- Email: [email protected]
Expression of a glycoprotein of the carcinoembryonic antigen family in normal and neoplastic sebaceous glands
CLINICAL AND LABORATORY STUDIES
Expression of a glycoprotein of the carcinoembryonic antigen family in normal and neoplastic sebaceous glands Limited role of carcinoembryonic antigen as a sweat gland marker Dieter Metze, M D , a H.-Peter Soyer, MD, b Bernhard Zelger, MD, c Michael Neumaier, MD, e Fritz Grunert, PhD, f Christine Hartig, MD, d Udo Amann, MD, a Ranjit Bhardwaj, PhD, a Christoph Wagener, MD, e and Thomas A. Luger, MD a Graz and Innsbruck, Austria, and
Miinster, Hamburg, and Freiburg, Germany Background: Carcinoembryonic antigen (CEA) is a well-known marker for sweat gland differentiation in adnexal neoplasms. Objective: The aim of this study was to examine the expression of glycoproteins of the CEA family, that is, CEA-180, nonspecific cross-reacting antigens (NCAs), and bitiary glycoprotein (BGP), in sebaceous glands and in neoplasms with sebaceous differentiation. Methods: Normal adult and fetal skin, hyperplasias, hamartomas, and neoplasms with sebaceous or follicular differentiation were stained immunohistochemicaIly with a panel of polyclonal and monoclonal antibodies highly specific for CEA-180, NCAs, and BGP. Double immunostaining was performed to correlate the CEA expression with that of epithelial membrane antigen (EMA), a glycoprotein consistently found in differentiating sebocytes. Results: Whereas sweat glands coexpressed CEA, NCAs, BGP, and EMA, sebaceous glands were exclusively labeled with the antibodies recognizing BGP or EMA. Staining of the sebaceous glands was restricted to mature sebocytes, sparing immature cells. At the ultrastmctural level immunoreactivity for BGP and EMA was demonstrable in the golgi area, in small vesicles, and along the cell membranes. During fetal development BGP was not found until the sebaceous glands matured. The expression of BGP and EMA was highly conserved in reactive, hamartomatous, and neoplastic profiferations of adnexal structures with sebaceous differentiation. Conclusion: The expression of BGP, a CEA glycoprotein, in differentiating sebocytes accounts for the reactivity of many anti-CEA antibodies with sebaceous glands and thus disqualifies the CEA family as a monospecific marker for sweat gland differentiation. (J AM ACADDERMATOL1996;34:735-44.)
Carcinoembryonic antigen (CEA) comprises a group of closely related glycoproteins that have been detected in normal as well as neoplastic epithelia of From the Departments of Dermatology, University of Mtinster, a University of Graz,b University of Innsbmck,c and Medical Center Mindend; and from the Department of Clinical Chemistry,e University Hospital Eppendorf, Hamburg, and the Institute of Immunobiology, University of Freiburg. f Presented in part at the 14th Congress of the International Society of Dermatopathology, Sienna, Italy, June 26-July 1, 1993. Accepted for publication Nov. 16, 1995. Reprint requests: Dieter Metze, MD, Department of Dermatology, University Mttnster, Von-Esmarchstrasse 56, D-48149 Mtinster, Germany. Copyright © 1996 by the American Academy of Dermatology, Inc. 0190-9622/96 $5.00 + 0 16/1/70667
the gut, fiver, lung, breast, and other tissues.I, 2 The CEA glycoproteins were identified as products of a complex gene family, which encodes the classical 180 kd CEA, 160 kd bifiary glycoprotein (BGP), and nonspecific cross-reacting antigens (NCAs) of 95 kd (NCA-95) or 90 kd (NCA-90). 3' 4 In addition, granulocytes have been previously demonstrated to constitutively express NCAs and BGP. 3 The CEA glycoproteins and their splice variants differ in size, number of immunoglobulin domains, posttranslational modification, membrane anchorage, and glycosylation. 3 They are part of the immunoglobufin supergene family that includes immunoglobufins, T-cell receptors, growth factor receptors, and cell adhesion molecules. 5 Members of this family are 735
7 3 6
Journal of the American Academy of Dermatology May 1996
Metze et al.
Table I. Reactivity for CEA glycoproteins in reactive, hamartomatous, and neoplastic proliferations with sebaceous differentiation Other CEA
Dermatofibroma With sebaceous differentiation (n = 2) Without sebaceous differentiation (n = 4) Seborrheic keratosis (n = 3) Tricholemmoma (n = 2) Yricbodiscoma (n = 1) Desmoplastic trichoepithelioma (n = 3) Trichoblastoma With sebaceous differentiation (n = 3) Without sebaceous differentiation (n = 7) Pilomatricoma (n = 6) BCC with follicular differentiation (n = 5) Follicular cyst Infundibular type (n = 6) Isthmus-catagen type (n = 6) Steatocystoma (n = 2) Nevus sebaceus (n = 5) Sebaceous hyperplasia (n = 25) Sebaceous adenoma (n = 11) Sebaceoma (n = 17) Sebaceous carcinoma (n = 2)
BGP
glycoproteins
+
-
+
_
_
(+)
_
_
(+)
+*
_
+*
+
-
+
+
-
+
+
--
_
_
(+)
_
_
(+)
m
m
_
m
m
m
+
--
+
+
--
+
+
--
+
+
--
+
+
--
+
+
--
+
+, R e a c t i v e ; - , u n r e a c t i v e ; (+), r e a c t i v e in e p i t h e l i a o t h e r t h a n w i t h s e b a c e o u s differentiation;
BCC, b a s a l
EMA
-I-
cell c a r c i n o m a .
* R e a c t i v i t y i n v a c u o l a t e d cells only, n o t i n c l e a r ceils d e v o i d o f v a c u o l e s .
essential in cell adhesion, cell surface recognition, protein sorting, and secretory mechanisms. 3 The expression of CEA glycoproteins has been shown to correlate with the differentiation and architecture of epithelia under normal and neoplastic conditions. 2 In the skin, " C E A " or CEA-like proteins have been described as a marker for sweat gland differentiation. Because conventional antibodies were not able to discriminate between the different members of the CEA gene family, immunoreactivity for " C E A " could not be further specified. 6-14 Using specific monoclonal antibodies we recently demonstrated that normal sweat glands contain CEA-180, NCA90, and BGP. 15 Accordingly, in sweat gland neoplasms a coexpression of at least two of the CEA glycoproteins has been shown to be highly conserved. 15a Many conflicting data have been reported on the presence of " C E A " or CEA-like proteins in sebaceous glands and neoplasms with sebaceous differentiation.ll, 12, 16, 17 Furthermore, no comparative studies on the expression of the individual CEA glycoproteins have been performed so far to prove the specificity of CEA glycoproteins for sweat gland
differentiation. Therefore we investigated normal sebaceous glands, hamartomas, and neoplasms w i t h sebaceous differentiation using a panel of well-characterized anti-CEA antibodies. The immunohistochemical results were correlated with the invariable expression of epithelial membrane antigen (EMA) in sebaceous glandsJ 6-18 Immunoelectron microscopy was performed to obtain further information regarding the possible function of these highly glycosylated proteins for sebum production.
