The Role of Molecular Analysis in Breast Cancer

8/12/2019 The Role of Molecular Analysis in Breast Cancer

1/12

A N C I L L A R Y T O O L S I N B R E A S T P A T H O L O G Y

The role of molecular analysis in breast cancer

FELIPE C. GEYER*, CATERINAMARCHIO`* AND JORGE S. REIS-FILHO

Molecular Pathology Laboratory, The Breakthrough Breast Cancer Research Centre, Institute of Cancer

Research, London, United Kingdom; *these authors contributed equally to this review

Summary

Breast cancer is a heterogeneous disease, encompassing a

wide variety of histological types and clinical behaviours.

Current histopathological classification systems for breast

cancer are based on descriptive entities that are of prognostic

significance. Few predictive biomarkers are currently avail-

able. High throughput molecular technologies are reshapingour understanding of breast cancer and a molecular

taxonomy that has stronger predictive power is slowly

emerging. Novel therapeutic targets and prognostic/predic-

tive gene signatures have been identified. This review will

address the contribution of molecular methods to our under-

standing of breast cancer and its precursors, their use in

breast cancer translational research and their impact on

diagnostic breast cancer histopathology.

Key words: Microarrays, comparative genomic hybridisation, prognostic

marker, predictive marker, genomic signature, histological grade, breast

cancer.

Received 20 July, revised 5 August, accepted 7 August 2008

INTRODUCTION

Breast cancer is a heterogeneous disease encompassing awide variety of pathological entities that are reported to havedistinct clinical behaviours.1,2 Pathologists have acknowl-

edged the complexity of breast cancer and endeavoured todevise classification systems to account for its diversity.However, current classification systems are descriptive,based on morphological entities that have been shown tohave prognostic implications. For the success of targeted

therapies and individualised medicine, a predictive, ratherthan purely prognostic, classification system is required.

Despite the translational research efforts of the 1980s and90s, only three biomarkers are routinely used in breastcancer patient management, namely oestrogen receptor(ER), progesterone receptor (PR) and HER2. Interestingly,all of these biomarkers have optimal negative predictivevalues (i.e., patients with ER negative breast cancer arehighly unlikely to respond to endocrine therapy; HER2negative breast cancers fail to respond to humanisedmonoclonal antibodies against HER2). However, theirpositive predictive value is rather limited, with a substantialproportion of patients with HER2 positive disease either

harbouring de novo resistance or developing resistance totrastuzumab over time.3

Molecular pathology has already had a great impact onthe diagnosis of haematological and soft tissue neoplasia.The contribution of molecular techniques to the classifica-

tion of carcinomas has not yet been so profound. In the lastfew years, breast cancer has been the epithelial malignancymost studied by molecular techniques; therefore, it is not

surprising that not only predictive markers, but also noveltherapeutic targets are emerging. The focus of this reviewwill be on the impact of molecular profiling analysis andmolecular genetics on our understanding of breast cancerand its precursors.

AN ARRAY OF CHANGES

The boom of high throughput technologies and the class

discovery studies pioneered by the Stanford group4,5 havenot only brought forward the fact that breast cancer is nota single entity but also provided a working model for amolecular taxonomy for breast cancer.48

Expression profiling analyses using microarrays have

demonstrated that breast cancers can be classified accord-ing to their expression patterns into at least five groups:48

luminal A, luminal B, normal breast-like, HER2 and basal-like. The most robust distinction observed by microarrayanalysis is between the transcriptome of ER positive and

ER negative breast cancers. Luminal tumours are describedas those that show expression patterns reminiscent ofnormal luminal epithelial cells of the breast, includingconsistent expression of low molecular weight cytokeratins8/18, ER and genes associated with an active ER path-way.48 At least two subgroups of luminal tumours havebeen identified: luminal A, which are usually of lowhistological grade, have an excellent prognosis and showhigh levels of expression of ER-activated genes; and

luminal B, which are more often of higher histologicalgrade, have higher proliferation rates and a poorerprognosis than luminal A tumours.48 Normal breast-likecancers are rather poorly characterised tumours; one of thedefining features of these lesions is that they consistentlycluster together with samples of fibroadenomas and normalbreast samples. The clinical significance of normal breast-like tumours is yet to be determined48 and some havesuggested that this subgroup may be a mere artefact ofexpression profiling (i.e., disproportionally high content ofstromal cells). HER2 tumours are usually ER negative andcharacterised by over-expression of HER2 and genesassociated with HER2 pathway and/or HER2 amplicon

on 17q12. HER2 cancers have very aggressive clinicalbehaviour; however, they are amenable to novel tailored

Pathology (January 2009) 41(1), pp. 7788

Print ISSN 0031-3025/Online ISSN 1465-3931 # 2009 Royal College of Pathologists of Australasia

DOI: 10.1080/00313020802563536


2/12

therapies using either humanised monoclonal antibodiesagainst HER2 or HER2 tyrosine kinase inhibitors.413

Although the vast majority (480%) of HER2 cancers asdefined by microarrays harbour HER2 gene amplificationor HER2 3 immunohistochemical expression,6,14 not alltumours that are HER2 amplified fall into the HER2

cluster by expression arrays analysis. There is also evidenceto suggest that some HER2 amplified, ER positive cancersfall within the luminal subtypes rather than the HER2microarray subtype.6 Basal-like cancers, another group ofER negative cancers, are so named because the neoplasticcells of this tumour type consistently express genes usuallyexpressed in normal basal/myoepithelial cells of the breast,including high molecular weight cytokeratins (5/6, 14 and

17), P-cadherin, caveolins 1 and 21522 and epidermalgrowth factor receptor (EGFR)17 and, in a minority ofcases, harbour EGFR gene amplification.7 These tumoursare usually of high histological grade and characterised by

high mitotic indices, the presence of central necrotic zones,pushing borders, conspicuous lymphocytic infiltrate and

typical/atypical medullary features.23,24 In addition, meta-plastic elements are not uncommonly found.25 Themorphological and immunohistochemical features ofbasal-like cancers are remarkably similar to those describedfor tumours arising inBRCA1germline mutation carriers21

and there is a growing body of evidence to suggest thatBRCA1 pathway is dysfunctional in sporadic basal-likecancers.26 In fact, engineered mouse models targeting Brca1and Trp53 genes in luminal or basal cells of the mousemammary gland resulted in the development of tumoursdisplaying morphological and immunohistochemical fea-tures that recapitulate those seen in human basal-like breastcarcinomas.27,28

BASAL-LIKE CARCINOMAS: BACK TO THEBASIS

Before discussing the implications of basal-like breastcarcinomas, it is worth mentioning that two misconceptionshave plagued the literature on this topic.

First, there is a pervasive misconception that basal-likecancers were discovered by gene expression profilinganalysis. In fact, the existence of a subgroup of aggressivebreast carcinomas showing features of myoepithelial/basaldifferentiation has been known since the 1960s16,2932 andthat these tumours more often express the immunohisto-chemical features now known to be characteristic of basal-

like cancers has been known since the 1990s.33

However, itwould be fair to say that basal-like breast carcinomas onlygained widespread interest after their rediscovery and

systematic classification by microarray-based expressionprofiling analysis.

Another important misconception is that basal-like andtriple negative (ER, PR and HER2 negative) cancers are

synonymous.34 Although the vast majority of triplenegative cancers are of basal-like phenotype3537 and thevast majority of tumours expressing basal markers are

triple negative,17,38 there is a significant number of triplenegative cancers that do not express basal markers and asmall, but still significant, subgroup of basal-like cancersthat express either hormone receptors or HER2.14,37,3941

Bertucci et al.39 have addressed this issue directly andconfirmed that not all triple negative tumours when

analysed by gene expression profiling were classified asbasal-like cancers (i.e., only 71% were of basal-likephenotype) and not all basal-like breast carcinomasclassified by expression arrays displayed a triple negativephenotype (i.e., 77% were of triple negative phenotype).Taken all together, these results are in accord with the

concept that the triple negative phenotype is not an idealsurrogate marker for basal-like breast cancers37,39,42 andcall for caution in the interpretation of ongoing therapeutictrials whose selection of patients was made on the basis oflack of ER, PR and HER2 expression.39 Furthermore,there are several lines of evidence to suggest that the groupof triple negative cancers is substantially more hetero-geneous than the group encompassed by basal-like breast

cancers.3739

Although the gold standard for diagnosis of basal-likecarcinomas remains microarray-based expression profiling,this technology is unlikely to be rolled out as a diagnostic

tool in most pathology departments. In fact, several groupshave endeavoured to develop surrogate markers for the

intrinsic gene list molecular taxonomy. Real-time PCRbased methods have already been developed and validatedin archival, formalin-fixed, paraffin-embedded tissue sam-ples.43,44 However, for routine management of breastspecimens, immunohistochemistry is likely to become themethod used to identify these subtypes. Although there isno internationally accepted immunohistochemical defini-tion for basal-like breast cancer,45 Nielsen et al. havedemonstrated that, using a panel of four immunohisto-chemical markers, basal-like carcinomas can be detectedwith a sensitivity of 76% and a specificity of 100%.17 Basal-like carcinomas are thus defined as being ER negative,HER2 negative and positive for expression of cytokeratin5/6 and/or EGFR. Importantly, this immunohistochemicalpanel has been shown to be of prognostic significance indifferent cohorts.38,42 Further refinement of this panel bythe same group38 has demonstrated that expression of basalmarkers (i.e., cytokeratin 5/6 and/or EGFR) in ER, PR andHER2 negative cases, identifies a subgroup of triplenegative cancers with a significantly worse outcome thanpatients with triple negative carcinomas that are negativefor basal markers.

