Current Breast Cancer Reports

, Volume 4, Issue 1, pp 48–55

Biology and Novel Targets in Metaplastic Breast Cancer

Authors

    • Department of Breast Medical Oncology, Unit 1354University of Texas M. D. Anderson Cancer Center
    • Department of Phase I ProgramUniversity of Texas M. D. Anderson Cancer Center
Systemic Therapy (J O’Shaughnessy, Section Editor)

DOI: 10.1007/s12609-011-0064-2

Cite this article as:
Moulder-Thompson, S.L. Curr Breast Cancer Rep (2012) 4: 48. doi:10.1007/s12609-011-0064-2
  • 72 Views

Abstract

Metaplastic breast cancer represents a spectrum of histologic subtypes with the common feature of divergent morphologic differentiation. Most of these subtypes are associated with chemotherapy resistance and an increased likelihood of developing distant metastatic disease, which has been associated with a poor prognosis. However, recent molecular characterization has indicated that some metaplastic cancers may respond to targeted therapy regimens currently undergoing evaluation in early phase clinical trials. In this review, the pathologic characteristics and epidemiology of metaplastic breast cancer are discussed along with novel therapeutic agents that may augment standard chemotherapy for this intriguing type of breast cancer.

Keywords

Breast cancerMetaplasticStem cellTriple-negative

Introduction

Metaplastic breast carcinoma is an uncommon and often aggressive subtype of breast cancer. Metaplastic breast tumors are characterized by distinct histologic areas of divergent morphologic differentiation that include epithelial elements admixed with squamous and/or mesenchymal elements. Although the morphologic components may vary considerably in individual tumors, as a group metaplastic cancers are often receptor negative (ER, PR, and HER2), and in some studies have been associated with a worsened prognosis compared to non-metaplastic tumors.

Additionally, there is growing evidence that metaplastic carcinoma represents a breast cancer subset enriched in epithelial to mesenchymal transition, with elevated levels of CD44+/24-/low cells compared to other breast cancer subtypes, including triple-negative breast cancers. These features, along with the lineage plasticity seen in metaplastic carcinoma, support the hypothesis that these tumors may arise from a more primitive cancer stem cell, which would suggest that this subset of breast cancers may be an excellent tumor type in which to study the effects of stem cell-targeted therapy.

In this review, the pathologic characteristics, epidemiology, and prognostic considerations of metaplastic breast cancer are discussed followed by an overview of the molecular characterization, including this subtype’s close resemblance to stem cell-driven tumors. Finally, emerging treatment options for this clinically challenging breast cancer subtype are described.

Pathology Characteristics

The metaplastic subtype of breast cancer encompasses a heterogeneous group of tumors with a great deal of histologic variation; however, all subtypes usually contain at least one area that has transitioned from an epithelial carcinoma to a non-epithelial phenotype [15]. The non-epithelial phenotype can include squamous differentiation, sarcomatous elements, and even osseous or neuroendocrine differentiation. Some tumors may have a low-grade appearance associated with an indolent behavior or may have a more aggressive, high-grade phenotype with a higher tendency for recurrence or metastasis. The lower grade, more indolent tumors include fibromatous-like metaplastic tumors and low-grade adenosquamous carcinomas, which have a bland-appearing morphology and very low propensity towards developing metastasis [69]. Metaplastic tumors with squamous differentiation may be difficult to distinguish from nonmalignant squamous metaplasia associated with response to injury, such as surgical resection or tumor biopsy [10]. Usually, these tumors are almost entirely composed of squamous cells and can often appear to be cystic. Although most tumors are high grade, the reported literature is mixed regarding prognosis and propensity for metastasis [4, 1114]. Matrix-producing carcinomas are defined as carcinomas with a transition to osseous or cartilaginous matrix [2, 15]. These tumors lack a spindle cell component and can resemble mucinous carcinomas due to their matrix production. The carcinomatous component of these tumors is usually high grade and this subtype has a high proclivity for metastasis [15]. The most common subtype of metaplastic carcinoma is the carcinosarcoma or sarcomatoid carcinoma subtype. These tumors are distinguished by a sarcomatoid component admixed with an invasive ductal component [1, 1618]. The component of sarcomatoid differentiation can be a small focus or the tumor may contain pure spindle cell morphology [16, 17]. Although the sarcomatous component is usually spindle cell, some tumors may have a rhabdomyosarcomatous, angiosarcomatous, or liposarcomatous appearance. Like matrix-producing carcinomas, these tumors have a high rate of metastasis [16, 17]. Of note, pleomorphic carcinomas of the breast that have a sarcomatous component have a much higher rate of disease recurrence than pleomorphic carcinomas lacking a sarcomatous component; thus, it has been advocated that this variation of pleomorphic carcinomas be considered metaplastic cancers [19].

