing prevention strategies for mitigating density-associated breast cancer risk

Human Cancer Biology

Relationship of Mammographic Density and GeneExpression: Analysis of Normal Breast Tissue SurroundingBreast Cancer

Xuezheng Sun1, Gretchen L. Gierach3, Rupninder Sandhu2, Tyisha Williams4, Bentley R. Midkiff2,Jolanta Lissowska5, Ewa Wesolowska5, Norman F. Boyd6, Nicole B. Johnson7, Jonine D. Figueroa3,Mark E. Sherman3, and Melissa A. Troester1,2

AbstractPurpose: Previous studies of breast tissue gene expression have shown that the extratumoral microen-

vironment has substantial variability across individuals, some of which can be attributed to epidemiologic

factors. To evaluate how mammographic density and breast tissue composition relate to extratumoral

microenvironment gene expression, we used data on 121 patients with breast cancer from the population-

based Polish Women’s Breast Cancer Study.

Experimental Design: Breast cancer cases were classified on the basis of a previously reported,

biologically defined extratumoral gene expression signature with two subtypes: an Active subtype, which

is associated with high expression of genes related to fibrosis and wound response, and an Inactive

subtype, which has high expression of cellular adhesion genes. Mammographic density of the contra-

lateral breast was assessed using pretreatment mammograms and a quantitative, reliable computer-

assisted thresholding method. Breast tissue composition was evaluated on the basis of digital image

analysis of tissue sections.

Results: The Inactive extratumoral subtype was associated with significantly higher percentage

mammographic density (PD) and dense area (DA) in univariate analysis (PD: P ¼ 0.001; DA: P ¼0.049) and in multivariable analyses adjusted for age and body mass index (PD: P ¼ 0.004; DA: P ¼0.049). Inactive/higher mammographic density tissue was characterized by a significantly higher

percentage of stroma and a significantly lower percentage of adipose tissue, with no significant

change in epithelial content. Analysis of published gene expression signatures suggested that Inactive/

higher mammographic density tissue expressed increased estrogen response and decreased TGF-bsignaling.

Conclusions: By linking novel molecular phenotypes with mammographic density, our results indicate

thatmammographic density reflects broad transcriptional changes, including changes in both epithelia- and

stroma-derived signaling. Clin Cancer Res; 19(18); 4972–82. �2013 AACR.

IntroductionMolecular profiling of gene expression of breast cancers

has shown that tumors are remarkably heterogeneous,which has profound influences on etiologic and clinicalresearch (1, 2). More recently, molecular analyses of themicroenvironment have shown similar heterogeneity (3–7), but the epidemiologic, clinical, and pathologic corre-lates of this variation are not well studied. Specifically,tandem analyses of breast cancers and the surroundingmicroenvironment may reveal important stromal–epithe-lial interactions. In fact, previous work suggests that stromalchanges may precede tumor invasiveness and may reflecttumor characteristics (8–13). Findings such as these haveled to speculation that the microenvironment may bedominant over tumor biology early in progression, wheninvasive cancers are still forming (8–13). Although theimportance of tumor microenvironment is increasingly

Authors' Affiliations: 1Department of Epidemiology; 2Lineberger Com-prehensive Cancer Center, University of North Carolina at Chapel Hill,Chapel Hill, North Carolina; 3Hormonal and Reproductive EpidemiologyBranch, Division of Cancer Epidemiology and Genetics, National CancerInstitute, Bethesda, Maryland; 4Department of Biology, Trinity University,San Antonio, Texas; 5Maria Sklodowska-Curie Memorial Cancer Centerand Institute of Oncology, Warsaw, Poland; 6Campbell Family Institute forBreast Cancer Research, Ontario Cancer Institute, Toronto, Ontario,Canada; and 7Department of Pathology, Beth Israel Deaconess MedicalCenter, Harvard Medical School, Boston, Massachusetts

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

X. Sun and G.L. Gierach contributed equally to this work.

CorrespondingAuthor:Melissa A. Troester, Department of Epidemiology,CB 7435, University of North Carolina at Chapel Hill, 135 Dauer Drive,Chapel Hill, NC 27599. Phone: 919-966-7408; Fax: 919-966-2089; E-mail:[email protected]

doi: 10.1158/1078-0432.CCR-13-0029

�2013 American Association for Cancer Research.

ClinicalCancer

Research

Clin Cancer Res; 19(18) September 15, 20134972

on March 18, 2021. © 2013 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst August 5, 2013; DOI: 10.1158/1078-0432.CCR-13-0029

http://clincancerres.aacrjournals.org/

established in the cancer biology literature (8, 14), theepidemiologic factors that affect the microenvironmentremain poorly understood.We recently reported an extratumoral signature, a so-

called Active signature, classifying extratumoral stromalmicroenvironments into two primary gene expression phe-notypes (Active and Inactive) based on unsupervised clus-tering on 72 normal tissue samples adjacent to invasivebreast cancer or ductal carcinoma in situ (Active, n ¼ 27;Inactive, n ¼ 45; ref. 6). The Active subtype had highexpression of genes involved in activation of fibrosis, cel-lular movement, increased TWIST expression, and positiveexpression of TGF-b signatures. The Inactive subtypeexpressed higher levels of cell adhesion and cell–cell contactgenes. Compared with the Inactive subtype, estrogen recep-tor–positive (ERþ) and hormone-treated patients with theActive subtype had poorer overall survival, suggesting pos-sible prognostic value. However, the Active subtype seemedto be independent of breast cancer subtype and standardclinicopathologic parameters, such as tumor size and grade(6). These findings raise the hypothesis that extratumoralsubtypes may be host factors rather than tumor-dependentfactors.We hypothesized that host factors influence the micro-

environment that exists before tumor development and thatthese changesmay be etiologically relevant. Thus, to furtherevaluate whether the Active/Inactive signature is related totumor factors, host factors, or both, we conducted geneexpression profiling on extratumoral non-neoplastic breasttissues from 121 patients with breast cancer of the popu-lation-based PolishWomen’s Breast Cancer Study (PWBCS;ref. 15). In particular, based on differential expression offibrosis, wound response, and cell adhesion genes in thetwo subtypes, we hypothesized that the Inactive microen-vironment would be associated with high mammographicdensity. We further hypothesized that because mammo-

graphic density is a radiologic reflection of variations inbreast tissue composition, high mammographic density/inactive microenvironment would be associated with highnonfatty stromaandepithelial content. PWBCSparticipantsare well characterized with respect to established breastcancer risk factors, including mammographic density, pro-viding us with the opportunity to link the Active/Inactivemolecular phenotype with mammographic density.