MATERIAL AND METHODS Tissue sampling and processing Samples of normal human sldn (n = 30) were obtained from various body sites including scalp, face, eyelids, axillary and inguinal areas, preputium, extremities, and trunk of both female and male patients undergoing surgery (age range, 3 to 75 years; mean, 35 years). Fetal skin was obtained from the trunks of four fetuses after spontaneous abortion at gestational weeks 4, 8, 12, and 21. Multiple specimens were obtained in each case and fixed in liquid nitrogen or paraformaldehyde and embedded in parablast. Paraffin-embedded biopsy specimens of
Journal of the American Academy of Dermatology Volume 34, Number 5, Part 1
Metze et al.
737
T a b l e I I . D i f f e r e n t i a l d i a g n o s e s o f s e b a c e o u s n e o p l a s m s : R e a c t i v i t y for C E A g l y c o p r o t e i n s in n e o p l a s m s w i t h c l e a r cells BGP
Syfingoma, clear-cell variant (n = 2) Clear-cell hidradenoma (n = 5) Mixed tumor, apocrine type (n = 1) Microcystic adnexal carcinoma (n = 1) Extramammary Paget's disease (n = 4) Mammary Paget's disease (n = 2) Clear cell acanthoma (n = 4) Seborrheic keratosis, clonal type (n = 2) Bowen's disease, pagetoid variant (n = 1) Lymphoepithelioma-like carcinoma (n = 1) Proliferating tricholemmal cyst (n = 2) Cutaneous metastases Breast carcinoma (n = l) Renal cell carcinoma (n = 1) Occult primary carcinoma (n = 1) Dermatofibroma with clear cells (n = 1) Balloon cell nevus (n = 2) Balloon cell melanoma (n = 1)
Other CEA glycoproteins
EMA
-
+
-
+*
+*
+* +*
+* +*
-
+
+
+
+
+
-
+
+
+
+
+
+, Reactive; -, unreactive. *Clear cells unreactive; staining is confined to the luminal surface of ductal lumina and cystic spaces.
Table
IlL Reactivity of antibodies Antibody
26/3/13 26/5/1 4/3/17 N1 9A6 80I-I3 4I)1 C2 192 47 Polyclonal Ab
CEA
NCA-90
NCA-95
BGP
+ + + + + +
+ + + +
+ + +
+ + + +
+, Reactive; -, unreactive; Ab, antibody.
reactive, hamartomatous, and neoplastic proliferations with sebaceous differentiation including 55 cases of sebaceous neoplasms were investigated (Table I). Thirteen specimens of sebaceous neoplasms were obtained from two patients with M u i r - T o n e syndrome. For differential diagnostic reasons, the study included 32 samples of benign and malignant neoplasms with a clear cell constituency (Table II). The diagnoses were made from examination o f hematoxylin-eosin-stained sections according to the criteria given in the literature. 19-26 For immunoelectron microscopy four biopsy specimens from normal skin of the axillae and trunk were fixed in a mixture of 4% paraformaldehyde and 0.5% glutaraldehyde, dehydrated in ethanol, embedded in LR-White
(London Resin Comp. Lt., Woking, England) and polymerized at 60 ° C.
Immunohistochemistry Light microscopy A panel of polyclonal and monoclonal antibodies27-3° specific for different members of the C E A family were used (for reactivities see Table HI). The specificity of the antibodies was determined against stable transfectant cell lines expressing individual members o f the C E A gene family 2s or according to an international comparative stndy.31, 32
For detection of E M A we used a well-characterized
738
Metze et aL
Fig. 1. In normal skin mature sebocytes and sebum are consistently stained with antibodies reactive for BGP, a glycoprotein of the CEA family~ (Imrnunoperoxidase method; original magnification x80.) monoclonal antibody (Dako, Glostmp, Belgium) as described by Heyderman et al.33 Deparaffmized parablastembedded sections and frozen sections were washed in 1% methanolic hydrogen peroxide, rinsed with phosphate-buffered solution (PBS), preincubated with 2% bovine serum albumin, and incubated with the primary antibodies diluted in 1% bovine serum albumin (dilutions: 26/5/1, 4/3/17, N1, and the polyclonal antibody, 1:400; 26/3/13 and 9A6, 1:1000; 192 and 47, 1:50; 80H3, 1:25; 4D1 C2, 1:60; EMA, 1:100). The sections were then developed by an indirect immunoperoxidase technique with the use of peroxidase-conjugated goat-anti-mouse or anti-rabbit antibodies (dilution 1:50; Dianova, Hamburg, Germany), 0.01% hydrogen peroxide, 3-amino-9-ethylcarbazole (Sigma Chemical Co., St. Louis, Mo.). For double immunostaining a colloidal gold/silver enhancement technique was combined with the immunoperoxidase technique.34 Briefly, incubation of the monoclonal EMA antibody was followed by rinsing and mounting of the sections with a 1 nm gold-labeled goatanti-mouse antibody (dilution 1:10) that was visualized by a commercial silver enhancement assay (Biotrend, Cologne, Germany). After washing and preincnhation with 2% bovine serum albumin the slides were incubated with the polyclonal CEA antibody (Dako) and stained as described for the routine immunoperoxidase staining. Counterstaining was performed with Harris' hematoxylin (Merck, Darmstadt, Germany). Electron microscopy Immunoelectron microscopy was performed with the
Journal of the American Academy of Dermatology May 1996
Fig. 2. Expression of CEA glycoprotein (visualized by red peroxidase staining) parallels reactivity for EMA, a well-established marker for sebaceous differentiation (labeled with black silver grains). (Double immunostaining; original magnifcafion x600.)
polyclonal CEA antibody and monoclonal EMA antibody that showed reactivity on methacrylate-embedded tissues. Ultrathin sections were mounted on uncoated copper grids. Preincubation with Millipore-filtered 4% ovalbumirl for 10 minutes was followed by overnight incubation with the primary antibodies (both CEA and EMA antibody diluted 1:100 in PBS). The grids were then thoroughly rinsed in PBS. Primary antibodies were visualized with a protein-A gold technique35 that used 15 nm gold particles (Biotrend) diluted 1:50 in PBS. After final PBS rinsing and short air drying, the sections were counterstained with uranyl acetate and subsequently examined in a Phillips CM10 electron microscope (Kassel, Germany). Controls Controls for immunohlstochemistry at light and electron microscopic level included preabsorption of the primary CEA antibodies with purified CEA (Biodesign, Kennebunkport, Me.), spleen or liver extracts (Sigma Chemical) that were rich in glycoproteins of the CEA gene family as identified by Western blot analyses (data not shown). For the anti-EMA antibody preabsorption was performed with centrifuged and diluted htunan breast milk. 36 In additional controls primary antisera were omitted or replaced with unrelated isotype-matched monoclonal antibodies (Dianova). To rule out cross-reactivity in the double immunostaining experiments, the primary and the species-specific secondary antibodies were mismatched. The specificity of the immunogold labeling was
Journal of the American Academy of Dermatology Volume 34, Number 5, Part 1
Metze et al.