As a group, metaplastic cancers are usually negative for estrogen receptor (ER), progesterone receptor (PR), and human epithelial growth factor receptor 2 (HER2) [20•]. Tumors are often larger at presentation than non-metaplastic invasive ductal carcinomas (IDCs); however, lymph node metastases are not commonly found at the time of surgical resection and tend to occur more frequently in carcinosarcomas or squamous carcinomas [20•]

Epidemiology and Prognosis

Given the heterogeneity and relative rarity of the subtypes of metaplastic breast cancer, it is not surprising that most published epidemiologic studies involve relatively small cohorts of patients with varied histologic tumor types. Pezzi et al. [20•] used the National Cancer Data Base to determine the incidence and clinical features of metaplastic breast cancer compared to non-metaplastic breast cancer. A total of 892 cases of metaplastic breast cancer and 255,164 cases of IDC were identified from January 2001 to December 2003 [20•]. Patients with metaplastic breast cancer were significantly older and were more likely to be African American or Hispanic. Metaplastic tumors were larger, had a higher American Joint Committee on Cancer (AJCC) stage at presentation (though less likely to have lymph node-positive disease), and were more likely to be poorly differentiated and ER-negative. These features have been confirmed by several smaller studies, some of which have also demonstrated that metaplastic carcinomas are often HER2 negative. Metaplastic patients were more likely to receive mastectomy instead of breast conserving surgery and more likely to receive adjuvant chemotherapy. Unfortunately, no data regarding survival outcomes was available from this publication.

Some groups have published outcome data in metaplastic breast cancer without specifying histologic type. Rayson et al. [21] retrospectively evaluated patients with metaplastic breast cancer (n = 27) treated at the Mayo Clinic from 1966 to 1997. In the patients with localized disease, there was a 40% 3-year disease-free survival (DFS), and the 3-year overall survival (OS) was 71%. In the patients who developed metastatic disease, only one partial response to doxorubicin (4 months in duration) was noted for 10 different chemotherapy regimens used to treat these patients. Not surprisingly, no responses were seen with tamoxifen. Other investigators have compared outcomes for metaplastic breast cancers to non-metaplastic breast cancers using either case-consecutive or case-control analyses. Although some of these studies have demonstrated no difference in outcomes for metaplastic patients [22], the majority have indicated an overall worse prognosis [2328].

In the classic series published by Wargotz and Norris [15], carcinosarcomas (n = 70) were reported to have the lowest 5-year survival rate (49%), whereas 5-year survival rates were slightly better for matrix-producing carcinomas (68%, n = 26), spindle cell carcinoma (64%, n = 100), squamous carcinoma of ductal origin (63%, n = 22), and carcinoma with osteoclastic giant cells (68%, n = 29). Researchers in Japan analyzed 53 cases of metaplastic breast cancer based upon histiologic subtype: spindle cell carcinoma (n = 11, 21%), squamous cell carcinoma (n = 27, 51%), carcinoma with osteocartilaginous elements (n = 7, 13%), and matrix-producing carcinoma (n = 8, 15%). No significant difference in disease-free periods were noted between the subgroups, but patients whose tumors contained high-grade spindle cell and pure squamous cell components were noted to have a higher rate of recurrence compared to tumors the investigators deemed low-grade spindle cell carcinomas and those with matrix-producing carcinomas [27].

