Background
Paget disease (PD) is a distinct intraepidermal adenocarcinoma with a pagetoid growth pattern. PDs are classified as mammary and extramammary subtypes according to their locations and their relationship to breast [
1,
2]. Mammary PDs account for 90% of the PDs occurring on the skin of nipple/areola complex and most of them represent tumor spread to the epidermis from an underlying invasive ductal carcinoma (53–60%) or ductal carcinoma in situ (24–43%). Compared to breast PD, primary extramammary PDs (EMPDs) are relatively uncommon and their histogenesis is less clear [
1,
2].
Primary EMPDS are found in areas rich in apocrine glands. The most common site of primary EMPDs is vulva followed by perianal skin, scrotum and penis, and axilla etc. [
1‐
6]. In women, more than 80% of primary EMPDs are in the vulva [
1‐
4,
6]. In men, approximately half of EMPDs are in the penoscrotal region [
4‐
8]. Most primary EMPDs are intraepithelial at their initial presentation (type Ia disease) but some have both intraepithelial disease and invasive adenocarcinoma i.e. invasive EMPDs [
1‐
11]. The invasive adenocarcinomas seen in primary EMPDs include those arising from intraepithelial EMPD (type Ib disease) and those giving rise to the intraepithelial disease (type Ic disease, underlying adenocarcinoma with subsequent epidermal involvement i.e. Paget disease as manifestation of an underlying adenocarcinoma) [
3]. Among patients with invasive EMPDs (type Ib and type Ic), 20% to 40% had lymph node metastasis [
4‐
7,
9,
11]. Up to 17% to 50% patients with invasive EMPDs also develop concurrent or subsequent distant metastasis [
4,
5,
7,
9‐
12].
Primary EMPDs should be distinguished from secondary EMPDs given their different treatment and prognosis [
3]. Secondary EMPD is usually the result of intraepithelial spread from a visceral carcinoma located elsewhere, with the gastrointestinal tract (colorectum) or urogenital tract (urinary bladder, prostate) being the most 2 common sources [
1‐
3,
9,
13‐
17]. EMPDs may also pose some diagnostic challenges in metastatic sites as they morphologically may mimic other tumors such as urothelial carcinoma and breast carcinoma. This diagnostic challenge is further complicated by the fact that patients with EMPDs have an increased risk of developing secondary primary tumors in which breast carcinoma, colorectal adenocarcinoma and urothelial carcinoma are among the most common ones [
4‐
6,
9,
14‐
17].
Given the overlapping morphologic features between primary EMPDs and secondary ones, and between metastatic EMPDs and their mimics in metastatic sites, immunohistochemical markers are often needed to facilitate the correct diagnosis. Several immunohistochemical markers, including cytokeratin 7, carcinoembryonic antigen, androgen receptor and c-erbB2 (HER2), have been used for diagnosing primary EMPDs, however, their specificity is relatively low [
18‐
20] and therefore limited their diagnostic utility in metastatic setting. Gross cystic duct fluid protein 15 (GCDFP15, also known as BRST-2) shows relatively high specificity for EMPDs but its sensitivity was only 60% to 85% and in many cases the staining was focal [
21‐
26]. Primary EMPD is analogous to breast Paget disease. Recently a transcription factor GATA-binding protein 3 (GATA3) has been identified as a very sensitive marker for breast carcinoma, both in both primary and metastatic sites [
27‐
31]. GATA3 was also reported to be highly expressed in apocrine glands and adnexal tumors [
30]. Apocrine gland has been proposed as the origin of primary EPMDs according to one theory [
1,
2,
21]. These findings suggest that GATA3 might be a sensitive marker for primary EMPDs. In the literature, there was only one recent report of GATA3 in 11 vulvar primary EMPDs [
32].
In this study, using immunohistochemical staining we investigated the expression of GATA3 in a large series of 72 primary EMPDs (45 with intraepithelial disease only, 26 with both intraepithelial disease and invasive adenocarcinoma including 14 also with lymph node metastasis, 1 with metastatic adenocarcinoma only for study) in male and female genital regions to explore the potential diagnostic utility of GATA3 in these tumors. We also compared GATA3 to GCDFP15 in these tumors for their sensitivity.