739
Table IV. Reactivity of anti-CEA antibodies with sebaceous glands Antibody
Fetal skin (4, 8, 12 wk)
Fetal skin (21 wk)
Adult skin
26/3/13 26/5/1 4/3/17 N1 9A6 80H3 4D1 C2 192 47 Polyclonal Ab
ND ND -
+ + ND ND +
+ + + +
+, Reactive; -, unreactive; Ab, antibody; ND, not done.
assessed by application of protein-A gold alone and by incubation with unlabeled protein-A before staining with protein-A gold.
any of the antibodies. Granulocytes present in blood vessels showed immunoreactivity for NCA-90, NCA-95, and BGP.
RESULTS Normal adult skin
Fetal skin
All biopsy specimens from adult human skin showed a consistent staining of the sebaceous glands with the antibodies recognizing BGP (Table IV). Labeling was strictly confined to mature sebocytes containing lipid vacuoles, whereas the germinative cells in the periphery of the lobules were not reactive. Disintegrating sebocytes and cellular debris in the opening of the sebaceous glands were also strongly stained for BGP (Fig. 1). A similar staining pattern in sebaceous glands was found for EMA. Double immunostaining confirmed the coexpression of E M A and BGP in the differentiating sebocytes (Fig. 2). No variations were observed with regard to body site or age of the patients. Sebaceous glands of terminal and vellus hair follicles as well as free sebaceous glands of the eyelids or preputium without accompanying hair follicles revealed equal reactivity. As sporadically found in the skin of older patients, cords of undifferentiated cells appeared on either side of the hair follicles in a "mantle"-like fashion that did not show any staining unless they contained vacuolated sebocytes (Fig. 3). Identical results were obtained in both parablastembedded and cryofixed material. In contrast, secretory portions of sweat glands were concurrently labeled for CEA, NCA-90, and BGP. Sweat ducts coexpressed CEA and NCA-90. Normal epidermis, hair follicles, blood vessels, and other cutaneous structures were not recognized by
Undifferentiated hair germs of embryonic skin were neither reactive for EMA nor for CEA glycoproteins. In the twelfth week of gestation staining for EMA could fn'st be observed in vacuolated cells in the center of hemispheric protuberances of the growing hair follicle. No labeling for BGP was detectable before full maturation of the sebaceous glands as seen in the specimen taken from a 21-week-old fetus.
Immunoelectron microscopy Immunoelectron microscopy of sebaceous glands confunaed the expression of a CEA glycoprotein and EMA in suprabasal differentiating sebocytes with lipid droplets (Fig. 4). Reactivity for both BGP and EMA was detectable in the perinuclear Golgi area, in small membrane-bound vesicles, and along the intercellular membranes. In contrast, larger nonmembrane-bound lipid vacuoles were not reactive. A diffuse cytoplasmic staining in terminally differentiated sebocytes paralleled the dissolution of cellular organelles.
Hamartomas, hyperplasias, neoplasms with sebaceous differentiation (Table I) Dermatofibroma with induction of folliculosebaceous units in the overlying hyperplastic epidermis showed consistent staining of scattered sebaceous cells and rudimentary sebaceous lobules with antibodies reactive for BGP and EMA (Fig. 5). Apart
740
Metze et aL
Journal of the American Academy of Dermatology May 1996
Fig. 3. Cords of undifferentiated cells that surround the hair follicle in the form of a "mantle" are not reactive unless they contain vacuolated sebocytes. (Original magnification x160.)
Fig. 5. Induction of sebaceous structures by dermatofibromas is associated with expression of BGP. (Original magnification x160.)
Fig. 4. Immunoelectronmicroscopydemonstrates CEA glycoprotein within small vesicles (arrows) of mature sebocytes. Lipid vacuoles (L). Undifferentiated cells are unreactive (U). (Postembedding immunogold technique; original magnification x6000.)
from some distinct membranous staining of keratinocytes for EMA in the stratum spinosum, no reactivity was found in dermatofibroma without sebaceous differentiation. Tricholemmoma, trichodiscoma, desmoplastic lrichoepithelioma, and trichoblastoma with signs of sebaceous differentiation revealed a cytoplasmic staining for BGP and E M A confmed to vacuolated cells. However, clear cells of tricholemmoma devoid of vacuoles were not stained. All the other neoplasms with follicular differentiation and seborrheic keratoses without sebaceous differentiation lacked reactivity for our panel of anti-CEA antibodies. Organoid and cystic hamartomas with sebaceous differentiation also exhibited labeling for BGP and
EMA. Differentiated sebocytes in rudimentary or well-developed sebaceous glands of nevus sebaceus expressed BGP and EMA irrespective of the age of the patient. In steatocystoma not only small sebaceous lobules, but also single scattered sebocytes within the cyst walls were distinctly demarcated by the BGP and EMA antibodies. The distribution of BGP and EMA within the sebaceous lobules of sebaceous hyperplasias was similar to normal sebaceous glands. However, in some cases multiple layers of unstained cells in the periphery of the glands signified slight alteration in the architecture. A pronounced defect in maturation was conspicuous in sebaceous adenoma. In the elongated lobules directly connected to the skin surface only a few sebocytes were labeled for B G P and EMA (Fig. 6). In sebaceoma aggregation of cells in all stages of sebaceous differentiation were associated with immunoreactivity for BGP and EMA. In addition, amorphous cellular debris secreted into ducts and cystic cavities was diffusely labelled (Fig. 7, A). The staining density in sebaceoma varied according to the degree of differentiation (Fig. 7, B). In sebaceous carcinoma immunoreactivity of the anti-BGP and anti-EMA antibodies was not confined to vacuolated cells. Moreover, in large areas a
Journal of the American Academy of Dermatology Volume 34, Number 5, Part 1
Metze et al.
