An Update on the Molecular and Clinical Characteristics of Apocrine Carcinoma of the Breast

Apocrine carcinoma of the breast is a rare malignancy. According to 2019 WHO classification, apocrine cellular features and a characteristic steroid receptor profile (Estrogen receptor (ER)-negative and androgen receptor (AR)-positive) define apocrine carcinoma. Her-2/neu protein expression is reported in ∼30-50% of apocrine carcinomas, while NGS analysis showed frequent PIK3CA/PTEN/AKT and TP53 mutations Followed by deregulation in the mitogen-activated protein kinase pathway components (mutations of KRAS, NRAS, BRAF ). A recent miRNA study indicates various miRNAs (downregulated hsa-miR-145-5p and upregulated 14 miRNAs such as hsa-miR-182-5p, hsa-miR-3135b, and hsa-miR-4417) may target the commonly altered pathways in apocrine carcinomas such as ERBB2/HER2 and mitogenactivated protein kinase signaling pathway. Although AR expression is a hallmark of apocrine carcinoma, little is known regarding the efficacy/resistance to antiandrogens. Success of bicalutamide, a non-steroidal anti-androgen, was reported in a case of Her2-negative apocrine carcinoma. Two recent studies, however, described presence of antiandrogen resistance biomarkers (a splice variant ARv7 and AR/NCOA2 co-amplification) in a subset of AR + apocrine carcinomas, cautioning the use of anti-androgens in AR + triple-negative breast carcinomas. Apocrine carcinomas rarely show biomarkers predictive of response to immune checkpoint inhibitors (PD-L1 expression, MSI-H status, and TMBhigh). Therefore, a comprehensive cancer profiling of apocrine carcinomas is necessary to identify potential therapeutic targets for a truly individualized treatment approach. Clinical Breast Cancer, Vol. 000, No.xxx, 1–10 © 2021 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ )


Introduction
Breast cancer is the leading malignancy among adult females worldwide, with a high mortality rate that is only preceded by lung cancer. 1 It is a heterogeneous and complex disease encompassing numerous and diverse histologic and molecular genetic types. 2 , 3 While invasive breast carcinoma of no special type (NST) constitutes ∼70% of all breast malignancies, the remaining 30% include various and rare (special) subtypes, defined by distinct morphology, molecular expressions, and/or genetic features; consequently, clinical course and treatment options vary significantly. 3 In the current review, we continue 4 , 5 to bring attention to the apocrine breast carcinoma, critically appraise and summarize the recent literature on molecular and clinical studies in the field.

WHO Definition of Apocrine Carcinoma
The 2019 WHO classification of breast tumors recognized apocrine carcinoma as a distinct, special type of breast cancer (under the name "carcinoma with apocrine differentiation"). 3 It is characterized by a distinct apocrine morphology (described and illustrated in Figure 1 ), which must be present in > 90% of cancer cells ( = essential criteria). As desirable criteria, the WHO proposed a characteristic steroid receptor profile: Estrogen receptor (ER)negative and androgen receptor (AR)-positive. When strictly defined using the essential and desirable criteria, apocrine carcinoma is a rare breast malignancy, constituting ∼1% of all breast cancers. 6 The WHO classification also proposed a diagnostic algorithm and differential diagnostic approach for breast tumors whose cells exhibit eosinophilic or foamy cytoplasm (Summarized and updated in Table 1 ). The diagnosis of apocrine carcinoma can occasionally be challenging due to the overlapping/similar/ morphology with some  other, even rarer neoplasms, such as oncocytic carcinomas. However, the combination of morphology and specific immunohistochemical biomarkers (eg, mitochondrial stains) can be helpful in such difficult cases ( Table 1 ). Other differential diagnoses (eg, granular cell tumors, histiocytic lesions) can also be ruled out using simple immunohistochemical algorithms ( Table 1 ). We believe that this approach will substantially improve the diagnosis of apocrine carcinoma, which has been a subject of discussion and controversy for a long time. The lack of clearly defined diagnostic criteria has also contributed to the contradictory and inconsistent data in the published literature, including diagnostics, clinical presentation, and outcome of the patients with apocrine carcinoma (please refer to the paragraphs on molecular and clinical characteristics, the results of these studies are summarized in Tables 2 and 3 ). A distinct "molecular apocrine carcinoma/tumor/" subtype was defined based on the analysis of the gene expression data and is characterized by the consistent AR activity and the lack of ER activity (with or without HER2 activity). 