Molecular Biology in the Breast Clinics—Current status and future perspectives

Abstract

Breast cancer is no longer considered a single disease, and with better understanding of cancer biology, its management has evolved over the years, into a complex individualized use of therapeutics based on variable expressions of predictive and prognostic factors. With the advent of molecular and genetic research, the complexity and diversity of breast cancer cells and their ability to survive and develop resistance to treatment strategies became more evident. At the same time, targeted therapies evolved, as specific targets were discovered such as HER2 receptor, and androgen receptor. More recent is the development of immunotherapy which aims at strengthening the host immune system to identify and kill the tumor cells. In breast cancer treatment, use of molecular tests has been a target of controversies, due to their high costs and inaccessibility in limited resource situations. Research in breast cancer is also proceeding at a rapid pace, but it is important to remember that breast cancer continues to be a complex interplay of alterations at molecular and genetic level, with the variability in expressions at protein level leading to difference in behavior and responses to treatment and overall outcome. In the succeeding paragraphs, we will try to review the available evidence in literature and attempt to understand the molecular complexity of breast cancer in order to simplify the art of treating the disease and improving outcomes.

Introduction

As was aptly foretold by Sir Bernard Fisher in 1992, “The science of breast cancer in 1992 is different from that of 1892, and the science of breast cancer in the next decade will differ from that of today. As the burgeoning science of molecular biology unfolds, it is unrealistic to expect that operative procedures and systemic therapy modalities now used to treat breast cancer will remain static, any more than it could have been assumed that surgery would remain unchanged during the past century, given the changes in our understanding of the disease during that time.” [1].Breast cancer is no longer considered a single disease, and with better understanding of cancer biology, its management has evolved over the years, into a complex individualized use of therapeutics based on variable expressions of predictive and prognostic factors. While surgery still remains the mainstay of breast cancer treatment with progressive refinement in surgical techniques (from the Halstedian era of radical surgery to current period of conservatism), Fisher’s work indicating breast cancer to be a systemic disease from inception showed the importance of adjuvant therapies especially chemotherapy. In the very early period, radiation therapy and chemotherapy were advised for inoperable cases without any surgery being offered in such cases. Currently, we know that early breast cancer has overall survival advantage with application of radiotherapy by preventing loco–regional recurrence.

For a considerable length of time, the only criteria which decided the use of adjuvant systemic chemotherapy and hormonal therapy (tamoxifen) were menopausal status and nodal disease. Various other factors discovered during the end of the twentieth century extending into the first part of the twenty-first century have radically changed the management of a patient with breast cancer. Probably, the most important discoveries of the twentieth century, after the discovery of estrogen receptor (ER) and tamoxifen, were the HER2 receptor and anti-HER2 therapy. The next important discovery in the last century was the understanding of biological subtypes of breast cancer by genome-wide molecular classification as defined by Perou et al. [2] in the year 2000, into the luminal A, luminal B, HER2 expressing, basal-like and normal-like, and its importance in appropriate treatment decisions and outcomes.

With the advent of molecular and genetic research, the complexity and diversity of breast cancer cells and their ability to survive and develop resistance to treatment strategies became more evident. At the same time, targeted therapies evolved, as specific targets were discovered such as HER2 receptor and androgen receptor. More recent is the development of immunotherapy which aims at strengthening the host immune system to identify and kill the tumor cells. These could be in the form of check point inhibition making it easier for the host immune system to detect cancer cells, or by engineering of the host immune system making them more efficient in identifying and killing the cancer cell as in T cell therapies.

In breast cancer treatment, use of molecular tests has been a target of controversies, due to their high costs and inaccessibility in limited resource situations. Research in breast cancer is also proceeding at a rapid pace, but it is important to remember that breast cancer continues to be a complex interplay of alterations at molecular and genetic level, with the variability in expressions at protein level leading to difference in behavior and responses to treatment and overall outcome. In the succeeding paragraphs, we will try to understand the molecular complexity of breast cancer in order to simplify the art of treating the disease and improving outcomes.

Biological or Molecular Subtypes of Breast Cancer

As first described by Perou et al. [2], breast cancers are divided into various molecular subtypes based on the genome-wide molecular expressions. These are luminal A, luminal B, Basal-like, HER2 expressing, claudin low, and normal. They also take into consideration estrogen receptor (ER), progesterone receptor (PgR), HER2, and Ki67.

Immunohistochemistry has long remained the basis of classification of subtypes of breast cancer and thereby governs the planning of adjuvant therapy. Evaluation of the receptors is of utmost importance and is described clearly within the ASCO-CAP guidelines [3, 4]. Handling of the tissue is very crucial in receptor identification including how it is fixed, transported, and stained to avoid cold ischemic damage and loss of receptors as these are proteins and can get denatured. It is recommended to keep the cold ischemic time short, mostly less than an hour, and to have adequate fixation time ranging from 6 to 72 h.

Risk Prediction Models in Breast Cancer

• Adjuvant!:Adjuvant! Online (Adjuvant!) [5] is a programmed tool to assist clinical decision on the most appropriate adjuvant therapy for breast cancer. It uses clinical factors like age, menopausal status, and stage, along with immunohistochemistry-detected hormone receptor status to provide estimates of the risk of disease recurrence or death at 10 years with and without a particular treatment along with side effects from therapy [6]. The data used for the baseline risk estimation has been derived from the SEER (Surveillance, Epidemiology and End Results) database 9 covering 14% of the US population [7]. The main limitation of this tool is that the estimates for efficacy of adjuvant therapy are largely based on reviews of randomized trials of adjuvant treatment during the late 1900s with the then available adjuvant therapies [8], and not all prognostic factors are accounted for and including small tumors.