Basal-like carcinomas have been shown to display arather aggressive clinical behaviour,15,18,41,46,47 with mostrecurrences happening within the first five years afterdiagnosis.41,47 Late recurrences are reported not to be ascommon. Basal-like cancers appear to respond to neoadju-

vant chemotherapy,14,48,49

however, patients with basal-liketumours, paradoxically, still have a worse outcome whencompared with those with tumours pertaining to other

molecular subgroups.14,15,18,35,46,48 However, there is sub-stantial circumstantial evidence to suggest that patients withbasal-like breast cancer who evolve to pathological com-plete response following neoadjuvant chemotherapy have

an excellent prognosis, whereas those who only display apartial response have remarkably poor outcomes.48,50 Thisunderpins the concept that basal-like cancers are a hetero-

geneous group in terms of their expression profiles,molecular genetic patterns51 and clinical behaviour.41,47

Interestingly, basal-like and tumours arising in BRCA1mutation carriers have recently been shown to preferentiallyexpress markers consistent with a cancer stem cell/cancer progenitor cell phenotype.52,53 However, it should

78 GEYER et al. Pathology (2009),41(1), January


3/12

be emphasised that not all basal-like cancers displayCD44/CD24 cells.52 Interestingly, although breast cancerstem cells are reported to be resistant to chemotherapy andcancer stem cells are more prevalent in basal-like breastcancers, the highest prevalence of pathological completeresponse following neoadjuvant chemotherapy is found in

the subgroup of basal-like cancers.

48,50

Although severaltheories have been advanced to reconcile these contrastingresults, they still remain purely theoretical.

GENE-EXPRESSION PROFILING: PROGNOSTICSIGNATURES

Pathology-derived details have long been used to tailortherapy for cancer patients. Their diligent use hascontributed to the reduction of breast cancer mortalityand morbidity, with a breast pathologist having a crucial

role in the treatment decision making. As an example,sentinel lymph node biopsy has been responsible for great

improvement in breast cancer treatment and its recognisedhigh negative predictive value for axillary nodal status54,55

depends on adequate histopathological examination.56,57

Adjuvant!Online (www.adjuvantonline.com, accessed July2008), Nottingham prognostic index, St Gallen guidelines,and National Institutes of Health (NIH) criteria,58,59 toolsoften used by oncologists to decide which patients shouldreceive chemotherapy, are based on tumour size, histolo-gical grade, ER status, vascular invasion and presence oflymph node metastasis. However, oncologists, pathologistsand surgeons would not disagree that the parameterscurrently available are not sufficient to capture thecomplexity of breast cancer and to tailor therapy forindividual patients. Therefore, several research groups havecarried out comprehensive microarray gene-expressionprofiling studies with the aim of improving on traditionalclinico-pathological parameters.

The first breast cancer prognostic signature described60

was developed through a class prediction analysis of aseries of 78 young patients with small (55 cm), lymph nodenegative cancers who did not receive adjuvant chemother-apy. By comparing the expression profiles of tumours frompatients who developed distant metastasis within 5 yearswith those who did not, the authors identified a multi-geneprognostic predictive score comprising 70 genes (Mamma-Print; Agendia, The Netherlands). Later the signature wasvalidated in larger cohorts with node positive and node

negative patients.61,62

A different approach has beenundertaken by Wang et al.63 who first identified 16 and60 genes associated with relapse in ER positive and ER

negative breast carcinomas, respectively, and assembledthem into a 76-gene prognostic signature (VDX2 genechips; Veridex LLC, USA). In the studies above, statisticalanalyses have arguably shown that the predictive power of

these signatures was significantly superior to the standardclinico-pathological parameters. However, subsequent sta-tistical analyses have called into question the actual

contribution of these signatures for breast cancer patientmanagement.6466

Following the attention these two prognostic signaturesattracted, several other prognostic signatures have been orare currently being developed. Fan et al.67 applied multiplesignatures (i.e., intrinsic subtypes, 70-gene signature,

wound response signature and 21-gene recurrence score)to a cohort of breast cancer patients and confirmed theirprognostic value.67 On the other hand, others have failed tofind a superior predictive value for the 70-gene signaturewhen compared with that of regular clinico-pathologicalfeatures.14,64,65 Although all of these signatures are based

on the expression of genes related to similar pathways andto some extent correlate with tumour proliferation,67,68

many were surprised by the fact that the overlap in terms ofgenes belonging to these signatures is negligible. As aconsequence, both pathologists and clinicians face thechallenge of which signature should be used. Disappoint-ingly, combining different signatures does not seem toresult in a significant improvement of accuracy.67

To complicate matters even further, many doubtsregarding the reliability and reproducibility of the techni-que have arisen.7,69,70 One of the reasons for this apparentfailure of microarrays in realising their potential stems from

the disparate paces of microarray technology developmentand the development of bioinformatics and statistics

applied to microarray analysis. For instance, in thebeginning of this century, data over-fitting was a poorlyunderstood concept and methods for power calculation formicroarray analysis were yet to be described.71 Fortunately,this field is maturing very rapidly and currently there areclear guidelines as to how a therapeutically significant genesignature should be developed and validated.72,73 Unfortu-nately, none of the microarray-based signatures describedto date fulfils all criteria required.

Although we believe that microarray-based technology,or most likely one of its derivatives, is likely to beincorporated in breast pathology practice, we argue thatthere are practical issues that need to be resolved beforegene expression profiling can be translated to routineclinical use. First of all, there are data to suggest that theprognostic power of the gene signatures reported to date israther limited in ER negative, high grade disease.68 Inaddition, although microarrays are quite reproducible andcan certainly be applied to class discovery studies andpreclinical analysis, their accuracy and reproducibility arenot sufficient for clinical use. For instance, up to 15% oferror in qualitative assessment of gene expression has beendemonstrated when the optimal samples were analysed withthe same platform and protocols in different laboratories.74

Furthermore, the applicability of microarrays to readilyavailable, formalin-fixed, paraffin-embedded material islimited, as this technology has been shown to perform

suboptimally when RNA extracted from this type of sampleis used. Although Illumina (USA; http://www.illumina.com/, accessed July 2008) has recently provided a method

for microarray-based gene expression analysis of formalin-fixed, paraffin-embedded samples (DASL gene expression),which has been shown to produce reasonable results whenapplied to matched frozen and formalin-fixed, paraffin-

embedded specimens,75 neither the 70-gene nor the 76-genesignature has been converted into this platform.

As stated above, we anticipate that one of the derivatives

of microarray analysis, rather than microarrays themselves,is likely to be incorporated into clinical practice. In fact,quantitative reverse transcriptase PCR (qRT-PCR)-basedsignatures and immunohistochemical panels have alreadybeen developed as alternative methods for expressionprofiling analysis. The prototype of this approach is

MOLECULAR ANALYSIS IN BREAST CANCER 79
http://www.adjuvantonline.com/http://www.illumina.com/http://www.illumina.com/http://www.illumina.com/http://www.illumina.com/http://www.adjuvantonline.com/


4/12

Oncotype DX (Genomic Health, USA),76 which is based onthe mRNA expression levels of 21 genes (16 cancer relatedgenes and five reference genes). These 16 genes comprisecomponents of ER pathway (ER, PGR, BCL2 and

SCUBE2), proliferation (Ki67, STK15, Survivin, CCNB1and MYBL2), HER2 amplicon (HER2 and GRB7),

invasion (MMP11 and CTSL2) and GSTM1, CD68 andBAG1. The analysis of the gene expression by qRT-PCRgives rise to the recurrence score, which has been shown tobe an independent prognostic factor in early stage,endocrine responsive, tamoxifen treated patients. Patients

with a high recurrence score are thought to benefit fromchemotherapy. Intermediate and low scores are assembledunder the good prognosis group and chemotherapy would

be of limited value. This assay has proven suitable for theanalysis of formalin-fixed paraffin-embedded samples,however the recurrence score was not validated in twostudies.77,78 Furthermore, some have argued that it only

recapitulates the information provided by grade, ER,HER2 and MIB179 and it remains to be determined if this

signature will be of any prognostic value for patientsreceiving aromatase inhibitors or anthracyline- and taxane-based therapeutic regimens.80

GRADING OF BREAST CANCER: THE END OFNOTTINGHAM GRADE?

Molecular studies have confirmed the importance ofhistological grade in breast carcinomas and this shouldnot be surprising, given that morphological features visiblein H&E sections such as nuclear atypia, tubule formationand mitosis represent the expression of thousands of genesand the interactions of tens of thousands of proteins.81 Ithas been demonstrated that grade, more than any otherclinico-pathological parameter or tumour intrinsic char-acteristic, is associated with the type, pattern and complex-ity of molecular changes seen in breast cancer and itsprecursors.1,2 Ma et al. analysed matched normal, pre-invasive and invasive breast lesions of different grades bymeans of expression profiling analysis and observed thatsamples preferentially clustered according to histologicalgrade rather than stage.82

Currently, histological grading of breast carcinomas isone of the cornerstones of treatment decision making. It isperformed by histopathological analysis using themodified Bloom and Richardson method,83 also known

as Nottingham grade. Despite its importance, concor-dance between pathologists has been reported to besuboptimal.84 In fact intra- and inter-observer agreement

ranges from 50% to 86%. The clinical significance of gradeI and grade III are well defined, but only about half ofbreast carcinomas are classified as such. Often oncologistshave to care for a patient with a grade II invasive ductal

carcinoma of no special type, which is not helpful in clinicaldecision making. This has prompted several groups toinvestigate methods that could refine the current grading

system, making it more reproducible and assigning highand low risk subgroups.