Materials and MethodsStudy population

The study population included 121 women from thePWBCS with available snap-frozen extratumoral breasttissues and mammographic density. The PWBCS is a pop-ulation-based case–control study conducted in two majorcities in Poland (Warsaw and Ł�od�z) during 2000 to 2003(15). PWBCS cases were women of ages 20 to 74 years withnewly diagnosed, pathologically confirmed in situ or inva-sive breast carcinoma identified through a rapid identifica-tion system organized at five participating hospitals and viacancer registries. Fresh tissues from invasive tumors, non-neoplastic adjacent breast tissue, and mammary fat tissuewere collected at the time of breast surgery and snap-frozenin liquid nitrogen. Tumor-adjacent breast tissues used inthis study were less than 2 cm from the tumor margin. Onthe basis of in vitro evidence of their distinctive microenvir-onments (16), basal-like and luminal tumors were over-sampled in this study. Information on clinicopathologic,demographic, and anthropometric factors was collectedfrommedical records and in-person interviews as describedpreviously (15). All the participants provided written inf-ormed consent under a protocol approved by the U.S.National Cancer Institute and local (Polish) InstitutionalReview Boards.

Mammographic density measurementPretreatmentmammogramsof the unaffected breastwere

sent to the Ontario Cancer Institute (Toronto, Canada)where they were digitized using a Lumisys 85 laser filmscanner. Patient identifiers were permanently deleted fromthe electronic images. Craniocaudal views of digitized filmswere used to assess mammographic density with Cumulus,an interactive computer-assisted thresholding programdeveloped at the University of Toronto (Toronto, Canada;ref. 17). One expert reader (N.F. Boyd) measured absolutedense area (cm2) and total breast area (cm2) using themethods as described previously (17); percentagemammo-graphic density was calculated by dividing the dense breastarea by the total breast area andmultiplying by100. A repeatset of 49 images was assessed for reliability. The intraclasscorrelation coefficients for percentagemammographic den-sity, dense area, and total breast area were 0.95, 0.93, and0.99, respectively, documenting excellent reproducibility.

Breast tissue composition measurementFrozen non-neoplastic breast specimens of approximately

100 mg were cut over dry ice and then used to cut frozensections. Sectionswere collectedatbothendsof the specimen

Translational RelevanceMammographic density is the strongest risk factor for

nonfamilial breast cancer among women, apart fromolder age, but its mechanistic underpinnings are poorlyunderstood.We recently reported two distinctmolecularsubtypes of normal tissue in patients with breast cancer.We hypothesized that mammographic density would beassociated with these subtypes based on their definingmolecular pathways (e.g., fibrosis and cell adhesion).Our results show that these well-defined molecularsubtypes of normal tissue are strongly associated withboth mammographic density and breast tissue compo-sition, establishing novel molecular correlates of mam-mographic density. Many of the pathways enriched inpatients with higher mammographic density are target-able, raising the possibility of developing preventionstrategies for mitigating density-associated breast cancerrisk.

Mammographic Density and Gene Expression

www.aacrjournals.org Clin Cancer Res; 19(18) September 15, 2013 4973




and then constructed into 20 mm slides. The central portionwas used for RNA extraction. After hematoxylin and eosin(H&E) staining, the slides were scanned into high-resolutiondigital images using the Aperio Scan-Scope XT Slide Scanner(Aperio Technologies) in the University of North Carolina(UNC) Translational Pathology Laboratory. After excludingslides with poor resolution or having folded tissues, slidesfrom 118 women (97.5%) were subjected to breast tissuecomposition analysis. To train the composition estimator inAperio’s Genie software, 15 representative digital slides wereselected and manually annotated for epithelial area, stromalarea, and total area (mm2) using Aperio ImageScope soft-ware. These digital area-based, quantitative estimates wereused to train Aperio’s Genie Classifier to partition epitheli-um, adipose tissue, nonfatty stroma, and glass into percen-tages. Examples of annotated digital images are presented inSupplementary Fig. S1. The regular H&E counterparts ofthese 15 digital slides were also evaluated by eye by apathologist who provided semiquantitative estimates of thepercentage of adipose tissue (10% bin width), epithelium(1% bin width), and nonfatty stroma (10% bin width). Toassess the performance of Genie classifiers, we compared theresults of three methods (by Genie, by pathologist digitalslide–based, and by pathologist regular H&E slide–based).The trained classifier was positively and strongly correlatedwithmanually scored area based on the digital images, for allthree-tissue compartments. The trained Genie classifier wasstrongly correlated with pathologist review based on regularH&E slides for stroma and adipose (Supplementary Table S1;Pearson correlation coefficient ranged 0.95–0.96), whereasrelatively lower for epithelium (Pearson correlation coeffi-cient ¼ 0.68). Compared with digital assessment, visualassessment (by human eye on regular H&E slides) of smallpercentage differences is weaker, such as epithelial tissue,which is sparse (<10%) in benign breast. Thus the digitalimage analysis data were used in analyses, and the trainedGenie Classifier was then applied to the remaining slides.

RNA isolation and microarraysAll microarrays on non-neoplastic breast tissue were

conducted at the University of North Carolina at ChapelHill (Chapel Hill, NC). The central section of fresh-frozennon-neoplastic tissue (as described earlier) was homoge-nized using aMagNALyser homogenizer (Roche), andRNAwas isolated by QIAzol extraction followed by purificationon an RNeasy column as described in Troester and collea-gues (5). RNA quality and quantity were analyzed on anAgilent 2100 Bioanalyzer and a ND-1000 NanoDrop spec-trophotometer, respectively, before running two-color 4 �44 K Agilent whole-genome arrays. Cy3-labeled referencewas produced from total RNA from Stratagene UniversalHuman Reference (spiked 1:1,000 with MCF-7 RNA and1:1,000 with ME16C RNA to increase expression of breastcancer genes) following amplificationwithAgilent lowRNAinput amplification kit. The same protocol was applied tototal RNA from breast tissues, with all patient sampleslabeledwithCy5.Datawere lowess-normalized, andprobesthat had a signal of less than 10 dpi in either channel were

excluded as missing. Probes that had more than 20%missing data across all samples were excluded from furtheranalysis. In expressiondata preprocessing,we (i) eliminatedthe probes without corresponding ENTREZ ID, (ii) col-lapsed the duplicate probes by averaging, (iii) imputedmissing data using k-nearest neighbors (KNN)methodwithk¼10, and (iv)median-centered eachgene.Microarraydatafor 121 specimens used in this analysis are publicly avail-able through the Gene Expression Omnibus (GSE49175).

Statistical analysisSamples were classified as having Active or Inactive extra-

tumoral subtype using unsupervised hierarchical clustering(average linkage) on the genes in the published Active/Inactive extratumoral signature (6). Of note, 324 of 3,518genes in the original Active signature were excluded in theclustering analysis as they did not pass filtering criteriadescribed earlier. Consensus clustering was used to evaluatewhether two clusters resulted in optimal segregation of thesample classes (18). The Active or Inactive extratumoralsubtype obtained by clustering was validated using theCreighton correlation method (19), described in Supple-mentary Fig. S2. Briefly, a standard vector corresponding toall genes in the Active/Inactive signature was constructed,with 1 assigned to upregulated genes and �1 assigned todownregulated genes. A Pearson correlation coefficient wascalculated for this standard vector versus the vector ofmedian-centered gene expression for each patient. The iden-tities of the 3,194 genes and their corresponding standardvector for Creighton correlation in the present study wereshown in Supplementary Table S2. Patientswere classified asActive if the Pearson correlation coefficient was greater thanzero, and Inactive if the coefficient was less than zero.

