Shaveta Vinayak, Hannah L. Gilmore, and Lyndsay N. Harris
INTRODUCTION
It has been said that cancer is a genetic disease and can be best understood by studying the DNA alterations that lead to the development of cancer. However, a deeper understanding of carcinogenesis requires insight into how these genetic changes alter cellular programs that lead to growth, invasion, and metastasis. This chapter is presented following the logical progression of DNA to RNA to protein, and it describes, at each step, the lesions that contribute to breast cancer carcinogenesis. The chapter also introduces concepts in epigenetics, microRNAs, and gene expression analyses that illustrate how new biologic discoveries and novel technologies have profoundly affected our understanding of breast cancer pathogenesis and influenced the treatment of patients.
GENETICS OF BREAST CANCER
Breast cancer is a heterogeneous disease fundamentally caused by the progressive accumulation of genetic aberrations, including point mutations, chromosomal amplifications, deletions, rearrangements, translocations, and duplications.1,2 Germ-line mutations account for only about 10% of all breast cancers, whereas the vast majority of breast cancers appear to occur sporadically and are attributed to somatic genetic alterations (Fig. 36.1).3
HEREDITARY BREAST CANCER
One of the most important risk factors for breast cancer is family history. Although familial forms comprise nearly 20% of all breast cancers, most of the genes responsible for familial breast cancer have yet to be identified. Breast cancer susceptibility genes can be categorized into three classes according to their frequency and level of risk they confer: rare high-penetrance genes, rare intermediate-penetrance genes, and common low-penetrance genes and loci (Table 36.1).4
High-Penetrance, Low-Frequency Breast Cancer Predisposition Genes
BRCA1 and BRCA2
BRCA1 and BRCA2 mutations account for approximately half of all dominantly inherited hereditary breast cancers. These mutations confer a relative risk of breast cancer 10 to 30 times that of women in the general population, resulting in a nearly 85% lifetime risk of breast cancer development.5 BRCA1 and BRCA2 mutation carriers are quite rare among the general population; however, the prevalence is substantially higher in certain founder populations, most notably in the Ashkenazi Jewish population, where the carrier frequency is 1 in 40.
More than a thousand germ-line mutations have been identified in BRCA1 and BRCA2. Pathogenic mutations most often result in truncated protein products, although mutations that interfere with protein function also exist.4,5Interestingly, the penetrance of pathogenic BRCA1 and BRCA2 mutations and age of cancer onset appear to vary both within and among family members. Specific BRCA mutations as well as gene–gene and gene–environment interactions as potential modifiers of BRCA-related cancer risk are areas of active investigation.6,7 Variation in risk may be explained by genetic modifiers that are different for both BRCA1 and BRCA2 mutation carriers. These alleles have been primarily identified from studies of the Consortium of Investigators of Modifiers of BRCA1/2 (CIMBA).8 The commonly identified single-nucleotide polymorphisms (SNP) that modify BRCA1/2 are listed in Table 36.2with their gene location, associated risks, and frequency. Evidence from published studies shows that these modifying SNPs combine multiplicatively, and therefore, may significantly alter a mutation carrier’s risk depending on the number of risk alleles present.9,10 In addition to retrospective studies, more recently, common breast cancer susceptibility alleles were evaluated prospectively to assess cancer risk conferred among unaffected BRCA1/2 mutation carriers.11 A risk score was constructed and divided into tertiles for sevenBRCA2-associated variants and four BRCA1-associated variants. Among BRCA2 mutation carriers, breast cancer risks were significantly different when stratified by tertiles; women in the highest tertile had a breast cancer risk of 72% by age 70 compared to 20% for those in the lowest tertile. There was no significant difference by risk score for BRCA1 mutation carriers.
BRCA1-related breast cancers are characterized by features that distinguish them from both BRCA2-related and sporadic breast cancers.4 BRCA1-related tumors typically occur in younger women and have more aggressive features, with high histologic grade, high proliferative rate, aneuploidy, and absence of estrogen and progesterone receptors and human epidermal growth factor receptor 2 (HER2). This triple-negative phenotype of BRCA1-related breast cancers is further characterized by a basallike gene expression profile of cytokeratins 5/6, 14, and 17, epidermal growth factor, and P-cadherin.12 Although BRCA1and BRCA2 genes encode large proteins with multiple functions, they primarily act as classic tumor suppressor genes that function to maintain genomic stability by facilitating double-strand DNA repair through homologous recombination.12,13 When loss of heterozygosity (LOH) occurs via loss, mutation, or silencing of the wild-type BRCA1 or BRCA2 allele, the resultant defective DNA repair leads to rapid acquisition of additional mutations, particularly during DNA replication, and ultimately sets the stage for cancer development.
The integral role of BRCA1 and BRCA2 in double-strand DNA repair holds potential as a therapeutic target for BRCA-related breast cancers. For example, platinum agents cause interstrand cross-links, thereby blocking DNA replication and leading to stalled replication forks. Poly (adenosine diphosphate [ADP]-ribose) polymerase-1 (PARP1) inhibitors additionally show promise as specific therapy for BRCA-related tumors. PARP1 is a cellular enzyme that functions in single-strand DNA repair through base excision and represents a major alternative DNA repair pathway in the cell.14,15 When PARP inhibition is applied to a background deficient in double-strand DNA repair, as is the case in BRCA-related tumor cells, the cells are left without adequate DNA repair mechanisms and ultimately undergo cell cycle arrest, chromosome instability, and cell death.4 Given their phenotypic similarities to BRCA1-related breast cancers, sporadic basallike breast tumors may display sensitivity to PARP inhibition as well.15 Phase II studies are currently under way to explore the use of PARP inhibitors in both BRCA- and basallike, non–BRCA-related breast tumors. There is much that remains to be understood about the optimal use of PARP inhibitors. Current challenges include, but are not limited to, identifying robust predictive biomarkers of response that can guide patient selection and understanding variations among PARP inhibitors in clinical development. Differences in potency and the mechanism of action have been well elucidated in recent preclinical studies,16–20 and the results of ongoing clinical trials will need to be interpreted in this context. Additionally, recent studies have also identified mechanisms of resistance to PARP inhibitors. One of these important mechanisms includes secondary mutations in the BRCA1/2 gene that can restore the open reading frame, and therefore, DNA repair functional activity,21–23 which renders tumors resistant to PARP inhibitors. Secondly, loss of tumor protein p53 binding protein 1 (TP53BP1) in BRCA1-deficient cells can restore DNA repair activity,24,25 and this may confer resistance.
Other High-Penetrance Genes
A small number of other high-risk, low-frequency breast cancer susceptibility genes exist, and they include TP53, PTEN, STK11/LKB1, and CDH1.4–6 These high-penetrance genes confer an eight- to tenfold increase in the risk of breast cancer as compared to noncarriers, but they collectively account for less than 1% of breast cancer cases. Like BRCA1 and BRCA2, these genes are inherited in an autosomal-dominant fashion and function as tumor suppressors.26 The hereditary cancer syndromes associated with each gene are usually characterized by multiple cancers in addition to breast cancer, as summarized inTable 36.1.
