Morphological Features and Immunohistochemical Profiling of Male Breast Gynaecomastia; A Large Tissue Microarray Study

Introduction Gynaecomastia is the commonest male breast condition accounting for approximately 85% of male breast lesions. There is minimal information on the immunohistochemical profile of gynaecomastia. We aimed to comprehensively profile a large series of gynaecomastia samples for putative mammary diagnostic, predictive and prognostic markers. Methods A total of 156 samples, were histologically reviewed, assembled onto tissue microarrays, and stained for oestrogen receptors (ERα, ERβ1, ERß2), progesterone receptors (total PR, PRα), androgen receptor (AR), basal & luminal cytokeratins (CK5/6, CK14, CK8/18) and the proliferation marker Ki67. Relevant cut offs for marker positivity were defined based on existing literature: AR (10%), ERα and PR (Allred score >3/8), ERß (10% and 20%), cytokeratins (10%) and Ki67 (10% and 20%). Results 108 samples from 86 patients aged 13-75 years were available for immunohistochemical assessment. 73.1% of the lesions were AR positive, compared to 99% for ERα and 100% for both ERß1 and ERß2. 98% of samples were positive for total PR and 97.1% for PRα. 69.8% expressed CK5/6 whilst 57% were CK14 positive. A tri-layered pattern of cytokeratin expression was also observed. Ki67 positivity was low with 17.1% and 6.7% classified as Ki67 positive using 10% and 20% cut off values respectively. A significant negative correlation was found between ERα expression and patient age (rs = -0.221, p=0.023). Bivariate correlations were produced, and comparisons made with previously published data regarding the immunohistochemical status in normal female breast tissue, proliferative and neoplastic breast diseases of the female and male breast. Conclusions Hormone receptors, including oestrogen receptor α and ß isoforms as well as androgen receptors were abundantly expressed within the intraductal luminal hyperplastic epithelium in gynaecomastia supporting the hormonal role in the pathogenesis and treatment. ERα, ERβ1 and ERβ2 were expressed in a higher proportion of cells compared with their expression in the female breast benign lesions which further characterises gynaecomastia biology. The identification of a low Ki67 proliferative index and the mixed cytokeratin profile in gynaecomastia differentiates this benign condition from male breast cancer. Therefore, Ki67 and cytokeratins can help in the differential diagnosis from histological mimics in the routine diagnostic work up.

Results: 108 samples from 86 patients aged 13-75 years were available for immunohistochemical assessment. 73.1% of the lesions were AR positive, compared to 99% for ERa and 100% for both ERß1 and ERß2. 98% of samples were positive for total PR and 97.1% for PRa. 69.8% expressed CK5/6 whilst 57% were CK14 positive. A tri-layered pattern of cytokeratin expression was also observed. Ki67 positivity was low with 17.1% and 6.7% classified as Ki67 positive using 10% and 20% cut off values respectively. A significant negative correlation was found between ERa expression and patient age (rs = -0.221, p=0.023). Bivariate correlations were produced, and comparisons made with previously published data regarding the immunohistochemical status in normal female breast tissue, proliferative and neoplastic breast diseases of the female and male breast.

INTRODUCTION
Gynaecomastia is a benign, non-neoplastic proliferative condition and is the most common lesion in the male breast. Pseudogynaecomastia, or lipomastia, refers to the accumulation of adipose tissue in the male breast without glandular hyperplasia (1). True gynaecomastia can be differentiated from pseudogynaecomastia by the presence of dense subareolar ductal tissue and fibrosis. There are several aetiological factors for true gynaecomastia including physiological causes, such as neonatal, adolescent, and elderly gynaecomastia, resulting from an imbalance between oestrogen stimulation and androgen inhibition (2). Pathological causes of gynaecomastia include Klinefelter's syndrome, where hypogonadism causes an increased oestradiol-to-testosterone ratio leading to the genesis of gynaecomastia (3). Drugs, obesity, and relative oestrogen excess (e.g., due to liver cirrhosis or prostate cancer) are amongst other causes of gynaecomastia. However, most gynaecomastia cases are regarded as idiopathic with no apparent cause. There is no proven link between gynaecomastia and the development of male breast cancer (4,5).
The histological features of gynaecomastia are well documented. Microscopically, the early stage of gynaecomastia (florid phase) is characterised by proliferative branching ducts with epithelial and stromal hyperplasia, and the late stage (fibrous phase) is characterised by collagenous stroma, less epithelial proliferation, and decreased vascularity (4).
Little is known about the immunohistochemical profile of male gynaecomastia and how it compares to the female breast due to the lack of large studies to date analysing the expression of putative diagnostic and prognostic markers within this lesion. Therefore, the aim of this study was to analyze the expression of several relevant proteins, including hormone receptors, cytokeratins (luminal and basal types) and proliferation markers in a large, well characterised cohort of surgically resected gynaecomastia specimens.