• Predict: Is another online prognostication tool to help in making appropriate decisions about postoperative adjuvant therapy. This model was derived from the Eastern Cancer Registration and Information Centre (ECRIC) dataset, comprising of 5,694 women treated in East Anglia from 1999 to 2003. This was validated with information from another set of 5,468 patients from the West Midlands Cancer Intelligence Unit (WMCIU) [9]. The parameters to be entered include age, tumor size, grade, number of positive nodes, ER status, and method of detection (screen or symptom-detected). Additional data on HER2 and Ki67 status were also included. Relative reduction in mortality is calculated to predict benefit with multiple lines of chemotherapy and hormone therapy and projected as values and graphs.

• The IHC4 + C score is a prognostic tool that integrates four immunohistochemical measures ER, PgR, HER2, and Ki-67, along with clinicopathological features to estimate the residual risk of distant recurrence at 10 years in post-menopausal women with ER-positive breast cancer who have received 5 years of endocrine therapy. Retrospective studies indicate that the test can identify a set of women that are at such low risk of recurrence that chemotherapy can be of little benefit [10].

RNA-Based Tests

Oncotype DX, Oncotype DX DCIS score, MammaPrint, RT-qPCR are all based on measuring RNA content and gene amplification. They carry an inherent risk of false positivity or negativity due to dilution with DCIS component, contamination with stromal fibroblasts and tumor heterogeneity [11–13]. This can be minimized by selective isolation of invasive component by micro-dissection. Hence, these RNA-based tests are not ideal for measuring receptor (proteins) expressions but are used to arrive at an overall risk score to decide adjuvant therapies. Drugs/chemotherapy used in the treatment of breast cancer is not without potential toxic effects and hence it is imperative to select appropriate patient subsets that are likely to benefit from it. This is particularly relevant in the setting of node-negative breast cancer where the risk-benefit ratio of adjuvant systemic therapy has to be judiciously evaluated, wherein these RNA-based tests are being applied.

• Oncotype DX 21 gene assay: This is a validated gene expression array that uses paraffin-embedded tumor tissue samples to test 16 cancer-related genes and 5 reference genes with qRT-PCR. It gives a Recurrence Score (RS) that predicts the risk of distant recurrence at 10 years and is characterized as low (< 18), intermediate (18–30), or high (> 30). This score was validated in node-negative, estrogen receptor–positive breast cancer patients, treated with 5 years of tamoxifen, who were enrolled in the National Surgical Adjuvant Breast and Bowel Project clinical trial B-14 [14]. The Kaplan-Meier estimates for the three groups were 6.8%, 14.3%, and 30.5% respectively and the difference in the low- and high-risk groups was statistically significant (P < 0.001). The utility of this score has been further confirmed in a retrospective analysis of the Arimidex, tamoxifen, and Alone or in Combination (ATAC) Trial [15] in postmenopausal women with hormone receptor–positive disease which demonstrated that RS is an independent predictor of distant recurrence in both node-negative and node-positive patients treated with anastrozole, providing significant additional prognostic information over standard parameters. The Oncotype DX RS assay has been recommended for use as a valid decision-making tool in node-negative hormone receptor–positive breast cancer patients by the American Society of Clinical Oncology [16] and the 2011 St. Gallen International Expert Consensus [17].

  TAILORx trial which has been recently published [18] has looked at the utility of Oncotype DX in predicting the benefit of use of chemotherapy over and above hormone therapy in the intermediate RS group. However, the authors have slightly redefined the boundaries of the intermediate group wherein low RS is < 10, intermediate RS is 11–25, and high RS is > 26. In all 9700 early breast cancer node-negative patients with hormone receptor–positive and HER2neu-negative disease were assigned risk scores. Patients with high-risk scores received adjuvant chemotherapy and hormone therapy, while only hormone therapy was offered to patients with low RS. The intermediate group that consisted of 6700 patients was then randomly assigned to receive hormone therapy alone versus chemotherapy followed by hormone therapy. The trial showed correlation between age and risk score with similarity of endocrine therapy and chemo endocrine therapy in women above 50 years of age with intermediate-risk scores. However, in women below 50 years of age in the intermediate-risk group, some benefit of chemotherapy was seen. Although this was inferred as avoidance of chemotherapy in 70% of the population, in Indian women, this comprises only 13.3% (personal communication of unpublished data) at a tertiary referral center.

• MammaPrint 70-gene profile: This prognostic 70-gene set was created by analysis of fresh frozen tumor samples at the Netherlands Cancer Institute and by comparing the gene expression profiles of tumors with and without subsequent distant metastasis at 5 years. Thereafter, it has been validated in 6 different studies [19–22]. This signature is able to predict prognosis regardless of lymph node status [21, 22] and also is a strong predictor of distant metastasis in the first 5 years. A major setback for use of this signature has been the need for large amounts of frozen tissue. However, this assay is now available for use on paraffin-embedded tissues, although there is no evidence of concordance with data from frozen samples. As per the results from the EORTC 10041/BIG 3-04 MINDACT trial [23], G Viale reported that molecular subtyping by MammaPrint was able to re-stratify 54% of patients with luminal B tumor by pathological subtyping to a low-risk luminal A–type group with comparable outcome. Among TN EBC, 5% were classified as luminal by molecular subtyping with luminal-like outcome. Molecular classification could help to identify a larger group of patients with low risk of recurrence compared with the more contemporarily used classification methodology including high-quality assessed Ki67.

