Introduction

Despite the increased incidence reported over the last 25 years, male breast cancer (MBC) is a rare disease accounting for less than 1 % of all breast cancer (BC) cases [1]. Its incidence rate rises gradually with age without the typical bimodal distribution of BC in women, and median age at diagnosis is higher in men than women (67 versus 62 years) [1]. Other than age, further risk factors include a series of medical conditions associated with an abnormal estrogen-to-androgen ratio (e.g., Klinefelter’s syndrome, obesity, liver diseases, testicular abnormalities), and germ-line mutations in BRCA1and BRCA2 [2, 3]. Additional potential genetic susceptibility factors are mutations in PALB2, androgen receptor (AR), CYP17, the cell-cycle checkpoint controller CHEK2, and RAD51B [2]. It is well known that MBC is an estrogen-driven disease. A population-based comparison showed that 92 % of MBCs were estrogen receptor (ER)-positive, compared to 78 % of ER-positive female BC (FBC) [4]. Such a difference, along with the observation that MBC occurs later in life, at a higher stage and lower grade, provided clues on the similarities between MBC and late-onset FBC. Since MBC has been traditionally considered as a disease closely resembling hormone receptor-positive FBC, anti-hormone therapies are the mainstay of treatment. Aromatase inhibitors (AIs), a class of compounds that prevents the conversion of androstenedione to 17b-estradiol (E2), are the gold standard endocrine therapy in both the adjuvant and metastatic setting in ER-positive postmenopausal women [5]. However, their use in MBC is mostly extrapolated from studies in female patients, due to the difficulty in carrying out prospective clinical trials in such a rare disease. Several pieces of evidence are questioning this approach, thus sparking the debate on optimal endocrine therapy for these patients. Firstly, a recent wave of molecular studies highlighted the existence of non-negligible differences between BC arising in men and women [6, 7]. Secondly, the hormonal milieu in men is different from that of age-matched women, and this leads to partly different mechanisms of hormone-related oncogenic stimulation [8]. Finally, whereas AIs replaced tamoxifen in postmenopausal BC following a series of head-to-head trials, in MBC tamoxifen seemed superior to AIs as adjuvant therapy [9]. Therefore, the translation of results from clinical trials conducted in FBC in the male setting might be an oversimplification and conceptually misleading. Herein, we analyze the spectrum of gender-related molecular and endocrine differences in BC and their implications for AI-based therapy in male patients.