7 Gene expression studies also revealed that these tumors exhibit predominantly luminal features (eg, expression of luminal cytokeratins and the  71 EGFR positive in 86.5% "molecular apocrine cases" (ER-/PR-/AR + ) 32% co-expressed EGFR and HER2 EGFR negatively affected the prognosis; correlated with AR and higher Ki-67 EGFR as a potential therapeutic target lack of basal features) and are therefore called "luminal androgen receptor"/LAR/ tumors. 7-10 However, molecularly defined apocrine carcinoma does not necessarily correlate with morphologically and immunohistochemically (ER-/AR + ) defined apocrine carcinomas with the estimated overlap of ∼70% to 80%. 8 In addition, a vast majority of LAR carcinomas are of triple-negative phenotype while 30% to 60% of morphologically and immunohistochemically defined apocrine carcinomas exhibit ERBB2/HER2 overex-pression. This was confirmed in a recent study by Bonnefoi et al, who showed the concordance between molecularly and immunohistochemically confirmed apocrine carcinomas to be 88%. They also found that 2/3 of these apocrine carcinomas were HER2 positive. 11 These data indicate that apocrine carcinomas are heterogeneous. Even when strictly defined by morphology and immunohistochemistry, two molecular subtypes of apocrine carcinomas exist (HER2positive and triple-negative). Within both molecular subtypes, a Clinical Breast Cancer 2022   It is still a common practice to diagnose apocrine carcinomas by their morphologic features. Given the recent recommendations from the WHO Classification of Breast Tumors, 6 we advise the practicing pathologists to adopt a new diagnostic algorithm, which combines a steroid receptor profile and the characteristic apocrine morphology. We believe that this approach will improve diagnostic accuracy and consistency in reporting of apocrine carcinomas and ultimately contribute to a better clinical characterization of this peculiar mammary malignancy.

Novel Molecular Characteristics and Biomarkers of Apocrine Carcinoma
The results of the recent molecular studies on apocrine carcinoma are summarized in Table 2 . AR expression is a characteristic, diagnostic hallmark of apocrine carcinoma but itself is not specific, as AR is expressed in a variety of breast carcinomas, both ER-positive (70%-90%) and ER-negative (20%-40%). [12][13][14] Numerous studies have also explored the prognostic value of AR expression in breast cancer. 12 , 14-17 Although earlier studies revealed promising therapeutic effects of antiandrogens in Clinical Breast Cancer 2022 AR + breast carcinomas, including apocrine tumors, 18 , 19 two recent studies, 20 , 21 showed low to modest therapeutic benefits of antiandrogens in a mixed group of AR + breast carcinomas. Both studies included ER-positive and ER-negative AR-positive breast carcinomas with a pretreatment assessment of the AR positivity. Furthermore, other important biomarkers, including those involved in a potential anti-AR resistance, have not been explored. Other clinical trials with anti-AR are explored in more detail in a recent review of Cipriano et al. 22 Genetic studies on the AR gene are rarely reported. A pivotal study by Kasami et al 23 explored the CAG repeat number of the AR gene in a cohort of fibroadenomas, DCIS, and invasive breast carcinomas. The authors found the highest CAG repeats in DCIS, particularly in DCIS with apocrine differentiation. 23 A study by Lee et al exploring the CAG polymorphisms of the AR gene revealed no association with the development of breast cancer, but patients with more (23 + ) CAG repeats of the AR gene had a poor prognosis. 24 Farmer et al found no significant differences in CAG repeats between molecular apocrine carcinomas (17)(18)(19) and basal (18) and luminal (20) subtypes. 7 Cremonini et al recently reported a small (n = 20), well-defined apocrine cohort exploring the status of the AR gene. They found AR loss (monosomy) in most of the tested cases, along with the retained transcriptional activity of several AR regulatory genes, including the MAGE family, UXT, and FLNA genes. 