Molecular oncology update: Breast cancer gene expression profiling

Luminal subtypes

Luminal A and luminal B express genes associated with luminal epithelial cells of normal breast tissue that typically express luminal cytokeratins 8 and 18. These are the most common subtypes and make up the majority of ER-positive breast cancer. The routine clinical assay (immunohistochemistry [IHC]) characterizes them by high expression of ER, progesterone receptor (PR), and other genes associated with ER activation. Further, the luminal A and luminal B subtypes have some important molecular and prognostic distinctions Table 1.

Human epidermal growth factor 2-enriched

The subtype represent up about 10–15% of breast cancers and is characterized by high expression of the HER2 and proliferation gene clusters, low expression of luminal clusters genes (CK-7, 8, 18, 19, X-box-binding protein 1, GATA-binding protein 3, hepatocyte nuclear factor 3, Annexin XXXI, and ER 1). These tumors are most often positive for HER2 and negative for ER and PR. About 30% of HER2-enriched tumors are clinically HER2-negative, and on the other, 50% of clinical HER2-positive breast cancers are HER2-enriched. They have a poorer prognosis compared with luminal A tumors and are predictive of response to herceptin.

Estrogen receptor-negative subtypes

ER-negative subtypes – this includes multiple subtypes, such as basal-like, claudin-low,¹ and interferon-rich,¹⁰ molecular apocrine. The basal-like subtype has some similarity in gene expression to that of the basal epithelial cells of normal breast tissue (expressing CK-5, 6, 17, integrin-β4, laminin, and fatty-acid binding protein 7.4) and represent about 15–20% of breast cancers. It is characterized by low expression of the luminal and HER2 gene clusters thus similar to triple-negative breast cancers (TNBCs). However, there is a significant discordance (up to 30%) between the basal-like and TNBC. These subtypes exhibit high frequency of BRCA1 mutations, increased genomic instability, high histologic grade, and have a poorer prognosis compared with luminal A tumors.

Claudin-low breast cancer

Claudin-low breast cancer represents 5% of breast cancers and is characterized by overexpression of genes associated with epithelial-mesenchymal transition-like chemokine ligand 12, vimentin and fibroblast growth factor 7, Krüppel-like factor 2, integrin α5, moesin, vascular endothelial growth factor-C, matrix metalloproteinase-9, CD79b, CD14, CD44⁺/CD24⁻, and ALDH1A1. This subtype frequently exhibits metaplastic and medullary differentiation and are similar to TNBC with a poorer prognosis compared with luminal A tumors.

Interferon-rich and molecular apocrine subtypes

These subtypes are rare, characterized by interferon-regulated genes, such as STAT1 and activation of androgen receptor signaling, respectively.

Normal-like subtypes

Normal-like subtypes are typified by similar gene expression pattern as the normal breast. It yet remains enigmatic as to whether it represents a separate subtype or a technical artifact introduced by low tumor cell composition of the sampled specimen.

Tools for molecular classification

For identification of four-basal-like, HER2-enriched, luminal A and luminal B subtypes, tools called single sample predictors (SSPs) are available including Blueprint, SSP203 (500 genes), SSP2006 (306 genes), and predictor analysis of microarray 50 (PAM50) (50 genes).^6,11,12 The SSPs utilizes large sets of tumor-intrinsic genes, which are analyzed by hierarchical clustering followed by nearest centroid classification. The limitation includes subtyping is dependent upon the type of SSP used for analysis, another intrinsic subtype called normal breast-like subtype is considered invalid and HER2 enriched subtype does not encompass all cases classified as HER2-positive by IHC/fluorescence in situ hybridization.

Oncotype DX

The 21-gene RS is well-validated prognostic assays. The assay calculates RS viz. low (RS < 18), intermediate (RS 18–30), or high score (RS > 30) using mathematical formula that includes sum of expression of 16 cancer-related genes plus 5 reference genes as shown in Table 2, generated using reverse-transcriptase polymerase chain reaction (PCR) on the formalin-fixed, paraffin embedded (FFPE) blocks, which are optimized to predict the locoregional and distant relapse despite tamoxifen therapy. Further, RS is able to predict relapse with aromatase inhibitors in postmenopausal women. The RS was validated in various large, independent retrospective multicenter trials involving over 3300 patients. The key study includes NSABP study B-14, NSABP B-20, Kaiser Permanente study, Southwest Oncology Group (SWOG) 8814 study. Currently, prospective trials (i) Oncotype DX Trial Assigning IndividuaLized Options for Treatment (Rx) (TAILORx) trial and (ii) RxPONDER trial (SWOG S1007) are ongoing to validate the predictive potential of Oncotype DX.^{13,14,15,16,17,18,19,20}

Oncotype DX is included in guidelines-St Gallen, European Society for Medical Oncology, American Society of Clinical Oncology, National Comprehensive Cancer Network as a decision tool enabling the identification of patients who are most likely to benefit from adjuvant chemotherapy and is indicated for women with node-negative, ER-positive breast cancer to determine the prognosis in patients recommended proceeding with at least a 5-year course of endocrine therapy.