The molecular landscape of male breast cancer

Over the past decade, we have witnessed an unprecedented evolution of novel biotechnologies that nowadays combine high-throughput and multiplex capability. Massive characterization efforts fragmented FBC into multiple molecular entities, each one characterized by a unique gene expression profile [10], a different incidence among ethnic groups [11], and a different set of both deregulated pathway nodes and genetic abnormalities [12]. To further complicate the picture, 10 integrative clusters defined by acquired somatic copy number aberrations were identified, splitting many of the intrinsic subtypes [13]. Furthermore, genome-wide comparison of multiple tumors, even including FBC, revealed the existence of 20 distinct mutational signatures, some “private,” being confined to a single tumor, and others shared by cancers of different origin [14]. Despite impressive progresses, the molecular scenario of MBC remained largely unexplored until recently, as evidence of specific molecular abnormalities were scattered and collected from studies analyzing only few molecular endpoints. Seminal reports exploiting comparative genomic hybridization analysis provided initial hints on chromosomal gains and losses, and suggested the existence of a similar pattern of chromosomal imbalances in BC arising in males and females [15]. Likewise, the evaluation of the methylation status of 25 genes revealed that the set of most frequently hypermethylated genes in MBC is similar to that of FBC, even though BRCA1 and BRCA2 promoter hypermethylation was less common in MBC [16]. Nevertheless, gender-related molecular differences are emerging. A study focusing on copy number changes of 21 BC-related genes found more copy number gains of EGFR and CCND1 and less copy number gains of EMSY and CPD in MBC versus FBC [17]. To a similar extent, in a comparative analysis of 56 MBC with an available FBC dataset, genomic gains were observed more frequently in MBC, whereas high-level amplifications were more frequent in FBC [18]. First attempts to provide a deeper molecular characterization of MBC have been presented, albeit with the lack of integrated genomic, transcriptomic, proteomic, and methylomic data analysis that recently allowed to dissect the molecular landscape of most common tumors. Unsupervised hierarchical clustering of gene expression profiling of 66 primary MBC coupled with tissue microarray for immunohistochemistry constructed for validation purposes (extended cohort) revealed the existence of two distinct subtypes, defined as luminal M1 (70 %) and luminal M2 (30 %) [6]. Gene ontology indicated the existence of different deregulated modules between the two subtypes. Up-regulated genes in the luminal M1 subtype, which was associated with worse prognosis, are involved in a variety of oncogenic activities spanning from cell migration and adhesion to angiogenesis, cell cycle, and cell division. Interestingly, even though luminal M1 tumors were almost all ER positive by immunohistochemistry, they presented a low score for the ER module. Conversely, luminal M2 tumors were enriched for immune response genes and with ER signaling-associated genes. According to the biological relevance of the protective role of the immune response, a positive correlation was reported between HLA positivity and better distant metastasis-free survival. From the biological standpoint, it is arguable that the three-step process of immunoediting [19], which leads from immune surveillance to immune escape, is not completed in HLA-expressing MBC. These two subgroups did not resemble any of the intrinsic subgroups identified in FBC, thus suggesting their male-restricted nature. The luminal subtypes partly overlap with a previous genomic imbalance-based classification identified within the same patient cohort, and dividing MBC in two genomic subtypes: male-simple and male-complex [18]. In more detail, 89 % of the luminal M1 tumors were classified as male-complex, whereas 47 % of the luminal M2 tumors as male-simple. In an independent gene expression profiling study, 37 ER-positive MBC and 53 ER-positive female BC, similar for clinical and standard pathological features, were analyzed [7]. Around 1.000 genes were found to be significantly differentially expressed between the two groups, resulting in distinct deregulated networks. Genes related to the AR pathway were overexpressed in MBC, along with genes mediating protein synthesis, cytoskeletal dynamics, and apoptosis. Conversely, a set of chemokines playing a crucial role during immune response was more expressed in female tumors along with other immune mediators. A partially different spectrum of mitogenic signals was also reported. The concept of a different gender-related landscape in hormone receptor pathways is further enforced by a tissue microarrays study in which 251 MBC and 263 FBCs matched for grade, age, and lymph node status were immunostained for ERα, various isoforms of ERβ and progesterone receptor (PR), AR, HER2, and a panel of cytokeratins [20]. Luminal B and HER2 phenotypes were not seen in males and, more importantly, two different clusters were isolated in relation to ERα expression. While ERα clustered with PR and its isoforms in FBC, in MBC ERα was associated with ERβ isoforms and AR. Next, by comparing microRNA expression profiles of 23 MBC and 10 female ductal breast carcinomas, 17 significantly deregulated microRNAs were isolated, 4 overexpressed and 13 underexpressed in MBC [21]. Immunohistochemistry for HOXD10 and VEGF was performed in order to evaluate the concordance between deregulated microRNAs and the expression of their targets. Down-regulation of microRNA-10b and microRNA-126 was accompanied by high expression of their targets HOXD10 and VEGF, respectively. Finally, despite the comprehensive genetic landscape of MBC has not been unveiled yet, with a consequent poor understanding of “driver,” “passenger,” and “actionable” mutations, a computational approach (CONEXIC: Copy Number and EXpression In Cancer) integrating comparative genomic hybridization and gene expression data to evaluate network perturbations in MBC and FBC yielded two different, gender-specific sets of candidate drivers [22]. While, on the one hand, these studies pointed out that hormone receptor pathways are among the driving forces in MBC, on the other a greater understanding of hormone receptor signaling networks and their vertical and lateral activators is required to overcome the critical hurdles of intrinsic and acquired resistance to endocrine therapies. Indeed, in FBC an extensive crosstalk exists between hormone receptor pathways and growth factor pathways (e.g., PI3K/Akt/mTOR and MAPK) [23]. Under hormone-deprived conditions, these signals either sensitize ER to ligand stimulation or activate the receptor in a ligand-independent manner [23]. Recently, ESR1activating mutations driving ER-dependent transcription have been described and associated with therapeutic resistance to hormonal therapy [24–27]. Having deciphered how intracellular pathway nodes interact with ER prompted the BOLERO-2 trial, which provided the proof-of-concept that endocrine resistance in postmenopausal hormone-receptor-positive BC can be antagonized by targeting intermediate effectors of canonical pathways [28]. It remains unclear whether these mechanisms also apply to MBC biology, albeit there are hints of potential gender-specific molecular modulators of AI-based therapy activity. As a paradigmatic example, gene expression profiling studies indicated that MBC is enriched for a different set of genes compared with FBC, mirroring different top-ranking deregulated biological functions that also encompass upstream activator of the PI3 K/Akt/mTOR cascade, such as HER2 [6]. To sum up, molecular studies presented so far, which are summarized in Table 1, provided the biological background supporting the use of AIs in MBC. However, a massive characterization of the disease illustrating the full spectrum of molecular abnormalities is essential in improving AI-based therapy. To this end, we believe two complementary strategies should be pursued. Firstly, fostering implementation of existing biobanks within a collaborative network for interdisciplinary research with a translational focus. Included biological samples should fulfill requirements for extensive annotation with cancer-related molecular features and clinical data including therapy administered and treatment outcomes. Systems biology approaches can indeed provide a more complete view of how coexisting molecular aberrations alter signaling networks and their impact on clinical outcomes. As a parallel strategy, establishing a collection of cell lines for gathering functional data from in vitro and in vivo studies focused on hormonal resistance in male MBC. This will enable investigators to explore the interplay between hormonal receptors and any current/potential targets placed within canonical and emerging signal transduction pathways, thus paving the way for biology-driven studies. To overcome the issue of the rarity of the disease, MBC patients might be included in clinical trials in FBC envisioning the use of targeted agents for restoring hormone sensitivity.