25 Based on these findings, the authors speculated that the patients with apocrine carcinoma might benefit from androgendeprivation therapy, but this requires a clinical validation. Androgen receptor variant 7 (ARv7), a splice variant of AR, is another essential biomarker closely related to anti-AR effectiveness (resistance) as confirmed in prostate carcinoma, 26 as well as in salivary duct carcinoma, which shares many morphologic and molecular similarities with apocrine breast carcinoma. 27 , 28 The AR-V7 encodes a truncated AR protein that possesses only the transactivating Nterminal domain without the C-terminal ligand-binding domain, resulting in constitutive activation of AR. 29 Similar effects have been shown in breast cancer. Thus, ARv7 variants might induce proliferation of the apocrine cell line MDA-MB-453 in the presence of antiandrogen enzalutamide. 30 ARv7 was recently described in a cohort of primary and metastatic breast cancers. 31 The overall frequency was ∼10% but was significantly higher (42%) in AR + carcinomas with apocrine morphology. Notably, ARv7 was also detected in primary, therapy-naïve breast carcinomas (without previous exposure to anti-androgens), indicating a potentially different mechanism of ARv7 activation in the breast compared with prostate carcinomas, but similar to the findings of ARv7 in salivary duct carcinomas. 28 The authors proposed routine ARv7 testing for all patients with AR-positive apocrine tumors that are being considered for the treatment with AR inhibitors. 31 Interestingly, ARv7 and co-amplification of AR and the nuclear co-receptor NCOA2, both of which are associated with anti-AR resistance in prostate cancer, 32 have been recently reported in a phase Ib/II clinical trial with AR-positive TNBC. 33 ARv7 was also identified among the non-responders to antiandrogens, indicating its role in the resistance to the therapy. Although LAR patients had better therapeutic response to the combined AR/PIK3CA inhibition than non-LAR TNBC patients, overall therapeutic benefit was limited. 33 This study is one of the properly designed clinical trials that included a comprehensive molecular assessment prior and after the targeted treatments. Owing to such a design, the authors were able to identify biomarkers of response and resistance.
Gross cystic disease fluid protein 15 (GCDFP-15 or PIP-3), along with AR expression, has been considered a biomarker of apocrine differentiation in the breast. 6 Nakamura et al recently reported that α-Methylacyl-CoA racemase (AMACR) is expressed with a marked preponderance in apocrine breast carcinomas (both in situ and invasive). AMACR has been a biomarker of prostate and several other cancers (eg, papillary renal cell carcinoma, urothelial carcinoma in situ). 34 , 35 They found that AMACR was expressed in 96% of apocrine ductal carcinoma in situ (DCIS) cases and 97% of invasive apocrine carcinomas, in contrast to non-apocrine carcinomas that exhibited AMACR positivity in only 22% of the cases. The study revealed a comparable sensitivity of AMACR with GCDFP-15 for apocrine carcinomas, whereas the AMACR specificity was significantly higher (78% vs. 32%). Notably, AMACR protein expression correlated well with its mRNA expression. 34 The expression of AMACR ( Figure 2 ) raises an interesting possibility of the role of diet and the interplay with the hormonal status and apocrine breast carcinomas. High animal fat consumption is associated with an increase in triple-negative breast cancer (TNBC) risk in premenopausal women. 36

Epithelial-to-Mesenchymal Transition (EMT) in Apocrine Carcinoma
Epithelial-mesenchymal transition (EMT) is one of the hallmarks of cancer progression 37 . The claudins are transmembrane proteins that regulate the tight junctions between epithelial cells and are involved in signaling between the epithelial cells and their environment 38 and are also involved in EMT. 37 Previous studies showed that claudins 1, 3, and 4 are consistently expressed in normal mammary epithelium. 39 In contrast, apocrine cells within apocrine metaplasia were positive for claudin 1 and consistently negative for claudin 4. 39 Sousha et al recently demonstrated claudin 1 and claudin 3 expressions and the lack of claudin 4 protein expression in a small cohort of apocrine lesions, including invasive apocrine carcinoma with the triple-negative phenotype. 40 The diagnostic and potential therapeutic utility of these findings remains unknown.
