In 2000, the era of genomic medicine in breast cancer was introduced when Perou et al¹ identified the “molecular portraits” of breast cancer. By analyzing the gene expression patterns of 8102 genes from 65 breast tumors, these authors identified four broad groups of tumors: (i) estrogen receptor (ER) positive/luminal-like, (ii) basal-like, (iii) ErbB-2 positive, and (iv) normal breast. These subtypes appeared biologically distinct as evidenced by their distinct genetic makeups and clinically distinct as evidenced by significant differences in outcome across the four groups. These initial subtypes have since been further refined by a number of groups, including The Cancer Genome Atlas Network,² and a consensus has emerged that there are four important subtypes, often referred to as luminal A, luminal B, HER2/neu positive, and basal-like, each with distinct features and clinical behavior.

The classification of breast cancer according to molecular subtypes has created a new paradigm for understanding and treating this disease. This new paradigm has opened the door to more personalized treatment and has identified opportunities for novel areas of investigation and clinical trials. This chapter will outline the characteristics of the different molecular subtypes of breast cancer and the clinical implications of each subtype with regard to treatment and prognosis.

Biological Characteristics

The luminal A subtype is the most common breast cancer subtype. Luminal A tumors are ER positive, progesterone receptor (PR) positive, and HER2/neu negative. They are characterized by low histologic grade and good prognosis. Luminal A tumors have low expression of Ki-67 and are most likely to retain activity of major tumor suppressors RB1 and TP53.^2–4

Luminal B tumors have a higher grade and higher Ki-67 expression than luminal A tumors and are associated with a worse prognosis. These tumors are typically ER and/or PR positive and HER2/neu negative. The frequency of TP53 mutations is higher in the luminal B subtype than in the luminal A subtype.^2–4

Basal-like tumors, which account for 10% to 20% of breast cancers, are often referred to as triple-negative breast cancers (TNBCs) because they are typically negative for ER, PR, and HER2/neu expression. Basal-like tumors have a high frequency of TP53 mutations (80%) as well as RB1 and BRCA1 mutations. Basal-like tumors are characterized by high Ki-67 expression, high histologic grade, and a poorer prognosis compared to the luminal subtypes.^2–4

The HER2/neu-positive subtype, which accounts for 15% to 20% of breast cancer cases, is divided into two subgroups: tumors that express ER genes and tumors that do not express ER genes. The HER2/neu-positive subtype is characterized by overexpression of HER2/neu and HER2/neu-associated genes, high Ki-67 expression, high histologic grade, and a poor prognosis. HER2/neu-positive tumors have higher expression of TP53 than do luminal-HER2/neu-positive tumors, which have higher expression of the luminal cluster genes.^2–4

Prognosis

Numerous studies have now confirmed the prognostic significance of these tumor subtypes,^3–6 with most studies demonstrating that luminal subtypes are associated with the best outcomes and basal-like tumors the worst outcomes. These findings hold true regardless of whether the subtypes are defined according to genomic profiles or biomarker surrogates (see the section Practical Considerations below); this fact underscores that the propensity to metastasize to distant sites is an inherent characteristic of tumors based on their genetic features. For patients with recurrence, locations of distant recurrence also appear to be characteristic of tumor subtype. Kennecke et al⁷ looked at the metastatic behavior of breast cancer subtypes and found that bone was the most common site of distant metastasis in all subtypes except basal-like. In multivariate analysis, compared with luminal A tumors, luminal-HER2/neu-positive and HER2/neu-positive tumors were associated with significantly higher rates of brain, liver, and lung metastases, and basal-like tumors were associated with higher rates of brain, lung, and distant nodal metastasis but significantly lower rates of liver and bone metastasis.⁷

Epidemiology

In addition to differing in biological characteristics and prognosis, the breast cancer subtypes differ in their associations with populations and demographic groups. The Carolina Breast Cancer Study⁶ examined cases of invasive breast cancer to determine the prevalence of subtypes within racial and menopausal subsets and to determine the associations between the breast cancer subtypes and tumor characteristics and survival. Subtypes were defined on the basis of immunohistochemical (IHC) assessment of ER, PR, HER2, cytokeratin 5/6, and HER1. The analysis provided important information about population-based differences in phenotypes that resulted in a poorer prognosis among young African American women.