741
diffuse cytoplasmic staining of undifferentiated epithelia could be observed (Fig. 8). A comparable staining pattern was found in the samples obtained from patients with Muir-Torre syndrome. Neoplasms with a clear cell constituency (Table II) A panel of neoplasms characterized by clear cells that were sometimes difficult to differentiate from sebocytes was either not labeled for BGP or, in addition to BGP, coexpressed other CEA glycoproteins. Beyond the different staining pattern, cells with a clear cytoplasm devoid of vacuoles were consistentlyunreactive. In clear-cell neoplasms staining for BGP was confined to glandlike structures, the luminal surface of ductal lumina, and cystic spaces, respectively (Fig. 9). Controls Preincubation of the antibodies with CEA, liver or spleen extracts, and human breast milk resulted in blocking or strong diminution of the immunohistochemical staining by the respective antibodies. Omission oftheprimary antibodies, isotype-matched antibodies, and the other control experiments applied for immunoelectron microscopy yielded no staining. In the double-immunostainingprocedure, mismatch of the secondary antibodies ruled out cross-reactivity of the detection system. DISCUSSION CEA has been widely accepted as a marker for sweat gland differentiation.6-9 However, our study demonstrates that sebaceous and sweat glands share the expression of BGP, a glycoprotein of the CEA gene family. Whereas sebaceous glands show reactivity for BGP only, normal sweat glands concurrently coexpress BGP, CEA, and NCA-90. The monoclonal antibodies applied in this study have been well characterized immunochemically.27-29, 31, 32Furthermore, theirimmunohistologic reactivity pattern has been confirmed at transcriptional level in various tissues. 37 Therefore our results seem to reliably reflect the appearance of individual CEA glycoproteins in cutaneous glands. The variable cross-reactivity of many commercially available antibodies with glycoproteins of the CEA family (personal observation) may account for the conflicting results as observed in sebaceous glands.11, la, 16, 17 In normal sebaceous glands of all body sites, BGP
Fig. 6. In sebaceous adenoma reactivity for BGP is reduced. (Original magnification x40.)
was strictly confined to mature sebocytes containing lipid vacuoles. Accordingly, cords of immature sebocytes that emanate from the base of the inftmdibulum and subsequently surrotmd the hair follicle in the form of a "mantle" are not reactive unless they contain vacuolated sebocytes. The "mantle" is discussed to represent the involution or the resting phase of a sebaceous gland cycle. 19,38 In addition, in fetal skin labeling for BGP was not detectable before complete maturation of the sebaceous glands. These findings confirm that differentiation of sebocytes is invariably associated with expression of this CEA glycoprotein. In sebaceous glands staining for BGP and EMA resulted in an identical distribution pattern. Only sebaceous glands of fetal skin expressed EMA in an earlier stage of development. EMA is a highly glycosylated protein fast described as a constituent of human milk fat globules involved in lipid secretion.39 EMA is detectable in various glandular tissues and neoplasms with epithelial differentiation. 18, 36 It appears reasonable that BGP and EMA play a role in sebaceous secretion. At ultrastmctural level BGP and EMA could be detected in small vesicles closely related to the Golgi zone that are known to contribute to the production of lipids.4°,41 The diffuse distribution of BGP within terminally differentiated sebocytes most likely can be attributed to the cellular disintegration during holocrine secretion. The expression of BGP was highly conserved in reactive, hamartomatous, and neoplastic proliferations with sebaceous differentiation. Dermatofibromas have the capacity to induce epidermal hyperplasia along with folliculosebaceous rudiments that
742
Metze et al.
Journal of the AmericanAcademyof Dermatology May 1996
Fig. 7. In sebaceoma aggregations of cells in all stages of sebaceous maturation and cystic stmc~xes are labeled for BGP (A). Reactivity in sebaceoma with low differentiation is confined to scattered sebocytes (B). (Original magnification; A, MOO; B, x60.)
Fig. 8. Sebaceous carcinoma shows diffuse staining of nonvacuolated ceils. (Original magnification x400.)
Fig. 9. Clear ceils as seen in solid-cystic hidradenoma are not reactive for BGP. Note staining of granulocytes in blood vessel. (Original magnification x200.)
showed staining for EMA and BGP. 19 Furthermore, in follicular neoplasms vacuolated epithelia consistent with sebaceous differentiation were labeled by the antibodies recognizing BGP and EMA. 2° All cases of neoplasms with sebaceous differentiation revealed an invariable reactivity for BGP and EMA that was strictly confined to vacuolated epithelia and secreted debris, sparing undifferentiated cells or ductal epithelia. Alternatively, a diffuse cytoplasmic labeling of undifferentiated epithelia in sebaceous carcinoma may be related to malignant
transformation. The consistent reactivity pattern of the antibody panel as summarized in Table I and the control experiments appear to exclude nonspecific staining because of cross-reactivity. Because there may be difficulties in the distinction of various neoplasms with a clear-cell constituency from sebaceous gland neoplasms, their staining patterns were compared. 19,21"24 However, sweat gland neoplasms, neoplastic proliferations ofkeratinocytes, cutaneous metastases, dermatofibromas, and melanocytic neoplasms containing clear cells could be
Journal of the American Academy of Dermatology Volume 34, Number 5, Part 1
clearly differentiated by their immunohistochemical profile (Table ID. In addition, clear cells devoid of vacuoles consistently lacked reactivity for BGP as opposed to sebocytes characterized by a vacuolated cytoplasm. For practical purposes, the accurate diagnosis of adnexal neoplasms with sebaceous differentiation is important because they serve as cutaneous markers for the Muir-Torre syndrome.42,43 This hereditary syndrome requires recognition because the patients are at high risk for multiple gastrointestinal and other malignancies.44, 45 In conclusion, the detection of BGP in sebaceous glands disqualifies the CEA family to be a monospecific marker for sweat gland differentiation. However, sweat and sebaceous glands are characterized by their differential expression of CEA glycoproteins. Accordingly, adnexal neoplasms can be distinguished by a panel of well-defined CEA antibodies. The monoclonal antibodies 47 and 192 were provided by F. Buchegger, University of Lausanne. The excellent technical assistance of U. Keller, C. Focke, and A. Wissel is gratefully acknowledged. REFERENCES
1. Thompson J, Zimmermann W. The carcinoembryonic antigen gene family:structure, expression and evolution. Tumour Biol 1988;9:63-83. 2. Thomas P, Toth CA, Saini KS, et al. The structure, metabolism, and function of the carcinoembryonic antigen gene family. Biochem Biophys Acta 1990; 1032:177-89. 3. Thompson JA, Grtmert F, Zimmermann W. Carcinoembryonic antigen gene family: molecular biology and clinical perspectives. J Clin Lab Anal 1991;5:344-66. 4. Beauchemin N, Turbide C, Afar D, et al. A mouse analogue of the human carcinoembryonic antigen. Cancer Res 1989; 49:2017-21. 5. Katz AM, Rosenthal D, Sander DN. Cell adhesion molecules: structure, function, and implication in a variety of cutaneous and other pathologic conditions. Int J Dermatol 1991 ;30:153-60. 6. Eto H, Nishiyama S. Sweat gland cells. In: Hiroaki U, Hideo Y, eds. A colour arias of dermato-immunohistocytology. Tokyo: Wolfe Medical Publications, 1989:40-1. 7. Penneys NS, Nadji M, McKinney EC. Carcinoembryonic antigen present in human eccrine sweat. J AM ACADDERMATOL 1981;4:401-3. 8. Cotton DWK. Immunohistochemical staining of normal sweat glands. Br J Dermatol 1986;114:441-9. 9. Graham RM, McKee PH, Chapman DV, et al. Intercellular canaliculi in eccrine sweat glands: an immunoperoxidase study. Br J Dermatol 1985;112:397-403. 10. Penneys NS. Immunohistochemistry of adnexal neoplasms. J Cutan Pathol 1984;11:357-64. 11. Kariniemi AL, Forsman LM, Wahlstr/Sm T, et al. Expression of differentiation antigens in benign sweat gland tumours. Br J Dermatol 1984; 111:175-82.
