describing this phenomenon, most of which rely on standard hematoxylin and eosin for diagnosis and margin evaluation [17].

With immunostaining using CK7, EMPD has been found to have significant involvement undetectable on routine hematoxylin and eosin staining. Precise margin control of the tumor with Mohs surgery using CK7 has elucidated the digitate, often haphazard growth pattern that EMPD can demonstrate [17]. With standard excision and even Mohs surgery without the use of CK7 immunostaining, these haphazard extensions may be missed, giving the impression of multicentric disease [18]. Authors have recently questioned the likelihood of truly multicentric EMPD [17]. Other theories raise the possibility of multiple foci of spontaneous tumor regression, leaving behind islands of normal skin that lead to a false negative margin on excision of EMPD [18].

are larger than keratinocytes and contain pleomorphic nuclei with pronounced atypia, prominent nucleoli, and mitotic figures [5, 19]. Paget cells typically have centrally located nuclei and abundant, finely granular cytoplasm [5]. Signet ring cells are not an uncommon finding. Reactive epidermal changes can be seen, including parakeratosis, hyperkeratosis, epidermal hyperplasia, and acantholysis. Most Paget cells are concentrated in the lower portion of the epidermis, in association with the pilosebaceous unit, but upward spread is not uncommon [5]. Paget cells may also be seen extending into sweat ducts, leading to uncertainty as to whether the neoplasm originates intraepidermally or from underlying cutaneous adnexal adenocarcinoma. Chronic inflammation with small capillary proliferation is also a common finding [11]. Over 90% of cases of EMPD demonstrate cytoplasmic mucin, staining positively with mucicarmine and PAS reagent.

Summary: Diagnosing EMPD/Disease

Pathophysiology

•Biopsy should be performed for any genital or extragenital dermatosis that fails to respond to the typical treatment for other possible diagnoses.

•Histology demonstrates Paget cells, which typically have centrally located nuclei; abundant, finely granular cytoplasm; and are usually located in the lower epidermis.

•Coexistent malignancy should be ruled out by performing a complete history and physical exam, review of systems, and appropriate diagnostic testing.

22.5Diagnosing EMPD/Disease Pathophysiology

22.5.1 Histology

A skin biopsy should be performed for any genital or extragenital lesion(s) that are suspicious for EMPD or fail to respond to a reasonable course of the typical treatments for other possible diagnoses. The diagnosis of EMPD must be confirmed by histology, and overly superficial biopsies should be avoided in order to prevent diagnostic confusion. Typical findings include intraepidermal aggregates of pale-staining cells that

22.5.2 Histologic Differential Diagnosis

Pagetoid spread, the upward migration of cells in the epidermis singly or in clusters, can also be seen in Bowen’s disease, superficial spreading melanoma, mycosis fungoides, Langerhans cell histiocytosis, sebaceous carcinoma, and Spitz nevus [5, 14]. Longstanding Bowen’s disease can demonstrate significant hyperkeratosis, elongation of the rete ridges, and reactive keratinocyte atypia. EMPD can occasionally be difficult to distinguish from Bowen’s disease, but mucin, signet cells, and glandular structures are useful features to distinguish EMPD [8]. Melanoma demonstrates a greater degree of nesting at the dermoepidermal junction, there is no acinar formation, and mucin is absent [7].

Immunohistochemical analysis can be helpful in difficult cases, using a panel of CK7, CEA, EMA, S100, HMB-45, and Cam5.2. One series reports that in 25% cases of EMPD, immunohistochemistry for lowmolecular weight cytokeratins, CEA, and S-100 protein was required to exclude Bowen’s disease and superficial spreading melanoma from the differential diagnosis [7].