The distributions of breast cancer risk factors by theActive/Inactive subtype were assessed, including age atdiagnosis (continuously and categorically as <50 and�50 years), body mass index (BMI; continuously andcategorically as <30 and �30 kg/m2), age at menarche(<12 and �12 years), parity (nulliparous and parous), ageat first full-term birth in parous women (<25 and �25years), use of oral menopausal hormone therapy (MHT;never and ever), menopausal status (premenopausal andpostmenopausal), family history of breast cancer in the first-degree relative (yes and no), previous benign breast diseasehistory (defined as if a subject ever had any biopsy or partialremoval procedure, but no cancer detected; yes and no),percentage mammographic density (continuously and cat-egorically as <25% and �25%), and dense area (continu-ously and categorically as <median and �median; median¼ 36.52 cm2). Moreover, the distributions of clinicopath-ologic characteristics by the Active/Inactive subtype wereevaluated, including tumor size (�2 and >2 cm), histologictype (ductal and others), differentiation stage (well/mod-erate and poor), number of positive axillary lymphnodes (0and �1), ER status, and intrinsic subtype. To identifypotential confounding factors, the distributions of clinico-pathologic characteristics and breast cancer risk factors bypercentage mammographic density and dense area were

Sun et al.

Clin Cancer Res; 19(18) September 15, 2013 Clinical Cancer Research4974




also examined. The corresponding associations were eval-uated using x2 tests for categorical variables and Student ttests for continuous variables. Moreover, to assess the mag-nitude of association between risk factor exposure (e.g.,obese vs. nonobese) and extratumoral microenvironmentsubtype, we computed ORs for the Inactive subtype (vs.Active, referent) and corresponding 95% confidence inter-vals (CI). Analysis of covariance (ANCOVA) was used toexamine the relationship between Active/Inactive subtypeand mammographic density, adjusting for potential con-founders, age and BMI, continuously.To explore the biologic meaning of microenvironment

subtypes, the gene expression profile in each sample wascompared with five previously reported gene expressionsignatures of related phenotypes, including young age(20), obesity (21), desmoid-type fibrosis (DTF; ref. 22),TGF-b (6), and estrogen [17b-estradiol (E2)] response (23).To ensure that each of these signatures was independent,overlap between gene sets was assessed and found to besmall (Supplementary Table S3). The median-centeredexpression profile of each patient was evaluated by calcu-lating Pearson correlation coefficients, using the method ofCreighton and colleagues described earlier (19). Moreinformation about the gene ID and direction of each sig-nature is listed in Supplementary Table S4. The associationsof existing signatures with the extratumoral subtype (Activeand Inactive), percentage mammographic density (<25%and �25%), and breast tissue composition (percentageadipose, epithelium, and stroma) were assessed by x2 testsand Student t test, respectively. Nonparametric exact meth-ods were used when expected cell count was less than 5.Logistic regression and generalized linear regression wereused to further evaluate the associations after adjusting forage and/or BMI. Probability values of less than 0.05 wereconsidered statistically significant. All analyses were con-ducted using R (version 2.9.2).

ResultsAssociations between Active/Inactive subtype andbreast cancer risk factorsUnsupervised hierarchical clustering (average linkage) on

the approximately 3,000 genes from the published Active/Inactive signature was used to classify extratumoral breasttissues from 121 cases into two homogeneous groups. Con-sensus clustering documented that two groups providedstable segregation of thedata (Supplementary Fig. S3).Usingthe Creighton correlation method to categorize the micro-environment subtype as described inMaterials andMethodsalso confirmed that the direction of association for Activeand Inactive groups by clustering was concordant with theprevious publication of the signature, showing high agree-mentof the two classificationmethodswith k statistic of 0.92(P < 0.0001). Approximately, 49% of patients (n¼ 59) wereActive in the current study, compared with 38% of patients(n ¼ 27) in Roman-Perez and colleagues (6).Consistent with our previous findings (6), we observed

no association between extratumoral subtype and ER status

and other tumor characteristics (Supplementary TableS5). Table 1 shows the distribution of participant char-acteristics, overall and categorized according to extratu-moral microenvironment subtype. Of all the breast cancerrisk factors evaluated, mammographic density measuresshowed the strongest differences by Active/Inactive sub-type. Compared with the Active group, mean percentagedensity was 9% higher and mean dense area was 8.8 cm2

more in the Inactive group (percentage density: P ¼0.0013; dense area: P ¼ 0.049). Substantial (but statisti-cally insignificant) associations were noted for age (age<50 vs. �50 years: OR, 2.00; 95% CI, 0.88–4.57) and BMI(BMI <30 vs. �30 kg/m2: OR, 1.76; 95% CI, 0.79–3.93),but no other breast cancer risk factors showed substantialassociations with Active/Inactive subtype.

We also evaluated potential confounders of the associa-tion between mammographic density and extratumoralmicroenvironment subtype. Both elevated percentage den-sity and dense area were significantly associated with youn-ger age, lower BMI, ever use of oral MHT, and premeno-pausal status (Supplementary Table S6). Of these factors,only age and BMI showed substantial associations withextratumoral subtype. Thus, we conducted multivariableanalysis to study mean differences in mammographic den-sity by extratumoral microenvironment subtype, control-ling for age and BMI (Table 2). In multivariable models,percentage density and dense area remained higher amongthe Inactive group (percentage density: P ¼ 0.0038; densearea: P ¼ 0.049).

Biologic relevance of Active/Inactive subtypeTo better understand the molecular characteristics of

extratumoral microenvironment subtype and mammo-graphic density–associated breast cancer risk, associationswere evaluated between Active/Inactive subtype, mammo-graphic density, and biologically relevant, published geneexpression signatures. As shown in Table 3, both the Inac-tive subtype and higher percentage density (defined as�25%) were significantly positively associated with ayoung-like age signature and an increased estrogen responseexpression signature (E2), whereas they were significantlyinversely associated with an obesity signature and a TGF-bsignature. After adjusting for patient age and/or BMI, mostof these associations remained statistically significant, butthe association of high percentage density with the young-like age signature and low TGF-b signature was no longersignificant (Supplementary Table S7). The previouslyreported signature derived from desmoid-type fibroidtumors and indicative of fibroblast response (DTF) wasonly associatedwith lower percentage density, but not afteradjusting for BMI and age.