Moderate-Penetrance, Low-Frequency Breast Cancer Predisposition Genes
Four genes have been identified that confer an elevated but moderate risk of developing breast cancer, namely CHEK2, ATM, BRIP1, and PALB2 (see Table 36.1). Each of these genes confers approximately a two- to threefold relative risk of breast cancer in mutation carriers, although this risk may be higher in select clinical settings.5 Mutation frequencies in the general population are rare, on the order of 0.1% to 1%, although some founder mutations have been identified. Together, these genes account for approximately 2.3% of inherited breast cancer. The moderate relative risk of breast cancer of these genes in conjunction with the low population frequency renders this class of genes very difficult to detect with typical association studies. However, these genes were specifically selected for study as candidate breast cancer genes based on their known roles in signal transduction and DNA repair in close association with BRCA1 and BRCA2.6
Low-Penetrance, High-Frequency Breast Cancer Predisposition Genes and Loci
Both candidate gene and genome-wide association studies (GWAS) have identified a low-risk panel of approximately 10 different alleles and loci in 15% to 40% of women with breast cancer (see Table 36.1).5Despite their frequency, the relative risk of breast cancer conferred by any one of these genetic variants alone is minimal, on the order of less than 1.5.4 Nevertheless, these alleles and loci may become clinically relevant in their suggestion of interactions with other high-, moderate-, and low-risk genes; these additive or multiplicative relationships could account for a measurable fraction of population risk. For example, association studies of fibroblast growth factor receptor 2 (FGFR2) and mitogen-activated protein kinase kinase kinase 1 (MAP3K1) within BRCA families showed that these SNPs conferred an increased risk in the presence of BRCA2 mutations.
Microsatellite Instability in Breast Cancer
There is emerging data that Lynch syndrome, an autosomal-dominant inherited disorder of cancer susceptibility caused by germ-line mutations in the DNA mismatch repair (MMR) genes including, MLH1, MSH2, MSH5, and PMS2,may increase the risk of breast cancer.27 Mutation carriers are at increased risk of colorectal and other cancers, but its association with breast cancer risk has been controversial. A prospective cohort study using the Colon Cancer Family Registry evaluated cancer risks among unaffected carriers and noncarriers with a pathogenic MMR gene mutation; notably, breast cancer risk was estimated to be fourfold among mutation carriers compared to the general population.27 A systematic review of breast cancer risk studies for Lynch syndrome mutation carriers showed mixed results; 13 studies did not observe an increased risk, whereas 8 studies observed an increased risk of breast cancer ranging from 2- to 18-fold compared to the general population.28 Further studies are needed to determine more precise estimates of breast cancer risk in Lynch syndrome carriers with longer follow-ups. These studies may also guide future breast cancer screening guidelines for this population.
MicroRNA and Cancer Susceptibility
Recent studies suggest that microRNA (miRNA) SNPs may also contribute to breast cancer susceptibility, and miRNAs appear to regulate many tumor suppressor genes and oncogenes via degradation of target miRNAs or repression of their translation. Thus, genetic variations in miRNA genes or miRNA binding sites could affect the expression of tumor suppressor genes or oncogenes and, thereby, affect cancer risk. For example, specific SNPs located within pre-mir-27a and mir-196a-2 genes have been associated with reduced breast cancer risk, which has been confirmed in a recent meta-analysis.29,30
SOMATIC CHANGES IN BREAST CANCER
The vast majority of breast cancers are sporadic in origin, ultimately caused by an accumulation of numerous somatic genetic alterations.1 Recent data suggest that a typical individual breast cancer harbors anywhere from 50 to 80 different somatic mutations.2 Many of these mutations occur as a result of erroneous DNA replication; others may occur through exposure to exogenous and endogenous mutagens. To date, hundreds of candidate somatic breast cancer genes have been identified through GWAS.31,32
Determining the role of each identified mutation in the development of breast cancer remains a substantial challenge. Data suggest that the vast majority of identified somatic DNA mutations in a given tumor are passengermutations, representing harmless, biologically neutral changes that do not contribute to oncogenesis.1,2 Conversely, driver mutations confer a growth advantage on the cell in which they occur and appear to be implicated in cancer development. By definition, driver mutations are found in candidate cancer genes (CAN).32 A comprehensive catalog of somatic mutations and CAN genes has been accumulated through multiple studies. When specific driver mutations are cataloged among several different breast tumors, a bimodal cancer genomic landscape appears, comprising a small number of commonly mutated gene mountains among hundreds of infrequently mutated gene hills.1,2Gene mountains correspond to the most frequently mutated genes found within breast tumors, such as TP53, CDH1, phosphatidylinositol 3-kinase (PI3K), cyclin D, PTEN, and AKT.6 Each individual gene hill, on the other hand, is typically found in less than 5% of breast tumors.1,33 This substantial heterogeneity of DNA mutations among breast tumors may explain the wide variations in phenotypes, both in terms of tumor behavior as well as responsiveness to therapy.
Historically, the focus of genetic research has been on the gene mountains, in part because they were the only mutations that available technology could identify. However, emerging data suggest that it is actually the gene hills that play a much more pivotal role in breast cancer, which is consistent with the idea that having a large number of mutations, each associated with a small survival advantage, drives tumor progression. Recent studies have shown that a substantial number of these infrequent somatic mutations sort out among a much smaller number of biologic groups and cell signaling pathways that are known to be pathogenic in breast cancer, thereby vastly reducing the complexity of the genomic landscape. Examples of such pathways include interferon signaling, cell cycle checkpoint, BRCA1/2-related DNA repair, p53, AKT, transforming growth factor β (TGF-β) signaling, Notch, epidermal growth factor receptor (EGFR), FGF, ERBB2, RAS, and PI3K. In short, it appears that common pathways, rather than individual gene mutations, govern the course of breast cancer development.33
Although recurrent point mutations are less common in breast cancer than other solid tumors, emerging data show that particular regions of the genome are commonly amplified, and these regions contain genes that drive cancer progression. The best example of an important amplified region is the 17q12 amplicon that harbors the HER2 oncogene. This amplicon leads to a more aggressive tumor phenotype, now the target of a highly successful antibody therapy, trastuzumab (Herceptin). It has been observed that RNA-mediated interference (RNAi) knockdown of coamplified genes within the 17q12 amplicon results in decreased cell proliferation and increased apoptosis.34 Thus, the 17q12 amplicon appears to encode a concerted genetic program that contributes to the oncogenesis.
There are several other amplicons, in addition to 17q12 (HER2), that seem to drive the cancer phenotype and have prognostic significance in breast cancers, for example, 11q13 (CCDN1) and 8q24 (MYC), 20q13.35 These regions contain gene sets that are important in DNA metabolism and in the maintenance of chromosomal integrity, suggesting that a response to DNA damaging agents used as anticancer therapy might be modulated by the presence of particular amplicons. Indeed, these coamplicons are frequent in HER2-amplified tumors and may modify tumor behavior and patient outcome.36,37 The contribution of these genomic alterations to functional consequences may lie not in the overexpression of individual genes, but of gene cassettes on the amplicon.
Direct clinical translation of the growing catalog of somatic alterations in breast cancer has yet to evolve. However, with advancing technology and further identification and categorization of genetic mutations, new opportunities for individualized diagnosis and treatment options are likely to emerge.
TRANSCRIPTIONAL PROFILING OF BREAST CANCER
The cellular programs that are encoded by DNA are enacted by transcription into messenger RNA (mRNA) and translated into protein. Not surprisingly, the DNA alterations described previously lead to either under- or overexpression of their associated mRNAs; consequently, abnormal gene expression patterns are a common finding in breast tumors. Gene expression profiling has been introduced into the clinical literature during the past decade because research suggests that assessing the expression of multiple genes in a tumor sample may reflect programs turned on by DNA alternations and predict tumor behavior. So-called molecular signatures hold promise for improving the diagnosis, the prediction of recurrence, and the selection of therapies for individual patients.