Sotiriou et al. undertook a hypothesis-driven approachfocusing on histological grade.85 They analysed grade Iand grade III ER positive breast carcinomas usingexpression arrays and proposed the gene expression

grade index (GGI). GGI is based on 97 genes, most ofthem are associated to cell cycle progression andproliferation. Indeed, proliferation has been reported tobe the most important predictor of outcome in breastcancer, especially in ER positive patients.86 GGI couldidentify grade I and grade III tumours in the validation

set with an accuracy of *

90% and was more stronglyassociated with relapse-free survival than histologicalgrade. Importantly, it was shown that GGI could stratifygrade II tumours into genomic low grade and genomichigh grade and that these groups were of prognosticsignificance: histological grade II GGI low tumours hadoutcomes similar to GGI low cancers, whereas histolo-gical grade II GGI high tumours had outcomes similar to

GGI high cancers. GGI has been subsequently validatedby the same group.87

Another genetic grade signature was proposed by Ivshina

et al.88 who profiled 347 breast invasive tumours with

expression arrays. Using two class prediction algorithms,six grade-associated genes were identified and accurately

classified grade I and grade III tumours in low and highgenomic grade, respectively. Grade II tumours were not amolecularly distinct group but could be separated intograde IIa (low grade-like) or grade IIb (high grade-like).Grade IIa patients had a significantly better outcome thangrade IIb patients, whereas no difference was observedbetween the grade IIa and grade I survival curves, or thegrade IIb and grade III curves. These results have beenfurther validated by the authors in datasets generated withdifferent microarray platforms.88 However, further analyseshave shown that there were significant molecularand clinico-pathological differences between the grade Iand grade IIa tumours and the grade IIb and grade IIItumours.

More recently, using a similar approach, Ma et al.reported a five-gene molecular grade index.89 In a way akinto the signatures described above, this molecular gradeindex could classify grade I and grade III tumours with89% accuracy and grade II tumours were stratified into twoclinically significant groups. The results were validated in aqRT-PCR assay suitable for formalin-fixed, paraffin-embedded clinical samples. Therefore, the authors arguedthat this signature could be incorporated in the dailypractice, as it removes the subjectivity and inter-/intra-observer variability associated with conventional histologi-cal grading. It has also been proposed that gradingsignatures could replace histological grade in currently

used prognostic models such as Adjuvant!Online andNottingham prognostic index.68,86,88 Ivshina et al. havedescribed that incorporating their genetic grade signature

into the Nottingham prognostic index, some patientsclassified as having worse prognosis shifted to the goodprognosis group. Hence, the use of genetic grade may helpimprove the identification of patients who could be spared

toxic adjuvant systemic therapy.88 Although there is a greatenthusiasm about genomic grading, additional independentvalidation of the above genomic grade signatures is still

required and their discriminatory power in ER negativetumours remains to be determined. Therefore, despite theissues related to reproducibility of Nottingham histologicalgrade, histological grade still remains a useful and lessexpensive tool to help tailor the therapy of breast cancerpatients.



5/12

PREDICTIVE SIGNATURES AND BIOMARKERS

Hormone receptors and HER2 are the only mandatorybiomarkers to be tested for breast cancer patient decisionmaking, and tamoxifen and trastuzumab represent two ofthe few successful examples of targeted therapy in breast

cancer. In recent years, a paradigm shift has occurred in theway mechanisms of response or resistance to therapy andthe predictive value of a biomarker are perceived. There aredata to suggest that the significance of predictive markersmay depend on the molecular subtype of the tumour.90,91

Biological pathways might be important in one particularmolecular subtype, such as TP53 mutation in ER positivechemotherapy resistant tumours;92 whereas the impact of

TP53 mutation in ER negative disease may differ. There-fore, studies designed to discover new predictive markers of

response to systemic therapy should take into account thedistinctive molecular features of the different subtypes,otherwise they might under- or over-estimate their perfor-mance (Fig. 1).86,90,91

Several groups have tried to identify reliable predictivefactors in the context of tamoxifen treated patients.93,94

Tamoxifen is the most frequently prescribed anti-oestrogen

agent in women with ER positive breast carcinomas in both

FIG. 1 Schematic illustration of data integration for prognostic/predictive markers and therapeutic targets identification based on high-throughput molecular

methods. The term oncogene addiction has been used to describe the phenomenon in which tumour cells become reliant upon the activity of specific oncogenes

for growth and progression.163 Consequently, identifying and inhibiting the activity or expression of the proteins encoded by these addictive oncogenes is a

highly attractive therapeutic strategy in human cancer.164 (A) Representative genome plot of DNA copy number and chromosome position of a HER2

amplified breast carcinoma. Log2 ratios are plotted on the Y axis against each clone according to genomic location on the X axis. The centromere is

represented by a vertical dotted line. BACs categorised as displaying genomic gains or amplification are plotted in green and those categorised as genomic

losses in red. (B) Representative image of a scanned microarray chip. Red, relative high expression; green, relative low expression. (C) Representative

micrograph of an invasive ductal carcinoma harbouring HER2 amplification by CISH. (D) Representative micrograph of an invasive ductal carcinoma

displaying HER2 over-expression by IHC.ArrayCGH, microarray comparative genomic hybridisation; FISH, fluorescent in situ hybridisation; CISH, chromogenic in situ hybridisation; qRT-PCR,

quantitative reverse transcriptase PCR; IHC, immunohistochemistry; siRNA, small interfering RNA; shRNA, short hairpin RNA.

early stage and advanced disease; however, a significantproportion of patients relapse due to intrinsic or acquiredresistance. Maet al. analysed the gene expression profile of60 early-stage tamoxifen-treated patients and identifiedthree genes associated with outcome: HOXB13, IL17BRand CHDH.94 High levels of HOXB13 mRNA and low

levels ofIL17BRmRNA are associated with recurrence. Asimple two-gene ratioHOXB13:IL17BRwas then proposedas a novel biomarker to predict recurrence in tamoxifentreated patients. This group developed a qRT-PCR assayand achieved comparable results. Although initial valida-tion studies failed to demonstrate the predictive power ofthis two-gene signature,95 subsequent studies validated theresults in two different cohorts and showed the ratio has

prognostic power and is predictive of tamoxifen re-sponse.96,97 In both cohorts the ratio was not significantin node-positive patients,96,97 explaining its lack of pre-dictive value in the study by Reid et al.95 Moreover, it has

also been shown that the ratio has complementaryprognostic value when combined with the five-gene

molecular grade index described above.89HER2 over-expression and HER2 gene amplification

are both predictors of response to humanised mono-clonal antibodies anti-HER2 or to HER2 tyrosine kinase



6/12

inhibitors.413 However, there is strong evidence to suggestthat current methods to determine which patients shouldreceive trastuzumab or HER2 tyrosine kinase inhibitorshave good negative predictive value, but suboptimalpositive predictive values. In the last years, several mechan-isms ofde novoand acquired resistance to trastuzumab have

been identified. It is currently accepted that PIK3CAactivating mutations, PTEN inactivation, IGF1R over-expression or expression of p95 HER2 isoform may all playroles in bothde novoor acquired resistance to therapies thattarget HER2.3,98 Furthermore, it has recently beensuggested thatHER2amplified breast cancers that harboura basal-like transcriptome are less sensitive to anti-HER2therapies.99 Therefore, one can anticipate that additional

molecular testing of HER2 positive cancers is likely to beincorporated as predictive markers for patients withtumours pertaining to this molecular subgroup.

In addition,HER2 has been reported to predict response

to anthracycline-based chemotherapy.100,101 AlthoughHER2 is considered to be the target gene for amplification

at chromosome 17q12-q21, the HER2 amplicon encom-passes several other genes, such as TOPO2A. TOPO2Agene encodes for topoisomerase-IIa, the direct moleculartarget of anthracycline,102 it is often co-amplified withHER2and its amplification seems to be restricted to HER2amplified tumours.103106 In this context TOPO2A hasemerged as a new putative predictor biomarker of response,as it has been shown that tumours with HER2/TOPO2Aco-amplification show a significantly better response toanthracycline-based chemotherapy.104,107,108

TUMOUR MICROENVIRONMENT

AND STROMAL SIGNATURESThe major focus of cancer research has been on themalignant cell itself. Most of the analyses described abovewere performed using whole tissue samples and thosejudged to possess insufficient tumour epithelial cell contentwere generally excluded. Thus, the specific contribution ofepithelial cells, stromal cells and extra-cellular matrix tothese tumour classifiers and prognostic/predictive signa-tures remains uncertain. However, some groups havefocused their studies on the tumour microenvironment inbreast cancer.109116

Comprehensive gene expression profile of each cell typehas shown that changes at the transcriptional level occur in

epithelial, myoepithelial and stromal cells, already evidentat the in situcarcinoma stage.109 However, genetic changesdetected by microarray-based comparative genomic hybri-

disation appear to be limited to cancer epithelial cells.109

The studies of epithelial-mesenchymal interactions havebrought some interesting insights about cancer progres-sion.112,114,116 As different groups have failed to detect

significant differences in genetic alterations between ductalcarcinoma in situ (DCIS) and invasive tumours,82 theprogression ofin situcarcinoma to invasive carcinoma may

not be due to the intrinsic properties of the tumourepithelial cells but determined by complex interactionsbetween epithelial cells and all the cell types that composethe tumour microenvironment.112,116 It has been hypothe-sised that progression to invasion may be promoted byfibroblasts and inhibited by myoepithelial cells.116

Based on the above concepts, it is perhaps not surprisingthat prognostic signatures derived from the profiles ofstromal cells have been developed. Using gene signatures oftwo fibroblastic tumours, desmoid-type fibromatosis (DTF)and solitary fibrous tumour (SFT), West et al.