Associations between subtype/mammographic densityand breast tissue composition

Given variation in stroma-derived signatures, we expectedthat stromal composition may vary by extratumoral micro-environment subtype. Indeed, as shown in Fig. 1, there weresubstantial differences in breast tissue composition by






Active/Inactive subtype. These differences paralleled thedifferences in tissue composition by percentage mammo-graphic density and dense area. The percentage of adiposetissue was significantly lower and the percentage of non-fatty stromal tissue was significantly higher among Inac-tive/mammographically dense patients (SupplementaryTable S8). No differences were detected in the percentageof epithelial tissue by extratumoral microenvironmentsubtype or measures of mammographic density.

Finally, because Active/Inactive subtype and mammo-graphic density seemed to be tracking cellular composition,the association between published gene expression signa-tures and cellular composition was also considered (Table3). Again mirroring microenvironment subtype and mam-mographic density, the percentage of nonfatty stroma wasassociated with young-like gene expression, decreasedexpression of obesity-associated genes, decreased expr-ession of a TGF-b signature, and increased expression of

Table 1. Participant characteristics by the Active/Inactive extratumoral microenvironment subtype

Total (n ¼ 121) Active (n ¼ 59) Inactive (n ¼ 62)

Characteristics n (%) n (%) n (%) Pa

Age at diagnosisMean (SD) 55.26 (9.78) 56.03 (8.72) 54.52 (10.70) 0.39<50 y 33 (27) 12 (20) 21 (34) 0.095�50 y 88 (73) 47 (80) 41 (66)

BMIMean (SD) 27.69 (5.30) 28.49 (5.48) 26.93 (5.06) 0.10<30 kg/m2 87 (72) 39 (66) 48 (77) 0.17�30 kg/m2 34 (28) 20 (34) 14 (23)

Age at menarche<12 y 40 (33) 18 (31) 22 (36) 0.50�12 y 80 (67) 41 (69) 39 (64)

Parity0 26 (21) 11 (19) 15 (24) 0.46�1 95 (79) 48 (81) 47 (76)

Age at first full-term birth in parous women<25 y 52 (55) 25 (52) 27 (57) 0.60�25 y 43 (45) 23 (48) 20 (43)

Use of oral MHTNever 80 (68) 37 (65) 43 (70) 0.52Ever 38 (32) 20 (35) 18 (30)

Menopausal statusPremenopausal 34 (28) 14 (24) 20 (32) 0.30Postmenopausal 87 (72) 45 (76) 42 (68)

Family history of breast cancerb

No 101 (84) 48 (83) 53 (85) 0.68Yes 19 (16) 10 (17) 9 (15)

Previous benign breast diseaseNo 89 (75) 45 (79) 44 (72) 0.39Yes 29 (25) 12 (21) 17 (28)

Percentage densityMean (SD) 27.25 (16.00) 22.56 (13.90) 31.70 (16.69) 0.0013<25% 57 (47) 35 (59) 22 (35) 0.0086�25% 64 (53) 24 (41) 40 (65)

Dense area (cm2)Mean (SD) 37.70 (24.59) 33.21 (23.08) 41.97 (25.40) 0.049<Medianc 60 (50) 33 (56) 27 (44) 0.17�Median 61 (50) 26 (44) 35 (56)

aP values for the comparisons between Active and Inactive subtypes were calculated by t test for continuous variables and x2 test forcategorical variables. Missing values were excluded from percentage calculations.bFamily history of breast cancer in first-degree relative.cMedian ¼ 36.52 cm2.

Sun et al.





estrogen response signatures. Epithelial composition wasmodestly associated with gene expression signatures for ageandobesity (higher proportion in young-like andnonobesegroups) after adjusting for BMI and age, respectively (Sup-plementary Table S7).

DiscussionBy linking gene expression patterns of non-neoplastic

breast tissue from patients with breast cancer to mammo-

graphic density measures of their unaffected breast, weobserved that distinct extratumoral microenvironmentswere independently associated with mammographic den-sity. In particular, higher percentage density and dense areawere related to the Inactive subtype. Histologically weconfirmed these findings and found that the Inactive sub-type and higher mammographic density were also stronglyand positively associated with the proportion of stromalcomposition. Further analysis of published gene expression

Table 2. Multivariable analysis of the association between mammographic density measures and theActive/Inactive extratumoral microenvironment subtype

Mammographic density Active (n ¼ 59) Inactive (n ¼ 62)

Model measures Mean (SE) Mean (SE) Pa

Crude mean Dense area, cm2 33.21 (3.16) 41.97 (3.09) 0.049Percentage density, % 22.56 (2.00) 31.70 (1.95) 0.0014

Age-adjusted mean Dense area, cm2 33.84 (3.01) 41.37 (2.94) 0.076Percentage density % 23.20 (1.71) 31.09 (1.67) 0.0013

BMI-adjusted mean Dense area, cm2 33.17 (3.19) 42.01 (3.12) 0.051Percentage density % 23.60 (1.81) 30.71 (1.77) 0.0060

Age and BMI-adjusted mean Dense area, cm2 33.41 (3.01) 41.78 (2.93) 0.049Percentage density % 23.77 (1.63) 30.56 (1.59) 0.0038

aP valueswere calculatedby t test for univariate analysis andANCOVA formultivariable analysis. AgeandBMIwereusedas continuousvariables.

Table 3. Relationship of biologically relevant signatures with the Active/Inactive extratumoralmicroenvironment subtype, mammographic density, and breast composition

Microenvironmentsubtypea

Percentagedensitya Breast compositionb

n (%) n (%) Mean (SE)

Signature (# genes used/# in signature)

Active(n ¼ 59)

Inactive(n ¼ 62)

<25%(n ¼ 57)

�25%(n ¼ 64)

Adipose(%)

Epithelium(%)

Stroma(%)

Age (779/802) Old 56 (95) 9 (15) 36 (63) 29 (45) 74.52 (2.64) 8.74 (0.87) 16.74 (2.56)Young 3 (5) 53 (85) 21 (37) 35 (55) 47.99 (2.77) 10.85 (0.82) 41.15 (2.69)P 6.2 � 10�21 0.049 3.6 � 10�10 0.079 3.1 � 10�09

BMI (677/760) Obese 56 (95) 4 (6) 35 (61) 25 (39) 78.07 (2.52) 8.55 (0.86) 13.38 (2.46)Nonobese 3 (5) 58 (94) 22 (39) 39 (61) 46.86 (2.44) 10.85 (0.83) 42.29 (2.37)P 1.0 � 10�26 0.014 7.3 � 10�15 0.055 8.4 � 10�14

DTF (581/758) Positive 31 (53) 27 (44) 36 (63) 22 (34) 65.37 (3.27) 10.22 (0.88) 24.41 (3.11)Negative 28 (47) 35 (56) 21 (37) 42 (66) 58.83 (3.11) 9.31 (0.83) 31.87 (2.96)P 0.32 0.0016 0.14 0.45 0.079

TGF-b (210/234) Positive 53 (90) 12 (19) 37 (65) 28 (44) 74.34 (2.61) 9.10 (0.82) 16.55 (2.50)Negative 6 (10) 50 (81) 20 (35) 36 (56) 47.72 (2.80) 10.47 (0.88) 41.81 (2.68)P 7.8 � 10�15 0.020 2.7 � 10�10 0.26 8.0 � 10�10

E2 (711/754) Positive 10 (17) 49 (79) 21 (37) 38 (59) 49.36 (2.80) 10.63 (0.85) 40.00 (2.70)Negative 49 (83) 13 (21) 36 (63) 26 (41) 74.08 (2.75) 8.88 (0.84) 17.04 (2.65)P 8.5 � 10�12 0.013 5.3 � 10�09 0.15 1.7 � 10�08

aP values were calculated by t test for continuous variables and x2 test for categorical variables except that when expected cell countwas less than 5, they were calculated by Fisher exact test.bThe sample size was 118 due to missing breast composition information for 3 patients.






signatures implicated estrogen response and TGF-b signal-ing pathways, providing clues about the biology underlyingmammographic density, a strong and consistent risk factorfor breast cancer.

