Tel.: 353-1-7165814, Fax: 353-1-2214428
Targeted therapy for triple-negative breast cancer: Where are we?
Version of Record online: 7 JUN 2012
Copyright © 2012 UICC
International Journal of Cancer
Volume 131, Issue 11, pages 2471–2477, 1 December 2012
How to Cite
Duffy, M. J., McGowan, P. M. and Crown, J. (2012), Targeted therapy for triple-negative breast cancer: Where are we?. Int. J. Cancer, 131: 2471–2477. doi: 10.1002/ijc.27632
- Issue online: 25 SEP 2012
- Version of Record online: 7 JUN 2012
- Accepted manuscript online: 14 MAY 2012 04:45AM EST
- Manuscript Accepted: 5 APR 2012
- Manuscript Received: 23 JAN 2012
- Science Foundation Ireland, Strategic Research Cluster Award. Grant Number: 08/SRC/B1410
- Molecular Therapeutics for Cancer Ireland, the Health Research Board Clinician Scientist Award. Grant Number: CSA/2007/11
- Cancer Clinical Research Trust
- triple-negative breast cancer;
- predictive markers;
Breast cancers that are negative for estrogen receptor (ER), progesterone receptors (PR) and HER2, using standard clinical assays, have been dubbed triple-negative (TN). Unlike other molecular subtypes of invasive breast cancer, validated targeted therapies are currently unavailable for patients with TN breast cancer. Preclinical studies however, have identified several potential targets such as epidermal growth factor receptor (EGFR), SRC, MET and poly ADP ribose polymerase 1/2 (PARP1/2). Because of tumor heterogeneity, it is unlikely that any single targeted therapy will be efficacious in all patients with TN breast cancer. The rational way forward for treating these patients is likely to be biomarker-driven, combination targeted therapies or combination of targeted therapy with cytotoxic chemotherapy.
Multiple classification systems have been proposed for breast cancer. Traditionally, these were based on criteria such as histology type, tumor grade, tumor size and presence or absence of lymph node metastasis. The most recent approach to the classification of breast cancer is based on molecular profiling, especially the expression of mRNA species. Thus, using gene expression profiling, several subtypes of breast cancer have been proposed.1, 2 These include luminal subtype A, luminal subtype B, HER2 subtype, normal breast subtype, basal subtype and claudin-low subtype.3–5
A large proportion of basal and claudin-low breast cancers were found to be negative for estrogen receptor (ER) and progesterone receptors (PR) and to lack overexpression/amplification of HER2. Breast cancers lacking these three markers were subsequently dubbed triple-negative (TN) tumors. Triple-negativity was later proposed as a surrogate for the basal subtype, although it should be pointed out that not all basal breast cancers are TN tumors and conversely, not all TN tumors are basal type. Overall, ∼80% of basal cell cancers are TN and ∼80% of TN exhibits a basal-like phenotype.6–8 TN tumors, however, are not a homogeneous group but differ in their gene expression, response to chemotherapeutic and biological therapies as well as prognosis.6–8
Although responsible for a minority (∼15% in Caucasian populations) of breast cancers, TN tumors cause a disproportionate number of breast cancer deaths. This poor prognosis is due to intrinsic aggressiveness and lack of treatment options, especially targeted therapies.6, 8 Currently, the only form of systemic therapy currently available for patients with TN disease is cytotoxic chemotherapy. Although some patients with TN breast cancers are initially relatively chemosensitive, and indeed, possibly more responsive to certain forms of chemotherapy than non-TN breast cancer, they nevertheless have a worse outcome.6, 8 Consequently, there is intense interest in identifying new targets for treating patients with TN breast cancer.
In recent years, several molecular alterations have been described in TN/basal breast cancer.6–8 The aim of this article is to discuss these alterations as possible targets for developing new treatments for patients with this form of breast cancer. In addition, we discuss potential predictive markers for several of the therapies undergoing preclinical and clinical evaluation.
