Advertisement
Commentary Free access | 10.1172/JCI73910
Center for Comparative Medicine and Department of Pathology and Laboratory Medicine, Davis School of Medicine, UCD, Davis, California, USA.
Address correspondence to: Robert D. Cardiff, Center for Comparative Medicine, UCD, Road 98 and Hutchison Drive, Davis, California 95616, USA. Phone: 530.752.2726; Fax: 530.752.7914; E-mail: rdcardiff@ucdavis.edu.
Find articles by Cardiff, R. in: JCI | PubMed | Google Scholar
Center for Comparative Medicine and Department of Pathology and Laboratory Medicine, Davis School of Medicine, UCD, Davis, California, USA.
Address correspondence to: Robert D. Cardiff, Center for Comparative Medicine, UCD, Road 98 and Hutchison Drive, Davis, California 95616, USA. Phone: 530.752.2726; Fax: 530.752.7914; E-mail: rdcardiff@ucdavis.edu.
Find articles by Borowsky, A. in: JCI | PubMed | Google Scholar
Published January 27, 2014 - More info
Accurate classification is essential for understanding the pathophysiology of a disease and can inform therapeutic choices. For hematopoietic malignancies, a classification scheme based on the phenotypic similarity between tumor cells and normal cells has been successfully used to define tumor subtypes; however, use of normal cell types as a reference by which to classify solid tumors has not been widely emulated, in part due to more limited understanding of epithelial cell differentiation compared with hematopoiesis. To provide a better definition of the subtypes of epithelial cells comprising the breast epithelium, we performed a systematic analysis of a large set of breast epithelial markers in more than 15,000 normal breast cells, which identified 11 differentiation states for normal luminal cells. We then applied information from this analysis to classify human breast tumors based on normal cell types into 4 major subtypes, HR0–HR3, which were differentiated by vitamin D, androgen, and estrogen hormone receptor (HR) expression. Examination of 3,157 human breast tumors revealed that these HR subtypes were distinct from the current classification scheme, which is based on estrogen receptor, progesterone receptor, and human epidermal growth factor receptor 2. Patient outcomes were best when tumors expressed all 3 hormone receptors (subtype HR3) and worst when they expressed none of the receptors (subtype HR0). Together, these data provide an ontological classification scheme associated with patient survival differences and provides actionable insights for treating breast tumors.
Sandro Santagata, Ankita Thakkar, Ayse Ergonul, Bin Wang, Terri Woo, Rong Hu, J. Chuck Harrell, George McNamara, Matthew Schwede, Aedin C. Culhane, David Kindelberger, Scott Rodig, Andrea Richardson, Stuart J. Schnitt, Rulla M. Tamimi, Tan A. Ince
Current breast cancer classification systems are based on molecular evaluation of tumor receptor status and do not account for distinct morphological phenotypes. In other types of cancer, taxonomy based on normal cell phenotypes has been extremely useful for diagnosis and treatment strategies. In this issue of the JCI, Santagata and colleagues developed a breast cancer classification scheme based on characterization of healthy mammary cells. Reclassification of breast cancer cells and breast cancer tissue microarrays with this system correlated with prognosis better than the standard receptor status designation. This scheme provides a major advance toward our understanding of the origin of the cells in the breast and breast cancers.
Considering all tumors of the breast to be equal in origin, diverging only based on clonal evolution, is akin to lumping all tumors of the alimentary canal together. Such gross characterization would ignore distinct differences of cell and tissue types: just as in the gastrointestinal tract, where the squamous epithelium of the esophagus transitions to the acid- and mucous-producing epithelia of the stomach, then to the absorptive villi of the small intestine and the gland crypts of the colon, the breast is lined by specialized cells and stroma that transition from the nipple ducts to the lactiferous sinuses, from the primary ducts to the branched ducts and terminal ducts, into individual terminal ductal lobular units (TDLUs). Within each of these sites, there is a mixed distribution of cells with specialized functions. Arguably, each site and each cell or its progenitor might have higher or lower susceptibility to cancer, with differing sensitivity or resistance to environmental cancer stimuli; therefore, these distinct regions might give rise to cancers with quite different clinical behaviors.
