Deciphering and Targeting Oncogenic Mutations and Pathways in Breast Cancer

Abstract

Advances in DNA and RNA sequencing revealed substantially greater genomic complexity in breast cancer than simple models of a few driver mutations would suggest. Only very few, recurrent mutations or copy-number variations in cancer-causing genes have been identified. The two most common alterations in breast cancer are TP53 (affecting the majority of triple-negative breast cancers) and PIK3CA (affecting almost half of estrogen receptor-positive cancers) mutations, followed by a long tail of individually rare mutations affecting <1%–20% of cases. Each cancer harbors from a few dozen to a few hundred potentially high-functional impact somatic variants, along with a much larger number of potentially high-functional impact germline variants. It is likely that it is the combined effect of all genomic variations that drives the clinical behavior of a given cancer. Furthermore, entirely new classes of oncogenic events are being discovered in the noncoding areas of the genome and in noncoding RNA species driven by errors in RNA editing. In light of this complexity, it is not unexpected that, with the exception of HER2 amplification, no robust molecular predictors of benefit from targeted therapies have been identified. In this review, we summarize the current genomic portrait of breast cancer, focusing on genetic aberrations that are actively being targeted with investigational drugs.

Implications for Practice:

Next-generation sequencing is now widely available in the clinic, but interpretation of the results is challenging, and its impact on treatment selection is often limited. This work provides an overview of frequently encountered molecular abnormalities in breast cancer and discusses their potential therapeutic implications. This review emphasizes the importance of administering investigational targeted therapies, or off-label use of approved targeted drugs, in the context of a formal clinical trial or registry programs to facilitate learning about the clinical utility of tumor target profiling.

Introduction

Breast cancer represents a complex and heterogeneous disease, encompassing several molecularly and clinically distinct entities that can be classified according to gene expression profiles and clinicopathological features [1, 2]. Hereditary disease accounts only for a small percentage of all breast tumors, whereas the majority of breast cancers are sporadic, resulting from the accumulation of acquired somatic alterations [3]. Recent advances in sequencing-based technologies have increased our understanding of genetic aberrations and dysregulated oncogenic pathways, involving growth signaling, stress response, metabolism, and cell-to-cell communication, which affect breast cancer development and progression. These somatic alterations, together with the host response to cancer, determine the clinical course of the disease. A large number of often individually rare, but potentially functionally important, molecular alterations are encountered in unique combinations in every cancer. How to best exploit the genomic anatomy of cancer for therapeutic advantage remains an important unanswered challenge [4–7]. In this review, we provide a portrait of currently known genomic alterations and related pathways in breast cancer.

Materials and Methods

We conducted a review of the literature using PubMed, Web of Science, and Embase databases from 2000 to May 2016 using the search terms “breast cancer,” “genetic mutations,” “genetic alterations,” “biomarkers,” “response to therapy,” “tumor heterogeneity,” and “targeted therapies.” Additional studies were identified through the references listed in review publications. Recent emerging clinical findings presented at scientific meetings in the form of abstracts were also reviewed and incorporated. We note that the reported frequency for gene-level mutations depends on platform, sequencing depth, sample size, and definition of actionability, which vary from study to study. In this review, we use the frequency numbers from the largest studies that we could identify. We also searched clinicaltrials.gov to identify relevant ongoing phase II–III clinical trials evaluating targeted agents in patients with breast cancer.

The Genomic Landscape of Breast Cancers

The genetic basis of susceptibility to breast cancer is related to multiple germline variations at different loci. These alleles have broadly variable frequencies and confer different levels of risk, leading to a classification as high-, moderate-, and low-penetrance alleles (Fig. 1). Even though the major rare high-penetrance genes, BRCA1 and BRCA2, which are involved in DNA repair processes, account for approximately 50% and 30% of familial breast cancers, respectively, further efforts are needed to unravel the impact of variants of unknown pathological significance or other susceptibility genes [3].

Sporadic breast cancers develop through the acquisition of novel malignant traits that confer a selective advantage to tumor cells. Although the accumulation of such genomic alterations is a requisite for the expansion and progression of a transformed cell population, recent insights that have emerged from comprehensive genomic studies have revealed additional heterogeneity in the genomic landscape of breast cancer [8–12]. The diversity of genetic pattern among breast tumors may explain the wide differences in biological features, clinical behavior, and response to therapies. The analysis of primary breast tumors has demonstrated that missense mutations are more common in estrogen receptor (ER)-positive/luminal and human epidermal growth factor receptor 2 (HER2)-positive subtypes, whereas triple-negative breast cancers (TNBCs) are enriched for nonsense, frameshift, and complex mutations (Fig. 2A) [13]. Similarly, a higher rate of genomic rearrangements was found in HER2-positive breast cancers and TNBCs [14]. Breast cancer subtypes also differ regarding the pattern of copy-number alterations (CNAs) (Table 1) [11]. Novel data on genomic spectra, signaling pathways, and gene expression profiles are leading to the refinement of the molecular classification of breast cancer (Fig. 2B; Table 1) [8–12, 14–18].

