Clear cells of Toker in accessory nipples


James E. Fitzpatrick MD, Department of Dermatology, Campus Box F703, 4200 East Ninth Avenue, Denver, CO 80262, USA
Tel: (720) 848-0585
Fax: (720) 848-0535


Background:  Clear cells of Toker are intraepithelial cells with clear to pale staining cytoplasm and bland cytologic features found in approximately 10% of normal nipples. Toker cells have been hypothesized as a precursor of extramammary Paget's disease (EMPD), although the distribution of Toker cells outside of the nipples has not been studied. Using immunohistochemistry, we studied 20 cases of accessory nipples for the presence of Toker cells.

Methods:  A retrospective study of 20 cases of accessory nipples was performed using routine hemotoxylin and eosin staining, as well as immunohistochemical staining for CK7, CK20, EMA, and GCDFP-15.

Results:  Thirteen out of 20 accessory nipples (65%) demonstrated Toker cells with CK7 staining. Toker cells in six of the 13 cases were also positive for EMA. Only one case with Toker cells showed immunoreactivity for antibodies to GCDFP-15.

Conclusions:  Toker cells occur outside the normal nipple epidermis in the epidermis of accessory nipples. The distribution of Toker cells along the milk line correlates with the distribution of most cases of EMPD along the milk line, especially in the groin and axillae. Further studies are necessary to define the relationship between Toker cells and EMPD.

Clear cells of Toker are intraepithelial cells with clear to pale staining cytoplasm and bland cytologic features found in approximately 10% of normal nipples using hematoxylin and eosin (H&E) staining.1 Toker cells are usually present along the basal layer of the epidermis or in the mid-epidermis. Occasionally aggregates of Toker cells form small tubular structures. Immunohistochemical stains for low-molecular-weight cytokeratins appear to be very sensitive in detecting Toker cells, and Toker cells have been demonstrated in up to 83% of normal nipples at autopsy using immunohistochemistry.2 Toker cells have been hypothesized as a precursor of extramammary Paget's disease (EMPD),3,4 although the distribution of Toker cells outside of the nipples has not been studied.

EMPD is histologically characterized by atypical pagetoid cells with pale cytoplasm arranged singly and as small nests throughout the epidermis. In some cases involving the perineum or perianal area, these cells may represent intraepidermal spread of an underlying rectal,5 cervical,6 transitional cell,2 prostatic,7 or apocrine gland carcinoma;8 however, a majority of cases are not associated with an underlying malignancy.6 The cell of origin of EMPD without an underlying carcinoma is unknown. Proposed progenitor cells include apocrine gland cells, keratinocytes, pluripotent germinative epithelial cells, and eccrine gland cells.9

Clinically EMPD is most often found in the perineum or anogenital region and is characterized by red patches or plaques with variable scale, erosion, and ulceration.6,10 The axillae are a less frequent site of involvement of EMPD, and cases have also been reported on the anterior chest, eyelid, ear, and in supernumerary nipples. A large series of EMPD cases in Japan showed 94% in the genital area, 4% in the perianal area, and 2% in the axillae.11 Cases of Paget's disease of the breast in which an underlying ductal carcinoma or carcinoma-in-situ was not identified have also been described and designated as EMPD of the breast.12 A salient clinical feature of EMPD is the distribution of most cases along the milk line, with a majority of cases occurring in the groin and axillae.

In an attempt to identify clear cells of Toker in sites other than normal nipples and correlate the distribution of Toker cells with the milk line and the distribution of EMPD, we studied 20 cases of accessory nipples for the presence of Toker cells.

Materials and methods

The surgical pathology files of the University of Colorado Health Sciences Center (UCHSC) and consultation cases of two of the authors (LEG and  JEF) were searched for cases of accessory or supernumerary nipples. After histologic review and confirmation of the diagnosis, 20 cases were identified. Pathology reports were reviewed for the locations of the lesions and age and sex of the patients. Immunohistochemical stains were performed on 5 µm paraffin-embedded sections using an automated Ventana ES immunohistochemistry system (Ventana Medical Systems, Tucson, AZ). Slides were heated at 60°C for 1 h after sectioning. The primary antibodies used and antigen retrieval employed are listed in Table 1. The primary antibody incubation was 32 min at 42°C. The Ventana diaminobenzidine detection system was used. Sections were counterstained with hematoxylin. Tissues that were previously shown to demonstrate immunoreactivity with the different antibodies were used as positive controls.

Table 1.  Primary antibodies and antigen retrieval
AntibodyManufacturer*CloneDilutionAntigen retrieval
  • *  

    DAKO Corporation, Carpinteria, CA; Signet Laboratories, Dedham, MA; Ventana Medical Systems, Tucson, AZ.

