Abstract
Extramammary Paget disease (EMPD) is a rare cutaneous malignant neoplasm arising in apocrine gland-rich areas. Although – like normal apocrine glands – EMPD frequently expresses androgen receptor (AR), the clinical significance of AR expression remains unclear. The present study investigated the clinicopathological impact of AR expression in EMPD. Immunohistochemistry for AR was performed in a retrospective cohort of 92 EMPD patients with 108 EMPD lesions, including 102 primary lesions, five lymph node [LN] metastases and one local recurrence. The total AR staining score was calculated as staining intensity score (IS 0–3) × positive-cell percentage score (PS 1–4). Expression levels were graded as Grade 1 (scores 0 and 1), Grade 2 (scores 2–4), and Grade 3 (scores 6–12). Higher expression grade was correlated with tumor thickness (P = 0.011), LN metastasis (P = 0.008), and higher EMPD stage (P = 0.023). Grade 1 EMPDs did not invade into the dermis and did not generate metastatic and/or recurrent lesions, whereas only Grade 2 or 3 EMPDs did so. AR expression in invasive components was significantly higher (P = 0.023) than in non-invasive components remaining within the epidermis. AR expression was further elevated in metastatic and/or recurrent lesions relative to locally invasive lesions (P = 0.014). These results clearly indicate that increased AR expression is associated with malignant progression of EMPD and that androgen blockade might be an effective therapy. Furthermore, AR expression assessed by immunohistochemistry may have potential for prediction of LN metastasis and local recurrence in EMPD.
1. Introduction
Extramammary Paget disease (EMPD) is a rare cutaneous malignant neoplasm arising in areas of apocrine gland-bearing skin, such as the genitoperineal region and axilla, and is classified as an apocrine gland neoplasm [1], [2]. Most EMPD cases are diagnosed as non-invasive (so-called carcinoma in situ) and usually show indolent disease progression. However, once invasive growth occurs in the dermis, regional lymph node (LN) and distant metastases frequently occur, and the prognosis worsens [3].
Normal apocrine glands of the skin contain no detectable estrogen receptor (ER) or progesterone receptor (PR) but consistently express androgen receptor (AR) [4]. Previous studies have indicated that most EMPD cases do not express ER and PR, but frequently express AR, similar to normal apocrine glands (54–90%) [5], [6], [7], [8], [9]. However, most previous immunohistochemical studies of AR in EMPD included a relatively small number of patients, perhaps mainly because EMPD is a rare condition. It is still controversial whether non-invasive and invasive lesions differ in their levels of AR expression [7], [10]. Moreover, those previous studies focused solely on primary lesions, and AR expression levels in primary lesions and metastatic and/or recurrent lesions have never been compared. Therefore, the clinical significance of AR expression has not been well clarified.
On the other hand, a previous case report has indicated that androgen blockade with a combination of bicalutamide and leuprolide acetate (a luteinizing hormone-releasing hormone agonist) for treatment of prostate cancer was able to reduce multiple bone metastases from EMPD [11], suggesting the potential effectiveness of such therapy for EMPD. However, before such therapy can be widely applied, it is necessary to clarify whether AR is truly associated with the development and progression of EMPD. The aim of the present study was to clarify the clinicopathological impact of AR expression in EMPD. We performed immunohistochemistry for AR in a large retrospective cohort of 92 EMPD patients with 108 EMPD lesions and examined in detail the correlation between immunohistochemical expression of AR and various clinicopathological factors. AR expression was also compared between non-invasive and invasive components and between primary lesions and metastatic and/or recurrent lesions.