Estrogen Receptor-Positive/Luminal Breast Cancers

ER-positive/luminal breast cancers have the highest number of recurrent mutations (Fig. 2B; Table 1) [9, 14]. In luminal A cancers, the most frequently mutated gene is PIK3CA (45%), followed by GATA3 (14%), MAP3K1 (13%), TP53 (12%), CDH1 (9%), MLL3 (8%), MAP2K4 (7%), NCOR1 (5%), and RUNX1 (5%) [9]. The PIK3CA and GATA3 mutations appear mutually exclusive, with the much rarer mutations in AKT1 (4%) and FOXA1 (2%), respectively [9, 19]. Luminal B tumors have a lower frequency of PIK3CA mutations (29%) and a slightly higher frequency of GATA3 mutations (15%) (Fig. 2B; Table 1) [9]. Conversely, the frequency of TP53 mutations (29%) is higher in luminal B cancers and is associated with worse outcome [20]. There is also higher frequency of ATM loss and CCND1, CDK4, CDK6, and MDM2 amplifications in luminal B cancers [9]. ER-positive cancers have relatively few recurrent CNAs compared with TNBCs (Table 1) [9, 10, 21, 22]. A common and potentially clinically relevant event in luminal B tumors is the amplification of fibroblast growth factor receptors (FGFRs) [9, 22].

HER2-Positive Breast Cancers

A core set of genomic regions consistently affected by CNAs has been identified in HER2-positive breast cancers (Table 1) [23]. In addition to the common amplification at 17q12 (containing the HER2 oncogene), other recurrent CNAs included a large number of gains, losses, and amplifications [24]. A source of genomic heterogeneity within HER2-positive tumors is dependent on the size of the HER2 amplicon, because multiple genes can be coamplified with HER2 [11]. Beyond the core of the amplicon, which includes at least HER2-C17orf37-GRB7 genes, coamplifications of other genes have been described [24, 25]. Particularly, the amplification of the HER2/TOP2A region has been correlated with the overexpression of several additional genes, such as CASC3, CDC6, RARA, and SMARCE1 [26]. It is worth noting that 20%–40% of HER2-positive cancers are TOP2A-deleted and that alterations in other growth factor receptors have been associated with the HER2 status, including EGFR, FGFR, and HER3 [9, 17, 27].

Triple-Negative Breast Cancers

TNBCs represent a molecularly highly heterogeneous group with substantial transcriptional and DNA copy number variability [28]. These cancers have the highest frequency of TP53 disabling mutations (up to 80%) [9, 20]. Mutations and/or deletions in RB1 (retinoblastoma 1; 20%) and mutations in PIK3CA (9%), MLL3 (5%), and GATA3 (2%), as well as amplification of the CCNE1 gene (9%), occur at relatively low frequencies (Fig. 2B) [9]. Other repeatedly observed, but rather rare, genetic aberrations include mutations in ATR, COL6A3, FBXW7, MYO3A, PARK2, SYNE1/2, and USH2A genes [17, 29]. Despite the overall low prevalence of PIK3CA mutations, the activation of the phosphatidylinositol 3-kinase (PI3K) signaling has been found in specific TNBC subtypes, especially because of loss of PTEN and INPP4B, or PIK3R1 mutations [9, 30]. The genomic landscape of TNBCs is characterized by frequent and numerous DNA copy number losses and gains, suggesting a form of genomic instability (Table 1) [9, 11, 31–33]. Gains/amplifications and overexpression of BRAF, CDK6, CDKN2A, CCNE1, E2F, EGFR, FGFR1, FGFR2, IGFR1, KIT, MET, MYC, PDGFRA, and PIK3CA are also often encountered in TNBC [9, 32, 33].