MATERIALS AND METHODS
Ethical approval was sought and granted from the Leeds (East) Research Ethics Committee (reference number 06/Q1205/156). Male patients who had undergone surgical excision for gynaecomastia at the Leeds Teaching Hospitals NHS Trust, during the period between 2000 to early 2011 were identified from the pathology database. Histological sections of those patients were collected and reviewed by two investigators (AB, AMS) to confirm the diagnosis, identify the morphological appearances, classify into early or late stage, and mark representative areas for tissue microarray (TMA) construction. Comprehensive clinical details were collected including patient age, presentation, laterality, history of previous medications, history of previous cancer and family history where available.

Full Face Immunohistochemistry
Prior to the construction of TMAs, representative full sections for 10% of the gynaecomastia cases studied were randomly selected and immunohistochemically stained for the complete panel to assess for staining heterogeneity. Staining was overall uniform across sections and tissue microarrays were deemed a suitable method for analysing the large sample numbers.

TMA Construction
Representative tissue cores from the donor blocks were assembled onto tissue microarrays.
Tissue microarrays (TMAs) were constructed using a manual tissue arrayer (Beecher Instruments, Inc., Sun Prairie, W1, USA) using representative 0.6mm tissue cores from representative marked areas. For each case, 4 cores, arranged in duplicates, were included. TMAs were sectioned at 3 microns and stained with the panel of immunohistochemical biomarkers. Cores of various other human tissues including female breast carcinoma, prostatic carcinoma, normal prostate, endometrium and appendix were assembled in the TMAs in an orderly fashion to serve as positive controls and also for orientation as previously described (6). For AR, ERa, total PR, CK5/6, CK8/18, CK14 and Ki67 optimised immunohistochemical protocols on Dako Autostainer Link 48 routinely used within the Leeds Teaching Hospitals NHS Trust diagnostic histopathology department and approved by United Kingdom National External Quality Assurance Scheme (NEQAS) were followed in accordance with standard operating procedures.
The initial immunohistochemical stage incorporated slide pre-treatment, whereby the formalin-fixed paraffin embedded tissue sections underwent deparaffinisation, hydration and heatinduced epitope retrieval (HIER) by immersion in the automated DAKO Envision Flex High pH Link solution system. During this timeframe refrigerated (4°C), commercial concentrated monoclonal mouse antibodies raised against human AR, ERa, total PR, CK5/6, CK8/18, CK14 and Ki67 tissue proteins were diluted in accordance with pre-determined, optimised dilution ratios ( Table 1) and placed within defined Dako reagent bottles. These diluted primary antibodies and ready-to-use reagents supplied within the DAKO Envision HRP/DAB+ Flex Plus High detection kit were placed within the Dako reagent racks. After completion of HIER, slides were washed in Envision Flex wash buffer. Slides and reagent racks were loaded onto the DAKO Autostainer Link 48 for integrated, automated section staining for tissue antigen visualisation using the Dako Autostainer Link 48, following manufacturer's instructions.
For ERb1, ERb2 and PRa a manual staining protocol using a Shandon sequenza rack was performed as previously described (7). Positive controls were included in each batch of staining ( Table 1). All sections were counterstained with Mayer's haematoxylin and Scott's tap water substitute.

Statistical Analysis
Data analysis was done using IBM SPSS Statistics version 27 programme. The percentages of samples positive for each receptor and combination of receptors were calculated. Correlations were two sided and considered as statistically significant when p ≤0.05 and highly significant when p <0.001. Pearson correlations were calculated between biomarkers, and between biomarkers and age. The non-parametric Mann-Whitney U test was used to assess the relation between age as a continuous variable and the presence/ absence of hyperplasia.