Potential Targets for Treating TN Breast Cancer
EGFR is one of four members of the HER family of receptors, the other family members being HER2, HER3 and HER4 (for review, see Ref. 9). All four proteins share a common structure that consists of a ligand-binding extracellular domain, a transmembrane domain, an intracellular kinase domain and a C-terminal signaling sequence. At least three of the HER family members, i.e., EGFR, HER2, HER3 have been implicated in cancer formation and/or progression, although the specific family member involved may vary with the tumor type. Thus, in breast cancer, the HER family member most frequently activated is HER2. In invasive breast cancer, this gene is either amplified or overexpressed in ∼15%–20% of newly identified cases. Patients with overexpression of the HER2 protein are usually treated with trastuzumab and lapatinib.10
Although TN breast cancers do not overexpress HER2, high expression of EGFR protein is present in a subset of these tumors. Indeed, expression of EGFR has been found to be higher in basal/TN breast cancer cell lines and breast cancers than in other types of breast cancer cells.11–16 Furthermore, phosphorylation of EGFR was shown to be higher in both basal/TN cell lines and human breast cancers than in luminal types, suggesting more active signaling.17 Additional evidence of a role for EGFR in TN breast cancers was the finding that high expression predicted adverse patient outcome in patients with this form of the disease.16
As with HER2, a number of therapies are currently available that target EGFR.10 These include monoclonal antibodies such as cetuximab and panitumumab, which prevent ligand binding and the tyrosine kinase inhibitors, gefitinib and erlotinib. Cetuximab and panitumumab are approved for the treatment of advanced colorectal cancers (CRC) lacking KRAS mutations,10 while gefitinib and erlotinib may be used for treating advanced non-small cell lung cancers (NSCLC) expressing certain mutant forms of EGFR.10
Several of these anti-EGFR therapies have undergone preclinical evaluation for potential anticancer activity in basal/TN breast cancer cells.11, 18, 19 At least two preclinical studies have shown that basal/TN cell lines are more sensitive to gefitinib and erlotinib than luminal cell lines.11, 18 Furthermore, in the basal/TN cell lines, synergy was observed for the combination of gefitinib and carboplatin,11, 19 as well as between docetaxel and gefitinib.19
Sensitivity to cetuximab in basal/TN cell lines, however, seems to be poor.11, 19 Thus, Hoadley et al.11 found that only one of the basal/TN cell line investigated, i.e., SUM-102, was sensitive to this monoclonal antibody. In this cell line, the combination of cetuximab and carboplatin was synergistic at low doses of each drug after long-term treatment. Short-term cotreatment with cetuximab and the platinum compound, however, was antagonistic. Unlike the report of Hoadley et al.,11 Corkery et al.19 were unable to show growth inhibitory effects for cetuximab on any of the TN cell lines investigated. Why gefitinib and cetuximab have different inhibitory effects is unclear.
Despite the paucity of preclinical data showing anticancer activity in basal/TN breast cancer cells, a number of clinical trials have investigated the use of cetuximab in patients with TN breast cancer. In one of these studies, patients with pretreated metastatic breast cancer were randomized to receive irinotecan and carboplatin with or without cetuximab.20 Response rates were superior with the combined antibody cytotoxic drug approach than with the chemotherapy alone (49% vs. 30%) in the TN subset of patients. Progression-free survival, however, was not significantly different in the two groups. In a second Phase II study involving TN breast cancer, use of cetuximab combined with carboplatin gave a response rate of 17%, compared to only 6% with cetuximab alone.21 In a further Phase II study, the addition of cetuximab to cisplatin increased both overall response rate (20% vs. 10%; p = 0.11) and progression-free survival (3.7 vs. 1.5 months; HR = 0.68, p = 0.032) compared to cisplatin alone.22 Overall survival data from this trial has not yet been published.
