Molecular Classification and Prognostic Signatures of Breast Tumors

Fig. 5.1

Breast cancer classification into five molecular subtypes. Hierarchical clustering of 115 tumor tissues and seven nonmalignant tissues using the “intrinsic” gene set. Experimental dendrogram showing the clustering of the tumors into five subgroups. Branches corresponding to tumors with low correlation to any subtype are shown in gray. (From Sorlie et al. [20])

The five major molecular subtypes identified in these studies differ not only in regard to their pattern of gene expression and clinical features but also in regard to the response to treatment and clinical outcome [5, 6, 20, 29, 30]. Patients with luminal tumors respond well to endocrine therapy; however, luminal A and luminal B tumors respond differently to the type of endocrine agent used (tamoxifen or aromatase inhibitors) and also exhibit a variable response to chemotherapy [31–34]. Patients with luminal A tumors present with an overall good prognosis with a 5 year survival rate of approximately 90 %. Patients with HER2-amplified tumors respond to trastuzumab antibody monoclonal therapy and to anthracycline-based chemotherapy; however, they generally have a poor prognosis and their 5 year survival rate can be as low as 20 % [35, 36]. Finally, patients with the basal-like tumor subtype have no response to endocrine therapy or trastuzumab therapy; however, they can be sensitive to platinum-based chemotherapy and poly(ADP-ribose) polymerase 1 inhibitors [37–39]. These tumors are especially common in African-American women and generally also have poor prognosis [40–42]. Interestingly, in the neoadjuvant setting, the intrinsic subtypes have also been found to exhibit different responses to treatment. The pathological complete response rates to standard chemotherapy based on anthracycline and taxane was approximately 7 % for the luminal A subtype, 17 % for the luminal B subtype, 36 % for the HER2-positive subtype, and 43 % for the basal-like subtype [32].

To study the utility of these subtypes in breast tumor classification, a total of 189 breast tumors across 1,906 “intrinsic” genes were analyzed by Parker et al. [32]. They identified a set of 50 genes that were further validated and compared for reproducibility of classification across different prediction methods and different patient cohorts. This analysis profiled by quantitative real-time PCR a total of 122 breast cancers from the 189 individuals into the “intrinsic” subtypes luminal A, luminal B, HER2-positive, basal-like, and normal-like. Owing to its high reproducibility, a standardized method of classification was developed, the Prediction Analysis of Microarray 50 (PAM50) Breast Cancer Intrinsic Classifier test, which is currently commercially available (ARUP Laboratories, Salt Lake City, UT, USA). The PAM50 assay offers the measurement of the expression level of 55 genes (50 classifier genes and five housekeepers) and is recommended for all patients diagnosed with invasive breast cancer, regardless of tumor stage or ER status.

The gene expression intrinsic subtypes were discussed for consideration at the last St. Gallen International Breast Cancer Conference [42]. A simplified clinicopathological classification that defines subtypes on the basis of the immunohistochemical analysis of ER, PR, and HER2 status and Ki-67 labeling index (Ki-67 is a cell proliferation marker), similar to what was proposed by Cheang et al. [31], was endorsed (Table 5.1). The breast cancer subtypes defined by this classification are similar but not identical to the five intrinsic subtypes and represent a convenient approximation that can be used in considerably less expensive and less complex assays. In general, the therapy recommendations for this classification follow the “intrinsic” subtype classification: luminal A tumor patients generally require only endocrine therapy, considering that they are mostly less responsive to chemotherapy; luminal B patients, in addition to endocrine therapy, should receive chemotherapy (both anthracycline-based and taxane-based); patients with HER2-positive tumors should receive chemotherapy and 1 year of treatment with trastuzumab; and patients with triple-negative tumors should be treated with chemotherapy (also anthracycline-based and taxane-based in addition to an alkylating agent, typically cyclophosphamide).