Patient Characteristics
A total of 118 patients fulfilled the inclusion criteria, of whom 38 had bilateral gynaecomastia. For the latter, tissues from each breast were reviewed and representative blocks selected. Cases diagnosed as gynaecomastia on core biopsy only without a surgical specimen over the 11 years search period were excluded.
Following histological review, surgical specimens comprising predominantly fatty tissue with little/no glandular elements were excluded from the immunohistochemical part of the study. Therefore, a total of 156 gynaecomastia samples were used for TMA construction and immunohistochemical analysis.
After excluding non-representative and missing cores, immunohistochemical data was available for 108 samples from 86 patients. 24 of these patients had bilateral samples taken. 48 samples were taken from the right breast, 59 samples were taken from the left breast. Laterality was not known for one sample. Patients' age ranged from 13-75 years, with a mean of 28.07 years ( ± SD:14.97).
Data of previous therapies were not available for 50% of patients. 30.6% had not received any previous therapies. 5.6% had a history of the selective oestrogen receptor modulator Tamoxifen, 3.7% had received Danazol therapy. One patient had Docetaxel therapy and another 2.8% received previous Casodex therapy, both of which were indicated for metastatic prostate cancer. Only one patient had a family history of breast cancer. 5.6% of patients had a history of illicit drug use (cannabis and cocaine) and one patient had previously used steroid body building supplements.

Morphological Appearances
All cases were examined for the characteristic morphological appearances of gynaecomastia on H&E sections. A total of 93 samples (86.1%) showed active (early) gynaecomastia with associated epithelial hyperplasia. There was no significant correlation between age and the histological appearance of gynaecomastia (Mann-Whitney U, p=0.306). Active gynaecomastia was characterised by hyperplastic branching ducts within epithelial proliferation and periductal stromal oedema ( Figure 1A). The stroma was generally cellular with and hyalinisation were occasionally noted. Morphologically distinct latent phase gynaecomastia was observed in the remaining smaller proportion of gynaecomastia cases. This latter phase of dormancy was characterised by stromal fibrosis and sparse glandular elements ( Figure 1B). No periductal oedema or prominent stromal vascularity was noted. While mammary lobules were not generally described as a feature of the male breast, examples of mammary lobules resembling those identified in the female breast were seen in a small proportion of cases ( Figure 1C).