Table 1 Molecular studies in MBC
Full size table

The hormonal background in aging male

Together with molecular factors, the different hormonal milieu between males and females represents a key determinant for interpreting AI efficacy in MBC. The major source of plasma E2 in males derives from peripheral aromatization of testosterone (T) [8]. With aging, the androgen/estrogen ratio shifts in favor of estrogens, as the decrease in testicular and adrenal T production is not coupled with parallel reduction of E2 levels [8]. Such an imbalance stems from the age-associated increase of both aromatase activity and fat mass [8]. As a result, E2 levels are significantly higher in aged males than in post-menopausal females [8]. Thus, the aromatase enzyme represents a crucial node in supplying cancer cells with oncogenic stimuli. Whether, on the one hand, this represents a sound background for the use of AIs in the clinical setting, additional endocrine factors should be taken into account for refining the therapeutic potential of these compounds. About 20 % of E2 is directly secreted by the testes, thus exerting tumor-promoting functions without the intermediate passage of aromatization. Moreover, pharmacological inhibition of the aromatase enzyme and the consequent drop in E2 level triggers the hypothalamic-pituitary feedback loop, which in turn might counteract the effects of AIs [29–32]. Indeed, prolonged administration of anastrozole to male adult rats led to a significant increase in testis weight coupled with increased levels of follicle-stimulating hormone (FSH), luteinising hormone (LH) and T [29]. Such AI-mediated increases in T levels might force the roadblock imposed by AIs, by fuelling T enzymatic conversion through an excess of substrate. Although the administration of AIs caused a decrease in E2 levels in healthy men, a parallel increase in FSH, LH and T was documented [30, 31], an association further strengthened by results coming from randomized, placebo-controlled clinical trials in hypogonadal elderly men [33, 34]. Therefore, adaptive endocrine changes occurring during AI therapy might paradoxically trigger two independent oncogenic routes. Hormone receptor pathway-mediated stimulation of cancer cells can be driven by both an excess of substrate for aromatization and the stimulation of AR-related signals. This latter mechanism, albeit not formally proven yet, can be argued by comparing molecular studies described above [7, 20], and evidence of tumor response with antiandrogens [35–37]. Along with host-related factors, tumor-associated endocrine factors further highlighted the central role of the aromatase enzyme in the hormonal network sustaining tumor growth. Tissue microarray documented intratumoral aromatase (ITA) expression in 12 out of the 45 evaluated specimens [38]. ITA positive MBC were associated with favorable pathologic features and improved 5 year overall survival. More recently, intratumoral E2 and T concentrations were reported to be higher in MBC than in FBC, and gene expression profile of laser capture-microdissected tumors, focused on estrogen-induced genes previously identified in a commercial cell line, showed that MBC and FBC formed independent clusters, as confirmed by immunohistochemistry for representative endpoints (RARα and RIP140) [39]. Even considering the small sample size of these studies, they raised the hypothesis that cancer cells are able to shape the local hormonal milieu to thrive.