  • †  

    Protease 1, proprietary endopeptidase (alkaline protease) of the serine protease family.

CK7DAKOOV-TL 12/301:800Microwave 10 min in pH 6.0 citrate buffer,
    Protease 1 (Ventana) digestion 4 min
CK20DAKOKs 20.81:800Microwave 10 min in pH 6.0 citrate buffer,
    Protease 1 (Ventana) digestion 4 min
GCDFP-15SignetBRST-21:50Decloaking chamber 5 min in pH 6.0
    citrate buffer
EMADAKOE291:100Protease 1 (Ventana) digestion 4 min


Twenty cases of supernumerary nipples were examined from 19 patients. Nineteen of 20 of the cases examined were from women and one was from a man. Two lesions were from the axilla (2/20), six were from the chest (6/20), three were from the breast (3/20), eight were from the abdomen (8/20), and one was designated as from the trunk (1/20). In 18 cases, the laterality of the lesion was given. Eleven cases were from the left side (11/18), and seven were from the right side (7/18). All of the examined cases showed a combination of a rugose, papillomatous epidermis with underlying sebaceous glands and disordered fascicles of smooth muscle. Sixteen out of 20 (16/20) cases also showed ductal structures in the mid- to deep reticular dermis.

Immunohistochemical staining for CK7 demonstrated Toker cells in 13 of 20 cases (65%). The number of Toker cells seen in an individual case ranged between two and 122 in a single section. The Toker cells were arranged singly or in small clusters in the lower half of the epidermis (Fig. 1). Most often the Toker cells were concentrated in the epidermis around the opening of an underlying duct or sebaceous gland. In the CK7-immunostained sections, some Toker cells showed a tadpole-like morphology. The intensity of CK7 staining in Toker cells was similar to that seen in the superficial ductal epithelial cells in cases which demonstrated both Toker cells and ductal structures in the dermis. Of the 16 cases that had ductal structures in the dermis, 10 had Toker cells. Three of four cases without ductal structures identified in the dermis had Toker cells.

Figure 1.

A) Epidermis of supernumerary nipple with scattered Toker cells showing paler cytoplasm and round bland nuclei. B) Sequential section stained for cytokeratin 7 showing positivity in Toker cells, some of which have a tadpole morphology.

On H&E staining, with correlation of CK7-stained sequential sections, several cases showed large, single, intraepidermal clear cells with round nuclei and clear to pale staining cytoplasm, consistent with Toker cells. These cells did not show significant cytologic atypia. The Toker cells were predominantly in the lower half of the epidermis, most often above the basal layer. In one case the cells formed a small tubule- or gland-like structure (Fig. 2). In some areas the cells that stained with CK7 were on sequentially stained H&E sections indistinguishable from normal keratinocytes.

Figure 2.

A) Cluster of pale staining Toker cells forming duct-like structure (→). B) Sequential section stained for cytokeratin 7 showing positivity in Toker cells.

CK20 immunoreactivity was not observed in Toker cells. Occasional single CK20-positive Merkel cells were identified in the basal layer of the epidermis.

Of 13 cases with CK7-positive Toker cells, six also showed immunoreactivity for antibodies to EMA in the Toker cells (Fig. 3). In one case EMA staining was strong, while the remainder showed only faint staining. Only one case showed Toker cells with immunoreactivity to GCDFP-15, and this case showed only a single clearly positive cell. A sequential section showed CK7 staining in the same cell.

Figure 3.

A) Epidermis of supernumerary nipple and underlying ductal structure. B) Sequential section stained for cytokeratin 7 showing positivity in Toker cells and ductal epithelium. C) Sequential section stained for epithelial membrane antigen (EMA) with positivity in Toker cells and superficial ductal epithelium.


Clear cells of the nipple epidermis were first identified in 1970 by Toker in approximately 10% of histologically normal nipples.1 These cells have bland cytologic features, with round nuclei, pale to clear cytoplasm, and scant chromatin. Toker cells are thought to be intraepidermal glandular epithelial cells, which may represent precursor, or primordial glandular epithelial cells in the epidermis.13 Toker speculated that in some cases Toker cells could give rise to Paget's disease by conversion into intraepidermal ductal carcinoma.1

Accessory, or supernumerary, nipples are congenital anomalies which are estimated to occur in 1–2% of the general population.14 Most commonly, accessory nipples develop along the milk lines, which arise from thickened ridges of ectoderm on the ventral sides of the embryo. Normal breast tissue arises in the milk lines over the chest, while the remainder of the milk lines regress. Accessory breast tissue or nipples result when part of the milk lines fails to regress.15 Much less commonly, accessory nipples or breast tissue may arise outside the milk line. Accessory breast tissue is subject to the same range of pathologic processes as normal breast tissue, including carcinoma and Paget's disease.4,16 However, accessory nipples have not been thoroughly studied for the presence of Toker cells.