Discussion
In this study, we investigated the immunohistochemical expression of GATA3 in a large series of 72 primary EMPDs in male and female genital regions. We found that GATA3 was highly expressed in the primary genital EMPDs. The high sensitivity of GATA3 is not only present in the intraepithelial disease (100%) but also in the invasive adenocarcinomas (96%) and metastatic adenocarcinomas (93%). These results indicate that GATA3 is a very sensitive marker for primary EMPDs in the genital regions.
GATA3 is a zinc-finger transcription factor involved in embryogenesis, cell proliferation and differentiation in multiple human tissues and organs, including breast, genitourinary system, parathyroid, skin, central nervous and hematopoietic systems [
33‐
36]. In 2007, Higgins et al. found that GATA3 was a sensitive diagnostic marker for urothelial carcinoma [
37]. Since then, there has been growing evidence that GATA3 could serve as a relatively sensitive diagnostic marker for breast carcinomas, parathyroid tumors, trophoblastic tumors, mesonephric adenocarcinomas, paragangliomas and pheochromocytomas etc. [
28‐
31,
38‐
42]. Other tumors with a less frequent expression of GATA3 include salivary gland tumors, malignant mesotheliomas, pancreatic adenocarcinomas, skin squamous cell carcinomas, skin adnexal tumors, renal oncocytomas, chromophobe renal cell carcinomas, and yolk sac tumors [
28‐
30]. Morbeck D et al. recently reported positive GATA3 expression in all 11 vulvar primary EMPDs (4 with invasive carcinoma) [
32]. They did not include any metastatic adenocarcinoma from vulvar Paget disease. They did not study male genital EMPDs, either. Our findings and that of Morbeck et al. [
32] add primary genital EMPDs to the list of tumors with
high expression of GATA3. High expression of GATA3 in EMPDs has some diagnostic implications, both for primary EMPDs and their metastasis.
Distinguishing primary from secondary EMPDs is clinically critical given their different treatment and prognosis [
3,
9]. Secondary EMPD in the genital region is usually the result of intraepithelial spread from a visceral carcinoma, with urogenital tract (urothelial carcinoma, prostate) and the gastrointestinal tract (distal colon, rectum) being the most 2 common sources [
1‐
3,
6,
9,
12‐
16]. In females, secondary EMPD in vulva caused by urothelial carcinoma typically involves periurethral vulvar vestibule but it may extend to the adjacent vulvar skin and it may also become invasive [
1‐
3,
6,
9,
12‐
16]. In males, both urothelial carcinoma and prostate carcinoma may involve scrotum in an intraepithelial pagetoid fashion [
43‐
45]. Rarely urothelial carcinoma [
43,
46] and prostate adenocarcinoma [
43,
47,
48] may recur in the penis as a secondary EMPD. Since both primary EMPDs and urothelial carcinomas are positive for GATA3, GAT3 is not useful in distinguishing primary EMPDs from secondary EMPD caused by urothelial carcinoma and other markers should be sought for this purpose. Urothelial carcinomas are often positive for uroplakin-III, p63 and p40 whereas EMPDs have an opposite immunohistochemical profile [
49‐
51]. GCDFP15 is often positive in primary EMPDs [
20‐
26] but it is only rarely positive in urothelial carcinoma [
43,
52]. Secondary EMPD caused by prostatic adenocarcinoma can be distinguished from primary EMPD by p501S (prostein) and GATA3. Prostatic adenocarcinoma is positive for p501S but negative for GATA3 whereas EMPD shows an opposite profile [
28‐
30,
43,
53]. Prostatic adenocarcinoma can be rarely positive for GCDFP15 and primary EMPDs can show positive prostate specific antigen (PSA) staining in as many as 30% cases [
43,
53]. Therefore, one cannot rely on PSA or GCDFP15 to distinguish primary EMPD from secondary PD caused by prostatic adenocarcinoma. Secondary EMPDs from anorectal adenocarcinomas typically extend from perianal skin to the vulva or scrotum [
1‐
6,
9,
43,
44]. GATA3 is negative in colorectal adenocarcinoma [
28‐
30] and therefore is useful to distinguish primary genital EMPD from secondary EMPD due to colorectal adenocarcinoma. It should be pointed out that primary EMPDs can be rarely positive for CDX2 (3%) [