Several technologies have been developed to generate molecular signatures, including cDNA and oligonucleotide arrays and multiplex polymerase chain reaction (PCR) technologies. These technologies and newly developed statistical methodologies now allow for evaluations of hundreds and even thousands of mRNAs simultaneously with groupings of samples based on coexpressedgenes.
Molecular Classification of Breast Cancer
The seminal work by Perou et al.38 and Sorlie et al.39 suggests a classification of breast cancer subtypes based on gene expression patterns they termed molecular portraits of breast cancer. Among the categories they defined were the luminal A and B tumor types (typically estrogen-receptor [ER] or progesterone-receptor [PR] positive), HER2 gene-amplified tumors, and a class termed basal-like due to the expression of basal keratins. Recent large scale efforts by The Cancer Genome Atlas Network (TCGA)40 and the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC)41 groups have confirmed these earlier findings in addition to providing more detailed molecular portraits.
Luminal Subtypes
The luminal subtypes comprise the majority of breast cancers and are characterized by the expression of genes that are normally expressed in the luminal epithelial of the breast such as cytokeratins 8 and 18 and the luminal expression signature (ESR1, GATA3, FOXA1, XPB1, and MYB). Luminal subtypes comprise the majority of clinical ER-positive breast cancers and can be divided into two subgroups: luminal A and luminal B. Luminal A tumors are more common and are characterized by high expression levels of ER-related genes and low expression of the HER2 cluster and proliferation-associated genes. In contrast, luminal B tumors are characterized by lower expression levels of ER-related genes, variable expression of the HER2 cluster, and higher levels of proliferation-associated genes. Luminal A tumors have an overall better prognosis than luminal B tumors.
HER2-Enriched Subtype
The HER2-enriched subtype comprises approximately 10% to 15% of all breast cancers and overexpresses both HER2 and proliferation-associated genes and has lower expression of ER-related genes. Interestingly, more recent work by the TCGA demonstrates that not all cancers that are clinically HER2-positive as defined by an immunohistochemical (IHC) analysis and/or fluorescent in situ hybridization (FISH) fall into the HER2-enriched molecular subtype and vice versa. The majority of clinically HER2-positive breast cancers that are not considered part of the HER2-enriched subgroup by gene expression profiling fall into the luminal intrinsic subtype with overexpression of HER2.
Estrogen Receptor–Negative Subtypes
The ER-negative subtypes comprise a heterogeneous group of tumors that clinically are termed triple negative breast cancer (TNBC) because they typically lack ER, PR, and HER2, and are often referred to astriple negative, although not all basallike tumors are triple negative and visa versa. The basallike category of the ER-negative subsets were first identified with first-generation microarray technology and show a high expression of proliferation genes and basal cytokeratins5,31,34 and a loss of genes associated with cell cycle control, which confer an overall poor prognosis. Though basallike tumors are the most common of the ER-negative subtypes (50% to 75% of all ER-negative tumors) and comprise 15% to 20% of all types of breast cancer, other ER-negative subtypes also exist, which include the recently described claudin-low group, as well as interferon-rich, androgen receptor, normallike groups. Though the claudin-low subgroup has some similarities to basallike breast cancer, it is distinct because these tumors have low expression of the claudin genes that are involved in epithelial cell tight-tight junctions. The claudin-low tumors have been of particular interest because they posses stem cell–like features of epithelial–mesenchymal transition (EMT).42 To further characterize the heterogeneity of TNBC, a recent study did a clustering analysis of gene expression profiling of primary tumors and identified six distinct subtypes.43 Major clusters included two basallike, an immunomodulatory, a mesenchymal, a mesenchymal stem–like, and a luminal androgen receptor subtype. Of further significance, this group provided preclinical evidence that these molecular subtypes were sensitive to different therapies.43 This has direct translational relevance and should be validated further.
Although the exact definition of molecular subtypes is an area of active debate, it is clear that these subtypes are reproducible in multiple, unrelated data sets, and their prognostic impact has been validated in these settings.38,40,44,45As a result, clinical trials are now being designed to subdivide patients by ER/PR and HER2 status to validate claims that therapeutic approaches should address these groups rather than the population of breast cancer patients as a whole. In 2011, the St. Gallen International Breast Cancer Conference recognized that breast cancer should not be treated as a single disease and recommended defining disease by molecular subtype using genetic array testing or approximated by ER/PR/HER2 status in conjunction with markers of proliferation, such as Ki-67. The panel reaffirmed this position again in 2013.46
Genetic Changes in Breast Cancer by Molecular Subtype
Mutational profiling of all types of breast cancer has demonstrated the marked heterogeneity that exists across the entire spectrum of tumors. Data from the TCGA highlights the fact that somatic mutations in just three genes (TP53, PIK3A, and GATA3) occurred at an incidence of greater than 10%.40 However, when the mutation profile of breast cancers is analyzed by intrinsic subgroup, certain patterns continue to emerge. Although the rate of significantly mutated genes is the lowest in the luminal subgroup, it is also the most heterogeneous group in terms of mutational spectrum. The most frequent mutation in luminal A tumors was in PIK3CA (45%) followed by MAP3K1, GATA3, TP53, CDH1, and MAP2K4. Like luminal A cancers, luminal B cancers also showed a wide range, with the most frequently mutation genes being TP53 and PIK3CA (both 29%). However, the TP53 pathway appears to be differentially inactivated with lower TP53 mutations in luminal A (12%) and higher mutations in luminal B (29%). Although the HER2-enriched subgroup also shows a high frequency of mutations in TP53 (72%) and PIK3CA(39%), unlike the luminal subtypes, HER2-enriched tumors appear to have a much lower frequency of other significantly mutated genes. Basallike tumors commonly harbor mutations in TP53 (80%), and there seems to be little overlap with the mutations seen in the luminal subtypes. In addition, the TP53 mutations present in the basallike group were mostly nonsense and frameshift–type mutations as opposed to more missense mutations seen in the luminal group. In fact, the mutations seen in the basallike group showed significant similarities to serous cancers of the ovary.42
Prognostic and Predictive Genomic Signatures
Prognostic Signatures
Gene expression molecular signatures are currently in clinical use for both defining prognosis and for determining the benefit of systemic therapies, including chemotherapy and endocrine treatment, for breast cancer. Van’t Veer et al.45 and van de Vijver et al.47 were the first to apply gene expression analysis to define a subgroup of breast cancer patients with an increased likelihood of metastasis. The estimated hazard ratio for distant metastases in the group with a poor prognosis signature, as compared to the group with the good prognosis signature, was 5.1 (95% confidence interval [CI], 2.9 to 9.0; p <0.001). The European Organisation for Research and Treatment of Cancer (EORTC) and the Breast International Group (BIG) are currently conducting a prospective clinical trial to validate the utility of this assay for sparing patients from systemic chemotherapy (the MINDACT study).48 In a preliminary analysis, the 70-gene profile signature was strongly prognostic, outperforming classic prognostic criteria such as those used by the St. Gallen consensus panel49; however, the magnitude of effect was much less than previously reported, with hazard ratios for time to distant metastases of 1.85 (1.14 to 3.0) and for overall survival of 2.5 (1.4 to 4.5). The 70-gene signature is now commercialized as the MammaPrint and has received clearance by the U.S. Food and Drug Administration (FDA) as a class 2, 510(k) product.