Intratumor Genomic Heterogeneity and Implications for Treatment

Tumor heterogeneity creates a potential framework to explain several clinical features of cancer [34–38]. Clinicians have long noticed variable tumor responses in different metastatic sites, suggesting variable treatment sensitivity in different locations. Within-tumor clonal heterogeneity and variable clonal composition of different metastatic lesions provide a simple explanation for this clinical phenomenon and is supported by recent results from tumor sequencing of metastatic sites and matching primary tumors. Clonal selection in response to therapy has also been demonstrated in several different cancer types and provides explanation for the development of treatment resistance over time [37, 39–41]. The evolutionary relatedness of primary tumors and multiple metastases from the same patient suggests multiple different ways that metastasis can arise (Fig. 3A) [42, 43]. Most metastatic lesions show large degrees of similarity to the primary tumor, indicating direct seeding from the primary tumor followed by modest genetic drift, whereas other lesions are more similar to one or more other metastatic lesions, suggesting that a metastatic site could give rise to further metastatic lesions (Fig. 3A) [42, 43]. Our understanding of how clinical drug resistance emerges during therapy has also evolved (Fig. 3B; Table 2) [44]. Some forms of resistance emerge through selection of intrinsically resistant subclones present in the cancer as a minority clone from the beginning (Fig. 3B) [12, 69, 73–75]. Other forms of resistance arise through de novo mutations that create new clones with drug resistance, not originally present in the cancer (Fig. 3B) [74, 76]. Several studies showed that appearance of new mutations, undetectable in the primary tumor, can arise after neoadjuvant chemotherapy (PTEN and TP53), HER2-targeted therapies (PIK3CA), and during endocrine therapy (ESR1) for metastatic breast cancer [50–54, 77–79]. Even though the relative proportion of drug-resistant subclones is generally higher in the post-treatment specimens, retreating cancers with drugs on which they previously progressed often continues to produce tumor response. Selectively removing a drug-sensitive subpopulation can lead to a temporary collapse of the entire population, but eventual recovery happens through new subclones stepping in to provide the critical function for the entire population [80]. Considering cancer as a society of neoplastic cells will allow for exploring new therapeutic strategies that could target between-cell communication or interrupt the complementary functions cells provide to each another.

The evolutionary relatedness of primary tumors and multiple metastases from the same patient suggests multiple different ways that metastasis can arise.

Molecular Predictors of Response to Targeted Therapy in Breast Cancer Subtypes

The fundamental premise behind “precision medicine” and tumor target profiling is that the targetable molecular abnormalities that are detected in a cancer will serve as predictors of treatment response to the appropriate targeted therapy. Several clinical trials suggest that the relationship between benefit from targeted therapies and molecular defects is more complex than hoped for.

For example, the predictive value of the PI3K pathway activation (i.e., PIK3CA mutations and loss of PTEN) in HER2-overexpressing tumors has been analyzed in different neoadjuvant clinical trials, generating variable results, leaving HER2 amplification as the only biomarker associated with anti-HER2 therapies benefit [64–68, 81, 82]. It is noteworthy that recent data have demonstrated the presence of HER2 mutations in nonamplified HER2 tumors, potentially advancing the development of novel anti-HER2 drugs in this clinical setting [83].

ER-positive tumors are also highly dependent on the PI3K signaling for cell growth and survival [84]. Everolimus, an mTOR (mammalian target of rapamycin) inhibitor that interrupts the PI3K-mediated signaling, is approved in combination with exemestane for the treatment of advanced postmenopausal ER-positive/HER2-negative breast cancer patients, but results from the BOLERO-2 trial showed that PIK3CA mutations did not predict benefit from everolimus. However, the PI3K assessment was performed on primary tumors rather than metastatic biopsies [85]. Conversely, in the same trial, circulating ESR1 mutations showed a potential predictive efficacy of the addition of everolimus to standard endocrine therapy [54, 85]. The addition of the AKT inhibitor MK-2206 to anastrozole also did not confer significant benefit to ER-positive/HER2-negative patients with PIK3CA mutations [86]. However, preliminary results from the phase III BELLE-2 trial indicated that only patients with circulating PIK3CA mutations benefited from a combination of the PI3K inhibitor BKM120 and fulvestrant [87]. Several larger phase II and III trials with α-specific PIK3CA inhibitors are still under way to further define the predictive function, and therapeutic target value, of PIK3CA mutations [88–90].

The CDK4/6 inhibitor PD-0332991 (palbociclib) was recently approved for the treatment of metastatic ER-positive breast cancer in combination with endocrine therapy, and several other CDK-targeted drugs are in clinical trials [91, 92]. Amplification of CDK4/6 and CCND1 and loss of the CDK4/6 inhibitors CDKN2A and CDKN2C are often detected in ER-positive cancers [9]. However, results from the PALOMA-1 trial showed that these genetic aberrations are not predictive for the efficacy of treatment with palbociclib [93].