Luminal A tumors (ER+, PR+/−, HER2−) and luminal B tumors (ER+, PR+/−, HER2+) were found in older patients than were the other subtypes. The luminal A subtype was the least common subtype among premenopausal African American women but the most common subtype overall, accounting for 51% of the cases in the study. Basal-like tumors (ER−, PR−, HER2−, cytokeratin 5/6+, and/or HER1+) were more common among premenopausal African American women than among postmenopausal African American women and non-African American women. In contrast, the incidence of HER2/neu-positive and ER-negative tumors did not vary by race or age.

The distribution of breast cancer subtypes has also been examined in the subset of women with BRCA-related breast cancers, which account for 5% to 10% of all breast cancers. Results show that the spectrum of subtypes of BRCA-related breast cancers differs from that of sporadic breast cancers. In BRCA1 mutation-related tumors, the predominant molecular subtype is the basal-like subtype.^8–10 Using 182 tumors obtained from BRCA1 mutation carriers and 109 controls collected as part of the Breast Cancer Linkage Consortium, Lakhani et al¹⁰ demonstrated that 70% of BRCA1-related tumors were basal-like tumors (ER−, cytokeratin 14 and/or cytokeratin 5/6 positive), compared to only 9% of matched controls. Results from the Consortium of Investigators of Modifiers of BRCA1/2 (CIMBA) also showed a strong association between BRCA1 mutations and TNBC (70% of BRCA1 mutation-related tumors were of TNBC subtype) and an odds ratio of 11 for risk of TNBC among BRCA1 mutation carriers.¹¹ However, the CIMBA group also noted that as the age at diagnosis of the women with BRCA1-related tumors increased, the proportion of TNBC tumors decreased. Consistent with these clinical observations, BRCA1-related tumors have been shown to consistently segregate with sporadic basal-like breast cancers in hierarchical clustering analysis of microarray expression profiling data.⁸

In BRCA2-related tumors, the primary phenotype appears to be luminal. In a retrospective study of 157 BRCA2-related cancers and 314 control tumors accrued through the Breast Cancer Family Registry, Bane et al¹² found that BRCA2-related tumors tended to be of higher grade than the control tumors. After adjustment for grade, BRCA2-related tumors were more likely than control tumors to be ER positive (68% vs. 62%), were less likely to express cytokeratin 5, a basal cytokeratin (13% vs. 20%), and were less likely to overexpress HER2 (6% vs. 12%). Interestingly, the CIMBA investigators, while confirming the higher rates overall of non-TNBC among BRCA2 mutation carriers (84%), noted that the proportion of TNBCs in this cohort of women increased with increasing age.¹¹

The strong association between BRCA1 and basal-like breast cancer has also prompted studies of germline BRCA mutations among women diagnosed with TNBC.^2,13,14 Data from The Cancer Genome Atlas show that 20% of basal-like breast cancers harbor a BRCA1 or BRCA2 mutation. However, the probability of a BRCA1 or BRCA2 mutation in a patient with basal-like breast cancer varies considerably by age, ethnicity, family history, and tumor histologic features. Among women of Ashkenazi Jewish heritage who present with TNBC, the probability of a BRCA mutation is 24% to 39%.^9,15 Among women with TNBC who also report a family history of breast cancer, the prevalence of a BRCA1 or BRCA2 mutation can be as high as 42%.¹³

While molecular classification of tumors according to gene expression profiles is the most accurate approach to assigning tumor subtype, this approach has limited applicability in clinical practice because of time, expense, and resource considerations. Therefore, large-scale clinical application of molecular subtypes has required adaptation of existing biomarkers to serve as surrogates for gene profiling data.^6,16 Specifically, grouping the readily available information about ER, PR, and HER2/neu status of the tumor has allowed approximation of the major genomically derived signatures.¹⁷ The groupings based on ER, PR, and HER2/neu status, referred to as constructed subtypes, are as follows: luminal A, ER positive, PR positive, and HER2/neu negative; luminal B, ER positive, PR positive, and HER2/neu positive; HER2/neu, ER negative, PR negative, and HER2/neu positive; and basal-like (commonly referred to as TNBC), ER negative, PR negative, and HER2/neu negative (Table 75-1).