Other groups have developed prognostic gene expression signatures, including the 76-gene Rotterdam signature, which identifies a high-risk group of node-negative patients, and the Genomic Grade Index (GGI), which distinguishes poor and good prognosis groups in breast tumors of intermediate histologic grade.50 The potential value of these signatures has yet to be clearly defined, but it emphasizes the role of gene expression profiling at distinguishing prognostic groups not otherwise recognizable by standard histologic or clinical parameters.
Predictive Signatures
Endocrine Therapy. Several groups have applied a gene expression profiling analysis to better define the likelihood of benefit from therapy. Such predictive signatures may have particular value as they help oncologists counsel patients about appropriate choices for treatment. Genomics Health Inc. (Redwood City, California) developed the Oncotype DX assay as a predictor of benefit from antiestrogen therapy using multiple real-time reverse transcriptase polymerase chain reaction (RT-PCR) assays in formalin-fixed paraffin-embedded tissue. The assay was developed from 250 candidate genes selected from published literature, genomic databases, and in-house experiments performed on frozen tissue. From these data, a panel of 16 cancer-related genes and 5 reference genes were used to develop an algorithm to compute a recurrence score, ranging from 0 to 100, that can be used to estimate the odds of recurrence over 10 years from the diagnosis.51 Paik et al.51 reported an analysis of two randomized controlled trials: the National Surgical Adjuvant Breast and Bowel Project NSABP-B14, in which node-negative patients with ER-positive tumors were randomly assigned to tamoxifen or nil; and NSABP-B20, in which node-negative patients with ER-positive tumors were randomly assigned to tamoxifen alone or with cyclophosphamide, methotrexate, and fluorouracil (CMF) chemotherapy. Using the tissue samples from NSABP-B20, patients were categorized into three recurrence score groups: low risk (recurrence score less than 18), intermediate risk (recurrence score 18 to 30), and high risk (recurrence score 31 to 100). Samples from NSABP-B14 were then analyzed and found to be 6.8% (4.0% to 9.6%), 14.3% (8.3% to 20.3%), and 30.5% (23.6% to 37.4%). Paik et al.51 further analyzed the performance of the Oncotype DX assay to include patients in the other arms of NSABP-B14 and NSABP-B20 and found that the Oncotype DX assay was a strong predictor of benefit from CMF in NSABP-B20, with little or no benefit from chemotherapy for patients with low or intermediate recurrence scores but substantial benefit for those with high recurrence scores. Conversely, in NSABP-B14, the benefit from tamoxifen versus observation was confined to the low and intermediate risk categories (p value for interaction of 0.001). These data suggest that in patients who have an apparent favorable prognosis based on clinical features (negative nodes, positive ER), the Oncotype DX assay helps determine those most likely to benefit from tamoxifen only (low recurrence scores) versus those most likely not to benefit from tamoxifen but likely to benefit from chemotherapy (high recurrence scores). The benefits of chemotherapy in the 25% of patients who have intermediate recurrence scores remains uncertain and are the basis of an ongoing prospective randomized trial (Tailor Rx) where those with high recurrence scores will receive endocrine therapy and chemotherapy, those with low recurrence scores will receive endocrine therapy alone, and those with intermediate recurrence scores are randomly assigned to endocrine therapy versus endocrine and chemotherapy. A study by Albain et al.52 suggested that a low recurrence score predicts a lack of benefit of fluorouracil (5-FU), adriamycin (doxorubicin), and cyclophosphamide (FAC) chemotherapy in node-positive breast cancer patients treated on Southwest Oncology Group SWOG-8814. Although these provocative data suggest a similar utility for Oncotype DX in node-positive patients, they require additional validation with modern-day regimens. The value of the Oncotype DX assay in predicting a benefit from hormonal therapy in patients treated with aromatase inhibitor therapy has recently been published, demonstrating that the assay performs equally with both tamoxifen and anastrozole but does not distinguish a benefit of one over the other.53 Given the independent prognostic utility of both the Oncotype DX recurrence score and the clinicopathologic factors, such as tumor size and grade, a recent study formally integrated each of these measures to determine whether prognostic and predictive value is improved over using a single measure.54 The use of an integrated score, recurrence score-pathology-clinical (RSPC), among ER-positive, node-negative patients provided a more significant prognostic value for distant recurrence when compared to the recurrence score alone. This score also resulted in better risk stratification and reduced the number of patients classified as intermediate risk. However, the addition of clinicopathologic factors to the recurrence score did not improve its predictive value for chemotherapy benefit.54
Several other additional predictors for ER-positive breast cancer include the Breast Cancer Index (AvariaDx Inc., Carlsbad, California), a quantitative RT-PCR–based assay that measures the ratio of theHOXB6 and IL17BR genes, and includes a proliferation score. It was shown to be a marker of recurrence risk in untreated ER-positive/node-negative patients55,56 in two studies of lymph node–negative, ER-positive, tamoxifen-treated patients with breast cancer and was more recently found to predict late recurrence after adjuvant endocrine therapy.57 The breast cancer index (BCI) was compared with Onctoype DX recurrence score, and IHC4, a score based on four protein markers detected by immunohistochemistry. In ER-positive, node-positive breast cancer patients given either tamoxifen or anastrozole in the Arimidex, Tamoxifen, or Alone or in Combination (ATAC) trial, the BCI index was the only prognostic index that identified populations at risk of both early and late recurrences.57 This may be of clinical value in postmenopausal patients, who have undergone 5 years of endocrine therapy because the test was defined in this population.
Chemotherapy. Defining predictors of response to chemotherapy and targeted therapies has been more challenging. Ayers et al.58 from the M.D. Anderson Cancer Center were the first to report that a multigene analysis of fine needle aspiration specimens predicts a response to neoadjuvant Taxol, 5-fluouracil, Adriamycin, and cyclophosphamide (TFAC) chemotherapy.59 Validation of gene signatures is of utmost importance in the future to determine the value of these expression profiles at predicting treatment response and clinical outcome in breast cancer patients. National organizations such as the American Society of Clinical Oncology, the National Comprehensive Cancer Network, and the College of American Pathologists have ongoing efforts to interpret the data from the burgeoning field of multigene biomarker tests to help the practicing clinician interpret their clinical utility.60
EPIGENETICS OF BREAST CANCER
Cells maintain their stable identity and phenotype over many generations without external stimuli or signaling events. This cellular memory is encoded in the epigenome, a collection of heritable information that exists alongside the genomic sequence. DNA methylation and chromatin modification are major epigenetic mechanisms in higher eukaryotes and are tightly coupled to basic genetic processes, such as DNA replication, transcription, and repair. It is well documented that cancers, including breast cancer, have altered patterns of DNA methylation and histone acetylation, leading to alterations in transcription that appear to be oncogenic.61,62 Recent work from TCGA demonstrates different patterns of methylation by breast cancer subtypes as defined by gene expression profiling. Among these subtypes, luminal B subtype had a hypermethylated phenotype, whereas basallike subtype had a hypomethylated phenotype.40 Ongoing initiatives, including the Epigenome Project, and further analyses of the TCGA data will likely enhance our understanding of epigenetics in breast cancer.