43]. GCDFP15 is negative in colorectal adenocarcinomas [
43]. Anorectal adenocarcinoma and primary vulvar EMPDs showed overlapping profiles in CK7 and CK20 though CK7 negativity favors the former and CK20 negativity favors the latter [
43]. In the genital area, rare pagetoid squamous cell carcinoma in situ can closely mimic intraepithelial EMPD [
54,
55] and may be misdiagnosed as such [
54]. Since some squamous cell carcinomas and normal epidermal cells are positive for GATA3 [
28‐
30], GATA3 is not useful to distinguish primary EMPD from pagetoid squamous cell carcinoma in situ. Instead P63 should be used in this scenario (p63 negative in primary EMPD but positive in pagetoid squamous cell carcinoma in situ) [
50,
51]. Lastly, melanoma in situ may closely mimic primary EMPD and rare pigmented primary EMPD has been reported [
56,
57]. Melanoma in situ was negative for GATA3 [
28‐
30] but positive for S100, melan-A and HMB45 whereas EMPD had an opposite immunoprofile.
Although most primary EMPDs are intraepithelial, approximately 4% to 20% primary vulvar EMPDs [
1‐
4,
9,
13,
15,
16] and 26% to 61% primary penoscrotal EMPDs were invasive at the time of presentation [
5,
7,
8,
10,
58‐
61]. Some of these invasive adenocarcinomas arise from the intraepithelial EMPD (type Ib primary EMPDs) whereas others are underlying adenocarcinomas which showed secondary epidermotropism (type Ic primary EMPDs) [
3,
6,
8,
9,
14‐
16]. In vulva, it is estimated that type Ic EMPDs account for at least 10–30% invasive EMPDs [
1‐
3,
8,
9,
15,
16,
62]. Rare type Ic primary EMPD in penoscrotum has also been reported [
63] and two of our cases belong to this category. Type Ic EMPDs were reported to be associated with a worse prognosis than type Ib EMPDs [
3,
9] and therefore pathologists should attempt to specify the subtypes of primary invasive EMPDs (Ib versus Ic). However, it is not always feasible to distinguish them. Our findings indicate that type Ib and apocrine type Ic diseases cannot be distinguished by their GATA3 and GCDFP15 immunoprofile given their similar profile for these two markers. Type Ic EMPDs are predominantly of apocrine type, but other types of adenocarcinomas may also rarely give rise to type Ic EMPDs including eccrine sweat gland adenocarcinoma [
64], Bartholin gland adenocarcinoma [
65], and adenocarcinomas of mammary-like glands [
66,
67] etc. One of the invasive adenocarcinomas in vulvar type Ic EMPDs in our study was an eccrine carcinoma. Cutaneous eccrine carcinomas were positive for GATA3 in 36% to 68% cases [
68,
69]. The only eccrine carcinoma in our study showed 4+ GATA3 staining (>75% cells). Thus, GATA3 immunostaining cannot distinguish type Ib EMPDs from apocrine and
eccrine type Ic primary EMPDs. Their distinction relies on morphology and other markers such as p63 and GCDFP15. Eccrine carcinomas were often positive p63 (85% to 89%) whereas primary type Ib EMPDs were not [
68,
69]. Eccrine carcinomas were only rarely positive for GCDFP15 (5%) [
69]. Adenocarcinoma of mammary-like gland in the vulva is rare and its diagnosis requires the presence of a transition zone between normal mammary-like glands and adenocarcinoma [
66,
67,
70]. Morphologically it is similar to breast carcinoma. Both ductal type [
66] and lobular-like [
67] mammary-like carcinomas with Paget’s disease (type Ic primary EMPD) have been reported. Although there has been no report of GATA3 in vulvar adenocarcinoma of mammary-like glands, it is conceivable that the vast majority of these tumors will be positive for GATA3 as in breast carcinoma. As expected, two thirds of vulvar mammary-like carcinomas were also positive for GCDFP15 [
70]. For these reasons, rare type IC primary EMPD due to mammary-like carcinoma cannot be distinguished from type Ib EMPD or type IC EMPD due to sweat gland adenocarcinoma by GATA3 and GCDFP15 immunostaining. Primary type Ic EMPDs caused by underlying apocrine carcinomas were often negative for ER and PR. In contrast, vulvar mammary-like carcinomas were often positive for these two markers [
66,
67,
70‐
72].