Mammographic density is believed to reflect the com-bined effects of cell proliferation and genetic damage toproliferating epithelial cells (24). This hypothesis is sup-ported by strong associations between mammographic

density and exogenous estrogen exposure (25, 26), and inthe current study, our data provide further molecularsupport for the role of estrogen in density. We observedstrong estrogen-response gene expression among thosewith dense tissue and among those with the Inactiveextratumoral microenvironment subtype. The strength ofthe molecular relationships is striking in light of recentfindings showing relatively weak molecular evidence of

Ad

ipo

se

pe

rce

nta

ge

Ep

ith

elia

l p

erc

en

tag

e

10

08

06

04

0

Active

<25% ≥ 25% <25% ≥ 25% <25% ≥ 25% <25% ≥ 25%

<Median ≥ Median<Median ≥ Median<Median ≥ Median<Median

Adipose Epithelial Stromal

≥ Median

Inactive

By microenvironment subtypeA

By mammographic percentage densityB

By mammographic dense area (median = 36.52 cm2)C

Active Inactive Active Inactive Active Inactive

20

01

02

03

0

Ep

ith

elia

l p

erc

en

tag

e

01

02

03

0

Ep

ith

elia

l p

erc

en

tag

e

01

02

03

0

Ad

ipo

se

pe

rce

nta

ge

10

08

06

04

02

0 Str

om

al p

erc

en

tag

e 80

60

40

20

0

Str

om

al p

erc

en

tag

e

80

60

40

20

0

Str

om

al p

erc

en

tag

e

Co

mp

ositio

n p

erc

en

tag

e

80

60

40

20

0

80

10

06

04

02

00

Co

mp

ositio

n p

erc

en

tag

e

80

10

06

04

02

00

Co

mp

ositio

n p

erc

en

tag

e

80

10

06

04

02

00

Ad

ipo

se

pe

rce

nta

ge

10

08

06

04

02

0

Figure 1. Breast tissue composition distribution by the Active/Inactive subtype and mammographic density. Using quantitative analysis of digital histologyimages, each sample from thePWBCS forwhich gene expressionwasmeasured alsowas assessed for composition.Nonfatty stromal percentagewashigherin Inactive subtype samples (A) andwomenwith highermammographic density (percentage, B; dense area,C). In contrast, adipose concentrationwas lower inthe Inactive/higher density patients. Epithelial content did not vary significantly by any of these variables.

Sun et al.





estrogen response in dense breast tissue. Haakensen andcolleagues found a signature of 24 genes differentiallyexpressed between high- and low-density breasts, includ-ing the gene coding for estrogen receptor (ESR1) andthree uridine 50-diphospho-glucuronosyltransferase (UGT)family genes (27) with postulated roles in protecting epi-thelial cells from genetic damage by local estrogen meta-bolites (27, 28). However, in a later study, Haakensen andcolleagues observed that while serum estradiol was associ-ated with mammographic density, none of the gene expres-sion changes related to serum estradiol was significantlyassociated with mammographic density (29). It is possiblethat the lack of significant associations with estrogenresponse in previous studies is attributable to differencesin the patient populations or due to the emphasis on fewergenes.Interestingly, we also observed a strong association

between estrogen response and stromal content (highernonfatty stroma/lower breast adiposity) in non-neoplasticbreast tissue, whereas the association between estrogenresponse and epithelial proportion was not significant. Thisnull result for epithelium may be attributed to lower var-iation in epithelial percentage across patients (interquartilerange for percentage composition was 4.4%–13.2%) andrandom measurement error in our estimation of epithelialpercentage, but our molecular data and the publishedliterature underscore the importance of stroma inmammo-graphic density (30–32). Few gene expression studies haveemphasized stromal biology in mammographic density,but one study by Yang and colleagues used canonicalpathway analysis of mammographic density–associatedgene expression to identify decreased TGF-b signaling indense breast tissue (33). Our findings confirmed this asso-

ciation by showing the strong association between TGF-bsignature and adipose and stromal composition. TGF-b isan important morphogen in normal mammary tissue, withinhibitory roles on proliferation of both ERþ and ER� cells(34). Thus, the decreased TGF-b pathway expression indense tissue may be upstream of differences in epithelialproliferation ormay reflect altered stromal–epithelial inter-action during oncogenesis (35–37).

Previous models to interpret the association betweenmammographic density and breast cancer risk have focusedon mitogens and mutagens, where luminal epithelial cellsare central in interpreting the mammographic density–breast cancer association. Although a breast tumor origi-nates from epithelial cells, the extended model suggests amore dominant role of stromal microenvironment andmorphogenesis in tumorigenesis (38), which supports theimportant role of interactions between stroma and epithe-lial cells in malignancy transformation proposed in theprevious studies (39). Consequently, studies focusing onthe association between mammography and breast tissuecomposition have helped to improve our understanding ofthe mammographic density–breast cancer association(30, 40–42). Given that themolecular findings of this studysupport and extend previously proposed mechanisms, wesuggest a revised conceptual model for the mammographicdensity–breast cancer association (Fig. 2; ref. 24). Indeed,other recent data also support the fundamental importanceof microenvironment in density (43), and further suggest acritical interplay between mechanics, morphogenesis, andmalignancy (44–46). Mammographic density may be apowerful biologic marker in its ability to comprehensivelysummarize the variation in mitogenesis, mutagenesis, andmorphogenesis of breast.

Hormones and growth factorsMITOGENESIS

MUTAGENESIS

MORPHOGENESIS

Cell proliferation

Mutation Breast cancer

Stromal composition

Mammographic density

Lipid peroxidation

Risk factors(genetic, dietary,reproductive,anthropometric,etc.)