• EndoPredict: This is a multigene signature based on expression of 8 genes which signify proliferation and hormone receptor signaling by qPCR using FFPE tissue samples. It predicts the risk of recurrence in ER-positive, HER2-negative tumors [24]. It is recommended for both node negative and 1-3 LN positive breast cancer that is low risk, ER positive, and HER2 negative. However, additional validation is needed for this. There is an ongoing prospective study that is assessing EndoPredict Genome test impact on shared decision of adjuvant chemotherapy in a set of eligible patients (ADENDOM).

• Breast Cancer Index: It is an assay that includes a 2-gene signature (HOX13:IL17BR) related with recurrence and a five-gene tumor molecular grade index (MGI) (BUB1B, CENPA, NEK2, RACGAP1, and RRM2) that is found to be raised in high-grade tumors [25, 26]. It predicts the risk of distant disease in node-negative hormone-positive tumors. In a retrospective study by Ma et al., the MGI differentiated low- from high-grade tumors with 86% accuracy. Patients were classified into risk groups based on a combination of the two scores which were classified as low/high (limit for MGI = 0 and HOXB13:IL17BR = 0.06): (a) “low”: any HOXB13:IL17BR with MGI low; (b) “intermediate”: HOXB13:IL17BR low and MGI high, and (c) “high”: both HOXB13:IL17BR and MGI high. The high-risk patients were found to have an eight times higher chance of developing distant metastasis than the low-risk patients.²⁶

• Prosigna (PAM50)-NanoString technology: Prosigna risk of recurrence score is based on a 50-gene-based Prediction Analysis of Microarray (PAM 50) and it can be used for the risk prediction of both node-negative and 1–3 node-positive patients of ER PgR-positive HER2neu-negative early breast cancer. Parker et al. [27] concluded that intrinsic subtype and risk predictors based on the PAM 50 gene set added significant predictive and prognostic information to staging, grade, and clinical molecular markers similar to Oncotype DX recurrence score. The risk of recurrence (ROR) score used in Prosigna is continuous and patients are assigned scores of < 10% (low), 10–20% (intermediate), and > 20% (high). Laenkholm et al. [28] validated Prosigna in 2558 patients from the Danish Breast Cancer Cooperative Group (DBCG) who were ER PgR-positive HER2neu negative and received 5 years of endocrine therapy alone, with 54.5% of these patients having 1–3 nodes positive for metastases. By this scoring, 26% of node-positive patients were deemed low risk while 17.8% of node-negative patients were reclassified as high ROR.

It must be noted that for node-positive patients, no multigene assay has shown to have sufficient evidence for recommendation as per the American Society of Clinical Oncology (ASCO).

Inflammatory Markers and Prognosis in Breast Cancer

Macrophages are the key cells in chronic inflammation and tumor-associated macrophages (TAMs) are often found in most advanced carcinomas. Macrophages are activated by Th1 cytokines and pathogens to form M1 macrophages—the classical or activation state. These have an anti-tumoral or beneficial activity. Anti-inflammatory mediators induce anti-inflammatory macrophages or M2 cells derived from Th2 microenvironment, which is responsible for humoral immunity [29]. M2 cells are associated with poor prognosis and directly influence tumor growth, migration, invasion, and metastasis by secreting a variety of growth factors, cytokines, and protease, which in turn let the tumor cells in through invasion of basement membrane. Neutrophils, fibroblasts, and myeloid suppressor cells have a similar mode of action [30]. With reference to breast cancer, presence of TAMs correlates to a poor prognosis [29].

Shou J et al. [31] analyzed the correlation between FOXP3+ tumor-infiltrating lymphocytes (TILs) and breast cancer prognosis in a total of 15 studies comprising 8666 breast cancer patients. Their results showed that higher FOXP3+ TIL level was significantly associated with poor prognosis in terms of overall survival (OS) (pooled HR, 1.60). They found that breast cancer with higher FOXP3+ TIL level positively and significantly correlated with CerbB-2-positive status and lymph node–positive status while there was a negative association with ER-positive status and PR-positive status. They concluded that the FOXP3+ TIL level is a promising poor prognostic factor in breast cancer.

The tumor microenvironment consists of extracellular matrix including collagen, hyaluronic acid, stromal cells, and inflammatory cells such neutrophils, lymphocytes, macrophages, and myeloid-derived suppressor cells. These immune cells secrete cytokines, chemokines, TNF alpha, IL 6, FGF, EGF, and HGF and have been implicated in playing an essential role in tumor progression, having a similar role in inflammation and epithelial mesenchymal transition. These cytokines, chemokines, proangiogenic factors, and ECM-modifying enzymes, such as matrix metalloproteinases, secrete certain bioactive molecules, which promote inflammation and other carcinogenic programs [30].

Molecular and Genetic Markers of Outcome in Breast Cancer

CA 15.3 and CEA

Both CA 15.3 and CEA have remained controversial as predictors and or prognosticators for very long. A very recent published meta-analysis of 36 studies in 12993 patients [32] showed that elevated CA 15.3 and CEA were statistically significant indicating a poorer DFS and OS (HR 2.03 and 1.79 respectively). Also, elevated CA15.3 was associated with advanced histological grade and younger age, while elevated CEA was related to non-triple-negative tumor type and older age. These two elevated markers were further associated with a higher tumor burden and thus indicate poor prognosis.