2. Materials and methods
2.1. Patients and tumor lesions
This study was approved by the Ethics Committee of Keio University School of Medicine (approval number: 20110159) and performed in accordance with the Declaration of Helsinki. We retrospectively enrolled 92 EMPD patients with 108 EMPD lesions: 102 primary lesions, five LN metastatic lesions, and one locally recurrent lesion. We collected biopsy samples for pathological diagnosis or surgical specimens obtained via total or palliative resection from patients with primary EMPD diagnosed between January 1999 and December 2021 at Keio University Hospital. In all cases, no underlying malignancywas found, ruling out the possibility of secondary EMPD, which is characterized by cutaneous involvement due to underlying carcinomas of the lower gastrointestinal, lower urinary, or lower female genital tracts [12]. Patients who had received any preoperative treatment such as chemotherapy and radiotherapy before total or palliative resection were excluded. Single, double, and triple primary lesions were seen in 84 (91.3%), six (6.5%), and two (2.2%) patients, respectively.
Available clinicopathological data were collected from hospital medical records and the pathology reporting system: demographic data (age at initial presentation and sex), clinical data (tumor site and tumor size), and histopathological features of the surgical specimen including tumor thickness and presence of lymphovascular invasion (LVI), the number of metastatic LNs at initial presentation or surgery, and distant metastasis at initial presentation. Tumor-node-metastasis (TNM) classification and EMPD staging were performed based on a system proposed by Ohara et al. [13], which is currently the only TNM staging system for invasive EMPD worldwide: tumor thickness was stratified into three groups: T0; tumor in situ, T1; tumor thickness ≤ 4 mm and no LVI, and T2; > 4 mm or LVI. LN metastasis was stratified into three groups: N0; no LN metastasis, N1; metastasis to one LN, and N2; metastasis to two or more LNs. Distant metastasis was stratified into two groups: M0, no distant or LN metastasis beyond the regional LN basin and M1, distant organ metastasis or LN metastasis beyond the regional LN basin: regional LN basin is defined as the lymph node where the first metastasis is frequently found, preceding the development of distant disease, such as the pelvic lymph node for EMPD in the genitoperineal region and the axillary lymph node for EMPD in the axilla.
In the present study, EMPD lesions were divided into two components: non-invasive components in the epidermis and invasive components in the dermis. In our cohort there were 78 lesions with only non-invasive components (76.5%), 10 lesions with non-invasive and invasive components (9.8%), and 14 lesions with only invasive components (13.7%). All patients with double or triple primary lesions showed no invasiveness in primary lesions and had no LN metastasis and no distant metastasis: the tumor site, tumor thickness and LVI were recorded for each lesion of patients who had double or triple primary lesions. The clinicopathological data for the 92 patients with 102 primary lesions are summarized in Table 1.
In addition, we obtained four LN metastatic lesions from four patients whose primary lesions were available, as well as one LN metastatic lesion and one locally recurrent lesion from one patient who had both LN metastasis and local recurrence. Although there were two patients with metastatic lesions of distant organs (brain and liver metastatic lesions, respectively) among the 92, we collected no metastatic lesions of distant organs because both patients had received adjuvant chemotherapy before resection of the metastatic lesions. Finally, in addition to the 102 primary lesions, we also enrolled five LN metastatic lesions and one locally recurrent lesion.
2.2. Immunohistochemistry
For primary lesions with and without an invasive component, formalin-fixed, paraffin-embedded blocks representing the deepest invasion and those representing the maximum tumor diameter were selected for analysis, respectively. For immunohistochemistry, each block was cut into 4-µm-thick sections. The primary antibody to AR (Clone SP107) was obtained from Roche Diagnostics (Penzberg, Germany). Although the antibody was ready-to-use, we applied it at a dilution of 1:2 after optimization using prostate cancer tissue sections showing AR expression. Immunohistochemical analysis was performed using a Leica Bond Max automated immunostainer (Leica Biosystems, Bannockburn, IL, USA). High-pH epitope retrieval buffer, EDTA-based pH 9.0 ER2 (Leica Biosystems; Catalog No.AR9640), was used for 20 min prior to application of the primary antibody (30 min). Detection was performed using a Leica Bond-detection kit with diaminobenzidine as the chromogen, followed by light hematoxylin counterstaining. Prostatic cancer tissue sections showing AR expression were used as positive controls. Only nuclear staining was considered positive. The results of immunohistochemistry were evaluated by a semiquantitative system using both staining intensity score (IS) and positive-tumor cell percentage score (PS), similar to the Allred score for evaluation of estrogen receptor positivity [14]. IS was defined as zero (no stain), one (weak), two (moderate), and three (strong) (Fig. 1). The percentage of positive cells was evaluated as the average for 10 random high-power fields. In lesions consisting of both non-invasive and invasive components, to appropriately select 10 random high-power fields, we first measured the areas of the non-invasive and invasive components in each lesion and calculated the ratio of the two areas. We then divided the 10 high-power fields into two groups representing the counts for the non-invasive and invasive components, respectively, based on the areal ratio of the two components. The PS was defined as zero, one (less than 10%), two (10% or more and less than 50%), three (50% or more and less than 75%), and four (75% or more) [15]. The total score (TS) was then calculated as the average of the product of IS and PS in the 10 random high-power fields. TS (IS × PS) ranged from 0 to 12, with only nine possible values (i.e., 0, 1, 2, 3, 4, 6, 8, 9, and 12). We then defined three subgroups: Grade 1 (TS 0 and 1), Grade 2 (TS 2–4) and Grade 3 (TS 6, 8, 9, and 12).