Clear cells of Toker have generated interest as a possible precursor cell of primary intraepidermal EMPD since morphologically they appear to be a benign counterpart of the malignant cells of Paget's disease. The occurrence of Paget's disease in the breast without an underlying carcinoma12 and in supernumerary nipples4,17,18 which may not contain breast tissue also gives support to the concept of an intraepidermal precursor such as a Toker cell leading to Paget's disease. Additionally, one case of EMPD arising in the breast was associated with multifocal hyperplasia of Toker cells, suggesting generalized hyperplasia of Toker cells as a precursor of Paget's disease.19

Immunophenotypically, Toker cells have been shown to have a similar profile to ductal epithelial cells of the breast, expressing CK7, AE1, and variable staining with EMA and GCDFP-15.2,19 Staining with antibodies to CK7 has been shown to be very sensitive in detecting Toker cells, demonstrating them in up to 83% of normal nipples.2

Our findings of Toker cells, demonstrated by CK7 staining, in supernumerary nipples expands the known distribution of Toker cells and gives further support to Toker cells as a possible cell of origin for EMPD. The demonstration of Toker cells occurring anywhere along the milk line is in concordance with the distribution of primary intraepidermal EMPD in the genital area, axillae, and in supernumerary nipples. Since most cases of EMPD occur in the groin and genital areas, further study to identify Toker cells in these areas would be informative. Preliminary unpublished data from our laboratory have demonstrated Toker cells along the full length of the milk line in human fetal tissue (Fitzpatrick JE, unpublished data).

Additionally, the cells of primary intraepidermal EMPD show a similar cytokeratin immunophenotype to Toker cells, being CK7-positive, CK20-negative in most cases.2,5,20 In the cases we describe, EMA positivity was predominantly faint and seen in 6/11 cases with Toker cells. Other reports have shown EMA positivity in 3/3 and 1/1 examples of normal or hyperplastic Toker cells.2,19 In our series, only a single Toker cell in one case was immunoreactive for GCDFP-15. GCDFP-15 has not been described in Toker cells. The case describing hyperplastic Toker cells showed positivity for HMFG-2 in the Toker cells,19 but GCDFP-15 was not specifically studied.

Immunohistochemical studies in EMPD have generally shown positivity for EMA and GCDFP-15. In several studies, GCDFP-15 reactivity was seen in 12 of 16 (75%),2 16 of 20 (80%),21 11 of 12 (92%),20 and three of five (60%)5 cases of EMPD limited to the epidermis. Strong positivity for EMA was seen in 13 cases of EMPD from the vulva and anal region.22 Reported cases of EMPD arising in the milk line on the chest23 and in supernumerary nipples4,18 have not described GCDFP-15 staining, and only one of these reports studied EMA expression, which was positive. The case of mammary Paget's disease with multifocal Toker cell hyperplasia19 was EMA-positive, both in the Toker cells and in the Paget's cells, but GCDFP-15 staining was not performed.

Methodological differences in antibody clones and concentrations, antigen retrieval, and immunostaining protocols could account for some differences in immunophenotypes, since these parameters may vary between laboratories. However, as seen in our series, there seems to be variability in the Toker cell phenotype for EMA with generally faint staining in positive cases. Except for a rare cell, Toker cells in our series were GCDFP-15-negative. This difference in immunophenotype between Toker cells and the cells of EMPD may represent different expression of antigens by benign and malignant counterparts. Alternatively, if Toker cells do represent a primitive precursor cell, they may not express antigens which are present on more differentiated ductal structures or their malignant counterparts.

Toker cells may also be related to the intraepidermal cells seen in lesions of clear cell papulosis. Clear cell papulosis is a rare entity characterized by hypopigmented papules and macules on the abdomen along the milk line in young children.24 Histologically these lesions show features similar to Toker cells, with plump, clear to pale staining cells with round nuclei in the epidermis along the basal layer and occasionally in the higher layers. The clear cells show immunoreactivity to AE1, with less reactivity to CEA, EMA, and GCDFP-15.25,26 It has been speculated that clear cell papulosis may be a benign counterpart of EMPD.27 Similar to the Toker cells in our report and the lesions of EMPD, the clear cells in clear cell papulosis are also distributed along the milk line.

In summary, Toker cells occur outside the normal nipple epidermis in the epidermis of supernumerary nipples, as demonstrated by CK7 staining. Toker cells may be present in lesions along the milk line, in a distribution similar to EMPD. Further studies to correlate the relationship of Toker cells and primary intraepidermal EMPD are necessary to demonstrate the relationship between these benign and malignant entities.