115 couldstratify a breast carcinoma dataset in at least two

subgroups with prognostic value independent of clinico-pathological risk factors, suggesting that there might bedistinct fibroblastic reaction patterns in breast cancer. Thefirst cluster expressed DTF genes and was associated with agood prognosis, whereas the second was less homogeneousand composed of two subgroups enriched for SFT genes.These results have been confirmed in different datasets and,at the protein level, breast cancers with strong expression of

one single gene (SPARC) of the DTF cluster have shown atrend for increased survival.110 However, it should be notedthat a previous study demonstrated that SPARC expressionin a large consecutive series of breast cancers was

associated with poor prognosis.117 One could hypothesisethat the DTF stromal reaction pattern would be similar to

that seen in the desmoplastic stromal pattern typicallyfound in tubular carcinomas; however, no correlationbetween molecular and morphological features has beendescribed as yet.

Using a different approach, Bergamaschi et al.111 grosslydissected 28 breast cancers, with some retention of extra-cellular matrix, and carried out an unsupervised hierarch-ical clustering analysis. This analysis revealed that tumourswere classified into four different groups according to theexpression of 278 selected extra-cellular matrix-relatedgenes. Interestingly, in the validation set, the authors founda strong association between one of the groups (Group 1)and high grade, ER negativity, TP53 mutations and pooroutcome. When the five previously defined molecularsubtypes4,5 were compared with the stromal signature, itwas observed that 81% of all basal-like, 4% of luminal Aand 23% of luminal B tumours fell into Group 1. Theluminal A and B tumours falling into this group displayed asignificantly worse outcome than other luminal tumours,indicating that the extra-cellular matrix signature can giveadditional information and is partly independent of theintrinsic subgroups. Furthermore, the authors identifiedthree genes out of the 278 previously selected that wereassociated with prognosis. Tumours harbouring highexpression levels of MARCO and low levels of PUNCand SPARCdisplayed a significantly worse outcome.

More recently, Finak et al.113 have isolated tumour

stroma and matched normal stroma from breast tumoursand derived a 26-gene signature strongly associated withoutcome called stroma-derived prognosis predictor

(SDPP). SDPP predicted prognosis with greater accuracythan the 70-gene signature (MammaPrint), could alsostratify several published whole-tumour derived datasetsand was independent of grade, age, lymph node involve-

ment, chemotherapy, hormonal therapy and both ER andHER2 status. The poor outcome cluster showed enrich-ment for markers of hypoxic and angiogenic response,whereas the good outcome cluster over-expressed immuneresponse genes.

Interestingly, an immune response gene expressionsignature has been shown to identify a good outcomesubtype of ER negative patients.118 Surprisingly, thissubtype was not related either to the extent of lymphocytic



7/12

infiltrate, medullary carcinomas or BRCAmutation status.The immune system seems to play an important role in asubgroup of ER negative patients who may benefit fromtreatments targeting tumour cells via the immune response,such as vaccine therapies.113 Interestingly, a small percen-tage of basal-like carcinomas, a group still without reliable

prognostic markers or signatures that reliably identify thosewith good or poor prognosis, was included in thissubtype.118

MOLECULAR GENETICS OF BREAST CANCER:FROM GENOMIC ARCHITECTURE TO BREASTCANCER GENETIC PATHWAYS

The concept that breast cancer encompasses a plethora ofentities with distinctive biological characteristics andclinical behaviour is also underpinned at the moleculargenetic level by a complex array of genetic alterations that

affect the function and control of individual genes andcellular processes.1,2 Not only expression profiling analysis,

but also the study of cancer genetics has had a profoundimpact on our understanding of the evolutionary pathwaysand causative factors in the initiation, development andprogression of breast cancer.

Microarray-based comparative genomic hybridisation(aCGH) enables researchers to perform rapid and highresolution screening of genomes and a detailed analysis ofthe global copy number changes.119 While expressionprofiling analysis looks at the transcriptional level, aCGH

scrutinises the genome-wide pattern of genetic alterations.Reliable results using DNA extracted from formalin-fixed,paraffin-embedded samples can be achieved, in particular ifBAC arrays are used. aCGH studies in breast cancer haveshown that copy number changes are associatedwith different clinico-pathological features to the gene-

expression subtypes previously described

4,5

and to differentoutcomes.120,121

Three patterns of genomic changes have been recentlydescribed, probably related to different mechanisms ofDNA repair defects found in breast cancers (Fig. 2).121 Thefirst pattern is called simplex and is defined by broadsegments of gains and losses, usually comprising entirechromosomes or chromosome arms. Tumours with this

pattern often display gains of 1q and losses of the whole16q arm. This pattern accounts for 60% of diploid tumoursand is related to low histological grade and luminal Aphenotype.122 The remaining tumours fall into a category

called complex. Tumours with a complex genomicarchitecture have been shown to have shorter survival in

retrospective analysis and to comprise at least two distinctpatterns: sawtooth (25%) and firestorm or amplifier(5%).121,122 The sawtooth pattern is characterised bymany segments of gains and losses of varying sizes, oftenalternating; copy number changes affect almost the entiregenome, however loci with high level amplification areusually not seen. The firestorm pattern resembles thesimplex, but shows at least one localised region of clustered,narrow peaks of amplification, with each cluster confined to

FIG. 2 Schematic illustration of the genomic patterns in breast cancer, as described by Hickset al.,121 and their association with molecular characteristics,

histological features and clinical variables. At the top representative genome plots of DNA copy number and chromosome position of the three distinct

genomic profiles. Log2 ratios are plotted on the Y axis against each clone according to genomic location on the X axis. The centromere is represented by a

vertical dotted line. BACs categorised as displaying genomic gains or amplification are plotted in green and those categorised as genomic losses in red.*Chinet al.122 **Basal-like carcinomas do not display as many high level amplifications as HER2 and luminal B carcinomas and they are predominantly of

sawtooth pattern; HER2 carcinomas are typically of complex-amplifier/firestorm pattern.



8/12

a single chromosome arm. Interestingly, complex patterns,in particular the firestorm pattern, were associated with aworse outcome.

Historically, breast cancer progression pathways wereseen as a multistep model, similar to that described byVogelstein et al.123 for colorectal cancer, where normal

breast epithelial cells would transform and progress througha series of morphologically identifiable precursors (i.e.,normal hyperplasia of usual type atypical ductalhyperplasia low grade DCIS high grade DCIS invasive carcinoma).2 However, in the last few years it hasbecome clear that breast cancer molecular pathways aremuch more complex than previously appreciated.1,2,124127

In fact, breast cancer molecular pathways encompass a

network of inter-related multi-step pathways, which can bebroadly classified into two groups/arms, based on histolo-gical grade.1,2 The low grade arm encompasses most of theprecursors pathologists are aware of, including columnar

cell changes, atypical ductal hyperplasia, lobular neoplasia(ALH/LCIS), and their invasive counterparts. These lesions

are characterised by low nuclear grade, consistent positivityfor hormone receptors and lack of HER2 expression, andquite simple diploid/near diploid karyotypes, with fewrecurrent changes: deletions of 16q and gains of 1q and16p.1,2 On the other hand, the high grade arm encompasseshigh grade DCIS and invasive high grade invasive ductalcarcinoma. These tumours frequently lack hormone recep-tors and express HER2 or basal markers. At the geneticlevel, tumours of the high grade arm are aneuploid, havecomplex karyotypes, with a plethora of unbalanced changesmapping to several chromosomal arms.1,2 Interestingly,510% of high grade breast cancers harbour deletions of16q and even when they do, the mechanisms leading to 16qdeletions are different from those described in low gradebreast cancer.128 Taken together, these lines of evidencesupport the idea that progression from low to high gradebreast cancer is an exceedingly rare biological phenomenon.This is further supported by the fact that recurrences ofDCIS have similar histological grade and molecular geneticfeatures when compared with those of index cases.129

Another line of evidence stems from expression arrayanalysis of matched atypical ductal hyperplasia/flat epithe-lial atypia, DCIS and invasive breast carcinomas.1,2 Asmentioned above, Maet al. have demonstrated that lesionsof similar histological grade rather than progression stagecluster together, suggesting that different histological gradesare associated with distinct gene expression signatures. Last

but not least, this is further corroborated by a recentobservation that a mixture of grade I, grade II and grade IIIDCIS is seen in only 9.2% of cases.130

Results of molecular genetic studies, on the other hand,have blurred the boundaries between low grade ductalcarcinomas and lobular neoplasia.1,2 Both show remark-ably similar immunohistochemical and molecular genetic

profiles, the main difference being the target gene of 16qlosses.1,2,131 Whilst in ductal lesions the target gene remainsto be identified,132134 in lobular carcinomas it has been

proven to be the CDH1 gene.135,136 This gene encodesE-cadherin, an adhesion molecule that mediates homo-phylic-homotypic adhesions. This gene is reported to beinactivated in lobular neoplasia and invasive lobularcarcinomas through a combination of genetic and epige-netic mechanisms.137,138 Moreover, identical CDH1 gene

mutations have been described in matched lobular neopla-sia and invasive lobular cancer,136 providing strongevidence to suggest that at least some examples of lobularneoplasia are non-obligate precursors of invasive lobularcarcinoma. Given the consistent down-regulation of E-cadherin in lobular lesions, anti-E-cadherin antibodies have

proven to be useful in the diagnosis of lesions withindeterminate phenotype and in the characterisation ofthe pleomorphic variant of lobular carcinoma.139,140 Ourgroup and others have demonstrated that pleomorphicLCIS and invasive pleomorphic lobular carcinoma arevariants of classic lobular cancer, harbouring similarmolecular genetic changes, including deletions of 16q141144 and E-cadherin down-regulation.141144 However, pleo-

morphic LCIS (PLCIS) and invasive pleomorphic lobularcarcinomas (iPLC) harbour additional genetic hits mappingto key oncogenes, such as MYC and HER2, which mayaccount for the high nuclear grade and the more aggressive

clinical behaviour reported.143,144 Given that PLCIS andiPLC are genetically advanced lesions, which are remark-

ably similar at the morphological, immunohistochemicaland molecular genetic levels, PLCIS should be considered adirect non-obligate precursor of iPLC143,144 and managedless conservatively than classic lobular neoplasia.