22.5.3 Evaluation for Internal Malignancy

After confirmation of the diagnosis of EMPD by histology, a workup to exclude underlying gastrointestinal,

genitourinary, and other less commonly associated distant malignancies is prudent. A comprehensive review of systems should be performed with this in mind, and a complete physical exam should be carried out, including genital and rectal exams, and a thorough skin exam to assess the degree of involvement. Careful palpation of lymph nodes is mandatory. Bilateral “underpants pattern” erythema has been associated with pronounced lymphovascular invasion and should alert the clinician to the likelihood of nodal disease [20].

In cases where extent of the tumor is questionable, multiple biopsies may be helpful to establish a more accurate estimate of the margins. If the anal area is involved, examination for anorectal or colon cancer could include fecal occult blood testing as well colonoscopy. Urine cytology and cystoscopy should be considered to screen for uroepithelial cancers. Women should also undergo breast and pelvic examination, Papanicolaou smear, as well as mammograms and pelvic imaging studies. Men should be screened for underlying malignancy through physical examination of the prostate and testicles and by prostate-specific antigen assay.

CEA has been used to screen for systemic disease in patients with EMPD [20]. Its sensitivity for detecting EMPD metastasis has been reported as 70% and specificity 93.8%. The marker has demonstrated utility in predicting prognosis, with elevated levels associated with higher likelihood of death from EMPD. Although the sensitivity of CEA is lower than ideal for a screening test, serum CEA level may be useful as a marker of systemic metastasis and response to treatment [20].

22.5.4 Sentinel Lymph Node Biopsy

The value of sentinel lymph node biopsy in EMPD is controversial. No large studies exist to provide an evi- dence-based approach. Results from a small series of patients with primary genital EMPD who underwent sentinel node biopsy found positive nodes in 4 of 23 patients, and 3 of the 4 patients with positive nodes developed distant disease [21]. None of the patients with a negative sentinel lymph node biopsy developed distant disease. None of the patients with intraepidermal disease had a positive lymph node biopsy. Complete node dissection was not carried out in all

patients, and no conclusion could be reached about any potential benefit of complete dissection. Further studies are needed to clarify the role of sentinel lymph node biopsy in EMPD.

22.5.5 Pathophysiology

There has been much debate about the pathophysiology of EMPD. Comparisons between mammary Paget disease and EMPD have been made in an effort to better understand the pathogenic mechanism. Mammary Paget disease is most often associated with underlying in situ or invasive ductal carcinoma of the breast, leading to the hypothesis that epidermotropic spread of neoplastic cells is responsible for cutaneous disease. The rare cases of mammary Paget disease without underlying breast carcinoma are referred to as primary Paget disease, and in these cases, the epidermis is felt to be the site of origin.

EMPD, on the other hand, is much less likely to be associated with an underlying neoplasm. Accordingly, it is believed that most cases of EMPD arise in the epidermis and have no association with an underlying cutaneous adnexal adenocarcinoma or internal malignancy.

Ackerman and colleagues reviewed the histology and clinical history of a series of cases of EMPD and concluded that EMPD is likely more than one disease, with the most common form originating in the epidermis and the less common subtypes originating in underlying glandular structures or contiguous internal organs with epidermotropic spread [22]. EMPD is now believed to exist in four clinical scenarios: first, disease originating in and contained within the epidermis; second, disease originating in the epidermis with secondary invasion of the dermis; third, epidermal disease secondary to underlying adnexal adenocarcinoma with epidermotropic spread (a controversial entity due to the difficulty in assessing origin of the tumor); and fourth, epidermal disease secondary to internal malignancy. In more recent classification schemes, the first three scenarios are classified as primary EMPD, while the fourth is classified as secondary EMPD [23].

22.5.6 Cell of Origin

The cell of origin of EMPD remains controversial. Theoretical candidate cells include epidermal/follicular

stem cells, Toker cells, and apocrine gland cells. Recent immunohistochemical analysis has led to the theory that Toker cells may play a role as the benign precursor to the neoplastic cells, giving rise to at least some cases of vulvar EMPD [24]. Clear cells of Toker are thought to be germinative cells of the lactiferous duct, identifiable by routine hematoxylin & eosin staining in approximately 10% of normal nipples, and also present in accessory nipples along the milk line. They have also been identified using CK7 staining in mammary-like glands found in normal genital skin and are felt to be an interface between the epidermis and the epithelium of lactiferous ducts of the breast, ectopic breast tissue, and mammary-like glands of the vulva [24].