Metze et aL
743
12. Penneys NS, Nadji M, Morales A. Carcinoembryonic antigen in benign sweat gland tumors. Arch Dennatol 1982; 118:225-7. 13. Penneys NS, Nadji M, Ziegels-Weissmann J, et al. Carcinoembryonic antigen in sweat gland carcinomas. Cancer 1982;50:1608-1 l. 14. Maiorana A, Nigrisoli E, Papotti M. Immunohistochemical markers of sweat gland tumors. J Cutan Pathol 1986; 13:187-96. 15. Metze D, Bhardwaj R, Amann U, et al. Glyeoproteins of the carcinoembryonic antigen (CEA) family are expressed in sweat and sebaceous glands of human fetal and adult skin. J Invest Dermatol 1996;106:64-9. 15a. Metze D, Gmnert F, Neumaier M, et al. Neoplasms with sweat gland differentiation express various glycoproteins of the carcinoembryonic antigen (CEA) family. J Cutan Pathol 1996;23:1-11. 16. Heyderman E, Graham RM, Chapman DV, et al. Epithelial markers in primary skin cancer: an immunoperoxidase study of the distribution of epithelial membrane antigen (EMA) and carcinoembryonic antigen (CEA) in 65 primary skin carcinomas. Histopathology 1984;8:423-34. 17. Friedman KJ, Boudreau S, Farmer ER. Superficial epithelioma with sebaceous differentiation. J Cutan Pathol 1987; 14:193-7. 18. Sloane JP, Ormerod MG. Distribution of epithelial membrane antigen in normal and neoplastic tissues and its value in diagnostic minor pathology. Cancer 1981;47:1786-95. 19. Steffen CH, Ackerman AB, eds. Neoplasms with sebaceous differentiation. Philadelphia: Lea & Febiger, 1994. 20. Ackerman AB, de Viragh PA, Chongchitnant N, eds. Neoplasms with follicular differentiation. Philadelphia: Lea & Febiger, 1993. 21. Weedon D. Turnouts of cutaneous appendages. In: Symmers WSC, ed. Systemic pathology: the skin; vol 9. Edinburgh: Churchill-Livingstone, 1992:817-72. 22. Brownstein MH, Shapiro L. The pilosebaceous minors. Int J Dermatol 1977;16:340-52. 23. Prioleau PG, Santa Cruz DJ. Sebaceous gland neoplasia. J Cutan Pathol 1984; 11:396-414. 24. Troy JL, Ackerman AB. Sebaceoma: a distinctive benign neoplasm of adnexal epithefium differentiating towards sebaceous cells. Am J Dermatopathol 1984;6:7-13. 25. Wick MR, Goellner JR, Wolfe JT, et al. Adnexal carcinomas of the skin: 1I. extraocular sebaceous carcinomas. Cancer 1985;56:1!63-72. 26. Nelson BR, Hamlet KR, Gillard M, et al. Sebaceous carcinoma. J AM ACAO DF2,MATOL1995;33:1-15. 27. Gmnert F, Abu Harfeil N, Schwarz K, et al. Two CEA and three NCA species, although distinguishable by monoclonal antibodies, have nearly identical peptide patterns. Int J Cancer 1985;36:357-62. 28. Daniel S, Nagel G, Johnson JB, et al. Determination of the specificities of monoclonal antibodies recognizing members of the CEA family using a panel of transfectants. Int J Cancer 1993;55:303-10. 29. Drzeniek Z, Lamerz R, Fenger U, et al. Identifcafion of membrane antigens in granulocytes and colonic carcinoma cells by a monoclonal antibody specific for biliary glycoprotein, a member of the carcinoembryonic antigen family. Cancer Lett 1991;56:173-6. 30. Nap M, ten Hoor KA, Fleuren GJ. Cross-reactivity with normal antigens in commercial anti-CEA sera, used for immunohistology: the need for tissue controls and absorption. Am J Clin Pathol 1983;79:25-31. 31. Hammarstrom S, Shively JE, Paxton RJ, et al. Antigenic
744
32.
33.
34.
35.
36.
37.
Journal of the American Academy of Dermatology May 1996
Metze et aL
sites in carcinoembryonic antigen. Cancer Res 1989;49: 4852-8. Gmnert F, Soubt M, Jantscheff P, et al. Specificity and epitope localization of CD 66/67 monoclonal antibodies. In: Schlossmann Sf, Boumsell L, Gilks W, et al, eds. Leukocyte typing V. White cell differentiation antigens. Oxford: Oxford Press. (in press) Heyderman E, Stmdiey I, Powell G, et al. A new monoclonal antibody to epithelial membrane antigen (EMA)E29: a comparison of its immunocytochemical reactivity with polyclonal anti-EMA antibodies and with another monoclonal antibody, HMFG-2. Br J Cancer 1985;52:35561. Lucocq JM, Roth J. Colloidal gold and colloidal silvermetallic markers for light microscopic histochemistry. In: Bullock GR, Petrusz P, eds. Techniques in immunocytochemistry; vol 3. New York: Academic Press, 1985:20336. Romano EL, Romano M. Staphylococcal protein A bound to colloidal gold: a useful reagent to label antigen-antibody sites in electron microscopy. Immunohistochemistry 1977;14:711-5. Heyderman E, Steele K, Ormerod MG. A new antigen on the epithelial membrane: its immunoperoxidase localisation in normal and neoplastic tissue. J Clin Pathol 1979; 32:35-9. Thompson J, M6ssinger S, Reichardt V, et al. A poly-
38. 39.
40.
41.
42.
43.
44.