Clinical experiences with AIs in metastatic MBC

The therapeutic potential of manipulating the hormonal background for treating MBC patients is rooted in tumor regressions observed with surgical procedures such as orchiectomy, adrenalectomy, and hypophysectomy [40]. Even though these procedures were associated with response rates spanning from 55 to 80 %, they have been widely replaced by more acceptable hormonal medical treatments. In the last decade, the use of AIs in metastatic MBC was prompted by their success in FBC patients [5], the initial evidence of target expression within the tumor [38], and the plethora of tamoxifen-associated side effects observed in men [41–43]. Despite the better tolerability of AIs compared to tamoxifen, the controversy surrounding their use in the metastatic setting was until recently fed by better outcomes reported with adjuvant tamoxifen than with AIs [9], and scattered evidence of antitumor activity [44–46]. Consistently, while on the one hand no objective responses were recorded in a cohort of five male patients with metastatic disease who received anastrozole [44], individual cases of response to letrozole were presented [45, 46]. Taking into account the intrinsic limitations of retrospective analyses, a first and more structured attempt aimed at collocating AIs into the clinical practice dates back to 2010 [47]. Fifteen metastatic patients were treated with an AI. Complete or partial responses (CR, PR) were recorded in six patients (40 %), and stable disease (SD) in two patients (13 %), translating into a disease control rate (DCR) of 53 %. The median progression-free survival (PFS) and overall survival (OS) were 4.4 months (95 % CI 0.1–8.6) and 33 months (95 % CI 18.4–47.6), respectively. There were no appreciable differences related to the type of AI used (non-steroidal versus steroidal), even though no firm conclusions can be drawn considering the small cohort examined. Beyond providing evidence of antitumor activity, endocrine analyses conducted in six patients revealed successful reduction of E2 levels. More importantly, in a patient with PR increased levels of E2, LH, and FSH were detected at tumor progression. Although anecdotic, this finding underlies the activation of the hypothalamic-pituitary feedback loop and the correlated counteraction of AI activity. The therapeutic potential of inhibiting the hormonal feedback loop for potentiating AI therapy stemmed from two PRs observed in two metastatic patients treated with either anastrozole or letrozole combined with leuprolide acetate [48]. However, only in 2013 two independent, retrospective studies tried to assess the clinical usefulness of such a combination therapy [49, 50]. Zagouri et al. [49] presented results from a cohort of twenty-three patients treated with an AI as first- or second-line, mostly letrozole or anastrozole, either alone or in combination with a GnRH analogue. Despite confirming the decade-long belief that AIs are an effective and safe treatment option for metastatic MBC patients, neither tumor response rate (SD:PR ratio with the administration of goserelin versus AI monotherapy: 64.7:17.7 and 33.3:50.0 %, respectively) nor OS favored the co-administration strategy. In the second study presented by our group [50], nineteen metastatic men were treated with letrozole combined with a GnRH analogue as a first- or second-line therapy. DCR (84.2 %), PFS (12.5 months), and OS (35.8 months) were fairly comparable with the study discussed above. However, among four patients for whom a GnRH analogue was introduced following tumor progression while on front-line AI monotherapy, we noted that three of them treated with such a sequential approach confirmed or improved the best overall response observed in the first-line. One patient with SD in the first-line experienced a PR following GnRH analogue introduction and the replacement of exemestane with letrozole at tumor progression, and two patients treated with first-line letrozole confirmed the PR and SD with the introduction of a GnRH analogue in the second-line setting. Although no firm conclusions can be drawn, also due to the lack of baseline and serial assessment of hormonal levels in our study, the suggestive hypothesis emerging is that introducing a GnRH analogue at tumor progression while on AI therapy might efficiently counteract the activation of the hormonal loop responsible for forcing the block imposed by AIs. Combining this observation with the not clearly proven superiority of front-line combination therapy, it is possible to hypothesize that a sequential approach might extend the period during which patients experience benefit from endocrine manipulations, thus delaying the use of chemotherapy. Table 2 and Table 3 summarize clinical experiences with AIs in metastatic MBC patients.