Researchers at the M.D. Anderson Cancer Center (MDACC) retrospectively studied outcomes for patients with sarcomatoid carcinoma and carcinosarcoma subtypes of metaplastic breast cancer using both the MDACC database (n = 100) and the Surveillance, Epidemiology and End-Results (SEER) data base (n = 313) [23]. In this analysis, patients with stage I–III disease had a 52% 5-year relapse-free survival (RFS) and a 64% 5-year OS, which was similar to that published by Wargotz and Norris [15]. The initial stage of the tumor was strongly associated with outcome, whereas the use of adjuvant chemotherapy or radiation therapy was not. Most tumors were either triple receptor negative or hormone receptor negative, HER2 unknown. In the patients treated with neoadjuvant therapy at MDACC, the pathologic complete response (pCR) rate was 10%, which is substantially lower than the 30% to 40% rate usually reported for triple-negative breast cancer, thus supporting the chemo-refractory nature of these tumors. The median survival from time of recurrent disease was 14 months (range, 1–55 months).

One of the largest case control studies from the MDACC group revealed that patients with localized metaplastic sarcomatoid carcinoma (n = 47) had significantly worse outcome than those with non-metaplastic triple-negative breast cancer who were matched for age, clinical stage, tumor grade, adjuvant chemotherapy, and radiation therapy [24]. In this study, 5-year DFS for patients with metaplastic breast cancer was 44% compared to 74% in patients with non-metaplastic triple-negative breast cancer (P = 0.018). Similar results were seen by investigators at the European Institute of Oncology when they studied outcomes for a variety of metaplastic cancers (n = 37), including matrix-producing, spindle cell, carcinosarcomas, squamous cell, and carcinoma with osteoclastic giant cells [25]. In this study, cases were matched with up to two controls (n = 72) based upon the year of surgery, tumor size, and the presence or absence of lymph node involvement. Overall survival was significantly worse in the metaplastic patients (hazard ratio [HR] = 5.0; P < 0.0001). Similar results were not seen by the Swedish Cancer Institute when they analyzed 24 metaplastic breast cancer patients from a prospectively collected database (years 1990 to 2005), matching two controls (n = 72) based upon date of diagnosis, age, tumor size, node status, ER/PR status, and HER2 status (when available) to each metaplastic case [22]. In this study of varying histologic categories of metaplastic breast cancer (including matrix-producing, spindle cell, carcinosarcoma, squamous, and “not stated”), the 5-year estimated RFS was 84% for metaplastic carcinomas and 93% for non-metaplastic breast cancers (P = 0.26), which is much higher than seen by other investigators. Considering the fact that most of the control patients would have larger tumors (mean, 2.1 cm) that were “triple negative” (approximately 92%), the 5-year estimated RFS is above that which would be expected from this group as well.

Two large case-consecutive analyses have also demonstrated a worse clinical outcome in patients with metaplastic breast cancer. Investigators from the National Cancer Center in Korea compared 35 metaplastic breast cancer patients to general IDC of the breast (n = 2839) and triple-negative breast cancers (n = 473) treated from 2001 to 2008 [28]. In this study, metaplastic breast cancer was a poor prognostic indicator for disease recurrence and overall survival by both univariate and multivariate analyses compared to IDC (HR = 3.89 for recurrence, P = 0.01; HR = 5.29 for death, P < 0.001) and triple-negative breast cancer (HR = 3.99 for recurrence, P = 0.01; HR = 3.14 for death, P = 0.02). Similar results were seen by investigators at the National Cancer Center in Japan when they compared the outcomes of 46 patients with metaplastic breast cancer to 6137 patients with IDC and 301 patients with invasive lobular carcinoma treated from1982 until early 2007 [26]. Patients with metaplastic breast cancer were 5.5 times more likely to develop disease recurrence compared to those with non-metaplastic cancers (95% CI, 3.2–9.6) and 4.2 times more likely to have disease specific death (95% CI, 2.2–8.1).

Molecular Characterization

As previously mentioned, metaplastic cancers are typically negative for ER, PR, and HER2 as measured by immunohistochemistry and/or fluorescence in situ hybridization (FISH). Metaplastic tumors have, however, been reported to frequently express high levels of the epidermal growth factor receptor (EGFR/HER1) [2934]. Several of the larger published series have documented overexpression of EGFR to occur in 66% to 76% of cases [29, 31, 32, 34]. Approximately one quarter of the tumors with high EGFR expression displayed EGFR FISH positivity; however, the majority of tumors displayed high EGFR copy number due to aneusomy rather than amplification in at least one publication [31, 34]. Metaplastic breast cancers also frequently express caveolin 1, α-crystallin, and vimentin [3538].