2.3. Statistical analysis
Pearson’s chi-squared test for categorical variables was used to evaluate the association between AR expression and clinicopathological parameters. Differences at P < 0.05 were considered statistically significant. All statistical analyses were performed using the statistical program RStudio (Rstudio Inc., Boston, MA, USA) and the R software package (R Foundation for Statistical Computing, https://www.r-project.org).
3. Results
3.1. Clinicopathological characteristics
The patients included 64 men and 28 women, with a mean age of 72.75 years (range 33–92 years). Tumors were localized predominantly in the genital area (71.6%), the perianal area (0.9%), axillae (11.8%), genital to perianal areas (spreading over two areas) (11.8%), and other areas (3.9%). In 17 (16.7%), 26 (25.5%), and 34 (33.3%) cases, the largest diameter of the primary lesion was less than 5 cm, 5 cm or more and less than 10 cm, and 10 cm or more, respectively. LVI was detected for eight (7.8%) lesions. With respect to tumor thickness, 78 (76.5%), eight (7.8%), and 14 (13.7%) lesions were categorized under T0, T1, and T2, respectively. Twelve (11.8%) lesions involved LN metastasis; one (1.0%) and 11 (10.8%) lesions were categorized as N1 and N2, respectively. Two (2.0%) lesions involved distant metastases. With regard to the EMPD stage classification, tumor in situ and tumor invasion were observed for 78 (76.5%) and 23 lesions, respectively; of the 23, seven (6.9%), four (3.9%), one (0.9%), nine (8.8%), and two (2.0%) lesions were categorized as stage I, II, IIIA, IIIB and IV, respectively. These clinicopathological data are summarized in Table 1.
3.2. Immunohistochemical determination of AR expression in primary EMPD
Among the 102 lesions, immunoreactivity was not observed in four (IS 0) (3.9%) and AR-positive Paget cells were observed in 98 (96.1%). Among the 98 lesions, 45 (44.1%), 39 (38.2%) and 14 (13.7%) showed weak (IS 1), moderate (IS 2), and strong (IS 3) staining, respectively. Thirteen (12.7%), 34 (33.3%), 17 (16.7%) and 34 (33.3%) lesions showed AR-positive tumor cell percentages of PS 1, PS 2, PS 3 and PS 4, respectively. Accordingly, there were 16 Grade 1 (TS 0 and 1) lesions (15.7%), 44 Grade 2 (TS 2–4) lesions (43.1%), and 42 Grade 3 lesions (41.2%). The results of immunohistochemistry are summarized in Table 2. There were no differences in morphology, i.e. cellular atypia, among the Grade 1, Grade 2, and Grade 3 Paget cells.
3.3. Correlation between AR expression and clinicopathological characteristics
AR expression grade was significantly correlated with tumor thickness (P = 0.011), LN metastasis (P = 0.008), and staging classification (P = 0.023): Grade 3 EMPDs exhibited greater tumor thickness, a higher number of LNs with metastases, and a more advanced stage. Correlations between AR expression and clinicopathological parameters are summarized in Table 3.