MicroRNAs (miRNAs) represent small non-coding RNAs that act as post-transcriptional regulators of various cellular functions. miRNAs negatively regulate gene expression by their binding to their selective messenger RNAs (mRNAs), causing either mRNA degradation or translational repression, depending on their complementarity with target mRNA sequences. 41 miRNAs have been extensively characterized in various cancers, including breast cancer. Recently, Koleckova et al 42 demonstrated that triple-negative breast carcinomas with apocrine and spindle cell (metaplastic) morphology exhibited a distinct miRNA profile compared with other breast cancers. In particular, they showed the downregulation of hsa-miRNA-143-3p and hsa-miRNA-205-5p and upregulation of the hsa-miR-22-3p, hsa-miRNA-185-5p, and hsa-miR-4443 ( Table 2 ). Apocrine carcinomas also had decreased expression of hsa-miR-145-5p and increased expression of additional 14 miRNAs, including hsa-miR-182-5p, hsa-miR-3135b, and hsa-miR-4417. The pathway  analysis revealed that these miRNAs closely interfere with several important signaling pathways, such as Wnt, ErbB/HER2, and MAPK pathways; the authors also speculated that these miRNAs might contribute to EMT in special types of TNBC -apocrine and spindle cell (metaplastic) carcinomas, concluding that further mechanistic studies are essential to confirm their observations. 42 Notably, we also demonstrated the active EMT in a case of morphologically apocrine DCIS (AR + ) harboring PTEN and HRAS mutations with progression to spindle cell metaplastic carcinoma that had the same mutational profile and a loss of AR expression. 43 EMT was supported by the loss of E-cadherin protein (without CDH1 gene mutations or loss) and nuclear ß-catenin expression in invasive spindle cell component. 43 Further studies are required to elucidate the EMT in apocrine carcinomas and its clinical relevance (therapy response and resistance).

ARTICLE IN PRESS
The recent studies indicate that EMT might be actively involved in the pathogenesis of apocrine carcinomas via several molecular mechanisms. However, these studies do not reflect the full spectrum of signaling pathways that are involved in EMT in apocrine tumors. 44 , 45 Also, the clinical relevance of the observed alterations in apocrine carcinomas should be confirmed.