The above findings suggest modest benefit from cetuximab in unselected TN breast cancer patients. However, greater benefit from therapy with the anti-EGFR antibody may exist for a subset of TN patients. Indeed, the key to the success of anti-EGFR therapy in patients with TNBC may depend on the use of biomarkers for predicting benefit. Extrapolating from studies in breast cancer, where HER2 levels were found to be predictive of response to the therapeutic monoclonal antibody, trastuzumab, it might be expected that levels of EGFR would predict response to anti-EGFR therapies. However, studies in both CRC and NSCLC have shown no significant relationship between levels of EGFR, as detected by immunohistochemistry and benefit from anti-EGFR therapies.23–26 Rather, measurement of specific mutations in the EGFR gene is used to predict response to the anti-EGFR tyrosine kinase inhibitors, gefitinib and erlotinib in patients with NSCLC, while specific mutations in KRAS are used to identify patients with advanced CRC, who are unlikely to benefit from cetuximab and panitumumab.10, 18, 24
The predictive markers used for anti-EGFR therapies in NSCLC and CRC, however, are unlikely to be of value in TN breast cancer as neither EGFR nor KRAS mutations appear to frequent in this malignancy. A search for predictive markers for anti-EGFR therapies in breast cancer is thus likely to focus on proteins downstream of EGFR. In this context, it should be noted that the absence of PTEN, a downstream negative regulator of cell signaling, was found to be associated with resistance to anti-EGFR therapies in patients with NSCLC and CRC.27–29 In recent studies, deletion of PTEN was found in ∼30% of TN breast cancer.30, 31 Moreover, loss was found to more frequent in TN than non-TN breast cancer.31 Provided that the preliminary findings for CRC and NSCLC can be confirmed and extrapolated to other malignancies, PTEN status is a potential predictive marker for anti-EGFR therapies in patients with TN breast cancer.
A further approach for enhancing the efficacy of anti-EGFR therapy in patients with TN breast cancer may be combination with a synthetic lethal partner. Synthetic lethality occurs between two molecules when loss of one function is compatible with viability but loss of both is detrimental. Evidence that this approach has clinical use comes from recent findings with PARP inhibitors in patients with BRCA1/2 mutation-associated breast and ovarian cancer, see below. Thus, identification of synthetic lethal partners could potentially improve response to EGFR antagonists.In an attempt to identify potential synthetic lethal partners for anti-EGFR agents, Astsaturov et al.32 treated tumor cells with > 600 siRNAs against EGFR signaling- proteins. Knockdown of > 60 genes sensitized the cell lines to anti-EGFR therapy. Targeting the protein products of three of these genes, i.e., protein kinase C, aurora kinase A and STAT3, acted synergistically with anti-EGFR treatments in reducing cell viability in vitro and tumor size in an animal model. Inhibition of other targets such as NOTCH,33 SRC17 and mTOR34 has also been shown to synergize with anti-EGFR therapies in TN breast cancer cells.
Like the EGFR protein, MET is a transmembrane receptor with an intracellular tyrosine kinase domain.35 However, unlike EGFR, the mature form of MET exists as a disulphide-linked heterodimer with an extracellular α chain and a transmembrane β chain. Also unlike EGFR which has multiple ligands, met seems to have only one, i.e., hepatocyte growth factor (HGF). Following HGF binding to met, autophosphorylation occurs on tyrosine residues Y1234 and Y1235 within the TK domain, thus activating kinase activity.35 Further phosphorylation, i.e., on Y1349 and Y1356, which are adjacent to the carboxy terminal end, generates docking sites for adaptors, resulting in downstream signaling. Activation of MET may be brought about by mutation, gene amplification, overexpression or by high levels of HGF. Activation results in increased proliferation, increased migration and decreased cell death.35
As with EGFR and MEK, considerable evidence has linked met with basal/TN breast cancers. Thus, two independent studies showed that expression of oncogenic forms of this gene in mice, induced breast tumors that were mostly of the basal type.36, 37 Consistent with this, both MET and phosphorylated MET were found at higher levels in basal compared to luminal cancer cell lines.17 Furthermore, in human breast cancers and cell lines, overexpression of MET was preferentially but not exclusively associated with the basal subtype.37, 38 Indeed, in one study, high MET levels predicted adverse patient outcome.36 Taken together, these findings implicate MET in the induction and/or progression of basal/TN breast cancers. MET may thus be a further new target for treating patients with this form of breast cancer.39
Several different strategies have been proposed for inhibiting HGF-MET signaling.40, 41 These include HGF competitive analogues that compete with ligand for binding to MET but that are unable to initiate signaling, antibodies against either HGF or MET and TKIs. Most of these TKIs inhibit additional kinases as well as MET. Thus, crizotinib blocks both ALK rearranged proteins and MET and is currently undergoing clinical trials in patients with NSCLCs containing EML4-ALK translocations.42, 43 Indeed, crizotinib has been FDA approved for the treatment of ALK-positive locally advanced or metastatic NSCLC. To date, little work has been published on the potential use of these anti-MET agents to treat TN breast cancer.