Microenvironment alteration

(e.g., TGF-β activation)

Figure 2. Conceptual model for potential mechanisms of the association between mammographic density and breast cancer risk with a focus on the role ofstroma. This figure is adapted from the biologic model proposed by Martin and Boyd (24), showing the underlying biologic processes linking risk factors tobreast cancer and suggesting the surrogate role of mammographic density. The genetic and environmental factors not only influence epithelial cellproliferation and induce genetic damage, but also regulate the surrounding microenvironment and stromal composition. The collective abnormalities inmitogenesis,mutagenesis, andmorphogenesis result in breast cancer development. By comprehensively summarizing the variation in these three processes,mammographic density works as a strong marker of breast cancer risk.






The observation that a strong component of genomicvariation in breast tissue is associated with a strong riskfactor for breast cancer, suggests that further investigation ofnormal tissue gene expression could yield novel insightsabout the biology of breast cancer risk andmammographicdensity–associated risk in particular. In fact, the normaltissue gene expression seems to reflect host factors and riskfactor exposure more strongly than tumor characteristics;none of the tumor characteristics evaluated (ER status,grade, breast tumor subtype, size, etc.) showed significantassociations with the Active/Inactive subtype. Althoughthere may be genes whose extratumoral expression is affect-ed by tumor subtype, the set of genes that determine themajor variation (andActive/Inactive subtype) are not tumorassociated. Previous epidemiologic studies evaluatingwhether mammographic density is related to breast cancercharacteristics are conflicting, with some suggesting thathigh density is associated withmore aggressive disease, andsome speculating that observed associations betweenmam-mographic density and tumor biology may be explained bydelayed diagnosis among high-density cases (i.e., maskingbias; refs. 47–50). Our current and previous research onActive/Inactive subtypes suggests no link with tumorcharacteristics, supporting the latter hypothesis that mam-mographic density does not have distinct influences onetiology of subtypes.

The strong association between mammographic densityand gene expression in the current study may be a result oflooking at broad transcriptional changes rather than indi-vidual gene-level correlates of density. Although a limitednumber (27, 33) of studies previously addressed genomicsignatures of mammographic density, these studies identi-fied very few density-associated genes (<100 genes), limit-ing their value in defining the biology of mammographicdensity. Weak associations with individual gene expressionchanges in previous studiesmay reflect low statistical powerto detect a small differentially expressed signature (51, 52),but it is striking that such a strong breast cancer risk factorshould produce suchweak genomic expression, when otherfactors such as age and BMI have broad effects on normaltissue gene expression (20, 21). Another explanation for thestrength of our pattern-focused analysis may be that mam-mographic density summarizes several biologically com-plex aspects of breast physiologic andpathologic conditionsand itself may not represent a singular biologic state. If thislatter explanation were true, analysis starting from mam-mographic density would be poorly correlated with indi-vidual gene expression values, as gene-by-gene supervisedanalysis works best in homogeneously defined exposures.In contrast, Active/Inactive subtype summarizes the effect ofvarious factors influencing breast microenvironment (i.e.,strong associations with age, obesity, TFG-b, and E2 signa-tures), and therefore better captures genomic variation.Future research should test other biologic themes for theirassociation with mammographic density and Active/Inac-tive microenvironment.

Our results should be interpreted with consideration ofseveral limitations. The non-neoplastic breast tissues used

in the present study were taken from patients with breastcancer. The influence of tumor on extratumoral micro-environment has been reported previously (5, 53) and wecannot exclude the possibility that the tissue of womenwith disease are systematically different from those with-out disease. However, as discussed earlier, the evidencethat Active/Inactive microenvironment is not associatedwith tumor characteristics has led us to hypothesize thatActive/Inactive subtype is a host-dependent factor. Cau-tion is necessary when extrapolating these findings to thenormal breast microenvironment of nondiseased wom-en. Future research on normal breast tissues from healthywomen is needed and will clarify the role of the Inactivemicroenvironment in disease development. Another keylimitation is sample size. We cannot rule out the possi-bility that negative results for associations between extra-tumoral environment subtype and parity, menopause,and MHT were a result of limited power. Besides inheritedfactors (54), other factors also play a role in mammo-graphic density and would be expected to show associa-tions with Active/Inactive subtype (55, 56). Indeed, ageand BMI, the two strongest identified endogenous factorsfor mammographic density, showed substantial associa-tions with Active/Inactive microenvironment (estimatedORs > 1.5). These associations were stronger when geno-mic surrogates for age and BMI, developed on the basis ofnormal breast and therefore capturing underlying biolog-ic influence on local organs, were examined. We also notethat we used mammograms of the unaffected breast toestimate density, avoiding the potential interference oftumor. Although this may have introduced some mea-surement error, mammographic density is highly corre-lated within a woman and density in the unaffected breasthas been found to be comparable with that in the affectedbreast (49, 56), so this is not expected to bias our results.

ConclusionIn summary, we evaluated extratumoral microenviron-

ment subtype for associations withmammographic densityand breast tissue composition. On the basis of strongassociations between the genomic subtypes and mammo-graphic density, it is likely that Active/Inactive subtype isalso associated with and reflective of breast cancer risk.Further research to better understand the molecular char-acteristics of mammographic density and microenviron-ment subtypes could identify pathways that are targetablein preventing mammographic-density associated risk.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

Authors' ContributionsConception and design: X. Sun, G.L. Gierach, J.D. Figueroa,M.E. Sherman,M.A. TroesterDevelopment of methodology: X. Sun, R. Sandhu, T. Williams, B.R.Midkiff, M.E. Sherman, M.A. TroesterAcquisitionofdata (provided animals, acquired andmanagedpatients,provided facilities, etc.):G.L.Gierach, R. Sandhu, T.Williams, B.R.Midkiff,J. Lissowska, N.F. Boyd, N.B. Johnson, J.D. Figueroa, M.E. Sherman, M.A.Troester

Sun et al.





Analysis and interpretation of data (e.g., statistical analysis, biosta-tistics, computational analysis): X. Sun, G.L. Gierach, R. Sandhu, T.Williams, B.R. Midkiff, J. Lissowska, J.D. Figueroa, M.E. ShermanWriting, review, and/or revision of themanuscript: X. Sun, G.L. Gierach,R. Sandhu, J. Lissowska, J.D. Figueroa, M.E. Sherman, M.A. TroesterAdministrative, technical, or material support (i.e., reporting or orga-nizing data, constructing databases): X. Sun, R. Sandhu, B.R. Midkiff, J.Lissowska, E. WesolowskaStudy supervision: J. Lissowska, M.A. Troester

AcknowledgmentsThe authors thank Lineberger Comprehensive Cancer Center’s Transla-

tional Pathology Laboratory for support of the image analysis work.

Grant SupportThis project was supported by grants from the National Cancer Institute

(U01-ES019472 and R01-CA138255), a Breast SPORE (P50CA058223)Career Development Award (to M.A. Troester), and a grant from the AvonFoundation. This work was supported (in part) by the Intramural ResearchProgram of the NIH, National Cancer Institute.

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.

Received January 4, 2013; revised May 30, 2013; accepted July 10, 2013;published OnlineFirst August 5, 2013.