Triple-negative breast cancer (TNBC) is defined by absent expression of estrogen receptors, progesterone receptors, and CerbB2 receptor by IHC and non-amplification of HER2 by fluorescent in situ hybridization (FISH). Defining AR positivity in TNBC has also been variable from 1 to 10% and is not very standardized [33, 34]. They are further classified by Androgen receptor (AR) expression and EGFR expression and CK5/6 expression by IHC into various categories with prognostic implication. Considering a cutoff of 1% for AR staining, nearly 41% TNBC were reported as AR positive in an exploratory study from Cedars-Sinai Medical Center, Los Angeles, California [35]. In fact, TNBC were classified based on age, tumor size, tumor grade, lymph node status, proliferation rate, immunopositivity for EGFR, CK5/6, Ki-67, and disease-free survival (DFS) into 3 categories. Low-risk (AR+ EGFR−) which represents the LAR molecular subtype with the best prognosis and may benefit the most from anti-androgen therapies; high risk (AR− EGFR+) which represents the basal molecular subtype with the worst prognosis and may benefit the most from chemotherapy regimens; and an intermediate-risk (AR+EGFR+ and AR−EGFR−) TNBC with an intermediate prognosis. These subgroups of TNBC need to be validated prospectively with disease outcomes as end point.

There are other classifications of TNBC that have been proposed. Lehman et al. [36] have further classified TNBC into four molecular subtypes namely basal-like1, basal-like2, mesenchymal, and luminal androgen receptor (LAR), based on different clinicopathological features and different driver signaling pharmacologically targetable pathways [36]. In another study, based on immune response, Jezequel et al. demonstrated three molecular subtypes of triple negative breast cancer namely basal with low immune response, basal with high immune response, and LAR [37].

Other Cell Targets in TNBC

Research in triple-negative breast cancer has advanced enough to indicate that there are other targets [38] in this subset of breast cancer not expressing the standard receptors that can open up a wide range of treatment options. In a gist, some of the main targets that are being investigated are (a) in the blood VEGF and VEGFR; (b) on the cell membrane EGFR/HER1, IGFBP, PDL1, c-kit, and c-met; (c) in cell cytoplasm PIK3CA/AKT/mTOR and PTEN; and (d) in cell nucleus BRCA1 and BRCA2, and TP53. VEGF family of isoforms leads to an increase of angiogenesis and the permeability of nearby vessels and lymphatics; the end result improved oxygen/nutrient transport, as well as increased tumor metastasis. VEGFA expression indicated a higher metastatic potential and poorer prognosis [39]. Although preclinical evidence suggested an overwhelming response to anti-VEGFA therapy (with bevacizumab), with significantly higher pCR, it was later found to be more effective in HER2-expressing tumors than in TNBC [40]. VEGFR levels may be more indicative of successful anti-VEGF-A therapy in TNBC, and TNBC with high VEGFR1 expression appear to have maximum benefit from anti-VEGFA therapy [41].

Programmed Cell Death 1 Ligand 1

Programmed cell death 1 ligand 1 (PD-L1) is a transmembrane protein encoded by CD274 that functions as a key checkpoint regulator in the immune response [42]. It is found in B cells, natural killer cells, and vascular endothelial cells and binds the programmed cell death protein 1 (PD-1) found on activated cytotoxic T cells. PD-L1 expression on tumor cells appears to be higher in TNBC than non-TNBC and is estimated to occur in ~ 20% of TNBC. This is another predictive target under research. The PD-1/PD-L1 pathway represents an adaptive immune resistance mechanism that is exerted by tumor cells in response to endogenous immune anti-tumor activity. PD-L1 is commonly over expressed on tumor cells or on non-transformed cells in the tumor microenvironment [43]. PD-L1 expressed on the tumor cells binds to PD-1 receptors on the activated T cells, which leads to the inhibition and deactivation of the cytotoxic T cells. The check point antibody inhibitors, such as anti-PD-1/PD-L1, like pembrolizumab, are a novel class of inhibitors that function as a tumor-suppressing factor via modulation of immune cell-tumor cell interaction to recruit the T cells and activate immune response and are the basis of immunotherapy trials in breast cancer.

These checkpoint blockers are rapidly becoming a highly promising cancer therapeutic approach that yields remarkable anti-tumor responses with limited side effects.

PIK3CA/AKT/mTOR Pathway

The PIK3CA/AKT/mTOR pathway has gained popularity as a potential route for TNBC resistance to chemotherapy. PIK3CA is an upstream catalytic enzyme that, when active, leads to cell growth and proliferation and in particular inhibition of cell death [44].

BRCA1 and BRCA2

BRCA1 and BRCA2 are protein-expressing spans of DNA found on chromosome 17q21 and 13:12.3, respectively. Mutations in BRCA1 and BRCA2 have been linked to increased lifetime breast cancer incidence, independent of other breast cancer-related genes [45]. Evidence suggests that BRCA1-mutated breast cancer has significant overlap with TNBC (BRCAness) [46]. Because BRCA1 and BRCA2 seem to be heavily involved in DNA repair, aberrations in BRCA1 and BRCA2 appear to sensitize TNBC to DNA-damaging platinum agents and PARP inhibitors. A meta-analysis from Chen et al. [47] examined the risk of remission rate in TNBC using standard neoadjuvant therapy with or without carboplatin. The results suggest that carboplatin improves pCR over than other agents used in TNBC treatment.

Tau Protein

Tau protein (50–64 kDa) is a product of a gene located on chromosome 17 (17q21) and belongs to the microtubule-associated protein (MAP) family. Microtubules are dynamic polymers composed of β-tubulin heterodimers and are essential components of the mitotic spindle and cytoskeleton. High tau mRNA expression in ER-positive breast cancer indicates an endocrine-sensitive disease [48]. Tau protein competes with taxane for controlling microtubule dynamic and loss of tau expression may render microtubules more vulnerable to the effects of taxanes [49]. Tau expression was also a better predictive marker of taxane resistance in ER-positive disease. A high tau protein expression correlated with lower pCR [48]. Tau protein is a weak prognosticator with high tau protein expression being associated with better prognosis, even after adjustment for grade, hormone receptor and HER2 expression, and nodal status [50].