These features would suggest that metaplastic cancers share characteristics of basal-like breast cancers and, indeed, genomic profiling using unsupervised hierarchical clustering grouped metaplastic tumors within the spectrum of basal-like breast cancers [39]. However, in this pivotal publication, Weigelt et al. [39] noted that significance analysis of microarrays (SAM) identified transcripts differentially expressed between metaplastic cancers and basal-like IDCs. Their analysis found that DNA repair pathways such as BRCA1, PTEN, and TOP2A were significantly downregulated in the metaplastic tumors compared to the basal-like IDCs. Further genomic analyses by Hennessy et al. revealed that tumors with squamous and sarcomatoid metaplasia more closely resembled the rare claudin-low subtype of breast cancers, which are characterized by enrichment of markers of epithelial-to-mesenchymal transition (EMT) and have absence of expression of luminal differentiation markers [40, 41•]. Interestingly, when a set of molecularly identified claudin-low tumors were histologically characterized, 10% of these tumors showed metaplastic differentiation [42].

Hennessy et al. [41•] also identified a high rate of mutations (47%) in PIK3CA and occasional mutations in PTEN (5%) in their sample set of metaplastic tumors, which was notably higher than the relatively low rate of mutations seen in basal-like IDC (8%). Reverse-phase protein array (RPPA) confirmed elevations in the phosphorylation of phosphatidylinositol 3-kinase (PI3K)/Akt pathway proteins, further supporting the genomic profiling analyses and suggesting agents that target the PI3K pathway may be a viable option for the treatment of certain types of metaplastic breast cancer [41•].

Until recently, the clonality of metaplastic cancers was the subject of debate, with most investigators stipulating that these cancers were derived from two independent malignant cell lines that collided at some point during tumorigenesis. Modern molecular analyses, however, have clearly determined that these tumors are monoclonal in origin, which has led to the suggestion that metaplastic tumors may result from malignant transformation of early precursor cells [4345]. Although molecular analyses of metaplastic carcinomas have established the clonal relationship of morphologically distinct components of the same tumor, it should be noted that some tumors contain additional genetic aberrations, such as copy number changes, gene amplification, and focal gene mutations or amplifications within the non-epithelial component of the cancer [45].

The dual staining of carcinosarcomas for both epithelial and myoepithelial markers led early investigators to hypothesize that this subtype of metaplastic breast cancer may have myoepithelial origins [3]. These initial observations were later confirmed by others using established myoepithelial markers such as CD10, p63, smooth muscle actin, and S-100 [37, 46]. Gene methylation signatures that segregate cell lines into epithelial and mesenchymal lineages have also supported this hypothesis, as the methylation signatures of both of the claudin-low and metaplastic breast cancer subtypes more closely resembled the mesenchymal expression signature, although the metaplastic tumors also expressed epithelial markers [47].

Metaplastic cancers also have significantly higher expression of genes involved in EMT compared to other types of breast cancer. For example, metaplastic tumors demonstrated upregulation of genes associated with extracellular matrix production and downregulation of genes that encode for cell-cell adhesion molecules and tight junctions, an effect more pronounced in tumors with spindle cell morphology [48]. SAM performed using gene expression profiles obtained from tumors with squamous and sarcomatoid metaplasia revealed upregulation of regulators of EMT such as TWIST1 and SNAIL2/SLUG [49, 50]. A close association of the metaplastic and claudin-low signatures with an “EMT-core signature” derived by overexpressing EMT-inducing transcription factors in cell lines has also been reported [51•]. Nuclear localization of β-catenin modulates genes involved in the induction of EMT and metaplastic breast cancers commonly display aberrant expression of β-catenin within the nucleus or cytoplasm [52]. Controversy exists regarding the source of this aberrant expression, with some investigators reporting mutations in CTNNB1 (the gene encoding β-catenin) in 26% of metaplastic breast cancers, whereas others have not detected any mutations in this gene [41•, 52, 53]. Activation of EMT has been assumed by many to be the reason metaplastic breast cancers have a higher rate of disease metastasis [48].