Table 5.1

Intrinsic and immunohistochemical (IHC) subtypes and type of treatment recommended (St.Gallen’s conference, 2011)

Intrinsic subtype	IHC subtype	Definition	Type of treatment
Luminal A	Luminal A	HER2 positive Ki-67 low	Endocrine therapy alone
Luminal B	Luminal B (HER2 negative)	HER2 negative ER positive PR positive Ki-67 high	Endocrine therapy with or without cytotoxic therapy
Luminal B	Luminal B (HER2 positive)	HER2 positive ER positive PR positive	Cytotoxic therapy plus anti-HER2 therapy plus hormonal therapy
HER2 overexpression	HER2 positive	HER2 positive ER negative PR negative	Cytotoxic therapy plus anti-HER2 therapy
Basal-like	Triple negative	HER2 negative ER negative PR negative	Cytotoxic therapy

Modified from Goldrisch et al. [42] and Perou et al. [3]

ER estrogen receptor, PR progesterone receptor, HER2 human epidermal growth factor receptor 2, ki-67 antigen ki-67- protein marker for cell proliferation

The St. Gallen panel [42] did not endorse the measurement of cytokeratin 5/6 or epidermal growth factor receptor for the determination of basal-like tumors for clinical decision making. In the future, this formal subtyping is very likely to be refined and expanded to include the measurement of novel tumor markers; presently, however, the St. Gallen consensus recommends the classification of breast cancer subtypes and the guide to therapeutic decisions to be based only on the four clinicopathological markers described above.

Although gene expression profiling has greatly contributed to the determination of breast cancer subtypes and their associated differential prognosis, presently this defined “intrinsic” molecular classification is not routinely used in clinical practice to classify the patient’s breast tumor subtype, and no new target therapies have yet resulted for these subtypes [43–45]. Several technical challenges limit its use in clinical practice, including not only the prohibitive costs of the equipment and reagents for the expression assays and the lack of suitable technical personnel to conduct the complex informatics data analysis, but mainly the lack of reproducibilty and uniformity among laboratories in relation to the selection of the “intrinsic” genes to be used. This latter limitation can be easily perceived by the existence of multiple versions of molecular classification systems developed or under development. In addition, most gene expression microarray analyses were performed by independent investigators using different methods and applied to different patient populations. Another important variability was the cellular composition of the tissue samples (stroma, tumor, and normal cells) in these studies [46, 47]. Cleator et al. [46], in evaluating the cellular composition of the classified samples, demonstrated that the percentage of invasive cancer cells within a sample influenced the expression profile; at least 10 % of the genes (144 genes) were found to correlate with cellular composition.

Other challenges include biological inherent facts, such as that the subtypes assigned by microarrays do not always correspond to the same subtype defined by the routine immunohistochemical (IHC) staining that is used in standard clinical assays [32, 48, 49]. In the retrospective analysis by de Ronde et al. [48], 195 stage II and stage III breast tumors from patients that received neoadjuvant treatment were classified by both IHC and messenger RNA expression analysis on then basis of the molecular classification. The IHC and molecular subtypes showed high concordance with the exception of the HER2 group, where 60 % of the HER2-positive tumors were not classified as the HER2 molecular subtype. In addition, for the ER-positive tumors, neither the PR status nor the endocrine responsiveness index (all the tumors showed similar degrees of response to chemotherapy) accurately distinguished the tumors into the luminal A and luminal B subtypes. In fact, several studies have suggested that these ER-positive subtypes may not be completely separate entities [50–52].

Once these and others critical challenges are overcome and a universally accepted signature for identifying breast cancer subtypes is established, assays that can maintain a similar level of analytical reproducibility and clinical utility can be developed and implemented for the molecular classification of a patient’s breast tumor. However such assays should not be expected, at least not soon, to replace the traditional breast cancer classification systems.

5.3 Prognostic Gene Expression Signatures

In the daily management of breast cancer, the selection of the most appropriate adjuvant treatment for an individual patient remains a challenge, despite the excellent assistance of the established therapy guidelines such as those of the St. Gallen consensus [1, 42], the National Institutes of Health [53], and Adjuvant! Online (http://www.adjuvantonline.com). The ability to identify breast cancer patients with either a very high or a very low risk of recurrence, who would need adjuvant systemic therapy, from those who could be spared such type of treatment is critical. The power of making this distinction at the time of diagnosis, from the analysis of the patient’s primary tumor, would substantially improve breast cancer survival.

Several multigene signatures that predict outcome and response to therapy in breast cancer have been developed through the data obtained from gene expression profiling (for reviews, see [5–9] (Table 5.2). In these studies, major prognostic factors, such as lymph node status and ER status, were addressed and have allowed subgroups of tumors with a very distinct clinical outcome that could not be predicted by conventional prognostic factors to be distinguished in the analysis of the patient’s primary tumors. The main objective in most of these studies was to predict which patients would benefit from a more aggressive treatment and which patients would be unlikely to respond and therefore for whom there would not be a significant survival benefit.