Clinical Importance of Molecular Classification of Breast Cancer

Molecular Subtypes, Genetic Profiling, and Neoadjuvant Therapy

The most common example of molecular subtyping and response to neoadjuvant treatment has been seen in HER2 overexpressing tumors to trastuzumab therapy [51]. WW binding protein 2 (WBP2) overexpression further correlates with response to trastuzumab-based neoadjuvant chemotherapy. Co-expression of WBP2 causes cyclin D inhibition and enhances trastuzumab effect [52]. A prospective study by Rouzier R et al. in 2005 [49] at M D Anderson Cancer Center reported post-chemotherapy response at molecular level by subtypes. There was a significant difference in the rate of pCR within subtypes. It was 45% for both basal-like (95% CI, 24–68) and CerbB2-positive subgroups (95% CI, 23–68), respectively, whereas it was only 6% (95% CI, 1–21) for luminal tumors. However, most of the basal-like and erbB2-positive tumors were high grade and ER negative, which are routine clinical and pathological variables of response prediction.

In the extraction of RNA for gene expression profiling and correlation with response to chemotherapy, both anthracyclines and taxanes have been reported in few studies. Ayers M et al. in 2004 [53] developed a 74-marker multigene model with a predictive accuracy of 78%. In another study by Gianni L et al. in 2005 [54], a group of 86 genes was proposed to have a significant correlation with pCR. A higher proportion of proliferation and immune-related genes along with a lower expression of ER-related genes was found to have an association with pCR. In a study that specifically looked at the response to docetaxel by Chang JC et al. in 2005 [55], tumors with higher response had an increased expression of genes involved in various processes like cell cycle, adhesion, cytoskeleton, protein transcription, modification and transport, and apoptosis. A higher expression of certain transcriptional and signal transduction genes led to resistant tumors. These are all study specific and have not reached the clinics as standards of care.

The literature is largely diverse on topoisomerase II-α and predicted response to chemotherapy [56, 57]. A recent trial [58] aimed to develop a gene expression signature using topoisomerase II-α to predict response or resistance to anthracycline-based chemotherapy in ER-negative tumors. Topoisomerase II-α, being a key enzyme in the process of DNA replication, makes it an important molecular target of anthracyclines as reported by Slamon D et al. [59]. They observed a significant association between topoisomerase II-α gene amplification and pCR, but not with topoisomerase II-α protein expression. They further proposed an ‘A score’ which integrated 3 molecular signatures, namely the topoisomerase II-α signature, stroma signature, and immune response signature, which have roles in the tumor microenvironment. This score was found to be significantly associated with pCR especially in patients who received anthracycline-based chemotherapy. The A score was reported to have a high negative predictive value of 0.98 (95% CI, 0.90–1.00) and the authors propose it to be useful in identifying the subset of patients who would not benefit from anthracyclines, thus avoiding the potential adverse effects.

The German GeparSixto study comparing chemotherapy + platinum versus chemotherapy as NACT did not show a difference in path CR (pCR) between the 2 arms in BRCA-mutated patients; the pCR rate was 65.4% with platinum and 66.7% without platinum. However, patients without pathogenic BRCA1 and BRCA2 alterations showed elevated disease-free survival rates when carboplatin was added (without carboplatin 73.5%; 95% CI, 64.1–80.8 versus with carboplatin 85.3%; 95% CI, 77.0–90.8; HR 0.53; 95% CI, 0.29–0.96; P = .04). In this study, patients without BRCA1 and BRCA2 germline mutations benefited from the addition of carboplatin and those with BRCA1 and BRCA2 mutations showed superior response rates without additive effects observed for carboplatin [60].

Thus, it would be interesting to unfold how upfront molecular profiling can add to the already existing armamentarium of clinicopathological factors in predicting response to therapy in a cost-effective way and guide decisions in the clinic.

Molecular Subtypes and Locoregional Therapy (LRT)

Locoregional therapy for breast cancer has undergone a paradigm shift over the last three decades, having moved from the Halstedian era of radical surgery to Fisher’s hypothesis of breast cancer being a systemic disease at inception and thereafter towards reducing extent of surgery to more conservative breast surgery. However, the essential basis of locoregional therapy still subscribes to the original Halstedian principle of complete eradication of all malignant cells from the breast and draining nodal basin [61].

Timing of Surgery Based on Molecular Subtype

Traditionally, neoadjuvant chemotherapy (NACT) in breast cancer has been reserved to down size tumors with the aim of breast conservation. Response to NACT and attaining a pathologic complete response varies by molecular subtype. A pooled analysis of data from 12 international trials, with close to 12,000 patients treated with neoadjuvant chemotherapy and surgery, found that the association between pathologic complete response and long-term outcomes was strongest in patients with triple-negative breast cancer and in those with HER2-positive, hormone receptor–negative tumors who received trastuzumab [62].

The CREATE-X trial and Katherine trial have suggested that with the more aggressive subtypes, i.e., the HER2-positive and triple-negative subtypes, not achieving a pathologic complete response is associated with inferior outcomes, indirectly suggesting that doing upfront surgery without chemotherapy may have negative impact, although this is not supported by any clear evidence currently. They clearly help to identify the good outcome of TNBC and HER2-expressing tumors.

The CREATE-X trial showed that disease-free survival was longer in the capecitabine group than in the control group, 74.1% versus 67.6%. Among patients with triple-negative disease, the rate of disease-free survival was 69.8% in the capecitabine group versus 56.1% in the control group (hazard ratio for recurrence, second cancer, or death, 0.58; 95% CI, 0.39 to 0.87), and the overall survival rate was 78.8% versus 70.3% (hazard ratio for death, 0.52; 95% CI, 0.30 to 0.90) [63].