Molecular genetic analysis has also helped clarify the roleof some putative breast cancer precursors. Some lesionsonce considered part of breast cancer progression pathways(e.g., hyperplasia of usual type) have been shown toharbour rare and fairly random chromosomal changesand their role as precursors have been called intoquestion.1,2,145147 On the other hand, other lesions suchas columnar cell changes/flat epithelial atypia,148 haveproven to be clonal and neoplastic, and to harbourmolecular genetic changes similar to those found in lowgrade ductal carcinomas, including 16q deletions. Inaddition, apocrine lesions have also been shown to harbourrecurrent unbalanced chromosomal changes;149 however,their actual role in breast cancer molecular pathways is yetto be determined.

Interestingly, a recent morphological study of low gradebreast lesions has given further support for the concept of alow grade breast neoplasia family, which encompassescolumnar cell changes/flat epithelial atypia, atypical ductalhyperplasia, low grade DCIS, lobular neoplasia andtheir respective invasive counterparts: tubular/tubular-cribriform breast carcinomas, grade I invasive ductalcarcinomas, invasive lobular carcinomas and tubulo-

lobular carcinomas.150

These tumours seem to have atypical luminal A phenotype and, as mentioned above, arecharacterised by positivity for hormone receptors, deletions

of 16q and have a rather indolent clinical course.2,7,15,46,148

Based on these lines of evidence it seems reasonable toconsider flat epithelial atypia, low grade ductal and lobularlesions as part of the same group of lesions. Furthermore, a

recent expression profiling analysis of special types ofbreast cancer has demonstrated the existence of a groupof ER positive cancers encompassing classic lobular

carcinomas and tubular carcinomas.131 On the otherhand, there are few similarities between low grade ERpositive and high grade ER negative ductal carcinomas interms of their morphological, immunohistochemical andmolecular features, suggesting that they constitute distinctentities.



9/12

BREAST CANCER SPECIAL TYPES:GENOTYPIC-PHENOTYPIC CORRELATION

Although histological grade has been extensively analysed

from a molecular perspective, histological type has notreceived the same attention. It should be noted, however,that an obvious example of genotypic-phenotypic correla-

tion in breast cancer is the secretory carcinoma of thebreast. This tumour, which has an indolent clinicalbehaviour and not uncommonly affects prepubertal pa-

tients, has been shown to consistently harbour a t(13;15)chromosomal translocation,151 involving the genes ETV6and NTRK3.151154 Subsequent studies have shown thatthis translocation in breast cancer is specific to secretorybreast carcinomas and that even tumours once thought tobe variants of this special type (i.e., acinic cell carcinomasof the breast) do not harbour this translocation.155

As mentioned above, there are several lines of evidence tosuggest a genotypic-phenotypic correlation betweenBRCA1 germline mutations and specific morphological

features of breast cancer. When compared with sporadiccontrols, BRCA1 tumours have been shown to be enrichedfor medullary cancers, to be of histological grade III, havehigher mitotic counts, pushing borders and brisk lympho-cytic infiltrate.156158 At the immunohistochemical level,BRCA1 cancers have been shown to more often lackhormone receptors and HER2 when compared withcontrols157159 and to preferentially be of basal-likephenotype.158,160,161 The link between BRCA1 mutationsand the phenotypic characteristics listed above is furtherconfirmed by two engineered mouse models developed bytwo independent groups, where Brca1 and Trp53 wereinactivated in the epithelial cells of the mouse mammarygland.27,28,158 Tumours developing in these animals were

characterised by the same constellation of morphologicalfeatures and immunohistochemical profile as described for

human BRCA1 tumours.A recent detailed expression profiling analysis of special

types of breast cancer demonstrated that some of thehistological types of breast cancer have a distinctive

transcriptome.131 One of the best examples of thisphenomenon is invasive micropapillary carcinoma, whichdisplayed a luminal phenotype, but consistently formed a

distinct cluster in hierarchical clustering analysis.131 Theseresults are in accord with our findings that demonstratethat micropapillary carcinomas of the breast have animmunohistochemical phenotype and a pattern of genetic

aberrations that are consistent with a luminal B pheno-type.162 On the other hand, Weigelt et al.131 demonstratedthat other special types of breast cancer, such as apocrinecarcinoma, may not constitute a distinct entity. From apathologists standpoint, the study of the molecularfeatures of special types of breast cancer may reveal notonly the basis for histological type, but also help identifynovel therapeutic targets for these tumours, given that eachsubtype has been proven to be more homogeneous at themolecular level than invasive ductal carcinomas of nospecial type.131

CONCLUSION

In conclusion, high throughput molecular techniques arereshaping the way breast cancer is perceived. In this era of

tailored therapies, a paradigm shift from morphologicaland prognostic classification systems to predictive models isof paramount importance. It is anticipated that bycombining morphological, immunohistochemial and mole-cular techniques, a more biologically and clinically mean-ingful classification of breast cancer and its precursors will

emerge.ACKNOWLEDGMENTS Jorge S. Reis-Filho and Felipe C.Geyer are funded by Breakthrough Breast Cancer. Caterina

Marchio` is part of the PhD programme Tecniche avanzate dilocalizzazione dei tumori umani (University of Turin) and isfunded in part by AIRC (Milam, Regional Grant 1182) andBreakthrough Breast Cancer.

The authors would like to thank the Microarray Laboratoryof the Breakthrough Breast Cancer Research Centre for thecourtesy of the microarray gene expression image.

Address for correspondence: Dr J. S. Reis-Filho, The Breakthrough Breast

Cancer Research Centre, Institute of Cancer Research, 237 Fulham Road,

London, SW3 6JB, United Kingdom. E-mail: [email protected]

References1. Reis-Filho JS, Simpson PT, Gale T, Lakhani SR. The molecular

genetics of breast cancer: the contribution of comparative genomichybridization. Pathol Res Pract 2005; 201: 71325.

2. Simpson PT, Reis-Filho JS, Gale T, Lakhani SR. Molecular evolutionof breast cancer. J Pathol2005; 205: 24854.

3. Nahta R, Esteva FJ. Trastuzumab: triumphs and tribulations.Oncogene 2007; 26: 363743.

4. Perou CM, Sorlie T, Eisen MB,et al. Molecular portraits of humanbreast tumours. Nature 2000; 406: 74752.

5. Sorlie T, Perou CM, Tibshirani R, et al. Gene expression patterns ofbreast carcinomas distinguish tumor subclasses with clinicalimplications. Proc Natl Acad Sci USA 2001; 98: 1086974.

6. Brenton JD, Carey LA, Ahmed AA, Caldas C. Molecularclassification and molecular forecasting of breast cancer: ready for

clinical application? J Clin Oncol2005; 23: 735060.7. Reis-Filho JS, Westbury C, Pierga JY. The impact of expression

profiling on prognostic and predictive testing in breast cancer.J ClinPathol2006; 59: 22531.

8. Sorlie T, Tibshirani R, Parker J, et al. Repeated observation of breasttumor subtypes in independent gene expression data sets. Proc NatlAcad Sci USA 2003; 100: 841823.

9. Geyer CE, Forster J, Lindquist D,et al. Lapatinib plus capecitabinefor HER2-positive advanced breast cancer. N Engl J Med2006; 355:273343.

10. Joensuu H, Kellokumpu-Lehtinen PL, Bono P, et al. Adjuvantdocetaxel or vinorelbine with or without trastuzumab for breastcancer.N Engl J Med2006; 354: 80920.

11. Piccart-Gebhart MJ, Procter M, Leyland-Jones B,et al. Trastuzumabafter adjuvant chemotherapy in HER2-positive breast cancer. N EnglJ Med2005; 353: 165972.

12. Romond EH, Perez EA, Bryant J,et al. Trastuzumab plus adjuvantchemotherapy for operable HER2-positive breast cancer. N Engl JMed2005; 353: 167384.

13. Slamon DJ, Leyland-Jones B, Shak S, et al. Use of chemotherapyplus a monoclonal antibody against HER2 for metastaticbreast cancer that overexpresses HER2. N Engl J Med 2001; 344:78392.

14. Rouzier R, Perou CM, Symmans WF,et al. Breast cancer molecularsubtypes respond differently to preoperative chemotherapy. ClinCancer Res 2005; 11: 567885.

15. Abd El-Rehim DM, Pinder SE, Paish CE,et al. Expression of luminaland basal cytokeratins in human breast carcinoma. J Pathol 2004;203: 66171.

16. Gusterson BA, Ross DT, Heath VJ, Stein T. Basal cytokeratins andtheir relationship to the cellular origin and functional classification ofbreast cancer. Breast Cancer Res 2005; 7: 1438.

17. Nielsen TO, Hsu FD, Jensen K, et al. Immunohistochemical andclinical characterization of the basal-like subtype of invasive breastcarcinoma. Clin Cancer Res 2004; 10: 536774.

18. Rakha EA, Putti TC, Abd El-Rehim DM,et al. Morphological andimmunophenotypic analysis of breast carcinomas with basal andmyoepithelial differentiation. J Pathol2006; 208: 495506.



10/12

19. SavageK, LambrosMB, Robertson D, etal. Caveolin1 is overexpressedand amplified in a subset of basal-like and metaplastic breastcarcinomas: a morphologic, ultrastructural, immunohistochemical,and in situ hybridization analysis. Clin Cancer Res 2007; 13: 90101.