Immune Checkpoint Inhibitors in Apocrine Carcinoma
Immune checkpoint inhibitors (ICI) have markedly improved the treatment options and outcome of various solid and hematologic malignancies, including triple-negative breast cancer (TNBC). Thus, FDA approved pembrolizumab in both neoadjuvant and adjuvant settings along with its companion diagnostic test for PD-L1 testing (22c3 pharmDx assay, Agilent Technologies). In contrast, atezolizumab and its CDx SP142 were initially approved in 2019 but were withdrawn in August 2021 from use in TNBC patients. 46 Most of the clinical trials and randomized studies with ICI have not specifically addressed the role of apocrine morphology among TNBC. In addition, the response of HER2-positive apocrine carcinomas to ICI remains largely unknown despite the recently published promising therapeutic effects of atezolizumab combined with anti-HER2 drugs on HER2-positive breast carcinomas. 47 Several predictive biomarkers of response to ICI have been validated, including PD-L1 expression (on cancer, immune cells or both), high tumor mutational burden (TMB-H), and high microsatellite instability status (MSI-H). In TNBC samples, PD-L1 expression (positivity defined as Combined Positive Score [CPS] ≥10) detected by immunohistochemistry has been approved as a companion diagnostic (CDx) test for pembrolizumab. 48 A few earlier studies have specifically explored predictive biomarkers to ICI in apocrine carcinomas (PD-L1), reporting conflicting results. 5 , 49 Three recent studies indicate that apocrine carcinomas show low PD-L1 expression in both tumor and immune cells, low tumor mutational burden, and are consistently microsatellite stable (MSS). [50][51][52] Although high TIL is a feature of TBNC, 53 the studies reported low TIL in apocrine carcinomas, which along with a low percentage of intratumoral CD8 + and CD3 + lymphocytes, and a loss of MHC class I (including PD-L1 + apocrine cases), make patients with this cancer less likely responsive to ICIs. 51 , 52 , 54

Other Targetable Biomarkers in Apocrine Carcinoma
Comprehensive molecular profiling aimed at identifying potentially targetable alterations in cancer has become the standard for precision oncology. 55 Numerous studies have been published on various subtypes of breast cancer, but those exploring molecular features specifically of apocrine carcinoma remain sparse. Sun et al profiled eighteen "pure" triple-negative apocrine carcinomas (apocrine morphology + AR positivity), revealing PIK3CA (72%) , PTEN (33%), and TP53 (28%) alterations as the most common in apocrine carcinomas. 50 A proportion of the cases also harbored genetic alterations within the MAPK pathway ( BRAF, HRAS, KRAS, MAP3K1 ), cell cycle regulators ( CDKN2A, CDKN2B, CDK6 ), and FGF pathways ( FGFR2 amplification and fusion) ( Table 2 ). Notably, one apocrine case had a well-described TERT gene promoter mutation (c.-124C > T), while another had a novel FGFR2-TACC2 fusion, not previously reported in breast cancer. 50 The authors concluded that a vast majority of apocrine carcinomas harbored potentially targetable but diverse genomic alterations, Clinical Breast Cancer 2022 making their detection a requirement for successful personalized medicine approach (eg, PIK3CA/mTOR inhibitors 50 ).
Based on the previous data and their own results, Lehmann et al 33 explored anti-AR Enzalutamide combined with PIK3CA inhibitor Taselisib in a small cohort of AR-positive metastatic TNBCs (phase Ib/II study, TBCRC032). Seventeen pretreated patients randomly received enzalutamide with or without taselisib. Although all the patients experienced disease progression at 16 weeks except for one patient with LAR who was on the combined treatment and had not progressed within 18 months when the study was terminated. 33 In addition, the authors found AR expression to be insufficient in predicting the response although the LAR carcinomas had a substantially higher clinical benefit (75%) compared with other TNBC molecular subtypes (12.5%). 33 Further and larger studies should definitely confirm the benefit of such combined targeted therapies in apocrine carcinomas.
The above-described molecular alterations in apocrine carcinomas generally align with the previously published studies. 5 , 56-58 Dysregulation of the cell cycle regulators (CDKN2A and B, CDK6) in a subset of apocrine carcinomas indicates a potential for the treatment with CDK4/6 inhibitors, as shown in the study of Asghar et al. 59 The authors performed comprehensive in vitro and in vivo experiments using various breast cancer cell lines, including the apocrine MDA-MB-453 cells. They demonstrated that the apocrine cells were highly sensitive to CDK4/6 inhibitors. More importantly and relevant to apocrine carcinomas, CDK4/6 inhibitors exhibited a synergistic effect with PIK3CA inhibitors in PIK3CA -mutant cell lines including MDA-MB-453, extending the use of combined treatment with both CDK4/6 and PIK3CA inhibitors. 59 Taken together, the recent data confirm the relevance of comprehensive molecular profiling in identifying the targetable biomarkers in apocrine carcinomas. Further translational and clinical studies (basket trials) are needed to verify the findings from the cell lines and molecular studies. These could pave new treatment modalities for patients with advanced disease.