c-SRC is a nonreceptor tyrosine kinase involved in signaling that culminates in the control of multiple biological functions including cell proliferation, cell differentiation, migration, angiogenesis and survival.44, 45 c-SRC is thus thought to play a key role in tumor formation and progression.44, 45 Several proteins with strong homology to SRC are present in mammalian cells, collectively known as the SRC family kinases (SFKs). These include in addition to SRC, FYN and YES, which are widely expressed, HCK, LCK, LYN, BLK and FGR, which are found mostly in haematopoietic cells and FRK-related kinases, which are primarily located in epithelial tissues.44, 45
As with EGFR and ERK, SFK signaling seems to be more active in basal/TN breast cancer than in certain other types of breast cancer cells. Thus, Hochgräfe et al.17 reported increased tyrosine phosphorylation of several SFK substrates including BCAR1 (p130Cas), BCAR3, CAV1, tensin-3 and STAT3, in basal/TN breast cancer cell lines compared to that in luminal breast cancer cells. Furthermore, protein levels of SRC as well as the SFK, LYN were found to be significantly higher in basal/TN human breast cancers than in other subtypes of breast cancers.17, 46
Several agents are currently available for inhibiting SRC and SRC-like enzymes.44, 45 Of these, the most widely studied is dasatinib. Dasatinib is an orally active tyrosine kinase inhibitor that targets SFKs as well as related kinases such as BCR-ABL, c-KIT, EPHA2 and PDGFR beta.44, 45 It is currently approved for the treatment of imatinib-resistant BCR-ABL-positive leukemia and is undergoing trials in other cancer types.47, 48
Again, similar to reports with EGFR and ERK inhibitors, a number of studies have shown that basal/TN breast cancer cell lines were more sensitive to SRC inhibitors, especially dasatinib, than luminal cancer cell lines.46, 49–52 In one of these studies, cotreatment with dasatinib and cisplatin was found to be synergistic in blocking growth in the TN cell lines investigated,46 while in another, combined treatment with dasatinib and etoposide additively inhibited cell growth.49 A potentially important finding with dasatinib in TN breast cancer was that its administration decreased the population of cells containing aldehyde dehydrogenase 1, suggesting elimination stem-like cells.50
Although dasatinib has been shown to inhibit the growth of basal/TN breast cancers in vitro, it is unclear whether this is mediated by inhibition of c-SRC or of related kinases that are also blocked by this compound, see above. Indeed, several reports have reported a lack of correlation between c-SRC/phosphorylated-SRC levels and dasatinib growth inhibitory ability in preclinical studies.46, 51–54
Although levels of c-SRC do not appear to predict sensitivity to dasatinib, the expression levels of other genes may do so. Thus, Huang et al.51 identified a six-gene signature that predicted in vitro sensitivity to dasatinib in a panel of breast cancer cell lines. These genes included EPHA2, CAV1, CAV2, ANXA1, PTRF and IGFBP2. The encoded proteins of all these genes are either targets for dasatinib or substrates for SRC kinases.51 Interestingly, expression of these genes was particularly high in basal/TN breast cancers.51 Finn et al.49 also reported an in vitro association between CAV1 mRNA expression and sensitivity dasatinib in basal/TN breast cancer cells. In addition, this group found that MSN (moesin) and YAP1 (yes-associated protein-1) mRNA were predictive of response to dasatinib. Another potential predictive test for response to dasatinib is the recently described gene signature reported by Moulder et al.55
Although approved for the treatment of chronic myeloid leukemia, preliminary results with dasatinib monotherapy in patients with breast cancer are disappointing. In a Phase II study, Finn et al.56 reported a confirmed response in only 2 and stable disease in 11 of 43 evaluable patients with locally advanced or metastatic TN breast cancer. Although no Grade 4 adverse events were seen, Grade 3 adverse events included fatigue (9%), diarrhea, pleural and dyspnea (all in 7%). Currently, several trials are investigating the combination of SRC inhibitors with cytotoxic and endocrine therapy for the treatment of unselected patients with breast cancer (for review, see Ref. 57).