Major epigenetic cancer drugs include DNA methyltransferase (DNMT) and histone deacetylase (HDAC) inhibitors. Preclinical studies show promise that HDAC inhibitors may have activity in breast cancer cells, and many clinical phase I and II studies are in progress.63–65
MicroRNAs
miRNAs are small noncoding RNAs that belong to a novel class of regulatory molecules that control the expression of hundreds of target mRNA transcripts66 in two ways. First, miRNAs that bind to protein-coding mRNA sequences that are exactly complementary to the miRNA induce the RNAi pathway. Messenger RNA targets are then cleaved by ribonucleases in the RNA-induced silencing complex (RISC). Second, miRNAs bind to imperfect complementary sites within the 3′ untranslated regions (3′UTR) of their target protein-coding mRNAs and repress the expression of these genes at the level of translation.67
miRNAs are known to be associated with breast cancer in both cell lines and clinical samples. For example, miR-21, miR-155, miR-7, and miR-210 are overexpressed in aggressive human breast cancers,68,69whereas let-7 and miR-125a have been shown to be downregulated in breast cancers.70 It has also been shown that miR-125a may function as a tumor suppressor by inhibiting ERBB2 and ERBB3. More recently, the TCGA identified seven subtypes by microRNA expression profiling. Among all of these microRNA clusters, only two of them had a positive correlation with TP53 mutation and overlap with the basallike subtype. No additional correlation with mutation status or mRNA-defined breast cancer subtypes was identified.40
MicroRNAs and Response to Cancer Treatment
miRNA misexpression patterns were found to be associated with cancer outcome and response to treatment, including radiation and chemotherapy. Certain miRNAs associated with hypoxia, such as miR-210, have been shown to be biomarkers of poor outcome in breast cancer.68 Furthermore, in vitro data show that certain miRNAs are associated with resistance to doxorubicin71 or tamoxifen.72 In patient samples, an association of miRNA’s tumor subtypes have specific miRNA patterns and this is associated with a poor outcome. Defining the role of miRNAs as biomarkers for prognosis and prediction, as well as their potential as targeted therapies, is an active area of research in breast cancer.
PROTEIN/PATHWAY ALTERATIONS
The molecular mechanisms that lead to cancer have been characterized as the hallmarks of cancer, as proposed by Hanahan and Weinberg and revised in 2011.73 They include sustained proliferative signaling, evading growth suppressors, resisting cell death, replicative immortality through telomerase inhibition, angiogenesis, invasion and metastasis, genomic instability, deregulated metabolism, and avoiding immune destruction. The effectors of genetic and epigenetic abnormalities are, in most cases, reflected in the abnormal levels, functions, and interactions of proteins and signaling pathways. Recent studies of the genome have generated new insights into the proteome associated with specific breast cancer subtypes and suggest important targets for therapy, in addition to those canonical drivers ER and HER2.40Undoubtedly, numerous alterations coordinate to result in the malignant phenotype; however, a number of key proteins and their pathways have emerged as critical drivers of breast cancer development and growth as well as potential therapeutic targets.
ESTROGEN RECEPTOR PATHWAY
Therapeutic Targets in Breast Cancer
Estrogen Signaling
Most breast cancers are intimately linked with exposure to estrogen and alterations in the estrogen receptor signaling pathway. Estrogen is a steroid hormone that exerts its actions by binding to the nuclear ER. Upon activation by its ligand, ER binds in a coordinated fashion with a number of coregulatory proteins to estrogen response elements in the promoter region of estrogen-responsive genes. This in turn directs the transcription of numerous growth-promoting genes, including PR. The level of ER expression is not only of biologic interest, but it is also a highly effective predictor for response to antiestrogens, which is a recommended treatment for all ER-expressing tumors.
Although ER is overexpressed in as many as 70% of invasive breast cancers, the precise mechanism by which this occurs is unclear. Amplification of the gene appears to be one mechanism (approximately 50% of cases with ER overexpression in one study), suggesting that transcriptional deregulation and posttranscriptional modifications (such as alteration of mRNA levels by miRNAs) may also play a role. In addition, recent studies suggest ER mutations can lead to constitutive activation of the pathway and may be a mechanism of resistance to antiestrogen treatment.74
Estrogen exerts its actions through both genomic (described previously) and nongenomic mechanisms. In contrast to the genomic actions of ER, nongenomic actions of ER are extremely rapid (within seconds to minutes of estrogen exposure) and are believed to result from the hormone-dependent activation of membrane-bound or cytosolic ERs. These nonnuclear ER actions result in rapid phosphorylation and activation of important growth regulatory kinases, including EGFRs, insulinlike growth factor 1R (IGF-1R), c-Src, Shc, and the p85α regulatory subunit of PI3K.5 This cross-talk between ER and growth factor receptors is bidirectional; for example, constitutive HER2 can increase ER signaling to the point where it is unresponsive to antiestrogen treatments. These findings suggest a role for HER2/IGF-1R/EGFR activation in both acquired and de novo resistance to treatment with antiestrogens.75
The ER pathway has proven to be an invaluable target for therapeutic treatments in breast cancer. A number of agents have been developed over the prior decades that can inhibit this pathway by either binding to the receptor itself (e.g., selective ER modulators such as tamoxifen, raloxifene, fulvestrant) or by decreasing the production of endogenous estrogen (e.g., aromatase inhibitors, ovarian ablation). Recent data suggest that a longer duration of tamoxifen (10 years) is superior to 5 years, and 5 years of Aromatase Inhibitor (AI) are standard of care after any duration of tamoxifen in postmenopausal women.76 Although these agents are highly effective and have made a significant impact on breast cancer morbidity and mortality, de novo and acquired resistance are also quite common. Recent studies suggest that the inhibition of growth factor pathways in conjunction with antiestrogen therapy can overcome resistance to these agents; for example, the mammalian target of rapamycin (mTOR) inhibitor temsirolimus with a steroidal inhibitor (exemestane) is a new standard of care after progression on a nonsteroidal aromatase inhibitor in the metastatic setting.77 The challenge for the oncology community is to define optimal biomarkers to predict patients most likely to benefit from longer tamoxifen or AI + mTOR therapy. As described previously, the Oncotype DX assay, IHC4, and Breast Cancer Index provide insight into the behavior of ER-positive tumors and help in treatment decision making.55–57
Growth Factor Receptor Pathways
Growth factor receptor pathways—in particular, tyrosine-kinase receptors—play an essential role in initiating both proliferative and cell survival pathways in tissues and are tightly regulated. In breast cancer biology, the ErbB family has been studied most extensively, but an expanding number of other growth factors, such as insulin-like growth factor receptors, have also been the subject of intense scrutiny in hopes of identifying effective therapeutic targets.78These growth factor receptor pathways can be constitutively activated by a number of mechanisms, including excessive ligand levels, gain-of-function mutations, overexpression with or without gene amplification, and gene rearrangements and resultant fusion proteins with oncogenic potential. This can ultimately lead to inappropriate kinase activity and growth promoting second messenger activation (Fig. 36.2).
Human Epidermal Growth Factor Receptor 2
HER2 (EGFR2 or ErbB2) is a member of a family of receptor tyrosine kinases that also includes EGFR (HER1, ErbB1), ErbB3, and ErbB4. Ligand binding to the extracellular domains of the ErbB1, ErbB3, or ErbB4 receptors induces homo- and heterodimerization and kinase activation. The HER2 protein exists in a closed conformation and has no ligand, but it is the preferred partner for dimerization with HER1, -3, and -4. At a molecular level, HER2amplification is associated with deregulation of G1/S phase cell cycle control via the upregulation of cyclins D1, E, and cdk6, as well as p27 degradation. HER2 also interacts with important second messengers, including SH2-domain–containing proteins (e.g., Src kinases).