45.
merase-chain-reaction assay for the specific identification of transcripts encoded by individual CEA gene family members. Int J Cancer 1993;55:311-9. Steffen.CH. Mantleoma. Am J Dermatopathol 1993; 15:306-10. Ormerod MG, Steele K, Westwood JI-I, et al. Epithelial membrane antigen:partialptnification assay and properties. Br J Cancer 1983;48:533-41. Strauss JS, Downing DT, Ebling FJ, et al. Sebaceous glands. In: Goldsmith LA, ed. Physiology, biochemistry, and molecular biology of the skin. Oxford: Oxford University Press, 1991:712-40. Bell M. A comparative study of the ultrastrucmre of the sebaceous glands of man and other primates. J Invest Dermatol 1974;62:132-43. Rothenberg J, Lambert WC, Vail JT Jr, et al. The Muir-Tone (Torre's) syndrome: the significance of a sop italy sebaceous tumor. J AM ACAD DERMATOL 1990;23: 638-40. Banse-Kupin L, Morales A, Barlow M. Torre's syndrome: report of two cases and review of the literature. J AM ACAD DERMATOL 1984;10:803-17. Cohen PR, Kohn SR, Kurzrock R. Association of sebaceous gland rumors and internal malignancy: the MuirTorre syndrome. Am J Med 1990;90:606-13. Schwartz RA, Torre DP. The Muir-Torre syndrome: a 25year retrospect. J AM ACAD DERMATOL1995;33:90-104.
Expression of a glycoprotein of the carcinoembryonic antigen family in normal and neoplastic sebaceous glands Limited role of carcinoembryonic antigen as a sweat gland marker Dieter Metze, M D , a H.-Peter Soyer, MD, b Bernhard Zelger, MD, c Michael Neumaier, MD, e Fritz Grunert, PhD, f Christine Hartig, MD, d Udo Amann, MD, a Ranjit Bhardwaj, PhD, a Christoph Wagener, MD, e and Thomas A. Luger, MD a Graz and Innsbruck, Austria, and
Miinster, Hamburg, and Freiburg, Germany Background: Carcinoembryonic antigen (CEA) is a well-known marker for sweat gland differentiation in adnexal neoplasms. Objective: The aim of this study was to examine the expression of glycoproteins of the CEA family, that is, CEA-180, nonspecific cross-reacting antigens (NCAs), and bitiary glycoprotein (BGP), in sebaceous glands and in neoplasms with sebaceous differentiation. Methods: Normal adult and fetal skin, hyperplasias, hamartomas, and neoplasms with sebaceous or follicular differentiation were stained immunohistochemicaIly with a panel of polyclonal and monoclonal antibodies highly specific for CEA-180, NCAs, and BGP. Double immunostaining was performed to correlate the CEA expression with that of epithelial membrane antigen (EMA), a glycoprotein consistently found in differentiating sebocytes. Results: Whereas sweat glands coexpressed CEA, NCAs, BGP, and EMA, sebaceous glands were exclusively labeled with the antibodies recognizing BGP or EMA. Staining of the sebaceous glands was restricted to mature sebocytes, sparing immature cells. At the ultrastmctural level immunoreactivity for BGP and EMA was demonstrable in the golgi area, in small vesicles, and along the cell membranes. During fetal development BGP was not found until the sebaceous glands matured. The expression of BGP and EMA was highly conserved in reactive, hamartomatous, and neoplastic profiferations of adnexal structures with sebaceous differentiation. Conclusion: The expression of BGP, a CEA glycoprotein, in differentiating sebocytes accounts for the reactivity of many anti-CEA antibodies with sebaceous glands and thus disqualifies the CEA family as a monospecific marker for sweat gland differentiation. (J AM ACADDERMATOL1996;34:735-44.)
Carcinoembryonic antigen (CEA) comprises a group of closely related glycoproteins that have been detected in normal as well as neoplastic epithelia of From the Departments of Dermatology, University of Mtinster, a University of Graz,b University of Innsbmck,c and Medical Center Mindend; and from the Department of Clinical Chemistry,e University Hospital Eppendorf, Hamburg, and the Institute of Immunobiology, University of Freiburg. f Presented in part at the 14th Congress of the International Society of Dermatopathology, Sienna, Italy, June 26-July 1, 1993. Accepted for publication Nov. 16, 1995. Reprint requests: Dieter Metze, MD, Department of Dermatology, University Mttnster, Von-Esmarchstrasse 56, D-48149 Mtinster, Germany. Copyright © 1996 by the American Academy of Dermatology, Inc. 0190-9622/96 $5.00 + 0 16/1/70667
the gut, fiver, lung, breast, and other tissues.I, 2 The CEA glycoproteins were identified as products of a complex gene family, which encodes the classical 180 kd CEA, 160 kd bifiary glycoprotein (BGP), and nonspecific cross-reacting antigens (NCAs) of 95 kd (NCA-95) or 90 kd (NCA-90). 3' 4 In addition, granulocytes have been previously demonstrated to constitutively express NCAs and BGP. 3 The CEA glycoproteins and their splice variants differ in size, number of immunoglobulin domains, posttranslational modification, membrane anchorage, and glycosylation. 3 They are part of the immunoglobufin supergene family that includes immunoglobufins, T-cell receptors, growth factor receptors, and cell adhesion molecules. 5 Members of this family are 735
7 3 6
Journal of the American Academy of Dermatology May 1996
Metze et al.
Table I. Reactivity for CEA glycoproteins in reactive, hamartomatous, and neoplastic proliferations with sebaceous differentiation Other CEA
Dermatofibroma With sebaceous differentiation (n = 2) Without sebaceous differentiation (n = 4) Seborrheic keratosis (n = 3) Tricholemmoma (n = 2) Yricbodiscoma (n = 1) Desmoplastic trichoepithelioma (n = 3) Trichoblastoma With sebaceous differentiation (n = 3) Without sebaceous differentiation (n = 7) Pilomatricoma (n = 6) BCC with follicular differentiation (n = 5) Follicular cyst Infundibular type (n = 6) Isthmus-catagen type (n = 6) Steatocystoma (n = 2) Nevus sebaceus (n = 5) Sebaceous hyperplasia (n = 25) Sebaceous adenoma (n = 11) Sebaceoma (n = 17) Sebaceous carcinoma (n = 2)
BGP
glycoproteins
+
-
+
_
_
(+)
_
_
(+)
+*
_
+*
+
-
+
+
-
+
+
--
_
_
(+)
_
_
(+)
m
m
_
m
m
m
+
--
+
+
--
+
+
--
+
+
--
+
+
--
+
+
--
+
+, R e a c t i v e ; - , u n r e a c t i v e ; (+), r e a c t i v e in e p i t h e l i a o t h e r t h a n w i t h s e b a c e o u s differentiation;
BCC, b a s a l
EMA
-I-
cell c a r c i n o m a .