20. Savage K, Leung S, Todd SK, et al. Distribution and significance ofcaveolin 2 expression in normal breast and invasive breast cancer: animmunofluorescence and immunohistochemical analysis. BreastCancer Res Treat 2008; 110: 24556.

21. Turner NC, Reis-Filho JS. Basal-like breast cancer and the BRCA1phenotype. Oncogene 2006; 25: 584653.

22. van de Rijn M, Perou CM, Tibshirani R, et al. Expression ofcytokeratins 17 and 5 identifies a group of breast carcinomas withpoor clinical outcome. Am J Pathol2002; 161: 19916.

23. Fulford LG, Easton DF, Reis-Filho JS,et al. Specific morphologicalfeatures predictive for the basal phenotype in grade 3 invasive ductalcarcinoma of breast. Histopathology 2006; 49: 2234.

24. Livasy CA, Karaca G, Nanda R, et al. Phenotypic evaluation of thebasal-like subtype of invasive breast carcinoma.Mod Pathol2006; 19:26471.

25. Reis-Filho JS, Milanezi F, Steele D, et al . Metaplastic breastcarcinomas are basal-like tumours. Histopathology 2006; 49: 1021.

26. Turner NC, Reis-Filho JS, Russell AM,et al. BRCA1 dysfunction insporadic basal-like breast cancer. Oncogene 2007; 26: 212632.

27. Liu X, Holstege H, van der Gulden H,et al. Somatic loss of BRCA1and p53 in mice induces mammary tumors with features of humanBRCA1-mutated basal-like breast cancer. Proc Natl Acad Sci USA

2007; 104: 121116.28. McCarthy A, Savage K, Gabriel A,et al. A mouse model of basal-like

breast carcinoma with metaplastic elements. J Pathol2007; 211: 38998.29. Dairkee SH, Mayall BH, Smith HS, Hackett AJ. Monoclonal marker

that predicts early recurrence of breast cancer. Lancet 1987; 1: 514.30. Gould VE, Koukoulis GK, Jansson DS,et al. Coexpression patterns

of vimentin and glial filament protein with cytokeratins in the normal,hyperplastic, and neoplastic breast. Am J Pathol1990; 137: 114355.

31. Malzahn K, Mitze M, Thoenes M, Moll R. Biological and prognosticsignificance of stratified epithelial cytokeratins in infiltrating ductalbreast carcinomas. Virchows Arch 1998; 433: 11929.

32. Murad TM, Scharpelli DG. The ultrastructure of medullary andscirrhous mammary duct carcinoma. Am J Pathol1967; 50: 33560.

33. Santini D, Ceccarelli C, Taffurelli M, Pileri S, Marrano D.Differentiation pathways in primary invasive breast carcinoma assuggested by intermediate filament and biopathological markerexpression.J Pathol1996; 179: 38691.

34. Kreike B, van Kouwenhove M, Horlings H, et al. Gene expressionprofiling and histopathological characterization of triple-negative/basal-like breast carcinomas. Breast Cancer Res 2007; 9: R65.

35. Banerjee S, Reis-Filho JS, Ashley S, et al . Basal-like breastcarcinomas: clinical outcome and response to chemotherapy. J ClinPathol2006; 59: 72935.

36. Rakha EA, El-Sayed ME, Green AR, et al. Prognostic markers intriple-negative breast cancer. Cancer 2007; 109: 2532.

37. Rakha EA, Tan DS, Foulkes WD, et al. Are triple-negative tumoursand basal-like breast cancer synonymous? Breast Cancer Res 2007; 9:404; author reply 405.

38. Cheang MC, Voduc D, Bajdik C, et al. Basal-like breast cancerdefined by five biomarkers has superior prognostic value than triple-negative phenotype. Clin Cancer Res 2008; 14: 136876.

39. Bertucci F, Finetti P, Cervera N,et al. How basal are triple-negativebreast cancers? Int J Cancer 2008; 123: 23640.

40. Calza S, Hall P, Auer G,et al. Intrinsic molecular signature of breastcancer in a population-based cohort of 412 patients. Breast Cancer

Res 2006; 8: R34.41. Jumppanen M, Gruvberger-Saal S, Kauraniemi P, et al. Basal-like

phenotype is not associated with patient survival in estrogen-receptor-negative breast cancers. Breast Cancer Res 2007; 9: R16.

42. Tan DS, Marchio C, Jones RL, et al. Triple negative breast cancer:molecular profiling and prognostic impact in adjuvant anthracycline-treated patients. Breast Cancer Res Treat 2008; 111: 2744.

43. Parker J, Mullins M, Cheang MC,et al. A supervised risk predictorof breast cancer based on biological subtypes. J Clin Oncol2008; 26:(in press).

44. Perreard L, Fan C, Quackenbush JF, et al. Classification and riskstratification of invasive breast carcinomas using a real-timequantitative RT-PCR assay. Breast Cancer Res 2006; 8: R23.

45. Rakha EA, Reis-Filho JS, Ellis IO. Basal-like breast cancer: a criticalreview.J Clin Oncol2008; 26: 256881.

46. Abd El-Rehim DM, Ball G, Pinder SE, et al. High-throughputprotein expression analysis using tissue microarray technology of alarge well-characterised series identifies biologically distinct classes of

breast cancer confirming recent cDNA expression analyses. Int JCancer 2005; 116: 34050.

47. Fulford LG, Reis-Filho JS, Ryder K, et al. Basal-like grade IIIinvasive ductal carcinoma of the breast: patterns of metastasis andlong-term survival. Breast Cancer Res 2007; 9: R4.

48. Carey LA, Dees EC, Sawyer L, et al. The triple negative paradox:primary tumor chemosensitivity of breast cancer subtypes. ClinCancer Res 2007; 13: 232934.

49. Dent R, Trudeau M, Pritchard KI, et al. Triple-negative breastcancer: clinical features and patterns of recurrence. Clin Cancer Res2007; 13: 442934.

50. Liedtke C, Mazouni C, Hess KR, et al. Response to neoadjuvanttherapy and long-term survival in patients with triple-negative breastcancer.J Clin Oncol2008; 26: 127581.

51. Jones C, Ford E, Gillett C,et al. Molecular cytogenetic identificationof subgroups of grade III invasive ductal breast carcinomas withdifferent clinical outcomes. Clin Cancer Res 2004; 10: 598897.

52. Honeth G, Bendahl PO, Ringner M, et al. The CD44/CD24-phenotype is enriched in basal-like breast tumors.Breast Cancer Res2008; 10: R53.

53. Wright MH, Calcagno AM, Salcido CD, et al. Brca1 breast tumorscontain distinct CD44/CD24- and CD133 cells with cancer stemcell characteristics. Breast Cancer Res 2008; 10: R10.

54. Krag D, Weaver D, Ashikaga T, et al. The sentinel node in breastcancera multicenter validation study.N Engl J Med1998; 339: 9416.

55. Veronesi U, Paganelli G, Galimberti V, et al. Sentinel-node biopsy toavoid axillary dissection in breast cancer with clinically negative

lymph-nodes. Lancet 1997; 349: 18647.56. Rutgers EJ. Sentinel node biopsy: interpretation and management of

patients with immunohistochemistry-positive sentinel nodes and thosewith micrometastases. J Clin Oncol2008; 26: 698702.

57. Viale G, Maiorano E, Mazzarol G, et al. Histologic detection andclinical implications of micrometastases in axillary sentinel lymphnodes for patients with breast carcinoma. Cancer 2001; 92: 137884.

58. Eifel P, Axelson JA, Costa J, et al. National Institutes of HealthConsensus Development Conference Statement: adjuvant therapy forbreast cancer, November 13, 2000. J Natl Cancer Inst 2001; 93: 97989.

59. Goldhirsch A, Wood WC, Gelber RD, et al. Meeting highlights:updated international expert consensus on the primary therapy ofearly breast cancer. J Clin Oncol2003; 21: 335765.

60. vant Veer LJ, Dai H, van de Vijver MJ, et al. Gene expressionprofiling predicts clinical outcome of breast cancer.Nature2002; 415:5306.

61. Buyse M, Loi S, vant Veer L,et al. Validation and clinical utility of a70-gene prognostic signature for women with node-negative breastcancer.J Natl Cancer Inst 2006; 98: 118392.

62. van de Vijver MJ, He YD, vant Veer LJ, et al. A gene-expressionsignature as a predictor of survival in breast cancer. N Engl J Med2002; 347: 19992009.

63. Wang Y, Klijn JG, Zhang Y,et al. Gene-expression profiles to predictdistant metastasis of lymph-node-negative primary breast cancer.Lancet2005; 365: 6719.

64. Dunkler D, Michiels S, Schemper M. Gene expression profiling: doesit add predictive accuracy to clinical characteristics in cancerprognosis? Eur J Cancer 2007; 43: 74551.

65. Eden P, Ritz C, Rose C, Ferno M, Peterson C. Good Old clinicalmarkers have similar power in breast cancer prognosis as microarraygene expression profilers. Eur J Cancer 2004; 40: 18371841.

66. Lu X, Lu X, Wang ZC, et al. Predicting features of breast cancerwith gene expression patterns. Breast Cancer Res Treat 2008; 108:191201.

67. Fan C, Oh DS, Wessels L,et al. Concordance among gene-expression-based predictors for breast cancer. N Engl J Med2006; 355: 5609.

68. Sotiriou C, Piccart MJ. Taking gene-expression profiling to the clinic:when will molecular signatures become relevant to patient care?NatRev Cancer 2007; 7: 54553.

69. Ioannidis JP. Microarrays and molecular research: noise discovery?Lancet2005; 365: 4545.

70. Michiels S, Koscielny S, Hill C. Interpretation of microarray data incancer.Br J Cancer 2007; 96: 11558.

71. Ein-Dor L, Zuk O, Domany E. Thousands of samples are needed togenerate a robust gene list for predicting outcome in cancer. Proc NatlAcad Sci USA 2006; 103: 59238.