Similarly, the Katherine study suggested that adjuvant T-DM1 after neoadjuvant trastuzumab and pertuzumab with chemotherapy reduced the risk of developing an invasive recurrence of the breast cancer or death by 50%, corresponding to an absolute improvement of a 3-year invasive disease-free survival rate by 11.3% (77% with trastuzumab and 88.3% with T-DM1) [64].

These trials have raised doubt on the timing of surgery based on subtype and the need to rethink the use of neoadjuvant therapy in these subtypes for selecting out those cases that will need prolonged adjuvant treatment due to significant residual disease to improve outcomes.

Risk of Local Failure Based on Subtypes

Numerous retrospective studies have documented differential local failure rates based on molecular subtypes. The luminal A subtype has shown a 5-year LRR rate ranging from 0.8 to 8% [65, 66] and the luminal B subtype has shown 5-year LRR rates ranging between 1.5 and 8.7% after breast conserving surgery and adjuvant radiotherapy [67, 68]. A positive HER2 status is known to confer a worse prognosis on the tumor with a 5-year LRR rate, ranging between 4 and 21% [69, 70]. Finally, LRR among patients with triple-negative breast cancer (TNBC) are documented to range between 3 and 17% [65, 66, 69]. This differential risk of LLR between the various subtypes mandates the need to consider the role of molecular subtypes in planning locoregional therapy for breast cancer patients (Table 1).

Table 1.

Risk of LRR by subtype

Molecular subtype	5-year risk of LRR
Luminal A	0.8–8%
Luminal B	1.5–8.7%
HER2 positive	4–21%
TNBC	3–17%

Breast conservation surgery with adjuvant radiotherapy is widely accepted for treatment of women with early-stage breast cancer. Local failure rates have been linked to young age (< 50), tumor size (> T2), negative hormone receptor status, and lymph node involvement [70].

There is emerging evidence in support of integrating molecular profiling while making decisions for locoregional treatment. Cheng et al. [71] explored the correlation between gene expression and LRR, and suggested that estrogen receptor and genomic predictive index are independent prognostic factors that affect LR control.

A large cohort study [72] evaluated the risk of LRR associated with choice of locoregional treatment (i.e., breast conservation versus mastectomy) in 768 women treated for triple-negative breast cancer. In this study, women who had undergone mastectomy had worse locoregional outcomes compared with those who had undergone BCT (a 5-year locoregional recurrence-free survival, 94% for BCS (n = 319) versus 85% for MRM (n = 287) or 87% for MRM+RT (n = 162), respectively). There was no statistical difference in LRR between MRM + RT and BCT group, thus suggesting an impact of RT in TNBC (HR 0.72; 95% CI, 0.36 to 1.43; P = 0.34). These results were similar to those published from multiple other studies where no significant difference was detected in LRR between TNBC and other subtypes with breast-conserving surgery [73, 74] (ER/PR positive, 2.3%; HER2 positive, 4.6%; and triple-negative, 3.2%).

Radiation and Molecular Subtypes

Kyndi et al. [75] retrospectively examined the impact of estrogen receptor, progesterone receptor, and HER2 status on outcomes in two Danish Breast Cancer Group trials of post-mastectomy radiotherapy. Although radiotherapy reduced the risk of local recurrence in all groups, rates of local recurrence remained highest in the triple-negative and HER2-positive subsets. This difference in local recurrence is observed even in cancers 1 cm or less in size and after treatment with neoadjuvant therapy [76]. The majority of local recurrences in triple-negative and HER2-positive cancers occur within the first 5 years after diagnosis, while local recurrence continues to occur in years 5–10 in patients with estrogen receptor–positive cancers [77].

Findings from a retrospective review of metaplastic breast cancer cases, which is a type of TNBC, from the 1998–2006 SEER database, by Tseng et al. [78] suggested that adjuvant radiation irrespective of the type of surgery played a role in improving overall survival. However, this needs to be confirmed in a prospective setting.

However, one must note that the available evidence in literature suggesting impact of locoregional treatment based on genotypic subtype is weak, because locoregional failure is associated with multiple patient- and tumor-related factors and it is difficult to conclusively show molecular profiling as an independent prognostic factor across all studies. Current data are somewhat conflicting, but together they suggest that outcomes for different molecular subtypes are dictated more by the inherent tumor biology not just extent of surgery and radiation therapy.

Management of the Axilla

Axillary lymph node status is the most significant prognostic factor for women with early breast cancer. Women with clinically positive axillary lymph nodes undergo a level I–II axillary dissection (ALND) and level III is cleared in cases with gross-positive level II/III nodes. In women with clinically negative axilla (cN0), there has been a paradigm shift towards conservation [79].

Sentinel lymph node biopsy or low axillary sampling is now the standard of care for cN0 cases. However, 30% of cN0 patients may be axillary node positive on pathology and they still do well without systemic therapy. The ACOZOG Z0011 and AMAROS trial challenged the age-old dogma of completion axillary dissection in women with 1–3 positive sentinel lymph nodes [80, 81]. The data from these studies is however useful mainly for women with low molecular risk of relapse. This is because in the ACOZOG Z11 study [80], most patients were ≥ 50 years of age (64%) and had clinical T1 tumors (68%), ER-positive tumors (77%), and only one positive SLN (60%). Only 16% of patients were ER negative; thus, the use of this data in ER-negative breast cancer is questionable. Also, data on HER2 positivity was not available. Similarly, the AMAROS study [81] included mostly T1 (17–18 mm) and HR-positive tumors (~ 78%), with 59.4–61.0% having micrometastatic disease in the SLN. The results from both trials that cannot be extrapolated to early breast cancer patients seen in India are mostly T2 with median tumor size 2.5 cm with 60% node positive, with 30% TNBC and 20% HER2neu positive. The cancers also are not screen detected and hence are relatively larger tumors with higher lymph node disease burden, mostly macro metastatic.