The PARPs are a family of enzymes that catalyze transfer of the ADP-ribose moiety from NAD+ to acceptor proteins. Of the 17 different PARP-like proteins known to exist, the most widely studied is PARP1. Although PARP1 has been implicated in several different biological functions,58 one of its best established roles is as a sensor of DNA damage and an initiator of the base excision repair (BER) pathway. After sensing single strand DNA breaks, PARP1 is rapidly recruited to the altered DNA, where its catalytic activity is greatly increased.58 This results in the addition of ADP-ribose polymer side chains to itself as well as to other proteins involved in BER such as XRCC1, DNA ligase and DNA polymerase-beta.58
Although PARP1 is best known for repair of single strand DNA breaks by BER, the proteins products of the BRCA1 and BRCA2 genes are necessary for repair of DNA double strand breaks in a process known as homologous recombination.59, 60 Subjects with an inherited BRCA1 or BRCA2 mutation exhibit normal DNA function, due to the presence of one functioning allele. Such subjects, however, are at increased risk of developing breast, ovarian and other cancers. Cancers occurring in BRCA1/2 mutation carriers have lost the wild-type allele of the BRCA gene and express only a mutant or truncated form. These cancers are thus unable to undergo BRCA1/2-mediated DNA repair.
As mentioned above, defective BRCA1/2 function can be combined with inhibition of PARP1 in a process known as synthetic lethality. This synthetic lethal interaction is believed to operate as follows.61, 62 Inhibition of PARP1 leads to the accumulation of single strand DNA breaks, which following replication, develop into double strand breaks. Normally, these double strand breaks are repaired by an error-free homologous-recombination double strand repair system involving BRCA1 and BRCA2. However, if BRCA1 and 2 are defective, as in the case of BRCA-associated cancers, repair may not occur. After cell division and DNA replication, these single strand breaks are converted into double strand breaks. In this situation, genomic instability and cell death occurs. In contrast, normal cells and cells heterozygote for BRCA1/2 mutations have a functioning DNA repair system and thus maintain cell viability. This is the likely explanation for the relative lack of toxicity of PARP inhibitors toward wild-type and BRCA1/2 heterozygote cells.62
Support for this hypothesis emerged in 2005, when two preclinical studies showed that cells deficient in either BRCA1 or 2 genes were considerably more sensitive than matched wild-type and heterozygous cells to a number of different PARP inhibitors.63, 64 Indeed, Farmer et al.63 showed that cells with defective BRCA 1 or BRCA2 genes were several orders of magnitude more sensitive to olaparib than cells lacking such defects. These findings were subsequently confirmed using the PARP inhibitor, AGO14699 (Pfizer, La Jolla, CA) in breast cancer cells with mutated or epigenetically silenced BRCA1/2 genes.65 Consistent with these preclinical studies, Phase 1/II clinical trials have shown that the benefit of olaparib in patients with advanced breast cancers is confined to those with BRCA1/2 mutation-associated malignancies.66, 67
Basal/TN breast cancers have several phenotypic characteristics that overlap with BRCA1 mutated tumors. These similarities include a tendency for high grade, negativity for ER, PR and HER2, expression of the basal cytokeratins 5 and 6, aberrant DNA repair and related gene expression signatures.68, 69 These shared characteristic suggest that basal/TN breast cancers have lower expression/dysfunction in BRCA1 or other homologous DNA repair genes and thus may be susceptible to PARP inhibitors as discussed above for BRCA1/2-mutation related cancers.62
To test this hypothesis, Hastak et al.70 investigated the effect of the PARP inhibitor, PJ34 (EMD Biosciences) on the in vitro growth of four TN and three luminal breast cancer cell lines. Significantly, all four TN cells lines were more sensitive to PJ34 than the luminal cell lines investigated. Furthermore, synergy was observed between PJ34 and gemcitabine and cisplatin in the TN cell lines. In contrast, antagonism was found between PJ34 and both the cytotoxic drugs with the luminal cells. Growth inhibition mediated by the combined treatments in the TN cell lines appeared to result from sustained DNA damage and inefficient DNA repair, which lead to P63/P73 mediated apoptosis.