3.4. Comparison of AR expression between non-invasive and invasive components
In the 10 lesions containing both non-invasive and invasive components, to clarify the difference in AR expression between the two components, AR grading was re-evaluated separately for each component using TS for the high-magnification fields assigned to the non-invasive component and such fields assigned to the invasive component, respectively. All 20 non-invasive or invasive components from the 10 lesions showed Grade 2 or 3 expression, and even the non-invasive component showed no Grade 1 expression among the 10 lesions. Pearson’s chi-squared test revealed that AR expression grades were significantly higher in invasive components (n = 10) than in non-invasive components (n = 10) (P = 0.023; Table 4).
Furthermore, we compared the two components within an individual lesion. The non-invasive components frequently showed focal and patchy (heterogeneous) staining, whereas the corresponding invasive components showed diffuse (homogeneous) staining. Paget cells showing Grade 2 or 3 expression within the non-invasive component were found to be contiguous with the invasive component (Fig. 2), indicating the possibility that Grade 2 or 3 Paget cells are able to invade into the dermis. In six lesions (60%), AR expression grade was higher in the invasive component than in the corresponding non-invasive component (Fig. 3A). Focusing on both expression grade and staining pattern, AR expression in the invasive component was found to be similar or higher than that in the corresponding non-invasive component in all 10 lesions.
3.5. Comparison of AR expression between primary lesions and metastatic and/or recurrent lesions
We further focused on four patients with LN metastasis and one patient with both LN metastasis and local recurrence. In LN metastatic lesions and locally recurrent lesions, the AR expression grade was also calculated based on TS for 10 random high-power fields. Pearson’s chi-squared test revealed that AR expression grades were significantly higher in metastatic and/or recurrent lesions (n = 6) than in primary lesions (n = 5) (P = 0.014; Table 5): there was a significant difference in AR expression between the metastatic and/or recurrent lesions and the non-invasive component of the primary lesions (P = 0.006), whereas there was no significant difference between the metastatic and/or recurrent lesions and the invasive component (P = 0.251) (Supplementary Table S1). In all five patients, primary lesions contained both non-invasive and invasive components. Then we compared the non-invasive components (n = 5), invasive components (n = 5) and then the metastatic and/or recurrent lesions (n = 6). In three of the five (60%) patients, AR expression grades were higher in the invasive component than in the non-invasive component, and such higher grades were inherited by metastatic and/or recurrent lesions (Fig. 3B). In Case 27, although AR expression grades did not differ between the non-invasive and invasive components, the grade in the LN metastatic lesion was higher relative to the primary lesion (Fig. 3B). In Case 38, the non-invasive component already showed Grade 3 expression, and similar findings were obtained in the invasive component and metastatic lesion (Fig. 3B).
4. Discussion and conclusions
In the present study, we investigated the correlations between the immunohistochemical expression of AR and various clinicopathological factors to clarify the clinical significanceof AR expression in the development and progression of EMPD. Although frequent AR expression has been reported in EMPD (54–90%) [5], [6], [7], [8], [9], previous studies simply evaluated the expression of AR immunohistochemically as positive or negative. Unlike such previous reports, in the present study we evaluated AR expression in detail on the basis of staining intensity and the percentage of AR- positive tumor cells in order to investigate the relationship of these parameters to clinicopathological characteristics. To this end, we divided the evaluated EMPDs into Grade 1, Grade 2, and Grade 3 groups. Moreover, as the study cohorts in previous studies had been small, probably due to the rarity of EMPD [5], [6], [7], [8], we collected 92 EMPD patients with 108 EMPD lesions, which we believe represents the largest such series of EMPD cases subjected to immunohistochemical and clinicopathological analyses hitherto.