72. Simon R. Roadmap for developing and validating therapeuticallyrelevant genomic classifiers. J Clin Oncol2005; 23: 733241.

73. Simon R, Radmacher MD, Dobbin K, McShane LM. Pitfalls in theuse of DNA microarray data for diagnostic and prognosticclassification. J Natl Cancer Inst 2003; 95: 148.

74. Shi L, Reid LH, Jones WD, et al. The MicroArray Quality Control

(MAQC) project shows inter- and intraplatform reproducibility ofgene expression measurements. Nat Biotechnol2006; 24: 115161.



11/12

75. Ravo M, Mutarelli M, Ferraro L, et al. Quantitative expressionprofiling of highly degraded RNA from formalin-fixed, paraffin-embedded breast tumor biopsies by oligonucleotide microarrays. LabInvest2008; 88: 43040.

76. Paik S, Shak S, Tang G,et al. A multigene assay to predict recurrenceof tamoxifen-treated, node-negative breast cancer. N Engl J Med2004; 351: 281726.

77. Esteva FJ, Sahin AA, Cristofanilli M, et al. Prognostic role of amultigene reverse transcriptase-PCR assay in patients with node-negative breast cancer not receiving adjuvant systemic therapy. ClinCancer Res 2005; 11: 33159.

78. Mina L, Soule SE, Badve S, et al. Predicting response to primarychemotherapy: gene expression profiling of paraffin-embedded corebiopsy tissue. Breast Cancer Res Treat 2007; 103: 197208.

79. Flanagan MB, Dabbs DJ, Brufsky AM, Beriwal S, Bhargava R.Histopathologic variables predict Oncotype DX recurrence score.Mod Pathol2008; 21: 125561.

80. Pusztai L, Cristofanilli M, Paik S. New generation of molecularprognostic and predictive tests for breast cancer. Semin Oncol2007;34: S106.

81. Rosai J. Why microscopy will remain a cornerstone of surgicalpathology.Lab Invest 2007; 87: 4038.

82. Ma XJ, Salunga R, Tuggle JT, et al. Gene expression profiles ofhuman breast cancer progression.Proc Natl Acad Sci USA 2003; 100:59749.

83. Elston CW, Ellis IO. Pathological prognostic factors in breast cancer.

I. The value of histological grade in breast cancer: experience from alarge study with long-term follow-up.Histopathology 1991; 19: 40310.

84. Robbins P, Pinder S, de Klerk N,et al. Histological grading of breastcarcinomas: a study of interobserver agreement. Hum Pathol1995; 26:8739.

85. Sotiriou C, Wirapati P, Loi S, et al. Gene expression profiling inbreast cancer: understanding the molecular basis of histologic gradeto improve prognosis. J Natl Cancer Inst 2006; 98: 26272.

86. Desmedt C, Sotiriou C. Proliferation: the most prominent predictorof clinical outcome in breast cancer. Cell Cycle 2006; 5: 2198202.

87. Loi S, Haibe-Kains B, Desmedt C, et al. Definition of clinicallydistinct molecular subtypes in estrogen receptor-positivebreast carcinomas through genomic grade. J Clin Oncol 2007; 25:123946.

88. Ivshina AV, George J, Senko O, et al. Genetic reclassification ofhistologic grade delineates new clinical subtypes of breast cancer.Cancer Res 2006; 66: 10292301.

89. Ma XJ, Salunga R, Dahiya S,et al. A five-gene molecular grade indexand HOXB13:IL17BR are complementary prognostic factors in earlystage breast cancer. Clin Cancer Res 2008; 14: 26018.

90. Andre F, Mazouni C, Hortobagyi GN, Pusztai L. DNA arrays aspredictors of efficacy of adjuvant/neoadjuvant chemotherapy in breastcancer patients: current data and issues on study design. BiochimBiophys Acta 2006; 1766: 197204.

91. Andre F, Pusztai L. Molecular classification of breast cancer:implications for selection of adjuvant chemotherapy. Nat Clin PractOncol2006; 3: 62132.

92. Tordai A, Wang J, Andre F,et al. Evaluation of biological pathwaysinvolved in chemotherapy response in breast cancer. Breast CancerRes2008; 10: R37.

93. Loi S, Haibe-Kains B, Desmedt C, et al. Predicting prognosis usingmolecular profiling in estrogen receptor-positive breast cancer treatedwith tamoxifen. BMC Genomics 2008; 9: 239.

94. Ma XJ, Wang Z, Ryan PD,et al. A two-gene expression ratio predicts

clinical outcome in breast cancer patients treated with tamoxifen.Cancer Cell2004; 5: 60716.

95. Reid JF, Lusa L, De Cecco L,et al. Limits of predictive models usingmicroarray data for breast cancer clinical treatment outcome. J NatlCancer Inst 2005; 97: 92730.

96. Goetz MP, Suman VJ, Ingle JN,et al. A two-gene expression ratio ofhomeobox 13 and interleukin-17B receptor for prediction ofrecurrence and survival in women receiving adjuvant tamoxifen.Clin Cancer Res 2006; 12: 20807.

97. Ma XJ, Hilsenbeck SG, Wang W, et al. The HOXB13:IL17BRexpression index is a prognostic factor in early-stage breast cancer.J Clin Oncol2006; 24: 46119.

98. Hudis CA. Trastuzumabmechanism of action and use in clinicalpractice.N Engl J Med2007; 357: 3951.

99. Harris LN, You F, Schnitt SJ, et al. Predictors of resistance topreoperative trastuzumab and vinorelbine for HER2-positive earlybreast cancer. Clin Cancer Res 2007; 13: 1198207.

100. Muss HB, Thor AD, Berry DA, et al. c-erbB-2 expression and

response to adjuvant therapy in women with node-positive earlybreast cancer. N Engl J Med1994; 330: 12606.

101. Paik S, Bryant J, Park C,et al. erbB-2 and response to doxorubicin inpatients with axillary lymph node-positive, hormone receptor-negative breast cancer. J Natl Cancer Inst 1998; 90: 136170.

102. Wang JC. DNA topoisomerases.Annu Rev Biochem1996; 65: 63592.103. Arriola E, Marchio C, Tan DS,et al. Genomic analysis of the HER2/

TOP2A amplicon in breast cancer and breast cancer cell lines. LabInvest2008; 88: 491503.

104. Arriola E, Rodriguez-Pinilla SM, Lambros MB,et al. TopoisomeraseII alpha amplification may predict benefit from adjuvantanthracyclines in HER2 positive early breast cancer. Breast CancerRes Treat 2007; 106: 1819.

105. Hicks DG, Yoder BJ, Pettay J,et al. The incidence of topoisomeraseII-alpha genomic alterations in adenocarcinoma of the breast andtheir relationship to human epidermal growth factor receptor-2 geneamplification: a fluorescence in situ hybridization study. Hum Pathol2005; 36: 34856.

10 6. Ja rv inen TA, Tanner M, Barlund M, Borg A, Isola J.Characterization of topoisomerase II alpha gene amplification anddeletion in breast cancer. Genes Chromosomes Cancer 1999; 26: 14250.

107. Orlando L, Del Curto B, Gandini S, et al . TopoisomeraseIIalpha gene status and prediction of pathological completeremission after anthracycline-based neoadjuvant chemotherapy inendocrine non-responsive Her2/neu-positive breast cancer. Breast2008; 17: 50611.

108. Tanner M, Isola J, Wiklund T, et al. Topoisomerase IIalpha gene

amplification predicts favorable treatment response to tailored anddose-escalated anthracycline-based adjuvant chemotherapy in HER-2/neu-amplified breast cancer: Scandinavian Breast Group Trial 9401.J Clin Oncol2006; 24: 242836.

109. Allinen M, Beroukhim R, Cai L, et al. Molecular characterization ofthe tumor microenvironment in breast cancer. Cancer Cell 2004; 6:1732.

110. Beck AH, Espinosa I, Gilks CB, van de Rijn M, West RB. Thefibromatosis signature defines a robust stromal response in breastcarcinoma.Lab Invest 2008; 88: 591601.

111. Bergamaschi A, Tagliabue E, Sorlie T, et al. Extracellular matrixsignature identifies breast cancer subgroups with different clinicaloutcome. J Pathol2008; 214: 35767.

112. Casey T, Bond J, Tighe S,et al. Molecular signatures suggest a majorrole for stromal cells in development of invasive breast cancer. BreastCancer Res Treat 2008; Mar 29 (Epub ahead of print).

113. Finak G, Bertos N, Pepin F, et al. Stromal gene expression predictsclinical outcome in breast cancer. Nat Med2008; 14: 51827.

114. Weigelt B, Bissell MJ. Unraveling the microenvironmental influenceson the normal mammary gland and breast cancer. Semin Cancer Biol2008; 18: 31121.

115. West RB, Nuyten DS, Subramanian S, et al. Determination ofstromal signatures in breast carcinoma. PLoS Biol2005; 3: e187.

116. Hu M, Yao J, Carroll DK, et al. Regulation of in situ to invasivebreast carcinoma transition. Cancer Cell2008; 13: 394406.

117. Jones C, Mackay A, Grigoriadis A, et al. Expression profiling ofpurified normal human luminal and myoepithelial breast cells:identification of novel prognostic markers for breast cancer. CancerRes2004; 64: 303745.

118. Teschendorff AE, Miremadi A, Pinder SE, Ellis IO, Caldas C. Animmune response gene expression module identifies a good prognosissubtype in estrogen receptor negative breast cancer. Genome Biol2007; 8: R157.