Clinical Studies on Apocrine Carcinoma
We systematically reviewed the recent literature ( ≥2018) on apocrine carcinoma exploring PubMed/MEDLINE, Scopus, and Web of Science Core Collection databases, using the following keywords: "Apocrine carcinoma", "carcinoma with apocrine differentiation", "molecular apocrine carcinoma", "luminal androgen receptor carcinoma", and "breast", "clinical characteristics/features", "outcome", and "survival". The studies exploring non-invasive carcinomas ( = apocrine DCIS) and benign apocrine lesions (eg, adenosis, metaplasia) were excluded from the analysis and review. Case reports and small case series ( < 5 patients) were also excluded.
Our literature search in the databases revealed 32 clinical studies that have been published since 2018 (the studies and their major results are summarized in Table 3 ). The number of the patients in the studies shows marked differences, varying from small studies involving 8 to 10 patients 60 , 61 to a large series ( > 1000 patients) retrieved from the publicly available databases such as the Surveillance, Epidemiology, and Ends of Results (SEER) and National Cancer Center Database (NCDB). Notably, most of the reported studies focused on the triple-negative apocrine carcinomas, while very few specifically explored HER2-positive apocrine carcinomas. 11 , 62 , 63 Similarly, most studies also reported the clinical outcome (overall-or disease-specific survival) of the patients with apocrine carcinoma alone or compared with the matched NST subgroup.
Consistent with the previous data, our literature survey confirms contradictory results about the response to chemotherapy and clinical outcome of patients with apocrine carcinoma. This is likely to be caused by the inconsistent diagnostic criteria used to define apocrine carcinomas. We believe that the new WHO definition of apocrine carcinoma with diagnostic utilization of essential and desirable criteria will help better define this category and hence identification of clinically useful information.
Very few studies specifically explored the effects of (neo) adjuvant chemotherapy in patients with apocrine carcinoma. [63][64][65][66][67][68] Several studies clearly pointed the limited response of molecularly defined triple-negative apocrine carcinomas (LAR) to neoadjuvant chemotherapy in comparison with other non-apocrine TNBC. 66 , 67 Both studies also reported significantly lower Ki-67 in LAR compared with non-apocrine TNBC, which is in line with previous studies. 13 , 57 Zhu et al and Mohammed et al showed that the lack of AR expression in TNBC independently predicted pCR among TNBC patients. 69 , 70 In addition, a systematic review of Trapani et al. revealed that triple-negative, AR + positive apocrine carcinomas had no benefit of adjuvant chemotherapy if treated in the early stage (pN0). 68 Instead, the authors proposed that antiandrogens should be considered for such patients in the adjuvant setting. 68 As discussed above, the response to anti-AR may also be limited due to the various resistance mechanisms as shown in a detailed clinical and molecular study by Lehmann et al. 33

Conclusions and Future Directions
Recent advances have contributed to the improved diagnostics of apocrine carcinoma of the breast. These efforts should also reduce the considerable variability and discrepancy in apocrine carcinomas' definition, molecular and clinical characteristics. Novel biomarkers have also been described, but their diagnostic and clinical (predictive and prognostic) utility has to be confirmed. The current evidence indicates that AR-positive breast carcinomas, including apocrine subtype, may have limited clinical benefit of (neo) adjuvant chemotherapy. Apart from the Her-2/neu target, advanced and/or metastatic apocrine carcinomas still have limited targeted treatment options. The role of antiandrogens in apocrine and other AR-positive breast carcinomas also require further research as limited data (small number of studies and small sample size) on the potential mechanisms of response/resistance are currently available. Therefore, a comprehensive genomic cancer profiling of apocrine carcinomas appears to be a promising approach that could reveal potential targets for an individualized therapeutic treatment.