In fact, some of the most useful cancer classifications are based on an adequate understanding of the normal cell types within a given tissue or organ. The prime examples and models for such classification are the hematopoietic malignancies. In this issue of the JCI, Santagata and colleagues have developed a breast cancer classification system based on phenotypes associated with normal mammary tissue (1).
From the early days of microscopic pathology, the pathology and natural history of breast cancer have been controversial (2). Waldeyer gets credit for the first description of a breast cancer in direct continuity with normal mammary epithelium (2, 3). Although his mentor Virchow favored the blastema hypothesis, which suggests that cancer cells originate from budding elements between normal tissues, Waldeyer’s concept that breast cancer originates from breast epithelium has survived to become universally accepted. For many years, Halstead’s concept that breast cancer metastasizes through centripetal spread led to increasingly radical mastectomies; however, dissenters such as Greenough (4) and Macdonald (5) contested his concepts of natural history and proposed that some tumors were more aggressive than others at initiation (6). Foote and Stewart began serious morphological subclassification of breast cancers with the descriptive misnomer lobular carcinoma (7); the early history of carcinogenesis (2) and the history of the conflicting views of breast cancer (6, 8) have previously been discussed in detail.
Evidence-based pathology dating back to the 1970s and 1980s resulted in lumping a heterogeneous array of tumor morphologies, which constitute the majority of breast cancers, under the heading of invasive ductal carcinoma, although this classification would be better named invasive mammary carcinoma of no special type. This era culminated in the current WHO morphological breast cancer classification based on histological pattern (9). Newer high-throughput molecular analyses of breast cancers have ignored morphological classifications and have simplified breast cancer classifications as being luminal or basal. These molecular classifications imply that there is no practical reason to separate these morphologies, and classification of special types of breast cancer with distinct morphological properties was regarded as a list of heterogeneous minority phenotypes without much value (1).
Unique morphological types have largely been discarded in current clinical practice. Instead, breast cancers are assayed for clinically useful targets based on the presence or absence of the magic targetable twosome, estrogen receptor (ER) and human epidermal growth factor receptor 2 (HER2). Tumors that do not contain either of the magic treatable receptors are categorized as triple-negative breast cancer (TNBC), encompassing a diverse group of tumors without druggable targets (10, 11). An unfortunate overlap of TNBC cancers with the genomic subset of basal-like cancers has cemented into place the concept that breast cancers are either luminal or basal, based on molecular criteria. Furthermore, simplifications about the nature and distribution of normal cell markers (such as cytokeratin 5, which is erroneously believed to be only in basal cells) continue to confuse the literature (12, 13).
In response to the need to be translationally relevant (i.e., fundable), basic breast cancer research tumor classifications have focused primarily on identifying cells with stem-like and progenitor-like properties. While these properties are hot topics, they do not define actual differentiation states of mature cells in the breast beyond conventional designations of basal verses luminal epithelial, and ductal verses lobular. Unfortunately, current descriptions of the human breast ductal/lobular tree taxonomy are oversimplified and often presented as possible or conjectured without empirical observations. In our view, the work to date on breast stem, progenitor, or “transit-amplifying” cells is limited by an incomplete understanding of the hierarchical taxonomy of mammary cells. Without this basis, it is not surprising that markers identifying cells with stem-like behavior in one context do not work for identifying cancer-driving cells in another. We anticipate, as proposed by Santagata et al. (1), that a hierarchy of undifferentiated, lineage-restricted, and lineage-committed stem and progenitor cells give rise to the array of luminal cell types outlined by their analyses, as well as a similarly complex array of basally oriented lining cells.
The paper by Santagata et al. represents a major advance toward a realistic, biologically based taxonomy of mammary cells and illustrates the utility of this taxonomy in evaluation of cancers of the breast (1). The authors painstakingly analyzed the normal lobules from 36 “normal” mammoplasties. Their studies led to a more comprehensive understanding of the distribution of antigens expressed in normal human mammary epithelium and to an evidence-based classification scheme. Immunofluorescence and immunohistochemistry were used to define cell groups based on hormone receptors (HRs; namely, ER, androgen receptor [AR; not progesterone receptor], and vitamin D receptor [VDR]), intermediate filaments (keratin 5 [K5], K6, K7, K14, and K18), the proliferation marker Ki67, and myoepithelial markers (CD10, p63, and SMA). Evaluation of these markers in luminal cells from normal human breast TDLUs could separate eleven luminal subtypes and two subsets within the myoepithelial compartment (1). Based on these insights, the authors proposed an ontology scheme based primarily on the number of HRs present (designated HR0–HR3), and a secondary categorical separation of HR0 tumors (which lack all three HRs) based on the presence or absence of K5. At the HR2 level, cells could be distinguished by the HR combination (i.e., which two of the three HRs were detected) as well as K5 status. Interestingly, proliferating cells (Ki67+) were HR0 and produced K18. Furthermore, HER2 was not expressed in normal TDLUs. These extensive observations will serve as a foundation for future understanding of human breast cancer.