Amsterdam 70-gene profile

US Food and Drug Administration approved commercially available test, known as MammaPrint, uses microarray technology to analyze 70 genes extracted from either FFPE tissue or fresh frozen tissue. It is recommended for early breast cancer of all ages with tumors size <5 cm, with the lymph node involvement up to three nodes. Unlike Oncotype DX, it can be used to determine the prognosis in patients with breast cancer regardless of hormone receptor-status and in patients with the HER2 positive disease. MammaPrint categorizes patient into two groups (i) low risk (ii) and high risk group for breast cancer distant relapse within 10 years of the initial diagnosis.

The 70 genes can be classified into various categories as shown in Table 3.

The clinical validity of the 70-gene prognostic profile has been demonstrated in multiple studies, including retrospective analysis of 295 consecutive invasive breast tumors from the tumor bank at the Netherlands Cancer Institute, 302 patient data contributed by the TRANSBIG consortium (5 participating European hospitals), prospective analysis of 427 women in The MicroarRAy PrognoSTics in Breast CancER (RASTER) trial. Currently, a multicentric prospective The Microarray in Node negative and 1–3 positive lymph node Disease May Avoid ChemoTherapy (MINDACT) clinical trial is ongoing which includes 6000 patients from 93 participating institutions in nine European countries, the results of which are awaited.^{21,22,23,24,25,26,27,28,29}

Predictor analysis of microarray 50 risk of recurrence score

The PAM50 is a 50-gene test utilizes microarray and quantitative reverse transcription PCR technology on FFPE tissue. The PAM50 test characterizes an individual tumor by intrinsic subtype and along with tumor size, generates the risk of recurrence (ROR) score that can stratify patients with ER-positive, node-negative disease into high, medium, and low subsets.^11,30,31

The clinical utility of the PAM50 and ROR score has been demonstrated in several studies^31,32,33,34 including ATAC trial (adjuvant tamoxifen or anastrazole), a retrospective analysis of 1017 postmenopausal patients and Austrian Breast Cancer Study Group 8 (ABCSG-8) trial, a retrospective analysis of 1478 postmenopausal patients, where ROR predicted the risk of distant recurrence at 10 years in both node-negative and node-positive disease and added significant prognostic information in all subgroups of patients.^34,35

Rotterdam 76-gene signature

The Rotterdam/Veridex 76-gene prognostic signature uses separate prognostic gene sets for ER-negative (16 genes) and ER-positive (60 genes) disease.³⁶ The 76 prognostic gene signature has been evaluated in an independent set of 171 tumors, demonstrating 93% sensitivity and 48% specificity predicting distant metastasis-free survival independent of clinical variables. Currently, the test is not commercially available and warrant more studies to validate the clinical utility of this test prior to its use in clinical practice.

Genomic grade index

Meant for improving prognostic value of tumor grading, which involves analysis of gene expression signature of 97 genes that distinguish histological Grade 1 from Grade 3 tumors³⁷ and reclassify Grade 2 tumors into two groups of high versus low ROR.³⁷ Subgroup analysis of 204 patients treated in the PACS01 trial, genomic grade index outperformed histologic grade and other proliferation markers (Ki67 messenger RNA and protein, mitotic activity index) as a predictor of disease-free survival.³⁸ Further studies are required for its clinical utility.

Molecular grade index

Another tumor grading multigene assay reclassifies Grade 2 tumors into Grade 1-like and Grade 3-like tumors. It involves 5-gene expression index to stratify patients into low or high ROR.³⁹

Breast cancer index

Breast cancer index (BCI) is a combination of two profiles, the HOXB13-to-IL17BR expression ratio (H: I ratio) and the molecular grade index. Various studies have validated the BCI as a predictor of endocrine responsiveness among patients with ER-positive breast cancer³⁹ and of prognosis.^39,40 Yet again, BCI clinical utility remains to be determined.

Endopredict

Involves RNA-based analysis of 11 genes (8 cancer related and 3 reference genes) to calculate a prognostic score. Its prognostic value is independent of conventional prognostic factors⁴¹ that was validated using the data from two ABCSG trials (ABCSG-6 and ABCSG-8). However, the performance of EP as compared with other prognostic tests has not been evaluated.