Molecular Subtypes and Chemotherapy

HER2 Enriched

Topoisomerase II-α gene aberration are mostly restricted to HER2-amplified tumors due to the physical proximity of the two gene loci. The reported range of the association in literature is 24–54% [82]. It has been shown in in vitro studies of ErbB2-amplified cell lines that co-amplification of Topo II-α gene leads to increased sensitivity to Topo II inhibitors like anthracyclines, while deletion leads to resistance [83].

HER2 positivity is described to predict a favorable response to anthracycline-based chemotherapy largely due to the co-amplification of Topo II-α [84–87]. Di Leo and Park used in situ hybridization techniques for diagnosing HER2 positivity, and in the cohort of HER2-positive tumors, with simultaneous amplification of Topo II-α, reported a better sensitivity to anthracyclines compared with non-anthracycline-containing regimens [84, 85]. Arriola et al. found a further significant association of the co-amplified cohort with ER positivity. On subgroup analysis, they noted better statistically significant survival of the HER2-amplified tumors that displayed co-amplification of Topo II-α [86]. All these studies propose topoisomerase II-α gene amplification as a reliable predictor of response to anthracyclines.

In accordance, subgroup analysis of early results of the BCIRG006 trial [82] (Breast Cancer International Research Group trial) suggest that tumors that lack HER2/Topo II-α co-amplification do not show any additional benefit of anthracyclines over non-anthracycline regimens that contain anti-HER2-targeted therapy. Patients whose tumors exhibited co-amplification of HER2/Topo II-α display equal benefit even without trastuzumab when anthracycline is part of the regimen. As a result, there are some isolated claims that HER2 non-amplified tumors may do without an anthracycline as part of their systemic therapy as the benefit of anthracyclines is restricted to HER2/Topo II-α co-amplified tumors [59, 87, 88]. However, they are based on isolated markers only and yet to be validated.

Another chemotherapy option from the long-term results of the same trial has been proposed for node-negative and low-risk node-positive HER2-amplified tumors with the addition of platinum and omission of anthracyclines. The 10-year follow-up of the BCIRG006 trial presented by Slamon DJ in 2015 [89] popularized the non-anthracycline TCH (docetaxel, carboplatin, trastuzumab) regimen as it showed similar DFS (73% versus 74.6% with AC-TH) and OS (83.3% versus 85.9% with AC-TH) benefit as the two-line chemotherapy regimen. The cardiotoxicity events were halved with the omission of anthracyclines while retaining the survival benefit.

TNBC

BRCA-mutated cancers frequently exhibit TNBC histology. They have a defective double-stranded DNA repair mechanism which is also seen in some sporadic TNBC cases [90]. Platinum agents act by causing breaks in double-stranded DNA which need functional BRCA proteins for their repair [91] and hence are found to be more effective in BRCA-mutated patients. It has been found in several studies that TNBC patients receiving platinum agents have demonstrated a superior pCR and hence better overall outcome [92–94]. Recently, Leong and colleagues have demonstrated that certain p53 family isoforms are expressed exclusively within about a third of TNBC tumors which through certain downstream pathways exhibit sensitivity to cisplatin [95]. Following which there has been a recent proposition that expression of p63 and p73 proteins could act as biomarkers indicative of sensitivity to platinum agents in TNBC cancers [96].

Higher sensitivity to anti-tubulin agents like ixabepilone could be attributed to the high expression of beta-III tubulin in TNBC tumors [97]. Ixabepilone in combination with other agents like capecitabine is being explored in TNBCs [98]. Eribulin, a semisynthetic non-taxane microtubule inhibitor, is shown to provide significant improvement in survival in heavily pretreated metastatic TNBCs [99].

PARP inhibitors have shown promise in the treatment of BRCA-mutated cancers and TNBCs [100–102]. PARP is an enzyme needed for base excision repair of single- and double-stranded DNA breaks. Through blockage of this mechanism and other roles, PARP inhibitors cause augmented cell death known as “synthetic lethality” [103]. The efficacy of these agents when combined with chemotherapy, toxicity profile, and timing of induction into the therapeutic armamentarium needs to be determined [104]. There is also a possibility that BRCA1-mutated tumors may be also less sensitive to taxanes [104].

mTOR inhibitors act through the PI3K/Akt signaling pathway that induces cell cycle arrest by suppressing the cellular proliferative responses [105]. Preclinical studies have suggested they maybe cytostatic [106]. A phase II study with everolimus in the metastatic setting reported low response rates with trends favoring ER-positive and HER2-negative subtypes [105, 107]. Studies of the efficacy of these agents in TNBC are underway [104].

Vascular endothelial growth factor (VEGF) is believed to be upregulated in TNBCs due to their high proliferation rate [108]. Certain small studies have shown beneficial effect of the addition of anti-VEGF agent bevacizumab to chemotherapy in the TNBC population [109]. The toxicities associated with these agents, however, are of concern.

Molecular Subtypes and Endocrine Therapy

With the advent of extended endocrine therapy, it is important to correctly select patients in whom it can be of maximum benefit keeping in mind the toxicity profile. The importance of understanding mechanisms of endocrine resistance helps in deciding and identifying correctly the responders.

Hormone Receptor and HER2-Positive Cancers

Retrospective studies suggest an inverse relationship between HER2 overexpression and hormone receptor positivity [110]. The proportion of breast cancer patients which express both HER2 and hormone receptor positivity (ER and/or PR) is approximately 10% [111]. Recent literature reports a complex interaction between the two pathways which may play a role in resistance to hormone therapy, particularly tamoxifen, and have an impact on clinical outcome [112, 113].