References1. PerouCM,Sorlie T, EisenMB, van deRijnM, Jeffrey SS, Rees CA, et al.

Molecular portraits of humanbreast tumours. Nature 2000;406:747–52.2. Sorlie T, Perou CM, Tibshirani R, Aas T, Geisler S, Johnsen H, et al.

Gene expression patterns of breast carcinomas distinguish tumorsubclasses with clinical implications. Proc Natl Acad Sci U S A 2001;98:10869–74.

3. Cichon MA, Degnim AC, Visscher DW, Radisky DC. Microenviron-mental influences that drive progression frombenign breast disease toinvasive breast cancer. J Mammary Gland Biol Neoplasia 2010;15:389–97.

4. Bissell MJ, Radisky DC, Rizki A, Weaver VM, Petersen OW. Theorganizing principle: microenvironmental influences in the normal andmalignant breast. Differentiation 2002;70:537–46.

5. Troester MA, Lee MH, Carter M, Fan C, Cowan DW, Perez ER, et al.Activation of host wound responses in breast cancer microenviron-ment. Clin Cancer Res 2009;15:7020–8.

6. Roman-Perez E, Casbas-Hernandez P, Pirone JR, Rein J, Carey LA,Lubet RA, et al. Gene expression in extratumoral microenvironmentpredicts clinical outcome in breast cancer patients. Breast Cancer Res2012;14:R51.

7. Finak G, Bertos N, Pepin F, Sadekova S, Souleimanova M, Zhao H,et al. Stromal gene expression predicts clinical outcome in breastcancer. Nat Med 2008;14:518–27.

8. Polyak K, Kalluri R. The role of the microenvironment in mammarygland development and cancer. Cold Spring Harb Perspect Biol2010;2:a003244.

9. Fukino K, Shen L, Matsumoto S, Morrison CD, Mutter GL, Eng C.Combined total genome loss of heterozygosity scan of breast cancerstroma and epithelium reveals multiplicity of stromal targets. CancerRes 2004;64:7231–6.

10. Weber F, Shen L, Fukino K, Patocs A, Mutter GL, Caldes T, et al. Total-genome analysis of BRCA1/2-related invasive carcinomas of thebreast identifies tumor stroma as potential landscaper for neoplasticinitiation. Am J Hum Genet 2006;78:961–72.

11. Fukino K, Shen L, Patocs A, Mutter GL, Eng C. Genomic instabilitywithin tumor stroma and clinicopathological characteristics of spo-radic primary invasive breast carcinoma. JAMA 2007;297:2103–11.

12. MaXJ,DahiyaS,RichardsonE, ErlanderM,SgroiDC.Geneexpressionprofiling of the tumormicroenvironment during breast cancer progres-sion. Breast Cancer Res 2009;11:R7.

13. Schedin P, Elias A. Multistep tumorigenesis and the microenviron-ment. Breast Cancer Res 2004;6:93–101.

14. Bissell MJ, HinesWC.Whydon't we getmore cancer? A proposed roleof the microenvironment in restraining cancer progression. Nat Med2011;17:320–9.

15. Garcia-ClosasM, Brinton LA, Lissowska J, ChatterjeeN, Peplonska B,Anderson WF, et al. Established breast cancer risk factors by clinicallyimportant tumour characteristics. Br J Cancer 2006;95:123–9.

16. Camp JT, Elloumi F, Roman-Perez E, Rein J, Stewart DA, Harrell JC,et al. Interactions with fibroblasts are distinct in basal-like and luminalbreast cancers. Mol Cancer Res 2011;9:3–13.

17. Byng JW, Boyd NF, Fishell E, Jong RA, Yaffe MJ. The quantitativeanalysis of mammographic densities. Phys Med Biol 1994;39:1629–38.

18. Wilkerson MD, Hayes DN. ConsensusClusterPlus: a class discoverytool with confidence assessments and item tracking. Bioinformatics2010;26:1572–3.

19. Creighton CJ, Casa A, Lazard Z, Huang S, Tsimelzon A, HilsenbeckSG, et al. Insulin-like growth factor-I activates gene transcriptionprograms strongly associated with poor breast cancer prognosis.J Clin Oncol 2008;26:4078–85.

20. Pirone JR, D'arcy M, Stewart DA, Hines WC, Johnson M, Gould MN,et al. Age-associated gene expression in normal breast tissue mirrorsqualitative age-at-incidence patterns for breast cancer. Cancer Epi-demiol Biomarkers Prev 2012;21:1735–44.

21. Sun X, Casbas-Hernandez P, Bigelow C,Makowski L, Joseph Jerry D,Smith Schneider S, et al. Normal breast tissue of obese women isenriched for macrophage markers and macrophage-associated geneexpression. Breast Cancer Res Treat 2012;131:1003–12.

22. Beck AH, Espinosa I, Gilks CB, van de Rijn M, West RB. The fibro-matosis signature defines a robust stromal response in breast carci-noma. Lab Invest 2008;88:591–601.

23. Oh DS, Troester MA, Usary J, Hu Z, He X, Fan C, et al. Estrogen-regulated genes predict survival in hormone receptor-positive breastcancers. J Clin Oncol 2006;24:1656–64.

24. Martin LJ, Boyd NF.Mammographic density. potential mechanisms ofbreast cancer risk associated with mammographic density: hypothe-ses based on epidemiological evidence. Breast Cancer Res 2008;10:201.

25. ChlebowskiRT,HendrixSL, LangerRD,StefanickML,GassM,LaneD,et al. Influence of estrogen plus progestin on breast cancer andmammography in healthy postmenopausal women: the women'shealth initiative randomized trial. JAMA 2003;289:3243–53.

26. Kerlikowske K, Cook AJ, Buist DS, Cummings SR, VachonC, Vacek P,et al. Breast cancer risk by breast density, menopause, and postmen-opausal hormone therapy use. J Clin Oncol 2010;28:3830–7.

27. Haakensen VD, Biong M, Lingjaerde OC, Holmen MM, Frantzen JO,Chen Y, et al. Expression levels of uridine 50-diphospho-glucuronosyl-transferase genes in breast tissue from healthy women are associatedwith mammographic density. Breast Cancer Res 2010;12:R65.

28. Guillemette C, Belanger A, Lepine J. Metabolic inactivation of estro-gens in breast tissue by UDP-glucuronosyltransferase enzymes: anoverview. Breast Cancer Res 2004;6:246–54.

29. Haakensen VD, Bjoro T, Luders T, Riis M, Bukholm IK, Kristensen VN,et al. Serum estradiol levels associated with specific gene expressionpatterns in normal breast tissue and in breast carcinomas. BMCCancer 2011;11:332.

30. Li T, Sun L, Miller N, Nicklee T, Woo J, Hulse-Smith L, et al. Theassociation of measured breast tissue characteristics with mammo-graphic density and other risk factors for breast cancer. CancerEpidemiol Biomarkers Prev 2005;14:343–9.