Limitation of the multigene assays

There is a lack of level I evidence. The evidence is based on the retrospective analysis; results of the two randomized prospective trials, MINDACT and TAILORx are still awaited. Based on the TAILORx interim analysis of the women in the lowest- risk group suggests that at 5 years, rates of distant relapse-free survival were 99.3 percent, of invasive disease free survival were 93.8 percent, and of overall survival were 98.0 percent. These results provide prospective evidence that the gene expression test identifies women with a low risk of recurrence who can be spared chemotherapy.¹⁹ The list of genes in various assays is not stable with negligible overlap but have similar performance and a good concordance. The bias of determining the prognosis correlates with the proliferation-/cell cycle-related genes, thus, multigene assay serves mere surrogates of proliferation. The application is limited for the ER-positive and HER2-negative breast cancers but cannot determine the prognosis for the early stage TNBC. Almost 40–60% of clinically intermediate-risk patients are found to be intermediate-risk RS group, thus the clinical dilemma where prognostication and management remains unclear. The prediction potential is valid for short-term distant recurrence (<5 years) but indeterminate prognostic potential after 5–10 years of follow-up. This multigene assay doesnot consider analysis of tumor tissue microenvironment including stromal cells, inflammatory infiltrate, and normal breast tissue. The percentage of nonneoplastic cells has a substantial impact on the final expression profile and meaningful prognostic signatures. Thus, next generation assays taking this analysis into account and overcoming the limitation are under development.

Second generation gene expression signatures

Most of the first generation signatures as discussed above were developed based on the epithelial cancer cells, and the contribution of stromal cells was not taken into account. The accuracy of gene predictors assay may be influenced by varying proportions of stromal components,⁴² and comprehensive GEP of each cell type has demonstrated that at the transcriptome level, changes occur in epithelial, myoepithelial, and stromal cells.⁴³ Various research groups have used prognostic signatures derived from the profile of stromal cells,^44,45,46 by analyzing the immune system^44,47,48 or cancer-related pathways^49,50,51 as shown in Table 4.

Multigene markers of response to chemotherapy

Oncotype DX-RS predicts the benefit from the addition of chemotherapy to tamoxifen. This was evident in NSABP B-20 trial where node-negative, ER-positive breast cancers were treated with Cyclophosphamide, Methotrexate Fluorouracil (CMF) in addition to tamoxifen. The 10 years distant disease-free (DDF) survival based on RS were high RS-88 versus 60% (relative risk [RR] 0.26, 95% confidence interval [CI] 0.13–0.53); low RS-96 versus 97% (RR 1.31, 95% CI 0.46–3.78); and intermediate RS – 90 versus 91% (RR 0.61, 95% CI 0.24–1.59). In SWOG 8814 trial node positive, ER-positive breast cancer were treated with CAF in addition to tamoxifen that resulted in DDF at 10 years high RS – DDF 55 versus 43%, (hazard ratio [HR] 0.59, 95% CI 0.35–1.01) and overall survival-73 versus 54%, (HR 0.56, 95% CI 0.31–1.02). These results were not seen among women with a low or intermediate RS.

Various research groups using supervised approaches have developed multigene signatures designed to predict response in patients receiving either chemotherapy or endocrine therapy. Several studies have attempted to chemotherapy by comparing gene expression profiles between high sensitivity and low-responsive tumors.^{52,53,54,55,56,57,58} The majority of the studies focused on identification of multigene signatures predicting response to neoadjuvant chemotherapy on the analyzed tumor samples obtained from biopsies taken at diagnosis before initiation of chemotherapy as shown in Table 5.

Despite the promising results, the signatures of chemotherapy sensitivity have so far had limited use in clinical practice. None of the different predictors of chemosensitivity is commercially available, and additional evidence is still required before they can be implemented in clinical practice.

Multigene markers of response to endocrine therapy

ER status has an important negative predictive value for evaluating the response to antiestrogen therapy. However, ER expression alone fails to predict which ER-positive tumor will respond or be resistant to different modalities of endocrine therapies. Various microarray-based gene expression signatures to predict the outcome of tamoxifen-treated patients have been developed by various research groups as shown in Table 6.

Sensitivity to endocrine therapy (SET) index was developed in a large series of ER-positive breast cancers was based on the principle that expression of genes correlated with ER may better predict response to endocrine treatment than ER expression alone.⁵⁹ Analysis of 165 genes coexpressed with ER, the SET index seems to be predictive of benefit from endocrine therapy independently of the inherent prognosis of the tumor and to identify a subset of tumors associated with an excellent prognosis and no relapse. Studies evaluating the clinical relevance of the SET index are warranted to expand its indications in clinical practice.

Financial support and sponsorship

Nil.