* R e a c t i v i t y i n v a c u o l a t e d cells only, n o t i n c l e a r ceils d e v o i d o f v a c u o l e s .
essential in cell adhesion, cell surface recognition, protein sorting, and secretory mechanisms. 3 The expression of CEA glycoproteins has been shown to correlate with the differentiation and architecture of epithelia under normal and neoplastic conditions. 2 In the skin, " C E A " or CEA-like proteins have been described as a marker for sweat gland differentiation. Because conventional antibodies were not able to discriminate between the different members of the CEA gene family, immunoreactivity for " C E A " could not be further specified. 6-14 Using specific monoclonal antibodies we recently demonstrated that normal sweat glands contain CEA-180, NCA90, and BGP. 15 Accordingly, in sweat gland neoplasms a coexpression of at least two of the CEA glycoproteins has been shown to be highly conserved. 15a Many conflicting data have been reported on the presence of " C E A " or CEA-like proteins in sebaceous glands and neoplasms with sebaceous differentiation.ll, 12, 16, 17 Furthermore, no comparative studies on the expression of the individual CEA glycoproteins have been performed so far to prove the specificity of CEA glycoproteins for sweat gland
differentiation. Therefore we investigated normal sebaceous glands, hamartomas, and neoplasms w i t h sebaceous differentiation using a panel of well-characterized anti-CEA antibodies. The immunohistochemical results were correlated with the invariable expression of epithelial membrane antigen (EMA) in sebaceous glandsJ 6-18 Immunoelectron microscopy was performed to obtain further information regarding the possible function of these highly glycosylated proteins for sebum production.
MATERIAL AND METHODS Tissue sampling and processing Samples of normal human sldn (n = 30) were obtained from various body sites including scalp, face, eyelids, axillary and inguinal areas, preputium, extremities, and trunk of both female and male patients undergoing surgery (age range, 3 to 75 years; mean, 35 years). Fetal skin was obtained from the trunks of four fetuses after spontaneous abortion at gestational weeks 4, 8, 12, and 21. Multiple specimens were obtained in each case and fixed in liquid nitrogen or paraformaldehyde and embedded in parablast. Paraffin-embedded biopsy specimens of
Journal of the American Academy of Dermatology Volume 34, Number 5, Part 1
Metze et al.
737
T a b l e I I . D i f f e r e n t i a l d i a g n o s e s o f s e b a c e o u s n e o p l a s m s : R e a c t i v i t y for C E A g l y c o p r o t e i n s in n e o p l a s m s w i t h c l e a r cells BGP
Syfingoma, clear-cell variant (n = 2) Clear-cell hidradenoma (n = 5) Mixed tumor, apocrine type (n = 1) Microcystic adnexal carcinoma (n = 1) Extramammary Paget's disease (n = 4) Mammary Paget's disease (n = 2) Clear cell acanthoma (n = 4) Seborrheic keratosis, clonal type (n = 2) Bowen's disease, pagetoid variant (n = 1) Lymphoepithelioma-like carcinoma (n = 1) Proliferating tricholemmal cyst (n = 2) Cutaneous metastases Breast carcinoma (n = l) Renal cell carcinoma (n = 1) Occult primary carcinoma (n = 1) Dermatofibroma with clear cells (n = 1) Balloon cell nevus (n = 2) Balloon cell melanoma (n = 1)
Other CEA glycoproteins
EMA
-
+
-
+*
+*
+* +*
+* +*
-
+
+
+
+
+
-
+
+
+
+
+
+, Reactive; -, unreactive. *Clear cells unreactive; staining is confined to the luminal surface of ductal lumina and cystic spaces.
Table
IlL Reactivity of antibodies Antibody
26/3/13 26/5/1 4/3/17 N1 9A6 80I-I3 4I)1 C2 192 47 Polyclonal Ab
CEA
NCA-90
NCA-95
BGP
+ + + + + +
+ + + +
+ + +
+ + + +
+, Reactive; -, unreactive; Ab, antibody.
reactive, hamartomatous, and neoplastic proliferations with sebaceous differentiation including 55 cases of sebaceous neoplasms were investigated (Table I). Thirteen specimens of sebaceous neoplasms were obtained from two patients with M u i r - T o n e syndrome. For differential diagnostic reasons, the study included 32 samples of benign and malignant neoplasms with a clear cell constituency (Table II). The diagnoses were made from examination o f hematoxylin-eosin-stained sections according to the criteria given in the literature. 19-26 For immunoelectron microscopy four biopsy specimens from normal skin of the axillae and trunk were fixed in a mixture of 4% paraformaldehyde and 0.5% glutaraldehyde, dehydrated in ethanol, embedded in LR-White
(London Resin Comp. Lt., Woking, England) and polymerized at 60 ° C.
Immunohistochemistry Light microscopy A panel of polyclonal and monoclonal antibodies27-3° specific for different members of the C E A family were used (for reactivities see Table HI). The specificity of the antibodies was determined against stable transfectant cell lines expressing individual members o f the C E A gene family 2s or according to an international comparative stndy.31, 32
For detection of E M A we used a well-characterized
738
Metze et aL
Fig. 1. In normal skin mature sebocytes and sebum are consistently stained with antibodies reactive for BGP, a glycoprotein of the CEA family~ (Imrnunoperoxidase method; original magnification x80.) monoclonal antibody (Dako, Glostmp, Belgium) as described by Heyderman et al.33 Deparaffmized parablastembedded sections and frozen sections were washed in 1% methanolic hydrogen peroxide, rinsed with phosphate-buffered solution (PBS), preincubated with 2% bovine serum albumin, and incubated with the primary antibodies diluted in 1% bovine serum albumin (dilutions: 26/5/1, 4/3/17, N1, and the polyclonal antibody, 1:400; 26/3/13 and 9A6, 1:1000; 192 and 47, 1:50; 80H3, 1:25; 4D1 C2, 1:60; EMA, 1:100). The sections were then developed by an indirect immunoperoxidase technique with the use of peroxidase-conjugated goat-anti-mouse or anti-rabbit antibodies (dilution 1:50; Dianova, Hamburg, Germany), 0.01% hydrogen peroxide, 3-amino-9-ethylcarbazole (Sigma Chemical Co., St. Louis, Mo.). For double immunostaining a colloidal gold/silver enhancement technique was combined with the immunoperoxidase technique.34 Briefly, incubation of the monoclonal EMA antibody was followed by rinsing and mounting of the sections with a 1 nm gold-labeled goatanti-mouse antibody (dilution 1:10) that was visualized by a commercial silver enhancement assay (Biotrend, Cologne, Germany). After washing and preincnhation with 2% bovine serum albumin the slides were incubated with the polyclonal CEA antibody (Dako) and stained as described for the routine immunoperoxidase staining. Counterstaining was performed with Harris' hematoxylin (Merck, Darmstadt, Germany). Electron microscopy Immunoelectron microscopy was performed with the
Journal of the American Academy of Dermatology May 1996
Fig. 2. Expression of CEA glycoprotein (visualized by red peroxidase staining) parallels reactivity for EMA, a well-established marker for sebaceous differentiation (labeled with black silver grains). (Double immunostaining; original magnifcafion x600.)