Importantly, HER2 amplification or protein overexpression (found in 20% of invasive breast cancers) is clearly associated with accelerated cell growth and proliferation, poor clinical outcome, and response to the monoclonal anti-HER2 antibody, trastuzumab. Numerous randomized trials have shown that the addition of trastuzumab to chemotherapy improves survival in both metastatic and early stage disease, leading to its inclusion in the standard of care for all patients with HER2-positive breast cancer.79 In addition, several other HER2-targeted agents have been approved for metastatic HER2-positive breast cancer and are being evaluated in the early stage setting. One of these, the monoclonal antibody pertuzumab, which targets the HER2-3 heterodimerization site, has recently been approved for use in the preoperative setting for stage II and III HER2-positive breast cancer.80 These rapid advances in the setting of targeted therapy for HER2-positive disease illustrate the profound effect that targeting an important molecular driver can have on clinical practice.
The precise mechanism of action of the FDA-approved HER2-targeted therapies trastuzumab, pertuzumab, lapatinib, and trastuzumab emtansine (TDM-1) are not well understood.79,81 In preclinical studies, the first three appear to inhibit signal transduction through the canonical signaling pathways for the HER2 receptor, but may vary in their degree of ras-MAPK versus PI3K-AKT pathway inhibition. This may largely be due to different degrees of inhibition of the coreceptors HER1, 3, and 4 that have a different predilection for each pathway. For instance, lapatinib inhibits both HER2 and EGFR (HER1) with a greater effect on the ras-MAPK pathway. Pertuzumab interferes with the HER3 heterodimer and hence has more effect on the AKT pathway. The new targeted therapy TDM-1 is an antibody–drug conjugate, and its effects are more cytotoxic than signal transduction. All three antibody therapies are thought to activate natural killer cells involved in antibody-dependent cellular cytotoxicity (ADCC); however, there is little evidence from clinical studies to support this or any mechanism of action. As a result, mechanisms of resistance are poorly understood; current hypotheses include activation of alternate receptors (e.g., IGF-1R, c-met) or the AKT pathway via a loss of phosphatase and tensin homolog (PTEN) or P13K mutation (see Fig. 36.2).78,79,81,82
Ras And Phosphatidylinositol 3-Kinase Pathways
Redundancies and cross-talk of numerous different signaling pathways are a common theme. Several downstream messengers, however, bear special consideration due to their functional importance and therapeutic implications. Recent data from TCGA Breast Cancer publication suggest that the PI3K/AKT and ras-MAP kinase pathways are particularly relevant in breast cancer based on frequent mutation, amplification, and/or activation of these pathways as measured by genomic technologies.40
P13K-AKT is a central signaling pathway downstream of many receptor tyrosine kinases and regulates cell growth and proliferation (see Fig. 36.2B). Activating mutations in the gene encoding the p110α catalytic subunit of PI3K (PI3CKA) may be an important contributing factor to mammary tumor progression, and the site of mutation differs depending on the breast cancer molecular subtype as noted previously. Activating mutations of the AKT gene family are seen in 2% to 4% of breast cancers, excluding the basallike subtype, where they are rare.42
PTEN dephosphorylates—and therefore inactivates—the p110 catalytic domain of PI3K and is either mutated or underexpressed (e.g., via methylation) in many breast cancers. Activation of the PI3K pathway, in turn, results in the 3-phosphoinositide–dependent kinase-mediated activation of several known kinases, including AKT1, AKT2, and AKT3. Interestingly, activated AKT1 appears to be antiapoptotic but also plays an anti-invasive role in tumor formation. In addition to the AKTs, downstream proliferative effectors of the PI3K pathway also include the mTOR complex 1 (TORC1), which consists of mTOR, raptor, and mLst8. It is currently believed that TORC1 mediates its progrowth effects through the activation of S6-kinase 1 and suppression of 4E-BP1, an inhibitor of cap-dependent translation. These observations all point to mTOR-raptor as a critical target in cancer therapy, and indeed, several mTOR inhibitors known as rapamycin analogs (e.g., CCI-779, RAD-001, AP-23576) are in clinical trials, and temsirolimus (Affinitor) has been approved for use with aromatase inhibitor therapy.
The ras/raf/MEK/MAPK pathway is also a critical signaling pathway for numerous growth factor receptors (see Fig. 36.2A). Thus far in breast cancer, agents that target the MEK pathway (e.g., raf inhibitor sorafenib) have had modest success as single agents, but studies in combination with other treatments hold more promise.
Angiogenesis
Angiogenesis is normally a tightly regulated process of vessel formation during physiologic events, such as wound healing and pregnancy. It has also been shown to be an important part of tumor growth and spread. In contrast to physiologic angiogenesis, tumor-associated angiogenesis is highly dysregulated with disorganized and distorted vasculature and increased vascular permeability. Thus, in recent years, angiogenesis has become a frequent target for the treatment of many cancers.
Central to this process is the proangiogenic factor, vascular endothelial growth factor (VEGF), which together with its receptors, regulate endothelial cell growth and new vessel formation.81 VEGF receptors (VEGFR), like EGFRs, are also tyrosine-kinase receptors. VEGF-A binds to both VEGFR1 (Flt-1) and VEGFR2 (KDR/Flk1). VEGFR2 appears to mediate most of the known cellular responses to VEGFs, whereas the function of VEGFR1 is less well defined. Bevacizumab, a humanized monoclonal antibody directed against VEGF-A, has been the most extensively studied thus far. To date, three large randomized trials have shown a statistically significant benefit in progression-free survival when bevacizumab was added to a variety of different chemotherapies in the first-line metastatic setting. However, the lack of survival advantage in any study has led to the withdrawal of bevacizumab’s approval, and studies in early stage breast cancer are required to be positive in order to reconsider its use in breast cancer. Multitargeted VEGFR tyrosine-kinase inhibitors such as sunitinib (VEGFR, platelet-derived growth factor receptor (PDGFR), and c-kit blockade) and sorafenib (VEGFR and RAF kinase blockade) have also been studied extensively in breast and other cancers. Despite the success of some of these agents, the identification of predictive factors for an antiangiogenic response have thus far proven to be elusive.
SUMMARY
Breast cancer is a heterogeneous malignancy, with multiple molecular subtypes and clinical presentations ranging from aggressive to indolent, and varying in distribution by age, menopausal status, and racial group. Recent molecular analyses have shed light on this heterogeneity by mapping patterns that correspond to clinical phenotypes. These studies include, but are not limited to, the study of germ-line variations in breast cancer susceptibility genes, gene expression, copy number and somatic mutations, epigenetic modifications, and alterations in protein pathways associated with the malignant phenotype. This review attempts to summarize the current state of the science with an emphasis on the insights provided by these studies on prognosis and prediction for therapy and potential new therapeutic targets. It is only through ongoing research that this disease will one day be cured.
REFERENCES
1. Bell DW. Our changing view of the genomic landscape of cancer. J Pathol 2010;2:231–243.
2. Wood LD, Parsons DW, Jones S, et al. The genomic landscapes of human breast and colorectal cancers. Science 2007;318:1108–1113.
3. Velculescu VE. Defining the blueprint of the cancer genome. Carcinogenesis 2008;29:1087–1091.
4. Turnbull C, Rahman N. Genetic predisposition to breast cancer: past, present, and future. Annu Rev Genomics Hum Genet 2008;9:321–345.
5. Foulkes WD. Inherited susceptibility to common cancers. N Engl J Med 2008;359:2143–2153.
6. Hirshfield KM, Rebbeck TR, Levine AJ. Germline mutations and polymorphisms in the origins of cancers in women. J Oncol 2010;2010:297671.