For TNBC, there are no targeted therapies specifically approved [47]. However, recent data suggest a significant clinical activity of poly(ADP-ribose) polymerase (PARP) inhibitors in germline BRCA mutant breast cancer with dysfunctional homologous recombination DNA repair [47–49].

Potential New Therapeutic Targets and Treatment Strategies Derived From Genomic Analysis of Breast Cancer

Despite very rare driver mutations that have been found in directly targetable genes (e.g., kinase genes), limiting the development of genotype targeted-therapies in breast cancer, several agents are currently used in the clinic, whereas others are not yet approved for clinical practice or are still under evaluation (Table 3). Furthermore, results derived from next-generation sequencing (NGS) studies have demonstrated novel genetic aberrations and pathways dysregulation that may represent future therapeutic targets (Table 4).

ESR1 mutations, leading to ligand-independent signaling, commonly arise under the selective pressure of endocrine therapy during the treatment of metastatic disease [53]. Several novel selective estrogen receptor degraders are in phase I (i.e., GDC0927, AZD9496, and RAD1901) and phase II (GDC0810) clinical trials that may turn out to be effective new therapies for these ESR1 mutant cancers (Table 3). Similar mutations conferring ligand-independent activity to HER2, in the absence of HER2 amplification, were also reported, and several clinical trials are testing the efficacy of HER2 kinase inhibitors in these patients, who are currently considered “HER2 normal” by immunohistochemistry of fluorescence in situ hybridization assay.

It is also increasingly recognized that PIK3CA inhibitors alone may be not sufficient to keep this important pathway inhibited because of redundancies, feedback loops, and simultaneous mutations in downstream pathway members. Indeed, concurrent perturbations of the PI3K/AKT/mTOR and the mitogen-activated protein kinase (MAPK) cascades are frequently observed in tumors. Several inhibitors of the Raf/MEK/ERK pathway are currently under clinical evaluation alone and in combination with PIK3CA/mTOR inhibitors (Table 3) [9, 58, 94, 95]. Furthermore, PI3K inhibition has been shown to boost the ERK pathway through the enhancement of the epidermal growth factor receptor (EGFR)/HER2/human epidermal growth factor receptor 3 (HER3) signaling, and HER3 may in turn promote the reactivation of the PI3K/AKT and MAPK pathways [96, 97]. The modest activity of EGFR inhibitors in TNBC may also be due to crosstalk between signaling pathways and compensatory feedback loops [98–101]. Thus, combination trials with other drugs, including NOTCH and AXL inhibitors, warrant further clinical investigation [99–101]. Amplifications of FGFR1 and the activation of related pathways have been observed in TNBC [9]. Dovitinib and lucitanib showed antitumor activity, and clinical trials in fibroblast growth factor (FGF)-deregulated breast cancer are ongoing (Table 3). However, the therapeutic potential of FGFR inhibition in breast cancer requires further evaluation, in light of uncertain clinical results, and the key role of FGFRs in ordinary cell physiology [102].

An emerging success story in targeted therapies is the significant clinical activity of PARP inhibitors in germline BRCA mutant breast cancer [45, 103, 104]. Several trials showed single-agent activity in BRCA mutant breast cancers, and numerous trials, including a large adjuvant trial with olaparib, are under way (Table 3).

APOBEC mutations can lead to instability and mutagenesis in breast cancer [105, 106]. The APOBEC3-mediated mutagenesis results in a specific pattern of nucleic acid changes, called APOBEC mutation signature, which has been found to be enriched in HER2-overexpressed, basal-like, and luminal B subtypes. This evidence suggests that APOBECs could be crucial factors for targeted therapies aiming to block mechanisms producing widespread genomic alterations that drive tumor development, progression, and likely therapy resistance [105, 106]. Besides APOBEC enzymes, other mutational processes involving specific DNA repair pathways (i.e., mismatch repair and homologous recombination) lead to distinctive DNA damage [107, 108]. The identification of different genomic signatures arising from distinct defective pathways can provide a powerful means of revealing the somatic mutational basis of breast cancer [107, 108].