Despite these great advances in breast cancer classification, the study by Santagata and colleagues (1) is primarily based on single-time-point analysis of clinically derived excisions and formalin-fixed, paraffin-embedded sections of breast tissue. Based on this snapshot analysis, one is left to wonder whether all luminal cells go through a defined sequence of immunnophenotypes that diverge at specific branch points, similar to hematopoietic differentiation, or are immunophenotypes that prove transient as each cell reacts and adapts to its microenvironment. For example, Ki67+ cells that are characterized as HR0 and K18+ must produce daughter cells belonging to all the other lineages if the mammary gland is to proliferate, or if breast tumors are to maintain their HR-dependent phenotype. Lineage definition through experimental manipulation will be almost impossible in humans; therefore, mammary cell lineage tracing will have to be examined using mammalian models, such as the mouse.
Moving forward on the assumption that their new taxonomy defines the cell of origin, Santagata et al. reclassified several collections of breast cancers and breast cancer tissue microarrays (TMAs). Excitingly, their classification scheme organized human breast cancers into categories that were relevant to prognosis (1). Comparison of the reclassified groups with the current categories revealed interesting mixtures of HR0–HR3 tumors within ER+ and HER2+ groupings (Table 1), which may explain why therapeutic strategies based on current standards of classification have failed.
In addition to using their taxonomy system to characterize breast cancer cells, Santagata and colleagues applied their classification to the collection of 50 NIH human breast cancer cell lines. Interestingly, nine of the cell lines contained none of the markers found to be associated with breast cancer (1). The cell lines tested included MDA-MB-231, SUM-159PT, MDA-MB-157, MDA-MB-436, HBL100, BT549, SUM131, 5M02, MDA-MB-435, and HS578T, which were originally thought to be representative of human breast cancers, but are no longer considered by most experts to be breast cancer cells. The fact that the HR designation also indicated that these cell lines are not representative of human breast cancer is encouraging. Scientists that use mouse models of breast cancer should be delighted to learn that the basal cell designation can be discarded, because human breast cancers did not contain this specific cell population.
The discontent between complex molecular profiling and simple morphological approaches can be seen in a plethora of other breast cancer–associated classification systems, which also rely on more recent molecular and immunohistochemical classifications to claim utilitarian value for their particular schemata. The shortcoming of all these taxonomies of human breast cancer has been a lack of biological basis, which is now provided by Santagata and colleagues (1). The authors cautiously point out that their work is a starting point and not yet a complete description of TDLU taxonomy. Meanwhile, they have provided ample clinically based data to justify the use of such a classification system. The question of whether the HR0–HR3 designation improves our understanding and treatment of breast cancer or simply adds to our current confusion remains to be answered. Although the present study is complex and provides valid clinical justifications for use of the HR0–HR3 classification scheme, the most important aspect of this report is that it provides a biological basis for testing various clinical and scientific hypotheses regarding the origins of breast cancers.
The authors’ work is supported by NCI grants U01 CA141541 and U01 CA141582 and by Athena Breast Health Network through the University of California Office of the President.
Address correspondence to: Robert D. Cardiff, Center for Comparative Medicine, UCD, Road 98 and Hutchison Drive, Davis, California 95616, USA. Phone: 530.752.2726; Fax: 530.752.7914; E-mail: rdcardiff@ucdavis.edu.
Conflict of interest: The authors have declared that no conflict of interest exists.
Reference information: J Clin Invest. 2014;124(2):478–480. doi:10.1172/JCI73910.
See the related article at Taxonomy of breast cancer based on normal cell phenotype predicts outcome.