7. Narod SA. Modifiers of risk of hereditary breast cancer. Oncogene 2006;25:5832–5836.
8. Chenevix-Trench G, Milne RL, Antoniou AC, et al. An international initiative to identify genetic modifiers of cancer risk in BRCA1 and BRCA2 mutation carriers: the Consortium of Investigators of Modifiers of BRCA1 and BRCA2 (CIMBA). Breast Cancer Res 2007;9:104.
9. Antoniou AC, Sinilnikova OM, McGuffog L, et al. Common variants in LSP1, 2q35 and 8q24 and breast cancer risk for BRCA1 and BRCA2 mutation carriers. Hum Mol Genet 2009;18:4442–4456.
10. Antoniou AC, Spurdle AB, Sinilnikova OM, et al. Common breast cancer-predisposition alleles are associated with breast cancer risk in BRCA1 and BRCA2 mutation carriers. Am J Hum Genet 2008;82:937–948.
11. Mavaddat N, Peock S, Frost D, et al. Cancer risks for BRCA1 and BRCA2 mutation carriers: results from prospective analysis of EMBRACE. J Natl Cancer Inst 2013;105:812–822.
12. Turner NC, Reis-Filho JS. Basal-like breast cancer and the BRCA1 phenotype. Oncogene 2006;25:5846–5853.
13. Venkitaraman AR. Cancer susceptibility and the functions of BRCA1 and BRCA2. Cell 2002;2:171–182.
14. Fong PC, Boss DS, Yap TA, et al. Inhibition of poly(ADP-ribose) polymerase in tumors from BRCA mutation carriers. N Engl J Med 2009;361:123–134.
15. Iglehart JD, Silver DP. Synthetic lethality—a new direction in cancer-drug development. N Engl J Med 2009;2:189–191.
16. Liu X, Shi Y, Maag DX, et al. Iniparib nonselectively modifies cysteine-containing proteins in tumor cells and is not a bona fide PARP inhibitor. Clin Cancer Res 2012;18:510–523.
17. Ma W, Halweg CJ, Menendez D, et al. Differential effects of poly(ADP-ribose) polymerase inhibition on DNA break repair in human cells are revealed with Epstein-Barr virus. Proc Natl Acad Sci U S A 2012;109:6590–6595.
18. Murai J, Huang SY, Das BB, et al. Trapping of PARP1 and PARP2 by clinical PARP inhibitors. Cancer Res 2012;72:5588–5599.
19. Patel AG, De Lorenzo SB, Flatten KS, et al. Failure of iniparib to inhibit poly(ADP-Ribose) polymerase in vitro. Clin Cancer Res 2012;18:1655–1662.
20. Shen Y, Rehman FL, Feng Y, et al. BMN 673, a novel and highly potent PARP1/2 inhibitor for the treatment of human cancers with DNA repair deficiency. Clin Cancer Res 2013;19:5003–5015.
21. Edwards SL, Brough R, Lord CJ, et al. Resistance to therapy caused by intragenic deletion in BRCA2. Nature 2008;451:1111–1115.
22. Norquist B, Wurz KA, Pennil CC, et al. Secondary somatic mutations restoring BRCA1/2 predict chemotherapy resistance in hereditary ovarian carcinomas. J Clin Oncol 2011;29:3008–3015.
23. Sakai W, Swisher EM, Karlan BY, et al. Secondary mutations as a mechanism of cisplatin resistance in BRCA2-mutated cancers. Nature 2008;451:1116–1120.
24. Bouwman P, Aly A, Escandell JM, et al. 53BP1 loss rescues BRCA1 deficiency and is associated with triple-negative and BRCA-mutated breast cancers. Nat Struct Mol Biol 2010;17:688–695.
25. Jaspers JE, Kersbergen A, Boon U, et al. Loss of 53BP1 causes PARP inhibitor resistance in Brca1-mutated mouse mammary tumors. Cancer Discov 2013;3:68–81.
26. Stratton MR, Rahman N. The emerging landscape of breast cancer susceptibility. Nat Genet 2008;40:17–22.
27. Win AK, Young JP, Lindor NM, et al. Colorectal and other cancer risks for carriers and noncarriers from families with a DNA mismatch repair gene mutation: a prospective cohort study. J Clin Oncol 2012;30:958–964.
28. Win AK, Lindor NM, Jenkins MA. Risk of breast cancer in Lynch syndrome: a systematic review. Breast Cancer Res 2013;15:R27.
29. Xu Q, He CY, Liu JW, et al. Pre-miR-27a rs895819A/G polymorphisms in cancer: a meta-analysis. PloS One 2013;8:e65208.
30. Hoffman AE, Zheng T, Yi C, et al. microRNA miR-196a-2 and breast cancer: a genetic and epigenetic association study and functional analysis. Cancer Res 2009;69:5970–5977.
31. Forbes SA, Bhamra G, Bamford S, et al. The Catalogue of Somatic Mutations in Cancer (COSMIC). Curr Protoc Hum Genet 2008;Chapter 10:Unit 10.11.
32. Stratton MR, Campbell PJ, Futreal PA. The cancer genome. Nature 2009;458:719–724.
33. Copeland NG, Jenkins NA. Deciphering the genetic landscape of cancer—from genes to pathways. Trends Genet 2009;25:455–462.
34. Kao J, Pollack JR. RNA interference-based functional dissection of the 17q12 amplicon in breast cancer reveals contribution of coamplified genes. Genes Chromosomes Cancer 2006;45:761–769.
35. Chin K, DeVries S, Fridlyand J, et al. Genomic and transcriptional aberrations linked to breast cancer pathophysiologies. Cancer Cell 2006;10:529–541.
36. Bentires-Alj M, Gil SG, Chan R, et al. A role for the scaffolding adapter GAB2 in breast cancer. Nat Med 2006;12:114–121.
37. Yu Q, Sicinska E, Geng Y, et al. Requirement for CDK4 kinase function in breast cancer. Cancer Cell 2006;9:23–32.
38. Perou CM, Sorlie T, Eisen MB, et al. Molecular portraits of human breast tumours. Nature 2000;406:747–752.
39. Sorlie T, Perou CM, Tibshirani R, et al. Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications. Proc Natl Acad Sci U S A 2001;98:10869–10874.
40. Cancer Genome Atlas Network. Comprehensive molecular portraits of human breast tumours. Nature 2012;490:61–70.
41. Curtis C, Shah SP, Chin SF, et al. The genomic and transcriptomic architecture of 2,000 breast tumours reveals novel subgroups. Nature 2012;486:346–352.
42. Perou CM. Molecular stratification of triple-negative breast cancers. Oncologist 2010;15:39–48.
43. Lehmann BD, Bauer JA, Chen X, et al. Identification of human triple-negative breast cancer subtypes and preclinical models for selection of targeted therapies. J Clin Invest 2011;121:2750–2767.
44. Gruvberger S, Ringnér M, Chen Y, et al. Estrogen receptor status in breast cancer is associated with remarkably distinct gene expression patterns. Cancer Res 2001;16:5979–5984.
45. van ’t Veer LJ, Dai H, van de Vijver MJ, et al. Gene expression profiling predicts clinical outcome of breast cancer. Nature 2002;415:530–536.
46. Goldhirsch A, Winer EP, Coates AS, et al. Personalizing the treatment of women with early breast cancer: highlights of the St Gallen International Expert Consensus on the Primary Therapy of Early Breast Cancer 2013. Ann Oncol 2013;24:2206–2223.
47. van de Vijver MJ, He YD, van’t Veer LJ, et al. A gene-expression signature as a predictor of survival in breast cancer. N Engl J Med 2002;25:1999–2009.
