The expression of AR is often detected in about 60% to 90% of ER + ve breast cancer cases [43,44]. In this sub-type of breast cancer, AR acts as a good prognostic factor. In a study of 931 patients, the survival curves demonstrated that the presence of AR in patients with ER + ve tumors showed a better outcome for disease-free survival (DFS) and overall survival (OS). A study of 1467 postmenopausal breast cancer patients showed similar results [45]. However, the presence of AR would be a poor prognostic factor for ER-ve patients [44]. AR expression in ER + ve/HER2-ve breast cancer was significantly associated with better breast cancer-specific survival (BCS), recurrence-free survival (RFS), and OS; however, AR expression became a poor prognostic factor in ER-ve patients [46]. A study that determined AR’s clinical significance in luminal-B breast cancers showed that the AR + ve cases would have better outcomes for time-to-relapse (TTR) and disease-specific survival (DSS) [47]. Another independent study revealed that high AR expression in ER + ve tumors was associated with less infiltration of lymphocytes, which is a sign of better prognosis, and better survival [48]. Several other studies have also revealed that the expression of AR in ER + ve breast cancer is associated with a smaller size, lower histopathological grading, and lower proliferative properties of the tumors, which might prolong the patients’ survival [44,49,50,51]. These clinical studies supported that AR expression could be a useful prognostic factor in breast cancers.

These findings suggest that AR likely functions as a tumor suppressor in ER + ve breast cancer. This raises the question to investigators: what is the connection between AR and ERα signaling in breast cancer? One of the possibilities is that activated-AR can antagonize the transcription activity of ERα by competitive binding to estrogen responsive elements (EREs). A recently published paper has clarified the detailed mechanism [52]. This study showed that AR activation could replace ERα from chromatin. AR then occupied over 40% of all ERα binding sites (ERBSs), leading to a loss of estrogen response elements (EREs) binding. Meanwhile, ERα was shown to gain new binding targets by relocating to some AR binding sites (ARBSs) to further regulate AR targeted genes, including tumor suppressor SEC14L2, EAF2, and ZBTB16 to inhibit the growth of cells. Furthermore, AR also competes with ERα for binding to a common co-activator, p300, which is essential for the activity of ERα. Since ERα needs a co-regulatory protein SRC-3 to recruit p300 while AR can bind to p300 directly, AR may obtain an advantage in the competition with ERα, resulting in suppression of ER signaling; the activation of AR, therefore, demonstrated an inhibitive effect on ER + ve breast cancer cells [52]. Moreover, AR can inhibit ERα indirectly by some mediator proteins. ERβ is a suppressor of ERα. Activated AR could up-regulate the expression of ERβ gene by binding to the ARE of its promoter region to suppress the activity of ER [53]. In summary, the activation of AR can suppress ER activity by different mechanisms. Since the ERα is a dominant pathway in promoting tumor growth in ER + ve breast cancers, suppressing the ERα can attenuate disease progression. Therefore, AR leads to the better outcome of patients with ER + ve breast cancers.

Approximately 70% of HER2 + ve/ER-ve breast tumors were detected as AR-positive [46]. In contrast to the ER + ve sub-type, AR + ve patients with a HER2 + ve/ER-ve feature reported a worse clinical outcome in studies. The previous research suggested that AR correlated to the poor DFS and OS in HER2 + ve/ER-ve breast cancer patients [44]. Another study reported that a high mRNA level of AR in HER2 + ve/ER-ve patients was associated with shorter DFS and OS [57]. Studies have demonstrated that AR can crosstalk with HER2 signaling. Such crosstalk could intensify the signaling pathways driven by both AR and HER2 through a positive feedback loop. In the WNT/β-catenin signaling pathway, AR induces the expression of WNT7B to activate the nuclear translocation of β-catenin; AR binds to β-catenin in the nucleus, with the help of FOXA1, leading to the AR/β-catenin complex translocating to the promotor region of HER3 to promote gene transcription, enhancing the activity of the HER3/HER2 heterodimer [61]. As mentioned earlier, HER2 can activate MAPK signaling [11]. The activated MAPK would induce the expression of AR, which in turn, can enhance HER2 expression. In this loop, AR is essential and adequate for HER2 activation, as AR favors the expression of HER3, while HER2 is crucial for the transduction of MAPK/AR signals [62]. Targeting AR by the shRNA and inhibitor could effectively suppress HER2 + ve/ER-ve breast cancer cell growth in vitro and in vivo [63]. These studies suggested that AR plays an oncogenic role in HER2 + ve breast cancer.