Table 2 Case reports with AIs alone or in combination with a GnRH analogue in metastatic MBC
Full size table
Table 3 Clinical studies with AIs alone or in combination with a GnRH analogue in metastatic MBC
Full size table

Conclusions and future directions

The importance of manipulating the hormonal milieu for treating MBC dates back to the 1940s, when orchiectomy was described as a treatment for skeletal metastases [40]. Since AIs have changed the treatment paradigm of ER-positive FBC in the post-menopausal setting, it is not surprising that these agents have been exploited for treating MBC patients. However, nowadays we have elements to judge such an extrapolation incomplete. Even though the secrets of the MBC genome have not been unrevealed yet, first attempts of molecular characterization together with knowledge on the endocrinology of aging allow to foresee that the greater the extent of hormonal deprivation via the inhibition of multiple nodes is (aromatase enzyme, the hypothalamic-pituitary axis, ER, and AR), the more pronounced the effects on cell viability should be. Pursuing this biological scenario might be, however, tricky, and the practical issue of adherence to therapy needs to be carefully considered. The lesson we learned is that AIs are active and produce tumor shrinkage and prolonged disease stabilization. In 2010, in a summary of a multidisciplinary international meeting on MBC panel members drew attention to some possible clinical trial designs [51]. In particular, a study aimed at comparing tamoxifen versus an AI with a safety endpoint was proposed. Beyond the intrinsic difficulty in conducting randomized clinical trials in a rare disease, as highlighted by the premature closure of the small-sized phase II study SWOG-S0511 (ClinicalTrials.gov; ID: NCT00217659), in our opinion this comparison is no longer necessary. Firstly, because most metastatic MBC patients have already received tamoxifen as adjuvant therapy, whose use in this setting is recommended and supported by clinical evidence, albeit retrospective [9]. From a biological standpoint, the efficacy of tamoxifen rechallenge might be hindered by tumor cell plasticity, as prolonged drug exposure elicits adaptive changes and induces clonal evolution, thus enabling cancer cells to evade therapy-induced death stimuli. Secondly, because at that time the three largest studies describing the antitumor activity of AIs were not published [47, 49, 50]. Thirdly and more importantly, because tamoxifen-related toxic effects and the correlated, non-negligible 20 % of discontinuation rate is a crucial clinical issue [43]. Even though AIs should be considered the mainstay of treatment in the metastatic setting and efforts should be focused on sharpening their potential, finding their exact collocation in the therapeutic continuum, and establishing their optimal use, is still posing a challenge due to the paucity of data available so far. Until results from prospective clinical trials or, alternatively, large case series will not be available, the choice of AI monotherapy versus dual hormonal (or sequential) therapy combining AIs with a GnRH analogue should be made taking into account multiple cancer- and patient-related factors, including molecular characteristics, medical history, patient needs, comorbidities, disease evolution and extension, and serial assessment of hormone levels. In approaching MBC patients, planning a long-term strategy implies to carefully weigh the endocrine-responsive nature of the disease on the one hand and the scant information available on the role of chemotherapy on the other hand. Indeed, clinical experiences with palliative chemotherapy envisioned outdated regimens used in the pre-taxane era [52]. Moreover, the magnitude of benefit deriving from chemotherapy is greater in endocrine-nonresponsive BC, and potential harms from using chemotherapy in elderly patients, in whom multiple comorbidities often coexist, need to be carefully considered. Delaying chemotherapy as long as possible is, therefore, a priority, at least in the absence of life-threatening or rapidly progressive lesions, or in the fraction of hormone-receptor-negative tumors. Combining these observations with the not already proven superiority of front-line combination therapy over AI monotherapy, in our opinion the sequential strategy discussed above should be considered, especially if supported by biochemical evidence of AI-induced hormonal changes. Finally, placing novel treatment modalities in a therapeutic framework aimed at manipulating the hormonal milieu with sequential endocrine therapies is of utmost importance. To this end, while ER is an established key oncogenic driver and relevant therapeutic target, as further confirmed by evidence of tumor response with fulvestrant [53], AR deserves further investigation as a therapeutic target in light of pathway components expression in MBC, evidence of antitumor activity with first-generation anti-androgens [35–37], and recent successes with novel anti-androgens in prostate cancer [54, 55].

Search strategy

Data for this review were found through searches of PubMed using the terms: “male breast cancer,” “metastatic,” “aromatase inhibitors,” “anastrozole,” “letrozole,” “exemestane,” “GnRH analogue,” “tamoxifen,” “gene expression profiling,” “comparative genomic hybridization,” “tissue microarray,” “microRNAs,” “mutations,” “subtypes,” “sex hormone pathways,” “estrogen receptor,” “androgen receptor.” We did not use a date limit. Only articles published in English were included. The reference list was selected on the basis of scientific and clinical relevance.