31. Yamaguchi Y. Microenvironmental regulation of estrogen signals inbreast cancer. Breast Cancer 2007;14:175–81.

32. Miki Y, Suzuki T, Sasano H. Controversies of aromatase localization inhuman breast cancer–stromal versus parenchymal cells. J SteroidBiochem Mol Biol 2007;106:97–101.

33. Yang WT, Lewis MT, Hess K, Wong H, Tsimelzon A, Karadag N, et al.Decreased TGFbeta signaling and increased COX2 expression in high






risk women with increased mammographic breast density. BreastCancer Res Treat 2010;119:305–14.

34. Moses H, Barcellos-Hoff MH. TGF-beta biology in mammary devel-opment and breast cancer. Cold Spring Harb Perspect Biol 2011;3:a003277.

35. Kuperwasser C, Chavarria T, Wu M, Magrane G, Gray JW, Carey L,et al. Reconstruction of functionally normal and malignant humanbreast tissues in mice. Proc Natl Acad Sci U S A 2004;101:4966–71.

36. Bhowmick NA, Chytil A, Plieth D, Gorska AE, Dumont N, Shappell S,et al. TGF-beta signaling in fibroblasts modulates the oncogenicpotential of adjacent epithelia. Science 2004;303:848–51.

37. DalyAC,Randall RA,Hill CS. Transforming growth factor beta-inducedSmad1/5 phosphorylation in epithelial cells is mediated by novelreceptor complexes and is essential for anchorage-independentgrowth. Mol Cell Biol 2008;28:6889–902.

38. Potter JD. Morphogens, morphostats, microarchitecture and malig-nancy. Nat Rev Cancer 2007;7:464–74.

39. Maller O, Martinson H, Schedin P. Extracellular matrix compositionreveals complex and dynamic stromal–epithelial interactions in themammary gland. J Mammary Gland Biol Neoplasia 2010;15:301–18.

40. Warren R, Lakhani SR. Can the stroma provide the clue to the cellularbasis for mammographic density? Breast Cancer Res 2003;5:225–7.

41. Alowami S, Troup S, Al-Haddad S, Kirkpatrick I, Watson PH. Mam-mographic density is related to stroma and stromal proteoglycanexpression. Breast Cancer Res 2003;5:R129–35.

42. Lin SJ, Cawson J, Hill P, Haviv I, Jenkins M, Hopper JL, et al. Image-guided sampling reveals increased stroma and lower glandular com-plexity in mammographically dense breast tissue. Breast Cancer ResTreat 2011;128:505–16.

43. DeFilippis RA, Chang H, Dumont N, Rabban JT, Chen YY, FontenayGV, et al. CD36 repression activates a multicellular stromal programshared by high mammographic density and tumor tissues. CancerDiscov 2012;2:826–39.

44. Levental KR, Yu H, Kass L, Lakins JN, EgebladM, Erler JT, et al. Matrixcrosslinking forces tumor progression by enhancing integrin signaling.Cell 2009;139:891–906.

45. Paszek MJ, Weaver VM. The tension mounts: mechanics meetsmorphogenesis and malignancy. J Mammary Gland Biol Neoplasia2004;9:325–42.

46. Butcher DT, Alliston T, Weaver VM. A tense situation: forcing tumourprogression. Nat Rev Cancer 2009;9:108–22.

47. Eriksson L, Czene K, Rosenberg L, Humphreys K, Hall P. The influenceof mammographic density on breast tumor characteristics. BreastCancer Res Treat 2012;134:859–66.

48. Ding J, Warren R, Girling A, Thompson D, Easton D. Mammographicdensity, estrogen receptor status and other breast cancer tumorcharacteristics. Breast J 2010;16:279–89.

49. Aiello EJ, Buist DS, White E, Porter PL. Association between mam-mographic breast density and breast cancer tumor characteristics.Cancer Epidemiol Biomarkers Prev 2005;14:662–8.

50. Gierach GL, Ichikawa L, Kerlikowske K, Brinton LA, Farhat GN, VacekPM, et al. Relationship between mammographic density and breastcancer death in the breast cancer surveillance consortium. J NatlCancer Inst 2012;104:1218–27.

51. Hackstadt AJ, Hess AM. Filtering for increased power for microarraydata analysis. BMC Bioinformatics 2009;10:11.

52. Bourgon R, Gentleman R, Huber W. Independent filtering increasesdetection power for high-throughput experiments. Proc Natl Acad SciU S A 2010;107:9546–51.

53. Graham K, Ge X, de Las Morenas A, Tripathi A, Rosenberg CL. Geneexpression profiles of estrogen receptor-positive and estrogen recep-tor-negative breast cancers are detectable in histologically normalbreast epithelium. Clin Cancer Res 2011;17:236–46.

54. Boyd NF, Dite GS, Stone J, Gunasekara A, English DR, McCredie MR,et al. Heritability of mammographic density, a risk factor for breastcancer. N Engl J Med 2002;347:886–94.

55. Butler LM, Gold EB, Greendale GA, Crandall CJ, Modugno F, Oes-treicher N, et al. Menstrual and reproductive factors in relation tomammographic density: the studyofwomen's health across thenation(SWAN). Breast Cancer Res Treat 2008;112:165–74.

56. Assi V, Warwick J, Cuzick J, Duffy SW. Clinical and epidemiologicalissues in mammographic density. Nat Rev Clin Oncol 2011;9:33–40.

Sun et al.





2013;19:4972-4982. Published OnlineFirst August 5, 2013.Clin Cancer Res Xuezheng Sun, Gretchen L. Gierach, Rupninder Sandhu, et al. Analysis of Normal Breast Tissue Surrounding Breast CancerRelationship of Mammographic Density and Gene Expression:

Updated version

10.1158/1078-0432.CCR-13-0029doi:

Access the most recent version of this article at:

Material

Supplementary

http://clincancerres.aacrjournals.org/content/suppl/2013/08/02/1078-0432.CCR-13-0029.DC1

Access the most recent supplemental material at:

Cited articles

http://clincancerres.aacrjournals.org/content/19/18/4972.full#ref-list-1

This article cites 56 articles, 18 of which you can access for free at:

Citing articles

http://clincancerres.aacrjournals.org/content/19/18/4972.full#related-urls

This article has been cited by 4 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://clincancerres.aacrjournals.org/content/19/18/4972To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/lookup/doi/10.1158/1078-0432.CCR-13-0029

http://clincancerres.aacrjournals.org/content/suppl/2013/08/02/1078-0432.CCR-13-0029.DC1

http://clincancerres.aacrjournals.org/content/19/18/4972.full#ref-list-1

http://clincancerres.aacrjournals.org/content/19/18/4972.full#related-urls

http://clincancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://clincancerres.aacrjournals.org/content/19/18/4972


Documents

Relationship of Mammographic Density and Gene Expression ......able, raising the possibility of developing prevention strategies for mitigating density-associated breast cancer risk