While these groupings are only approximations, they yield results similar to those of studies utilizing full gene profiles. However, there are limitations to using receptor status as a surrogate for molecular subtypes. One limitation is that approximately 20% of results of IHC ER and PR testing worldwide may be inaccurate because of variations in preanalytic variables, thresholds for positivity, and interpretation criteria.¹⁸ In an effort to improve the accuracy of IHC ER and PR testing in breast cancer, the American Society of Clinical Oncology (ASCO) and the College of American Pathologists (CAP) convened and developed a guideline for ER and PR testing.¹⁸ Among the recommendations in this guideline is a cutoff of ≥1% cells positive to distinguish “positive” from “negative” cases.¹⁸ Another limitation is the problem of how to classify specimens with equivocal findings for HER2, defined as 2+ by IHC analysis. This problem has been addressed with the complimentary use of fluorescence in situ hybridization (FISH). The current ASCO-CAP guideline recommends classifying specimens as HER2 positive if they are 3+ by IHC analysis or FISH positive, HER2 negative if they are 0 or 1+ by IHC analysis or FISH negative, and equivocal if they are 2+ by IHC analysis, in which case FISH is required to determine final HER2− status.¹⁹

In addition to the aforementioned technical challenges, there are other limitations to using receptor status as a surrogate for molecular subtype. Only about 85% of TNBCs behave as basal-like tumors on gene expression arrays, and 15% to 45% of basal-like tumors express at least one clinical receptor.²⁰ Additionally, only 30% to 50% of luminal B tumors are HER2 positive by IHC analysis or FISH, resulting in some being erroneously categorized as luminal A on the basis of receptor status determined by IHC analysis.⁶ Therefore, in an effort to improve the correlation between genomic subtypes and their clinical approximations, investigators are researching ways to further distinguish basal-like breast cancer from non-basal-like breast cancer from TNBC.^21,22 In particular for differentiating basal-like from non-basal-like breast cancer, expansion to a 5-biomarker panel has been proposed. Using data from over 3000 breast cancer patients, Cheang et al²³ compared the outcomes of women with breast cancer, defined on the basis of either a 3- or a 5-biomarker panel. In the 5-biomarker panel, positive staining for either cytokeratin 5/6 or EGFR was required in addition to the ER-negative, PR-negative, HER2-negative (TNBC) phenotype. The authors demonstrated that the definition of basal tumors according to the expanded panel of markers better predicted breast cancer survival. This finding suggested that the addition of cytokeratin 5/6 or EGFR to ER, PR, and HER2/neu provided a more specific clinical definition of basal-like tumors.²³ In an effort to refine classification of TNBC, Lehmann et al²⁴ identified six TNBC subtypes, and Chen et al²⁵ created a web-based subtyping tool for these TNBC subtypes.

Identification of different breast cancer subtypes and elucidation of differences between these subtypes in underlying biology and clinical outcomes have led to increasing emphasis on treatments targeted to specific breast cancer subtypes.

ER-Positive Breast Cancer

Tamoxifen, a selective ER modulator, has been shown to reduce recurrence of ER-positive breast cancer by nearly 50% with 5 years of treatment. In 1982, the National Surgical Adjuvant Breast and Bowel Project (NSABP) initiated a randomized clinical trial, NSABP B-14, to assess the effectiveness of adjuvant therapy with tamoxifen in patients with histologically negative lymph nodes and ER-positive tumors.²⁶ The disease-free survival through 10 years of follow-up was significantly higher in women treated with tamoxifen than in women treated with placebo (69% vs. 57%). In fact, tamoxifen therapy not only reduced the rates of local and distant recurrence but also decreased the occurrence of second primary tumors in the contralateral breast by 37%. Subsequently, an Early Breast Cancer Trialists’ Collaborative Group overview analysis confirmed the benefit of tamoxifen in reducing the annual recurrence rate by almost half for ER-positive tumors.²⁷ The advantages were attained with an acceptable incidence of adverse effects from tamoxifen. Further studies investigating the optimal duration of tamoxifen therapy have found no additional benefit to continuing tamoxifen for more than 5 years.