Around 10% of breast cancer belong to the TNBC sub-group. This sub-type of breast cancer is more aggressive and has a high recurrence risk. The expression of AR was detected in 10–50% of TNBC [42]. In a clinical study, AR + ve TNBC patients were shown to have a decreased survival rate compared with AR-ve TNBC patients [45]. In a study of 559 TNBC cases, the results indicated that AR expression was associated with a worse prognostic outcome in terms of OS; for patients without lymph node metastasis, AR + ve patients had poor OS and DFS, in which the risks of mortality and recurrence were three times higher compared with the AR-ve patients [58]. Similarly, the expression of AR was found commonly in lymph node metastatic TNBC, but rarely in non-lymph node metastatic tumors [64]. Another study showed that AR + ve TNBC patients were more likely to develop a disease recurrence than those unexpressed patients [59]. A study of 263 TNBC patients supported that AR + ve patients would have worse outcomes in five-year distant disease-free survival (DDFS) [60]. Clinical research has associated AR with an inadequate response to neoadjuvant chemotherapy, suggesting the contribution of AR to drug resistance [65]. An in vitro study indicated that AR could promote the survival of TNBC cells, expression of invasion related genes, and thus, metastasis; the inhibition of AR suppressed the metastatic potential of TNBC cells [66]. AR can form a complex with SRC, by recruiting the SRC substrate, focal adhesion kinase (FAK), and the PI3K regulatory subunit, p85α, thus rapidly activating the SRC/PI3K/FAK pathway and its downstream gene, thereby driving cell metastasis [67]. These results suggest that AR can promote the tumor progression of TNBC. In TNBC, activating PIK3CA mutations were frequently detected in AR + ve patient samples and cell lines. The PI3K pathway has been revealed to contribute to breast cancer development, while the combined inhibition of AR and PI3K significantly suppressed cell propagation in cell models [68]. These results support that AR can be involved in the pathogenesis of TNBC. The inhibition of AR might suppress progression and reduce the aggressiveness of the disease.

Earlier studies highlighted that TNBC patients might benefit from the presence of AR with an improved five-year survival rate, OS, DFS, higher disease-specific survival, and low recurrent risk [69,70,71,72], while the cases with the absence of AR would have a higher risk of tumor metastasis [73,74,75]. A meta-analysis involving 2826 TNBC patients revealed AR expression was related to better DFS and lower tumor grade, but a higher incidence of lymph node metastasis, and no impact on OS [76]. However, another more recent study that analyzed 4914 TNBC patients from 27 studies showed that there was no correlation between AR and patients’ DFS, OS, DDFS, or disease relapse-free survival [77]. The reasons for these contradictory results are still under investigation. Noteworthy, TNBC patients can be further classified into different sub-types by their intrinsic gene profiles. AR + ve luminal TNBCs, known as luminal AR (LAR) sub-type, shows unique characteristics [78]. It has been demonstrated that LAR cancers displayed molecular features similar to luminal A and B breast cancers (ER + ve), including multiple highly reactive hormone-regulated pathways [79]. Interestingly, resembling AR + ve/ER + ve breast cancers, studies have emphasized that patients with LAR type cancers had a favorable prognostic outcome with lower KI-67 levels, lower tumor grade, and higher OS. Moreover, TNBC sub-types were associated with different pathological complete response (pCR) rates to neoadjuvant chemotherapy, with LAR having the worst response, while the basal-like, another TNBC sub-type, had the best response [80]. Furthermore, the differences in correlation between AR with OS among different races and ethnicities has also been reported [81,82]. In around one-third of TNBC cases, the overexpression of ERβ was observed in patient samples, which could suppress the activity of PI3K and AR by upregulating phosphate and tensin homolog (PTEN), further suppressing the cell growth [83]. EGFR and BRCA1 may also affect the function of AR in breast cancers. It has been reported that the EGFR expression level and the frequency of BRCA1 deficiency are higher in TNBC [84]. The co-inhibition of AR and EGFR showed an additive growth suppression [85]. BRCA1 was reported as one of the AR co-activators, while a deficiency in BRCA1 may downregulate the expression of AR, and thus the activity of AR [86]. Therefore, the crosstalk of AR, EGFR, and BRCA1 may affect the significance of AR in breast cancers, especially in TNBC. In prostate cancer, the methylation of CpG islands located in the AR promoter and microRNA modulation leading to the silencing of gene transcriptional activity was reported [87,88]. Whether AR’s expression level and activity in breast cancer are also related to epigenetic modification is poorly understood. A study suggested that 5’ untranslated region mutation (T105A) of AR promotor was identified from AR-negative breast cancer patients, and could affect AR expression [89]. MicroRNAs, for example, miR-34, miR-205, and miR-320, have been reported to modulate the expression of AR in prostate cancer [90]. There should be a similar regulatory mechanism of AR expression in breast cancer. MiR-34 [91] and miR-205 [92] are tumor suppressors in breast cancer. However, the information showing whether these miRNAs would modulate the expression of AR is missing. We do believe some miRNAs would be the upstream regulators of AR expression. Therefore, addressing the upstream regulators of AR will be important in breast cancer. These results may partially explain the conflicting results. In addition, AR-targeted antibodies and the cut-off point for AR positivity (Table 1) used among different studies were diverse. Collectively, these suggest that a more authoritative guidance is needed for determining AR activity in order to help evaluate the clinical significance of AR in TNBC patients.