Recurrent mutations of TP53 make this gene a target of particular interest in TNBC. Considering that the direct targeting of TP53 is challenging, an alternative option may be to focus on activated signaling molecules downstream of TP53. For instance, CHEK1/2, ATM, and ATR are emerging as promising targets (Tables 3, 4) [109]. TP53-deficient cells rely on ATR/CHEK1 to arrest cell-cycle progression after DNA-damaging chemotherapy. Thus, the inhibition of CHEK1 and/or ATR in TP53-deficient cancers may lead to mitotic catastrophe and apoptosis [103, 104]. Several other TP53-associated kinases, such as polo-like kinases, aurora kinases, and, recently, the MPS1 kinase, have shown promising results in vitro and in vivo, suggesting their potential as suitable candidates for further development toward clinical studies (Table 4) [110]. The TP53-associated pathway is also defective in an important quote of luminal B tumors because of TP53 mutations, amplification of MDM2, or ATM loss [9]. Several molecules that inhibit the ubiquitin ligase MDM2, or that disrupt the MDM2-TP53 interaction, are currently in early clinical evaluation in patients with advanced malignancies other than breast cancer [111]. Other proteins involved in the ubiquitin-proteasome system, such as the tumor suppressor FBXW7 and the SKP1 ligase, have been found to be mutated or dysregulated in breast cancer (especially TNBC/basal-like), affecting a complex network of oncogenic processes, including cell differentiation and genomic stability [17, 112, 113]. In particular, loss of SKP1 has been shown to deregulate multiple targets, including EGFR and SRC tyrosine kinase in basal-like breast cancer [113]. An integrated genomic-proteomic approach has recently allowed the identification of novel phosphosite markers in PIK3CA- (i.e., EIF2AK4 and RPS6KA5) and TP53- (i.e., CHEK2, EEF2K, and MASTL) mutated tumors, and other important amplicon-associated phosphorylated kinases (i.e., CDK12, PAK1, PTK2, RIPK2 and TLK2) (Table 4) [113]. Overall, the analysis of activated pathways and the understanding of dysregulated signaling circuitries can provide novel insights into the functional consequences of somatic mutations and help identify novel therapeutic targets in breast cancer.

Overall, the analysis of activated pathways and the understanding of dysregulated signaling circuitries can provide novel insights into the functional consequences of somatic mutations and help identify novel therapeutic targets in breast cancer.

Genes identified as oncogenic drivers in other tumor types may also represent potential therapeutic targets. For instance, tumors carrying activating mutations and amplification in BRAF, KRAS, and MET genes might be sensitive to specific inhibitors. Furthermore, other potential driver mutations in genes that are involved in chromatin remodeling and epigenetic regulation of gene expression have been reported in breast cancer [8, 9, 17, 114]. Patients with such mutations may be potentially enrolled in clinical trials using drugs that target epigenetic modifications, particularly DNA methyltransferase and histone deacetylase inhibitors (Tables 3, 4).

Tumor cells dynamically interact with a heterogeneous microenvironment through a bidirectional signaling network. An outstanding ability of the immune system, especially adaptive immunity, is to recognize tumor-specific antigen. Different gene mutations generate new antigenic peptides that could be recognized by T lymphocytes and trigger antitumor immune responses [115]. The identification of specific antigens resulting from mutations in driver genes, as well as chromosomal translocations and gene amplifications, could provide the rationale for the development of novel immunotherapeutic strategies for patients with breast cancers [116]. Remarkably, the majority of mutated antigens contributing to the immunogenicity of human tumors results from alterations in nonrecurrent genes, suggesting that infrequent mutations identified from large-scale sequencing studies may represent tumor-specific antigens and hold promise for use in cancer immunotherapy (Table 4) [115].

Discussion

Identifying the functional relevance of individually often-rare variants is a daunting challenge for laboratory investigators and clinical trialists. Studying the functional interaction between the multiple somatic alterations and inherited germline polymorphisms is going to be key for the formulation of a more comprehensive model of carcinogenesis. The importance of interaction between the host immune system and the cancer is also increasingly recognized as an important determinant of clinical outcome. Somatic mutations, even without functional impact on biological pathways, can represent novel antigens that could activate adaptive immunity in the tumor microenvironment. A major challenge in the implementation of molecular target-directed therapeutic trials is the rarity and diversity of potentially actionable mutations in breast cancer, so that very large numbers of patients need to be screened to find the 20–100 patients required for a typical phase II trial [117–119].

The presence of multiple potentially targetable alterations in most cancers suggests that individualized combination therapies may be the most effective strategy, which presents a fundamental challenge to any traditional clinical trial design and raises unprecedented regulatory issues. Furthermore, the repertoire of biologically and therapeutically important mutations may change during the evolution of a cancer or perhaps even from metastatic site to site. Although several studies have demonstrated the existence of “private” mutations (i.e., mutations unique to a particular tumor site while absent in other sites), the biological and therapeutic relevance of these observations is unknown.