119. Tan DS, Lambros MB, Natrajan R, Reis-Filho JS. Getting it right:designing microarray (and not microawry) comparative genomic

hybridization studies for cancer research. Lab Invest2007; 87: 73754.120. Bergamaschi A, Kim YH, Wang P, et al. Distinct patterns of DNA

copy number alteration are associated with differentclinicopathological features and gene-expression subtypes of breastcancer.Genes Chromosomes Cancer 2006; 45: 103340.

121. Hicks J, Krasnitz A, Lakshmi B, et al. Novel patterns of genomerearrangement and their association with survival in breast cancer.Genome Res 2006; 16: 146579.

122. Chin K, DeVries S, Fridlyand J, et al. Genomic and transcriptionalaberrations linked to breast cancer pathophysiologies. Cancer Cell2006; 10: 52941.

123. Vogelstein B, Fearon ER, Hamilton SR, et al. Genetic alterationsduring colorectal-tumor development. N Engl J Med1988; 319: 52532.

124. Buerger H, Otterbach F, Simon R, et al. Comparative genomichybridization of ductal carcinoma in situ of the breast-evidence ofmultiple genetic pathways. J Pathol1999; 187: 396402.

125. Buerger H, Otterbach F, Simon R,et al. Different genetic pathways in

the evolution of invasive breast cancer are associated with distinctmorphological subtypes. J Pathol1999; 189: 5216.



12/12

126. Roylance R, Gorman P, Harris W, et al. Comparative genomichybridization of breast tumors stratified by histological grade revealsnew insights into the biological progression of breast cancer. CancerRes1999; 59: 14336.

127. Stange DE, Radlwimmer B, Schubert F, et al. High-resolutiongenomic profiling reveals association of chromosomal aberrations on1q and 16p with histologic and genetic subgroups of invasive breastcancer. Clin Cancer Res 2006; 12: 34552.

128. Cleton-Jansen AM, Buerger H, Haar N,et al. Different mechanisms ofchromosome 16 loss of heterozygosity in well- versus poorly differentiatedductal breast cancer.Genes Chromosomes Cancer 2004; 41: 10916.

129. Waldman FM, DeVries S, Chew KL,et al. Chromosomal alterationsin ductal carcinomas in situ and their in situ recurrences. J NatlCancer Inst 2000; 92: 31320.

130. Allred DC, Wu Y, Mao S, et al. Ductal carcinoma in situ and theemergence of diversity during breast cancer evolution. Clin CancerRes2008; 14: 3708.

131. Weigelt B, Horlings H, Kreike B, et al. Refinement of breast cancerclassification by molecular characterization of histological specialtypes.J Pathol2008; 216: 14150.

132. Kumar R, Neilsen PM, Crawford J,et al. FBXO31 is the chromosome16q24.3 senescence gene, a candidate breast tumor suppressor, and acomponent of an SCF complex. Cancer Res 2005; 65: 1130413.

133. Rakha EA, Green AR, Powe DG, Roylance R, Ellis IO.Chromosome 16 tumor-suppressor genes in breast cancer. GenesChromosomes Cancer 2006; 45: 52735.

134. van Wezel T, Lombaerts M, van Roon EH, et al. Expression analysisof candidate breast tumour suppressor genes on chromosome 16q.Breast Cancer Res 2005; 7: R9981004.

135. Cleton-Jansen AM. E-cadherin and loss of heterozygosity atchromosome 16 in breast carcinogenesis: different genetic pathwaysin ductal and lobular breast cancer? Breast Cancer Res 2002; 4: 58.

136. Vos CB, Cleton-Jansen AM, Berx G, et al. E-cadherin inactivation inlobular carcinoma in situ of the breast: an early event intumorigenesis.Br J Cancer 1997; 76: 11313.

137. Droufakou S, Deshmane V, Roylance R, et al. Multiple ways ofsilencing E-cadherin gene expression in lobular carcinoma of thebreast.Int J Cancer 2001; 92: 4048.

138. Sarrio D, Moreno-Bueno G, Hardisson D, et al. Epigenetic andgenetic alterations of APC and CDH1 genes in lobular breast cancer:relationships with abnormal E-cadherin and catenin expression andmicrosatellite instability. Int J Cancer 2003; 106: 20815.

139. Jacobs TW, Pliss N, Kouria G, Schnitt SJ. Carcinomas in situ of thebreast with indeterminate features: role of E-cadherin staining incategorization.Am J Surg Pathol2001; 25: 22936.

140. Maluf HM. Differential diagnosis of solid carcinoma in situ.SeminDiagn Pathol2004; 21: 2531.

141. Nishizaki T, Chew K, Chu L, et al. Genetic alterations in lobular breastcancerby comparative genomichybridization. IntJ Cancer 1997; 74: 5137.

142. Palacios J, Sarrio D, Garcia-Macias MC,et al. Frequent E-cadheringene inactivation by loss of heterozygosity in pleomorphic lobularcarcinoma of the breast. Mod Pathol2003; 16: 6748.

143. Reis-Filho JS, Simpson PT, Jones C, et al. Pleomorphic lobularcarcinoma of the breast: role of comprehensive molecular pathologyin characterization of an entity.J Pathol2005; 207: 113.

144. Simpson P, Reis-Filho J, Lambros M, et al. Molecular profilingpleomorphic lobular carcinomas of the breast: evidence for a commonmolecular genetic pathway with classic lobular carcinomas. J Pathol2008; 215: 23144.

145. Boecker W, Buerger H, Schmitz K, et al . Ductal epithelialproliferations of the breast: a biological continuum? Comparative

genomic hybridization and high-molecular-weight cytokeratinexpression patterns. J Pathol2001; 195: 41521.

146. Boecker W, Moll R, Dervan P, et al. Usual ductal hyperplasia of thebreast is a committed stem (progenitor) cell lesion distinct fromatypical ductal hyperplasia and ductal carcinoma in situ. J Pathol2002; 198: 45867.

147. OConnell P. Genetic and cytogenetic analyses of breast cancer yielddifferent perspectives of a complex disease. Breast Cancer Res Treat2003; 78: 34757.

148. Simpson PT, Gale T, Reis-Filho JS,et al. Columnar cell lesions of thebreast: the missing link in breast cancer progression? A morphologicaland molecular analysis. Am J Surg Pathol2005; 29: 73446.

149. Jones C, Nonni AV, Fulford L, et al. CGH analysis of ductalcarcinoma of the breast with basaloid/myoepithelial celldifferentiation. Br J Cancer 2001; 85: 4227.

150. Abdel-Fatah TM, Powe DG, Hodi Z, et al. High frequency ofcoexistence of columnar cell lesions, lobular neoplasia, and lowgrade ductal carcinoma in situ with invasive tubular carcinomaand invasive lobular carcinoma. Am J Surg Pathol 2007; 31: 41726.

151. Tognon C, Knezevich SR, Huntsman D, et al. Expression of theETV6-NTRK3 gene fusion as a primary event in human secretorybreast carcinoma. Cancer Cell2002; 2: 36776.

152. Diallo R, Schaefer KL, Bankfalvi A,et al. Secretory carcinoma of thebreast: a distinct variant of invasive ductal carcinoma assessed bycomparative genomic hybridization and immunohistochemistry. HumPathol2003; 34: 1299305.

153. Diallo R, Tognon C, Knezevich SR, Sorensen P, Poremba C.

Secretory carcinoma of the breast: a genetically defined carcinomaentity. Verh Dtsch Ges Pathol2003; 87: 193203.

154. Makretsov N, He M, Hayes M, et a l . A fluorescence in situhybridization study of ETV6-NTRK3 fusion gene in secretorybreast carcinoma. Genes Chromosomes Cancer 2004; 40: 1527.

155. Reis-Filho JS, Natrajan R, Vatcheva R,et al. Is acinic cell carcinomaa variant of secretory carcinoma? A FISH study using ETV6 splitapart probes. Histopathology 2008; 52: 8406.

156. Lakhani SR, Jacquemier J, Sloane JP, et al. Multifactorial analysisof differences between sporadic breast cancers and cancers involvingBRCA1 and BRCA2 mutations. J Natl Cancer Inst 1998; 90:113845.

157. Honrado E, Benitez J, Palacios J. The molecular pathology ofhereditary breast cancer: genetic testing and therapeutic implications.Mod Pathol2005; 18: 130520.

158. Tan DS, Marchio C, Reis-Filho JS. Hereditary breast cancer: frommolecular pathology to tailored therapies. J Clin Pathol 2008; 61:107382.

159. Lakhani SR, Van De Vijver MJ, Jacquemier J,et al. The pathology offamilial breast cancer: predictive value of immunohistochemicalmarkers estrogen receptor, progesterone receptor, HER-2, and p53in patients with mutations in BRCA1 and BRCA2. J Clin Oncol2002;20: 23108.

160. Lakhani SR, Reis-Filho JS, Fulford L, et al. Prediction of BRCA1status in patients with breast cancer using estrogen receptor and basalphenotype. Clin Cancer Res 2005; 11: 517580.

161. Foulkes WD, Stefansson IM, Chappuis PO,et al. Germline BRCA1mutations and a basal epithelial phenotype in breast cancer. J NatlCancer Inst 2003; 95: 14825.

162. Marchio C, Iravani M, Natrajan R, et al . Genomic andimmunophenotypical characterization of pure micropapillarycarcinomas of the breast. J Pathol2008; 215: 398410.

163. Weinstein IB. Cancer. Addiction to oncogenesthe Achilles heal ofcancer.Science 2002; 297: 634.

164. Weinstein IB, Joe AK. Mechanisms of disease: Oncogene addiction

a rationale for molecular targeting in cancer therapy. Nat Clin PractOncol2006; 3: 44857.


The Role of Molecular Analysis in Breast Cancer

Documents