Metaplastic Carcinomas as Surrogates for Cancer Stem Cell-targeted Therapy

It is hypothesized that some breast cancers may contain a subpopulation of cells with the ability to self-renew, differentiate, and undergo EMT [54]. It remains unclear if this cancer stem cell (CSC) subpopulation is composed of tumor cells that have acquired stem cell-like characteristics or if the cells originated through malignant transformation of a mammary stem cell [55, 56].

Several preclinical models have implicated these putative CSCs in resistance to common therapies against breast cancer, including radiation, chemotherapy, and targeted therapy [5761]. Clinically, the expression of stem cell markers in primary tumors has been associated with poor prognosis and increased risk for disease recurrence or metastasis [62, 63]. Additionally, genomic analysis of CSC-driven, tumorigenic cell lines has identified a gene expression signature that is enriched in residual disease after neoadjuvant therapy, supporting the role of these cells as a source of intrinsic resistance to chemotherapy and endocrine therapies [64]. Of note, this CSC signature is weighted for PI3K activity, which supports preclinical data implicating the PI3K/mTOR pathway in CSC survival and tumorigenicity [65, 66]. Moreover, treatment with the mTOR inhibitor, rapamycin, reduced CSC fractions in cell culture and decreased tumorigenicity in mouse models [65]. Although there are substantial preclinical data to support clinically available agents to target CSC-driven tumors, there are several challenges to validating this approach to therapy using clinical trials. First, CSCs are likely responsible for therapy resistance in a subset of breast cancers, so predictors of response need to be developed to select patients for clinical trials. Secondly, developing predictors of response in the metastatic setting or using tissue from untreated primary tumors could under-represent the true population of CSC-positive tumors due to limitations in sampling or analytical techniques that may miss the small fraction of CSCs within a heterogeneous tumor. Additionally, CSCs constitute a small fraction of cells within a tumor, so clinical trials that rely solely upon response to therapy as the primary endpoint may miss a regimen effective against CSCs. Finally, increasing heterogeneity in metastatic tumors could lead to a resistant subclone of cells unresponsive to CSC-targeted therapy. This suggests the need to use CSC-targeted therapy early in the course of the disease; however, agents are not usually introduced into the adjuvant or neoadjuvant setting without evidence of response in metastatic disease.

Molecular characterization has indicated that metaplastic breast cancers (MpBCs) display characteristics of CSCs, including features such as EMT [41•]. Like CSCs, MpBCs have activation of the PI3K pathway and demonstrate a relatively high rate (> 50%) of mutations in PI3K or loss of PTEN compared to non-metaplastic breast cancers [41•, 67]. MpBCs also show strong correlation with the CSC-derived genomic profile associated with therapy resistance in the neoadjuvant setting [41•, 64]. Both CSCs and MpBCs also express high levels of vascular endothelial growth factor (VEGF) and hypoxia-induced factor 1α (HIF-1α), a transcription factor important for vasculogenesis [6871]. The Wnt signaling pathway is involved in the regulation of stem cell proliferation and self renewal [72]. Metaplastic cancers display high rates of Wnt pathway deregulation, which may be the source for aberrant β-catenin localization found in this cancer subtype. These common features suggest that MpBCs may arise from a more primitive CSC and/or may act as a surrogate in which to study the effects of stem cell-targeted therapy.

Emerging Systemic Therapies for Metaplastic Breast Cancer

Although data concerning systemic therapy for metaplastic breast cancer are limited, the available information suggests that systemic chemotherapy is less effective for this subtype compared to non-metaplastic triple-negative breast cancers [73]. For instance, a retrospective analysis of patients treated with biphasic metaplastic sarcomatoid breast cancer (n = 100) found the pCR rate to anthracycline-based chemotherapy to be 10% among the 21 patients treated neoadjuvantly, a response rate markedly lower than would be expected for high-grade, triple-negative, non-metaplastic breast cancer. Of the 7 patients with metastatic metaplastic breast cancer treated with 10 chemotherapy regimens in the Mayo retrospective experience, 1 patient developed a partial response of 4 months in duration when treated with doxorubicin [73]. The median survival after disease recurrence in this study was 8 months. Other investigators have described similar clinical experiences with systemic therapy [3, 74].