48. Piccart M, Loi S, Van’t Veer L, et al. Multi-center external validation study of the Amsterdam 70-gene prognostic signature in node negative untreated breast cancer: Are the results still outperforming the clinical-pathological criteria? Presented at: 2004 27th San Antonio Breast Cancer Symposium; December 2004; San Antonio, TX.
49. Goldhirsch A, Wood WC, Gelber RD, et al. Meeting highlights: updated international expert consensus on the primary therapy of early breast cancer. J Clin Oncol 2003;21:3357–3365.
50. Sotiriou C, Wirapati P, Loi S, et al. Gene expression profiling in breast cancer: understanding the molecular basis of histologic grade to improve prognosis. J Natl Cancer Inst 2006;98:262–272.
51. Paik S, Shak S, Tang G, et al. A multigene assay to predict recurrence of tamoxifen-treated, node-negative breast cancer. N Engl J Med 2004;351:2817–2826.
52. Albain KS, Barlow WE, Shak S, et al. Prognostic and predictive value of the 21-gene recurrence score assay in postmenopausal women with node-positive, oestrogen-receptor-positive breast cancer on chemotherapy: a retrospective analysis of a randomised trial. Lancet Oncol2010;11:55–65.
53. Mamounas EP, Tang G, Fisher B, et al. Association between the 21-gene recurrence score assay and risk of locoregional recurrence in node-negative, estrogen receptor-positive breast cancer: results from NSABP B-14 and NSABP B-20. J Clin Oncol 2010;10:1677–1683.
54. Tang G, Cuzick J, Costantino JP, et al. Risk of recurrence and chemotherapy benefit for patients with node-negative, estrogen receptor-positive breast cancer: recurrence score alone and integrated with pathologic and clinical factors. J Clin Oncol 2011;29:4365–4372.
55. Ma XJ, Hilsenbeck SG, Wang W, et al. The HOXB13:IL17BR expression index is a prognostic factor in early-stage breast cancer. J Clin Oncol 2003;28:4611–4619.
56. Ma XJ, Wang Z, Ryan PD, et al. A two-gene expression ratio predicts clinical outcome in breast cancer patients treated with tamoxifen. Cancer Cell 2004;5:607–616.
57. Sgroi DC, Sestak I, Cuzick J, et al. Prediction of late distant recurrence in patients with oestrogen-receptor-positive breast cancer: a prospective comparison of the breast-cancer index (BCI) assay, 21-gene recurrence score, and IHC4 in the TransATAC study population. Lancet Oncol 2013;14:1067–1076.
58. Ayers M, Symmans WF, Stec J, et al. Gene expression profiles predict complete pathologic response to neoadjuvant paclitaxel and fluorouracil, doxorubicin, and cyclophosphamide chemotherapy in breast cancer. J Clin Oncol 2004;22:2284–2293.
59. Acharya CR, Hsu DS, Anders CK, et al. Gene expression signatures, clinicopathological features, and individualized therapy in breast cancer. JAMA 2008;299:1574–1587.
60. Harris L, Fritsche H, Mennel R, et al. American Society of Clinical Oncology 2007 update of recommendations for the use of tumor markers in breast cancer. J Clin Oncol 2007;33:5287–5312.
61. Barski A, Cuddapah S, Cui K, et al. High-resolution profiling of histone methylations in the human genome. Cell 2007;129:823–837.
62. Veeck J, Esteller M. Breast cancer epigenetics: from DNA methylation to microRNAs. J Mammary Gland Biol Neoplasia 2010;15:5–17.
63. Fiskus W, Ren Y, Mohapatra A, et al. Hydroxamic acid analogue histone deacetylase inhibitors attenuate estrogen receptor-alpha levels and transcriptional activity: a result of hyperacetylation and inhibition of chaperone function of heat shock protein 90. Clin Cancer Res2007;13:4882–4890.
64. Zhou Q, Shaw PG, Davidson NE. Inhibition of histone deacetylase suppresses EGF signaling pathways by destabilizing EGFR mRNA in ER-negative human breast cancer cells. Breast Cancer Res Treat 2009;117:443–451.
65. Yardley DA, Ismail-Khan RR, Melichar B, et al. Randomized phase II, double-blind, placebo-controlled study of exemestane with or without entinostat in postmenopausal women with locally recurrent or metastatic estrogen receptor-positive breast cancer progressing on treatment with a nonsteroidal aromatase inhibitor. Clin Oncol 2013;31:2128–2135.
66. Lee Y, Ahn C, Han J, et al. The nuclear RNase III Drosha initiates micro-RNA processing. Nature 2003;425:415–419.
67. Reinhart BJ, Slack FJ, Basson M, et al. The 21-nucleotide let-7 RNA regulates developmental timing in Caenorhabditis elegans. Nature 2000;403:901–906.
68. Camps C, Buffa FM, Colella S, et al. hsa-miR-210 Is induced by hypoxia and is an independent prognostic factor in breast cancer. Clin Cancer Res 2008;5:1340–1348.
69. Foekens JA, Sieuwerts AM, Smid M, et al. Four miRNAs associated with aggressiveness of lymph node-negative, estrogen receptor-positive human breast cancer. Proc Natl Acad Sci U S A 2008;105:13021–13026.
70. Iorio MV, Casalini P, Tagliabue E, et al. MicroRNA profiling as a tool to understand prognosis, therapy response and resistance in breast cancer. Eur J Cancer 2008;18:2753–2759.
71. Kovalchuk O, Filkowski J, Meservy J, et al. Involvement of microRNA-451 in resistance of the MCF-7 breast cancer cells to chemotherapeutic drug doxorubicin. Mol Cancer Ther 2008;7:2152–2159.
72. Miller TE, Ghoshal K, Ramaswamy B, et al. MicroRNA-221/222 confers tamoxifen resistance in breast cancer by targeting p27Kip1. J Biol Chem 2008;283:29897–29903.
73. Hanahan D, Weinberg RA. The hallmarks of cancer. Cell 2000;100:57–70.
74. Oesterreich S, Davidson NE. The search for ESR1 mutations in breast cancer. Nature Genet 2013;45:1415–1416.
75. Massarweh S, Schiff R. Unraveling the mechanisms of endocrine resistance in breast cancer: new therapeutic opportunities. Clin Cancer Res 2007;13:1950–1954.
76. Davies C, Pan H, Godwin J, et al. Long-term effects of continuing adjuvant tamoxifen to 10 years versus stopping at 5 years after diagnosis of oestrogen receptor-positive breast cancer: ATLAS, a randomised trial. Lancet 2013;9869:805–816.
77. Baselga J, Campone M, Piccart M, et al. Everolimus in postmenopausal hormone-receptor-positive advanced breast cancer. N Engl J Med 2012;366:520–529.
78. Yarden Y, Sliwkowski MX. Untangling the ErbB signalling network. Nat Rev Mol Cell Biol 2001;2:127–137.
79. Hudis CA. Trastuzumab—mechanism of action and use in clinical practice. N Engl J Med 2007;1:39–51.
80. Jhaveri K, Esteva FJ. Pertuzumab in the treatment of HER2+ breast cancer. J Natl Compr Canc Netw 2014;12:591–598.
81. Ellis LM, Hicklin DJ. VEGF-targeted therapy: mechanisms of anti-tumour activity. Nat Rev Cancer 2008;8:579–591.
82. Hynes NE, Dey JH. PI3K inhibition overcomes trastuzumab resistance: blockade of ErbB2/ErbB3 is not always enough. Cancer Cell 2009;5:353–355.