Molecular alterations are also being investigated as response predictors to targeted therapies. However, no single gene alterations emerged as clinically useful predictors to any therapy (other than HER2 amplification for HER2-targeted therapies). There is only weak, or no, association between PIK3CA mutations and response to mTOR or PI3K inhibitors in clinical trials. Similarly, amplifications in the CDK genes were not predictive of benefit from CDK4/6 inhibitors, and FGFR inhibitors also had only modest activity in FGFR1-amplified tumors. Conversely, activating mutations in HER2 emerged as a potential new marker and target to extend the use of anti-HER2 kinase inhibitor therapy. Several ongoing clinical trials are examining the therapeutic target value of this abnormality. ESR1 mutations that render the receptor constitutively active are also frequently encountered in metastatic biopsies, whereas this mutation is very rare in primary cancers. Efforts are under way to develop therapies that target the mutant ER.

Some, but not all, of the somatic mutations present in a cancer can also be detected as circulating-free tumor DNA. It is hoped that sequencing of circulating tumor DNA that can easily be obtained from blood may provide a tool to monitor risk of recurrence after potentially curative therapy, detect early recurrence, and monitor response to therapy [120–124].

Besides the importance of variants in protein-coding regions, the majority of the alterations occur in noncoding portions of the genome [125]. Different noncoding sequence variants, ranging from single-nucleotide variants to small insertions and deletions, to large structural variants have been demonstrated in human cancer [125–127]. Somatic variations can lead to gain of transcription factor-binding sites responsible of tumorigenesis. Fusion events because of genomic rearrangements, variations in noncoding RNAs and their binding sites, and variants in pseudogenes modulating gene expression have been linked to cancer [125]. Like somatic variants, germline noncoding affects gene expression through several mechanisms (e.g., promoter mutations, single-nucleotide polymorphisms in enhancers and noncoding RNAs and their binding sites, and variants in introns) [125]. Recently, recurrent mutations have been found in the promoter of TBC1D12 and WDR74, as well as in the long noncoding RNAs MALAT1 and NEAT1 in breast cancer [108]. However, the identification of noncoding drivers is complex and requires additional studies. Overall, it is evident that cancer results from a more complex interplay of inherited germline and acquired somatic variants, including also noncoding region alterations. As for breast cancer, which seems to harbor more alterations in the noncoding regions compared with other tumors, the identification of noncoding driver mutations/alterations remains crucial to enable therapeutic approaches that target the specific linked proteins. The constant evolution of technologies allows the ever-increasing identification of novel mutated cancer genes (e.g., FOXP1, MED23, MLLT4, XBP1, and ZFP36L1) [108]. This ongoing update of cancer genes and noncoding regions’ alterations progresses toward a comprehensive understanding of the source and consequences of somatic mutations, ultimately enabling the development of novel targeted therapies for breast cancer.

Conclusion

Recent insights from comprehensive genomic characterizations of cancer revealed large-scale heterogeneity in the genomic landscape of breast cancers. The diversity of genomic aberrations and the resulting dysregulation of a broad range of biological pathways help to explain the varied clinical behavior of breast tumors. Better understanding the biological effects of these aberrations remains a major focus of research and may pave the way for rational combination targeted therapies.

Acknowledgments

This work was supported by Associazione Italiana per la Ricerca sul Cancro Grant 6251 to L.S.; Fondazione Italiana per la Ricerca sul Cancro Fellowship 18328 to G.B.; and the Breast Cancer Research Foundation to L.P.

Author Contributions

Conception/Design: Libero Santarpia, Lajos Pusztai

Provision of study material or patients: Libero Santarpia, Giulia Bottai, Catherine M. Kelly, Balázs Győrffy, Borbala Székely, Lajos Pusztai

Collection and/or assembly of data: Libero Santarpia, Giulia Bottai, Catherine M. Kelly, Balázs Győrffy, Lajos Pusztai

Data analysis and interpretation: Libero Santarpia, Giulia Bottai, Catherine M. Kelly, Balázs Győrffy, Borbala Székely, Lajos Pusztai

Manuscript writing: Libero Santarpia, Lajos Pusztai

Final approval of manuscript: Libero Santarpia, Giulia Bottai, Catherine M. Kelly, Balázs Győrffy, Borbala Székely, Lajos Pusztai

Disclosures

The authors indicated no financial relationships.

References

Author notes