Table 5.2

Commonest prognostic gene expression breast cancer signatures commercially available

Gene expression signatures	Patient population	Prediction	Number of genes	Material	Assay	Company
Oncotype Dx	ER positive/negative LN negative Tamoxifen treated	Risk of recurrence	21 genes	FFPE	RT-PCR	Genomic Health (Redwood City, CA, USA)
MammaPrint	ER positive/negative LN negative Tumor size < 5 cm Age < 61 years	Risk of distant metastasis	70 genes	Frozen	Microarray	Agendia (Huntington Beach, CA, USA, and Amsterdam, The Netherlands)
PAM50	LN negative ER positive/negative No systemic therapy	Risk of relapse	55 genes	Frozen/FFPE	qRT-PCR/microarray^a/nCounter^b	ARUP Laboratories (Salt Lake City, UT, USA): (qRT-PCR format) Nanostring Technologies (Seattle, WA, USA): nCounter format
MapQuant DX	ER positive/negative LN positive/negative	Molecular grading	97 genes	Frozen/FFPE	Microarray	Ipsogen (New Haven, CT, USA, and Marseilles, France)
Breast Cancer Index	ER positive LN negative	Risk of late recurrence Response to endocrine therapy	2 genes, HOXB13:IL17R molecular grade index	FFPE	RT-PCR	BioTheranostics (San Diego, CA, USA)

ER estrogen receptor, LN lymph node, FFPE formalin-fixed paraffin-embedded, RT–PCR real-time PCR, qRT–PCR quantitative real-time PCR

^aPAM50, marketed by Arup Laboratories as a breast cancer classifier

^bIn development

Vant’veer et al. [54], some of the pioneers of these studies, proposed a prognostic gene signature to identify a group of good prognosis patients with minimal risk of development of distant metastasis within 5 years after diagnosis. The expression of 25,000 genes was analyzed in primary breast tumors, and a set of 70 genes with differential expression profiles separated the patients into two categories, “poor” and “good” signature groups, on the basis of their risk of developing distant metastasis. Among the genes that were upregulated in the poor signature group were genes involved in the cell cycle, angiogenesis, invasion and metastasis, and signal transduction, such as CCNE2, MCM6, MMP9, MP1, RAB6B, PK428, ESM1, and the vascular endothelial growth factor receptor FLT1. Subsequent studies confirmed the reproducibility of the initial 70-gene signature as a predictor of outcome independently of traditional clinical–histopathological prognostic markers [55–57]. This validation analysis led to the development of the commercial test MammaPrint developed by Agendia (Amsterdam, The Netherlands). This test is approved by the Food and Drug Administration for use in patients less than 61 years old, who are lymph-node-negative, and who present with a tumor smaller than 5 cm in size. This signature is currently being evaluated in a large clinical trial, MINDACT (Microarray In Node-Negative and 1–3 Positive Lymph-Node Disease May Avoid Chemotherapy Trial), which is performed in breast cancer patients with ER-positive, lymph-node-negative disease with long-term follow-up and known clinical outcome. The primary end point is to test its robustness and clinical applicability in identifying patients who could be spared the use of chemotherapy without affecting the survival outcome. On the basis of an independent validation study [58], this trial now also includes patients with one to three positive axillary lymph nodes.

The other prognostic signature also commercially available is the Oncotype DX breast cancer assay (Genomic Health, Redwood City, CA, USA). This assay was developed on the basis of the identification of 250 selected genes with different expression profiles [59–61], initially tested in patients from the National Surgical Adjuvant Breast and Bowel Project (NSABP) B-20 clinical trial [62]. After statistical analysis and clinical validation, 21 genes (16 cancer-related genes and five reference genes) were selected and their expression analysis was translated into a “recurrence score” (RS), which was then used to assign the patients to one of three groups, on the basis of the risk of developing distant metastasis: low risk (RS < 18), intermediate risk (RS ≥ 18 and RS < 31), and high risk (RS ≥ 31) [63

Only gold members can continue reading. Log In or Register to continue