A small series of patients with metaplastic cancer (n = 5) treated in a phase I trial of liposomal doxorubicin, bevacizumab, and the mTOR inhibitor, temsirolimus (DAT regimen) demonstrated a clinical benefit rate (complete response + partial response = stable disease ≥6 months) of 60%, including one complete response of greater than 1 year in duration (to date, the patient continues in remission) [67]. Considering the high rate of PI3K/PTEN mutations and activation of the PI3K pathway in metaplastic tumors, and the response seen to the DAT regimen, it is anticipated that drugs that target the PI3K pathway may show promising activity in metaplastic cancers (Table 1). Interestingly, the patient with the complete response to DAT did not have a detectable PIK3CA mutation or loss of PTEN.
Table 1

Drugs that target the PI3K pathway

Drug

Mechanism of action

Documented activity in metaplastic breast cancer?

FDA approved

Sirolimus

mTOR inhibitor

No

Everolimus

mTOR inhibitor

No

Temsirolimus

mTOR inhibitor

Yes, in combination with liposomal doxorubicin and bevacizumab [67]

Under development in clinical trials

Deforolimus

mTOR inhibitor

No

INK128

mTOR inhibitor

No

AZD8055

mTOR inhibitor

No

OSI-027

mTOR inhibitor

No

XL147

PI3K inhibitor

No

GDC-0941

PI3K inhibitor

No

GSK615 (1059615)

PI3K inhibitor

No

CAL-101

PI3K inhibitor; delta specific

No

NVP-BKM120

PI3K inhibitor

No

PX-866

PI3K inhibitor

No

GSK690693

Akt inhibitor

No

Perifosine

Akt inhibitor

No

MK-2206

Akt inhibitor

No

GSK2141795

Akt inhibitor

No

GSK2110183

Akt inhibitor

No

LY2780301

Akt inhibitor

No

GDC-0068

Akt inhibitor

No

Triciribine

Akt inhibitor

No

RX-0201

Akt inhibitor (antisense oligonucleotide)

No

SF1126

PI3K/mTOR inhibitor

No

NVP-BEZ235

PI3K/mTOR inhibitor

No

NVP-BGT226

PI3K/mTOR inhibitor

No

XL765

PI3K/mTOR inhibitor

No

GDC-0980

PI3K/mTOR inhibitor

No

PF-04691502

PI3K/mTOR inhibitor

No

PKI-587

PI3K/mTOR inhibitor

No

Taking into account the similarities in metaplastic tumors and CSCs, drugs that target CSC properties, such as the ability to self-renew or undergo EMT, may also show activity in patients with metaplastic breast cancer. Signaling pathways such as Wnt, Notch, and Hedgehog have been implicated in breast cancer carcinogenesis and are interesting targets for drug development in metaplastic breast cancer. Drugs targeting Notch and Hedgehog are currently under clinical investigation in phase I and phase II trials, whereas those targeting Wnt are in preclinical development (Table 2).
Table 2

Stem cell-targeted therapies in clinical trials

Drug

Mechanism of action

MK0752

Inhibition of Notch cleavage by γ-secretase

R04929097

Inhibition of Notch cleavage by γ-secretase

PF-03084014

Inhibition of Notch cleavage by γ-secretase

LY450139

Inhibition of Notch cleavage by γ-secretase

GDC-0449

Hedgehog pathway inhibitor

BMS-833923 (XL139)

Hedgehog pathway inhibitor

LDE225

Hedgehog pathway inhibitor

PF-04449913

Hedgehog pathway inhibitor

IPI-926

Hedgehog pathway inhibitor

Conclusions

Much has been learned and remains to be learned about the biology of metaplastic breast cancers. Although in the past this subtype has been more of a “catch all” category, it is anticipated that further testing using complex molecular analyses will help to guide therapy for the subcategories of metaplastic breast cancers. This has been supported by at least one example of a clinical experience where molecular analyses demonstrated a target of interest (PI3K) for sarcomatoid carcinoma, which subsequently led to responses seen withPI3K-targeted therapy for the treatment of this subtype of metaplastic cancer. It is clear that the study of this rare cancer subtype will require the effort of multiple clinical trial groups; however, advances made in this tumor subtype may eventually translate into improvements for CSC-driven breast cancers.

Disclosure

No conflicts of interest relevant to this article were reported.

Copyright information

© Springer Science+Business Media, LLC 2012