Introduction
Luminal breast cancers represent over 70% of cases [
1]. At least 1% of their cells express estrogen (ER
+) or progesterone (PR
+) receptors or both [
2], driving estrogen (E)-dependent growth. Despite progress toward early diagnoses and advances in treatment, 20% to 30% of all patients with breast cancer and 40% to 50% of patients with luminal breast cancer experience relapses that include distant metastases [
3],[
4]. This tends to occur within the first 5 years for patients with basal-like ER
−PR
− or HER2
+ disease and later for patients with luminal disease [
5]. In one study, median 15-year distant relapse rates were 27.8% for luminal A and 42.9% for luminal B [
5]. Because molecular properties of primary tumors may be preserved in metastases [
6], adjuvant endocrine therapies can improve initial survival rates even in patients with advanced luminal disease [
7]. Nevertheless, the survival curve for luminal disease declines steadily after 5 years, overtaking more aggressive breast cancer subtypes after about 15 years [
5],[
8]. Therefore, since they represent the most common forms of the disease, luminal tumors are responsible for most breast cancer deaths. Explanations for prolonged luminal tumor dormancy and their slow but inexorable recurrence and lethality remain unclear, and roles of cellular heterogeneity and hormones in this process, if any, are poorly understood.
The Women’s Health Initiative (WHI) report on postmenopausal hormone replacement therapy (HRT) showed that the risks of combined E plus progestin (P), unlike those of physiological E alone, outweighed the benefits [
9]. Widespread acceptance of the WHI data led to a general decrease in HRT use. Concurrent reductions in the incidence of invasive luminal cancers indirectly validated the WHI conclusions [
10]. However, explanations for the deleterious effects on the breast of physiological E and P in combination HRT remain unclear partly because hormonal effects on carcinogenesis versus proliferation are often conflated, and the term “risk” intimates that the hormones are causative. P appears to have no effect on long-term tumor growth
in vivo [
11] but expands normal adult mammary stem cells and cancer stem cells [
12]–[
14]. Regarding WHI, we therefore postulated that for E+P, the P component, in a non-proliferative step, reactivates cancer stem cells in pre-existing but undiagnosed, perhaps dormant, disease [
15]. That said, little is known about the roles of E and P in metastasis and recurrence from dormancy.
Clinically, the major sites of luminal metastases are bone (>49%), followed by pleura/peritoneum, liver and lung (~20%), distant nodes (~14%), and brain (6%) [
5]. The ER and PR status of the primary tumor may be reflected in bone metastases [
5],[
16], explaining the use of endocrine therapies to treat disseminated disease that is rarely reanalyzed for biomarker expression. Few solid tumor models exist for detailed studies of luminal metastases and their hormone regulation. One interesting new model uses serially transplanted patient-derived luminal tumor xenografts, three of which demonstrate E-dependent growth and retain luminal markers and gene expression profiles [
17]. The xenografts metastasize to lungs and lymph nodes (LNs) [
17], but the role of hormones, if any, in tumor-cell dispersal is unclear. Lacking efficient solid tumor models, a recent study used systemically injected ER
+PR
+ MCF-7 cells to show that they can generate metastases in an E-dependent manner but that the initial homing and seeding steps with development of micrometastases do not require E [
18]. Additionally, two dormancy models that included luminal cells were recently described [
19],[
20]. However, they do not address the role of hormones in metastasis or recurrence [
4].
We previously demonstrated that in E-replete states, ER
+PR
+ orthotopic xenografts metastasize to distant LNs and occasionally to other organs [
21]. Detailed immunohistochemistry (IHC) analyses of such tumors showed that during their expansion in mice, initially pure ER
+PR
+ cells develop cellular heterogeneity. At necropsy, presumptive “luminal” tumors contain at least one cell subpopulation we call “luminobasal” that is ER
−PR
− and expresses cytokeratin 5 (CK5), a protein usually associated with basal-like cancers [
11],[
13],[
22]. In clinical samples of luminal disease, similar basal-like, ER
−PR
−CK5
+ cells, whose numbers increase with hormone therapies, can be found [
23]. These hormone-resistant, possibly chemo-resistant cells are likely to have a poor prognosis. The heterogeneity raises questions about the identity of cell subpopulations in primary luminal disease that are responsible for metastatic engraftment and growth.
In this study, we develop a luminal metastases model and assess the role of E and P in metastatic engraftment and recurrence from dormancy. To short-circuit the cellular heterogeneity issue and study engraftment by each cell population independently, T47D-derived solid tumor xenografts were partitioned into their luminal (called E3) and luminobasal (called EWD8) subpopulations [
22]. These, plus established luminal (MCF-7) and basal (MDA-MB-231) breast cancer cell lines, were tagged with luminescent and fluorescent markers for
in vivo and
ex vivo analyses. Cells were injected into the left ventricle of ovariectomized (ovx’d) immunocompromised mice, and their colonization and proliferation in distant organs was monitored in the absence of hormones or following E or E+P repletion. We found that luminobasal and basal cells generate metastases regardless of the hormonal state. In contrast, luminal cells rarely form metastases unless E or E+P is restored. The organs colonized by luminal and basal cells are similar and mimic the clinical pattern, dominated by bone. Despite initial injection of
in vitro pure luminal cells, their metastases
in vivo exhibit cellular heterogeneity, including outgrowth of ER
−PR
−CK5
+ luminobasal subpopulations. These cells proliferate more slowly than surrounding luminal cells. Notably, although luminal cells seldom generate macrometastases in the absence of hormones, viable dormant micrometastases engraft at distant sites. If mice harboring such occult tumor cells are subsequently hormone supplemented, overt metastases materialize.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
NO, NGM, BSB, SKA, JMH, MPP, PJ, KJ, PH, and KBH contributed to experimental conception and design, data acquisition, data analysis, and interpretation. NO and KBH drafted the manuscript and revisions and are responsible for intellectual content. All authors read and approved the final draft.