In the present study, we found a high frequency of AR immunopositivity (TS 1–12) in primary lesions (96.1%). To our knowledge, this study is the first to have demonstrated significant correlations between AR expression and tumor thickness, LN metastasis, and staging classification. Moreover, it is the first study to have demonstrated significantly higher AR expression in the invasive component of EMPD than in the non-invasive component, as well as significantly higher AR expression in metastatic and/or recurrent lesions than in primary lesions. Although the difference was not statistically significant, LVI was observed more frequently in Grade 3 lesions (11.6%) than in Grade 1 (0%) and Grade 2 (7.0%) lesions. These data clearly showed that increased immunohistochemical expression of AR was associated with malignant progression of EMPD.
Regarding treatment for unresectable or metastatic cases, wide local excision has long been the standard treatment for EMPD. Although several chemotherapeutic regimens have been proposed and docetaxel monotherapy and low-dose 5-fluorouracil/cisplatin therapy have been frequently used, no standardized chemotherapy has yet been established, possibly because of the rarity of EMPD [10]. Given the overexpression of human epidermal growth factor receptor 2 (HER2) in the tumors of some patients with advanced or metastatic EMPD, it was suggested that the anti-HER2 antibody trastuzumab, when combined with platinum- or taxane-based chemotherapy, is therapeutically beneficial [16]. On the other hand, hormonal therapeutic strategies have been established for hormone receptor-positive cancers such as prostate and breast cancers. Recently, a Phase II trial of combined androgen blockade for AR-positive salivary duct carcinoma, which is a rare malignant tumor like EMPD, was conducted in Japan, and the results suggested that – in comparison to conventional chemotherapy – this approach had equivalent efficacy and less toxicity in patients with AR-positive recurrent/metastatic or unresectable locally advanced salivary gland carcinoma [17]. In the context of EMPD with high AR expression, a previous experimental study has shown that an established transplantable tumor from skin metastasis of EMPD expressing ER and AR in nude mice could be stimulated by injection of testosterone, dihydrotestosterone, diethylstilbestrol, and 17β-estradiol, suggesting that it was hormone-dependent and capable of estrogen and androgen signaling [18]. In the context of clinical data, androgen blockade therapy has been reported to reduce multiple bone metastases in one EMPD patient [11], and another study has shown that two types of agent for hormonal therapy – tamoxifen targeting estrogen and bicalutamide targeting androgen – were beneficial in a case of metastatic EMPD with ER and AR expression [19]. However, no clinical trial of androgen blockade therapy for EMPD has been conducted worldwide because the clinical significance of AR expression has not been well characterized in a large number of clinical cases. Since the present results suggest that AR plays a role in disease progression, androgen blockade therapy would appear to have therapeutic potential for EMPD with AR expression.
In the present study, Grade 1 AR expression was not observed in any of the 20 components of 10 lesions comprising both non-invasive and invasive components, indicating that any component likely to become invasive, or a component that is already invasive, would show Grade 2 or 3 expression. AR-positive Paget cells in the dermis were observed just below the non-invasive component showing Grade 2 or 3 expression (Fig. 2), suggesting that only Grade 2 or 3 Paget cells in the epidermis are able to invade into the dermis. Grade 1 expression was not observed in any of the 15 components or lesions from 5 patients with metastatic and/or recurrent lesions, suggesting that only Grade 2 or 3 Paget cells are capable of metastasis or recurrence.
In conclusion, the present data suggest that Grade 2 or 3 AR positivity may be a predictor of invasion, metastasis and recurrence in EMPD. Immunohistochemistry for AR in biopsy and surgically resected specimens may have potential for prediction of metastasis and recurrence through close follow-up based on prognostication.
Funding
This study was supported by JSPS KAKENHI Grant Number 21K16236.
CRediT authorship contribution statement
Junko Kuramoto: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review & editing. Kenta Kobayashi:Investigation, Formal analysis, Methodology, Writing – review & editing. Ikuko Hirai:Investigation, Methodology, Writing – review & editing. Yoshio Nakamura：Conceptualization, Formal analysis, Funding acquisition, Methodology, Project administration, Writing – review & editing. Takeru Funakoshi：Conceptualization, Formal analysis, Methodology, Writing – review & editing. Yae Kanai：Conceptualization, Methodology, Writing – review & editing.