Mechanisms of resistance to endocrine therapy include loss of ER expression, altered micro-RNA expression, and epigenetic mechanisms among others. There is a suggestion that by targeting specific pathways involved in this “crosstalk,” there may be a possibility of overcoming this resistance [114]. Preclinical models have shown that dual targeting of ER and HER2 could reverse and delay the development of drug resistance [115–117]. There is evidence to show that combined treatment improves response rates and prolongs progression-free survival in the metastatic setting [118–120].

In a study that included hormone receptor–positive postmenopausal metastatic breast cancer, patients were randomized to receive the aromatase inhibitor letrozole with either lapatinib or placebo [118]. In the group with HER2 overexpression, lapatinib plus letrozole combination led to a better PFS of 8.2 months versus 3 months in the non-lapatinib arm. In another study [119] which included both HER2- and HR-positive patients only, randomization was to receive anastrozole with or without trastuzumab. The group receiving both anastrozole and trastuzumab experienced significant PFS benefit (hazard ratio = 0.63, 95% CI = 0.47 to 0.84, median PFS = 4.8 versus 2.4 months, P = 0.0016). The slightly higher side effects in the combination arm were managed without any major concern. This paves the way for early institution of combination therapy when possible and is now part of the European School of Oncology–European Society for Medical Oncology second international consensus guidelines for advanced breast cancer.

Newer agents are also being explored for this subgroup. The combination of letrozole and trastuzumab along with everolimus is being tested by Wheler JJ and colleagues (ClinicalTrials.gov identifier: NCT02152943) [121]. The ongoing NA-PHER2 trial by Lousberg L et al. (ClinicalTrials.gov identifier: NCT02530424) trial is testing the combination of trastuzumab, pertuzumab, palbociclib, and fulvestrant in the neoadjuvant setting [122]. It will be interesting to see the outcomes of these studies for future options.

Pitfalls of Molecular Classification

Results of a comparative analysis by Mitch Dowsett’s group [123] published in Lancet Oncology in 2010 revealed that microarray gene profiling does not accurately identify the group of HER2-positive breast cancers. Since this is a group that has tremendous survival benefit potential with targeted therapy options, it would be worthwhile to not rely on gene profiling to take treatment decisions for this subgroup.

They have also raised doubts about the reproducibility of tumor classification by microarray-based gene expression profiling due to standardization issues [124, 125]. Ioannidis et al. [126] evaluated the reproducibility of microarray-based gene expression profiling methodology from 18 articles. Majority of the analyses could not be reproduced leading to concerns on the implementation. The main reason cited by the group was incomplete information. They emphasized the regulatory need to make data availability public along with details of processing and analysis. Another issue to think is the discordance in prognostication of cases by genomic signatures and with models that use clinicopathological variable, reported to be as high as 30% [19].

Indian Experience

The molecular tests are primarily used in a population of low-risk disease, primarily node-negative ER and or PgR-positive and HER2neu negative. The patient population treated in year 2017 at a high volume tertiary referral center in India was evaluated to see the number of patients that could benefit from a molecular profiling technique and thereafter avoid chemotherapy.

One thousand sixty-three women early breast cancer patients with information available were evaluated for their eligibility to the TAILORx trial, 593 (55%) were node negative, of which 189 (17.7%) were ER/PgR-positive and HER2-negative [127] grade III tumors comprised of 95 out of the 189 (50.2%). Of 189 eligible, 75 (40%) were < 50 years of age, 114 (60%) were > 50 years. Applying the same ratios as in the published results of TAILORx [18], 40% of < 50 years (31/75) would probably fall in the strata of recurrence risk score < 15 and all 114 who were 50 years or older would be benefitted by being able to avoid chemotherapy. Thus, only 145 of 1063 (13.6%) women in our clinic with early breast cancer would really benefit with use of OncotypeDx, by being able to avoid chemotherapy, and not 70% as was reported in the study population of TAILORx trial. As shown by J Cuzick et al. [128], the IHC-4 score which consists of ER, PgR, HER2neu, and Ki67, the information contained in these commonly performed IHCs, is as good as the GHI RS score. They tested the IHC4 score on 1125 patients in a cohort from ATAC trial patients and found it to be comparable with the GHI RS score in predicting disease progression.

Conclusion

Molecular classification of breast cancer provides a new framework for the study and understanding of breast cancer, but how many robust molecular subtypes exist and how best to assign a molecular class to new cases are currently unknown. Multigene signatures can be used to help guide therapy and predict prognosis and response to preoperative chemotherapy. The extent to which multigene signatures improve patient outcome compared with current clinicopathological variable-based predictions is yet to be determined in head-to-head comparison prospective clinical trials [129].

For molecular subtype classification to be incorporated into routine clinical practice and treatment decision-making, stringent standardization of methodologies and definitions for identification of breast cancer molecular subtypes is needed [123]. Clinical variables measuring the extent of tumor progression, such as tumor size and nodal status, still add independent prognostic information to proliferation genes [130].

What still remains for these tests to enter the standard armamentarium of breast cancer management decisions are the financial implications. With the high costs, the wide applications of these tests remain limited, even if all rest is standardized. This also sets the platform for development of low-cost prognostic tools. One such tool in development is the image-based histomorphometric scoring or image-based risk scoring (IbRiS) [131]. This tool reiterates the importance of phenotypic expression in histology, with aid of artificial intelligence, in defining cancer behaviour towards treatment and outcomes If validated, it would bring down costs for helping to decide on adjuvant therapy in early and low-risk breast cancer.. An improved version is currently being tested and validated in Indian patients currently in an ongoing study at a tertiary cancer center funded by the Department of Science and Technology.