Among patients with invasive EMPDs (type Ib and Ic), some will develop metastatic disease at the time of presentation or in their subsequent disease courses. In the SEER data, 17.1% patients with invasive EMPDs have lymph node metastasis (male 16.0%, female 17.6%) and 2.5% have distant metastasis (male 3.8%, female 1.9%) at presentation [
4]. In a recent Japanese study of 301 primary invasive EMPDs (both male and female), 114 (37%) had metastasis including 20% node metastasis and 17% distant metastasis (16% with both nodal and distant metastasis) [
12]. Lymph nodes metastasis typically involved inguinofemoral nodes but pelvic and para-aortic nodes were also involved in some patients [
9‐
11,
22,
58‐
62]. Distant metastatic sites include bone, lung, liver, lung, brain and muscle [
9,
58‐
62]. Invasive EMPDs were morphologically similar to other types of tumors especially breast carcinoma and urothelial carcinoma, and therefore they may pose some diagnostic difficulty in metastasis, which can be further complicated by the fact that patients with primary EMPDs have an increased risk of developing other types of secondary primary tumors (overall 5–8% chance). Breast carcinoma and urothelial carcinoma are among the most common secondary tumors in these patients, and they can occur either before or after the diagnosis of primary EMPDs [
6,
9,
12,
14‐
17,
58‐
60,
62]. In patients with both a primary invasive EMPD and another type of tumor (particularly urothelial and breast carcinoma), the differential diagnosis for a metastatic tumor with positive GATA3 staining should also include metastatic primary EMPDs in the list of differential diagnosis. A panel of immunohistochemical markers should be used to facilitate the correct diagnosis.
GCDFP15 was a useful marker for primary EMPDs but its sensitivity was 60% to 85% [
20‐
26]. In this study, we showed that GATA3 is relatively more sensitive than GCDFP15 for primary EMPDs, especially in male patients. Our study is the largest series of primary EMPDs with GCDFP15 staining. It is interesting to note that GCDFP15 stains a higher percentage of primary EMPDs in female patients than male patients.
Although GATA3 is a sensitive marker for primary genital EMPDs, it should be pointed out that it is not specific for these tumors. As described above and reviewed elsewhere, several other types of tumors including urothelial carcinoma, breast carcinoma, paragangliomas/pheochromocytomas, trophoblastic tumors, and mesonephric adenocarcinomas are often positive for GATA3 [
28‐
31,
38‐
42]. In this sense, GATA3 is less specific than GCDFP15 for primary genital EMPDs. In difficult cases particularly in metastasis, both GATA3 and GCDFP15 should be used in conjunction to avoid misdiagnosis.
Lastly, high expression of GATA3 in primary EMPDs may also help shed some lights on the histogenesis of these tumors. Currently there are 3 theories: intraepidermal origin of adnexal origin such as apocrine glands, multipotent stem cells in the epidermis or infundibular stem cells from hair follicles [
1,
2,
20,
73]. Positive staining for both GATA3 and GCDFP15 in primary EMPDs probably favors the first theory.
One limitation of our study is that we did not include genital secondary EMPDs. Secondary EMPDs are rare and it is difficult to collect a meaningful number of cases to do a comparison study. The two most common types of carcinomas that cause secondary EMPDs are urothelial carcinoma and colorectal carcinoma [
1‐
3,
9,
13‐
17]. As discussed above, GATA3 immunoreactivity was seen in most urothelial carcinomas but not in colorectal carcinomas [
28‐
30,
37].