To test the clinical relevance of the above in vitro findings, O'Shaughnessy et al.71 investigated inipirib (a small molecule originally thought to have PARP-inhibitory properties) for the treatment of advanced TN breast cancer patients. In a Phase II multicenter trial, 123 women were randomized to receive carboplatin plus gemcitabine or both these drugs plus inaparib. Final analysis showed that the addition of iniparib to the chemotherapy was associated with an improved objective response rate (52% vs. 32%, p = 0.02) and an improved rate of clinical benefit (56% vs. 34%, p = 0.01). Furthermore, addition of the PARP inhibitor increased progression free-survival from a median of 3.6 months to 5.9 months (p = 0.01) and median overall survival from 7.7 to 12.3 months (p = 0.01). Unfortunately, a follow-up phase III trial failed to meet the prespecified criteria for significance with respect to progression-free and overall survival.72 Since publication of this study, it has been shown that iniparib acts by mechanisms unrelated to direct competitive inhibition of PARPs.73 The results of the above trial72 may thus not apply to classical competitive inhibitors of PARPs.
Similarly, Gelmon et al.,74 found no response to olaparib monotherapy in 24 patients with advanced TN disease. In this study however, 7/17 (41.2%) of BRCA mutation-related ovarian cancer and 11/46 (24%) of non-BRCA mutation-related ovarian cancers showed objective response with the inhibitor.
The role of PARP inhibitors in the treatment of TN breast cancer and indeed in other sporadic cancers is thus presently unclear. Their efficacy in this subset of breast cancer patients may, however, depend on the identification of predictive markers. Potential predictive markers include methylation status of BRCA1/2 genes,65 levels of BRCA1/2 proteins75 or PTEN status.76–78 Measurement of these putative markers should be incorporated into trials involving PARP inhibitors. Potential pharmacodynamic markers for monitoring response in patients receiving PARP inhibitors include formation of RAD51 foci, phosphorylated histone H2AX (gamma H2AX) and poly ADP ribose.79–81 Ideally, these parameters should be assessed in tumor biopsy material from patients undergoing treatment. However, as such tissue may not always be available, measurement in surrogate cells/tissues such as in blood peripheral blood mononuclear cells, circulating tumor cells, hair follicle cells or skin biopsies may suffice.82
Other potential targets for treating TN breast cancer
A list of other potential therapeutic targets for TN breast cancer is shown in Table 1.
Despite promising data from preclinical studies, there are still no validated biological/targeted therapies available for breast cancer patients that are negative for ER, PR and HER2. Indeed, because of the heterogeneity of these tumors, it is unlikely that any single treatment will be efficacious in all patients with TN breast cancer. The most effective treatment approach for these patients is likely to be a combination of biological/targeted therapies or combined biological/targeted therapy with cytotoxic agents. Ideally, both the choice of biological/targeted therapies and cytotoxic drugs should be biomarker-driven.
- 8Directed therapy of subtypes of triple negative breast cancer. Oncologist 2010; 15(Suppl 3): 8–15..
- 14Epidermal growth factor receptor and vascular endothelial growth factor receptor 2 are specific biomarkers in triple-negative breast cancer. Results from a controlled randomized trial with long-term follow-up. Breast Cancer Res Treat 2010; 120: 491–8., , , et al.
- 21TBCRC 001: EGFR inhibition with cetuximab added to carboplatin in metastatic triple-negative (basal-like) breast cancer. J Clin Oncol 2008; 26(Suppl 15):abstract 1009., , , et al.
- 22The addition of cetuximab to cisplatin increases overall response rate (ORR) and progression free survival (PFS) in metastatic triple-negative breast cancer (TNBC): results of a randomized phase II study (BALI-1). Ann Oncol 2010; 21(Suppl 8): viii96., , , et al.
- 25Analysis of epidermal growth factor receptor expression as a predictive factor for response to gefitinib ('Iressa', ZD1839) in non-small-cell lung cancer. Br J Cancer 2004; 19: 208–12., , , et al.
- 51Identification of candidate molecular markers predicting sensitivity in solid tumours to dasatinib: rationale for patient selection. Cancer Res 2007; 6; 2226-38., , , et al.
- 72A randomized phase III study of iniparib (BSI-201) in combination with gemcitabine/carboplatin (G/C) in metastatic triple-negative breast cancer (TNBC). J Clin Oncol 2011; 15_Suppl (May 20 Supplement): 107., , , et al.
- 84ADAMs as new therapeutic targets for triple negative breast cancer. J Clin Oncol 2011; 29(Suppl; abstr 1062)., , , et al.
- 85ADAM17 a novel therapeutic target for treatment of triple negative breast cancer? Cancer Res 2011; 71(24 Suppl); 416s., , , et al.