Hormone Receptor Expression in Gynaecomastia
All hormone receptors analyzed, including ERß and PR isoforms, were highly expressed in gynaecomastia lesions (  (Figures 2A-L). The tri-layered pattern of basal cytokeratin expression enclosing a luminal layer has been identified in a patchy fashion ( Figures 2J, K). The Ki67 proliferation index was low ( Table 3). 10.5% of samples showed 100% ERa expression. 23.1% of samples showed no (0%) AR expression and 4.6% showed 100% AR expression. The majority of samples (41.7%) showed moderate staining of AR in the luminal cells.
When co-expression of hormone receptors was analyses, 66.7% of samples were positive for AR, ERa and PR. 68.5% were positive for AR, ERß1 and PR, compared to 66.7% for AR, ERß2 and PR ( Table 4).  The correlations between age and expression of each biomarker were also studied. There was a significant negative correlation between ERa expression and age (rs= -0.221 p=0.023); a feature opposite to that seen in the female breast. ERa was the only biomarker to show a significant correlation with age. ERa also showed a highly significant positive correlation with AR (0.499, p<0.001), PR (0.305, p=0.002), PRa (0.340, p<0.001), and significantly correlated with CK8/18 (0.219, p=0.025). High ERß1 significantly correlated with lower CK14 expression (-0.224, p=0.025). ERß2 showed a significant positive correlation with PRa (0.272, p=0.006). Highly significant correlations were noted between AR and both PR (0.281, p=0.005) and PRa (0.340, p<0.001). Ki67 showed no significant correlation with any of the studied markers. The correlations between percentage expression of the studied biomarkers are shown in Table 5.  results. Here, we detail the immunohistochemical profile of a large series of male breast gynaecomastia lesions represented on TMAs and the association between different protein markers.
Hormones have long been implicated in the pathogenesis of gynaecomastia and endocrine treatment, including androgens, anti-oestrogens, and aromatase inhibitors are currently in use for the medical management of those lesions (14,15). When gynaecomastia persists after a period of reassurance and observation and/or discontinuation of exposure, Tamoxifen (a selective Estrogen receptor modulator, SERM) is the first medical treatment option (14). Tamoxifen is known to reduce pain and size of gynaecomastia. Examples of lesions that have completely resolved following Tamoxifen treatment have also been reported (16). Danazol, a weak androgen, has been also used but is less effective than Tamoxifen and may indeed exacerbate gynaecomastia due to its conversion to oestrogen and/or the resultant weight gain (17).
We report a predominance of various types of nuclear steroid receptors in gynaecomastia with over two thirds of the lesions expressing androgen receptor. A previous small study reported 80% AR positivity in those lesions (18). An earlier study reported AR positivity in 100% of gynaecomastia samples, compared to 87% positivity in male breast carcinoma (19). A strong correlation between ERa and AR was noted in this study (0.499, p<0.001), which is similar to a previously recorded close association noted between the two markers (p<0.01) (19).
AR immunoreactivity has been shown to range between 38% and 81% in MBC (20) and the presence of it has shown to correlate with better overall survival in MBC (21). The AR expression shown in this study supports the use of nonaromatising androgen therapies, such as dihydrotestosterone, which has shown good response rates in patients with prolonged pubertal gynaecomastia (22).
One intriguing finding in our study is the inverse correlation between ERa expression and age with a decrease in its expression with the increase in patient's age. To our knowledge, this is a novel finding and is different from the female breast where ERa expression has been shown to increase in mammary epithelium of post-menopausal women, likely a response to decreased circulating levels of oestrogen and increased sensitivity of the receptors (23,24). The reason for this finding in the male breast awaits further studies but may be due to the relative increase of serum estradiol relative to androgen in elderly men or disturbances in the local breast tissue response to estrogen due to the decreased inhibitory effect of androgen as a result of aging. Table 6 summarises the expression of hormone receptors in non-neoplastic and malignant breast lesions in men and women as reported in the literature.
Within our current study, 99% of samples were deemed ERa positive. High percentages of ERa positivity have been documented in both gynaecomastia and MBC at 100% and 87% expression respectively (30). Cases of MBC are often found to have a high percentage of ERa expression compared to female breast cancer (83% and 68% respectively) (30).
Another novel finding is the high expression of ERß isoforms, ERß1 and ERß2, in all the gynaecomastia lesions examined. In addition to the localisation in the lesional epithelial cells, ERß isoforms were also expressed in the stromal cells suggesting a role of ERß in the pathogenesis of gynaecomastia. Nicoletti et al. confirmed high levels of ERß RNA expression in primary cultured cells from 50 examples of male pubertal gynaecomastia including in stromal cells (31). They concluded  that the data support a role of ERß in the pathophysiology of pubertal gynaecomastia. However, that same study noted a significant inverse correlation between AR and ERa (p<0.01) in carcinoma (31). An inverse correlation was also noted between AR and PR (p<0.01) in carcinoma cases. In contrast, our study noted a significant positive correlation between AR and PR (0.281, p=0.005), indicating that the relationship between biomarkers within gynaecomastia is distinctly different from malignant male breast lesions.
Little information is available on the basal and luminal cytokeratin expression in gynaecomastia. Here, we confirm the previously reported tri-layered pattern of expression of outer and inner positive basal cells enclosing a middle luminal layer. Within this tri-layer epithelium of gynaecomastia lesions, CK5/ 6 and CK14 are commonly expressed, 67% and 21% respectively in the inner luminal layer, 6% and 1% respectively in the intermediate luminal layer, and 84% and 99% respectively in the outer myoepithelial layer 2 . Our study showed similar results of 68.9% expression of CK5/6 and 57% expression of CK14 within glandular epithelium. Basal cytokeratins therefore can be used in the diagnostic setting to confirm hyperplasia. In the female usual ductal hyperplasia, the expression of basal cytokeratins occurs in a heterogeneous/mosaic fashion with positive basal like cells admixed with negative luminal cells (4). Awareness of the different patterns of the immunohistochemical expression of CK5/6 and CK14 in the male and female hyperplastic lesions is important to avoid mistaking the trilayered pattern in male breast for Pagetoid spread/atypia. It is of note that, in male breast cancer, a basal profile is extremely uncommon (5,32,33).
In conclusion, we report on the morphological and immunohistochemical features of a large cohort of male breast gynaecomastia lesions received in a large single institution over 11 year period. The majority of the lesions showed hyperplasia, exhibited high levels of oestrogen and progesterone receptors as well as androgen receptors. We confirm, for the first time, the abundance of ERß isoforms, both in the epithelial and stromal elements, and the negative association between ERa and age. The findings suggest a major role of hormonal factors in the pathogenesis of gynaecomastia and the value of anti-hormonal therapy in the medical management of this common condition. Using hormone receptors basal cytokeratins and ki67 can help differentiate florid examples of hyperplastic male breast lesions from the malignant mimics.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding author.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Leeds (East) Research Ethics Committee (reference number 06/Q1205/156). Written informed consent from the participants' legal guardian/next of kin was not required to participate in this study in accordance with the national legislation and the institutional requirements.

AUTHOR CONTRIBUTIONS
PP: analyzed the data and wrote the first draft; AB: performed the laboratory work, collected patient data, analyzed the data and contributed to writing up; VS: ethical approval, contributed to writing up; AS: conceived the idea, supervised experimental work, oversaw writing up. All authors approved the final manuscript.