The finding of Toker cells in genital skin of females has led to the hypothesis that the pathogenesis of EMPD may be similar to that of mammary Paget’s disease. Following this line of thinking, EMPD may arise by three mechanisms: first as a primary, intraepithelial neoplasm originating in the epidermis from Toker cells, with the potential for invasive spread; second as a neoplastic process arising from carcinoma of anogenital mammary like glands; and third as carcinoma arising from skin appendages, Bartholin glands, or other organs like the anus, rectum, urethra, or cervix [25].

22.5.7Distinguishing Primary and Secondary EMPD by

Immunohistochemical Techniques

At present, distinguishing primary from secondary EMPD is possible only by extensive medical evaluation to exclude underlying malignancy. Given the significant difference in prognosis between patients with primary and secondary EMPD related to underlying malignancy, interest has been increasingly focused on utilizing immunohistochemistry to distinguish primary and secondary EMPD, with the goal of gaining a reproducible and rapid method of establishing the likelihood for the presence or absence of an underlying malignancy.

Multiple immunostains have been examined in EMPD, including CK7, CK20, HER-2/neu, BRST-2, CDX2, androgen receptor, and cyclin D1. Some studies have suggested that a CK7+, CK20−, and BRST-2+ immunohistochemical phenotype is consistent with

primary EMPD, while a CK7+, CK20+, and BRST-2− phenotype is consistent with secondary EMPD.

CK7 is a cytokeratin found in apocrine glands, sebaceous glands, secretory coils of eccrine glands, and in the epithelium of breast, lung, and genitourinary tract, but not the gastrointestinal tract. CK7 staining has been found to be near universally positive in cases of both primary and secondary EMPD. CK20 is a cytokeratin found in Merkel cells as well as most colorectal adenocarcinomas. BRST-2 is a monoclonal antibody that detects a glycoprotein expressed by apocrine glands, some eccrine glands, and minor salivary glands. Breast adenocarcinomas with apocrine features also express the BRST-2 antigen. HER-2/neu is a transmembrane growth factor receptor that is overexpressed in 25–30% of primary breast carcinomas, with high staining sensitivity for mammary Paget’s disease and variable staining for both primary and secondary EMPD [19]. CDX2, a transcription factor that regulates the development of intestinal epithelium, has shown value in differentiating primary EMPD from EMPD secondary to anorectal malignancy. In the majority of cases, CDX2 stains positive in EMPD secondary to anorectal malignancy and is negative in primary anogenital EMPD [19].

A recent evaluation of a series of EMPD cases examined using CK7, CK20, HER-2/neu, BRST-2, CDX2, androgen receptor (AR), and cyclin D1 found CK7+ staining in 100% of cases (both primary and secondary EMPD) [19]. CK20 was positive in 22% of primary EMPD and 50% of secondary EMPD, while BRST-2 showed 48% positivity in primary EMPD and 25% in secondary EMPD. The authors conclude that the panel consisting of CK7, CK20, and BRST-2 has limited utility in distinguishing primary and secondary EMPD due to the variable staining pattern of CK20 and BRST-2. They also conclude that adding HER-2/neu and CDX2 may help differentiate primary anogenital EMPD from anogenital EMPD associated with anorectal adenocarcinoma, but fails to distinguish primary EMPD from EMPD secondary to uroepithelial or prostatic malignancy. Ber-EP4 is another immunohistochemical marker that has shown promise in distinguishing primary and secondary EMPD, but more research must be done to establish its utility as well as the overall value of immunostains in the diagnosis of EMPD.