polyclonal CEA antibody and monoclonal EMA antibody that showed reactivity on methacrylate-embedded tissues. Ultrathin sections were mounted on uncoated copper grids. Preincubation with Millipore-filtered 4% ovalbumirl for 10 minutes was followed by overnight incubation with the primary antibodies (both CEA and EMA antibody diluted 1:100 in PBS). The grids were then thoroughly rinsed in PBS. Primary antibodies were visualized with a protein-A gold technique35 that used 15 nm gold particles (Biotrend) diluted 1:50 in PBS. After final PBS rinsing and short air drying, the sections were counterstained with uranyl acetate and subsequently examined in a Phillips CM10 electron microscope (Kassel, Germany). Controls Controls for immunohlstochemistry at light and electron microscopic level included preabsorption of the primary CEA antibodies with purified CEA (Biodesign, Kennebunkport, Me.), spleen or liver extracts (Sigma Chemical) that were rich in glycoproteins of the CEA gene family as identified by Western blot analyses (data not shown). For the anti-EMA antibody preabsorption was performed with centrifuged and diluted htunan breast milk. 36 In additional controls primary antisera were omitted or replaced with unrelated isotype-matched monoclonal antibodies (Dianova). To rule out cross-reactivity in the double immunostaining experiments, the primary and the species-specific secondary antibodies were mismatched. The specificity of the immunogold labeling was
Journal of the American Academy of Dermatology Volume 34, Number 5, Part 1
Metze et al.
739
Table IV. Reactivity of anti-CEA antibodies with sebaceous glands Antibody
Fetal skin (4, 8, 12 wk)
Fetal skin (21 wk)
Adult skin
26/3/13 26/5/1 4/3/17 N1 9A6 80H3 4D1 C2 192 47 Polyclonal Ab
ND ND -
+ + ND ND +
+ + + +
+, Reactive; -, unreactive; Ab, antibody; ND, not done.
assessed by application of protein-A gold alone and by incubation with unlabeled protein-A before staining with protein-A gold.
any of the antibodies. Granulocytes present in blood vessels showed immunoreactivity for NCA-90, NCA-95, and BGP.
RESULTS Normal adult skin
Fetal skin
All biopsy specimens from adult human skin showed a consistent staining of the sebaceous glands with the antibodies recognizing BGP (Table IV). Labeling was strictly confined to mature sebocytes containing lipid vacuoles, whereas the germinative cells in the periphery of the lobules were not reactive. Disintegrating sebocytes and cellular debris in the opening of the sebaceous glands were also strongly stained for BGP (Fig. 1). A similar staining pattern in sebaceous glands was found for EMA. Double immunostaining confirmed the coexpression of E M A and BGP in the differentiating sebocytes (Fig. 2). No variations were observed with regard to body site or age of the patients. Sebaceous glands of terminal and vellus hair follicles as well as free sebaceous glands of the eyelids or preputium without accompanying hair follicles revealed equal reactivity. As sporadically found in the skin of older patients, cords of undifferentiated cells appeared on either side of the hair follicles in a "mantle"-like fashion that did not show any staining unless they contained vacuolated sebocytes (Fig. 3). Identical results were obtained in both parablastembedded and cryofixed material. In contrast, secretory portions of sweat glands were concurrently labeled for CEA, NCA-90, and BGP. Sweat ducts coexpressed CEA and NCA-90. Normal epidermis, hair follicles, blood vessels, and other cutaneous structures were not recognized by
Undifferentiated hair germs of embryonic skin were neither reactive for EMA nor for CEA glycoproteins. In the twelfth week of gestation staining for EMA could fn'st be observed in vacuolated cells in the center of hemispheric protuberances of the growing hair follicle. No labeling for BGP was detectable before full maturation of the sebaceous glands as seen in the specimen taken from a 21-week-old fetus.
Immunoelectron microscopy Immunoelectron microscopy of sebaceous glands confunaed the expression of a CEA glycoprotein and EMA in suprabasal differentiating sebocytes with lipid droplets (Fig. 4). Reactivity for both BGP and EMA was detectable in the perinuclear Golgi area, in small membrane-bound vesicles, and along the intercellular membranes. In contrast, larger nonmembrane-bound lipid vacuoles were not reactive. A diffuse cytoplasmic staining in terminally differentiated sebocytes paralleled the dissolution of cellular organelles.
Hamartomas, hyperplasias, neoplasms with sebaceous differentiation (Table I) Dermatofibroma with induction of folliculosebaceous units in the overlying hyperplastic epidermis showed consistent staining of scattered sebaceous cells and rudimentary sebaceous lobules with antibodies reactive for BGP and EMA (Fig. 5). Apart
740
Metze et aL
Journal of the American Academy of Dermatology May 1996
Fig. 3. Cords of undifferentiated cells that surround the hair follicle in the form of a "mantle" are not reactive unless they contain vacuolated sebocytes. (Original magnification x160.)
Fig. 5. Induction of sebaceous structures by dermatofibromas is associated with expression of BGP. (Original magnification x160.)
Fig. 4. Immunoelectronmicroscopydemonstrates CEA glycoprotein within small vesicles (arrows) of mature sebocytes. Lipid vacuoles (L). Undifferentiated cells are unreactive (U). (Postembedding immunogold technique; original magnification x6000.)
from some distinct membranous staining of keratinocytes for EMA in the stratum spinosum, no reactivity was found in dermatofibroma without sebaceous differentiation. Tricholemmoma, trichodiscoma, desmoplastic lrichoepithelioma, and trichoblastoma with signs of sebaceous differentiation revealed a cytoplasmic staining for BGP and E M A confmed to vacuolated cells. However, clear cells of tricholemmoma devoid of vacuoles were not stained. All the other neoplasms with follicular differentiation and seborrheic keratoses without sebaceous differentiation lacked reactivity for our panel of anti-CEA antibodies. Organoid and cystic hamartomas with sebaceous differentiation also exhibited labeling for BGP and
EMA. Differentiated sebocytes in rudimentary or well-developed sebaceous glands of nevus sebaceus expressed BGP and EMA irrespective of the age of the patient. In steatocystoma not only small sebaceous lobules, but also single scattered sebocytes within the cyst walls were distinctly demarcated by the BGP and EMA antibodies. The distribution of BGP and EMA within the sebaceous lobules of sebaceous hyperplasias was similar to normal sebaceous glands. However, in some cases multiple layers of unstained cells in the periphery of the glands signified slight alteration in the architecture. A pronounced defect in maturation was conspicuous in sebaceous adenoma. In the elongated lobules directly connected to the skin surface only a few sebocytes were labeled for B G P and EMA (Fig. 6). In sebaceoma aggregation of cells in all stages of sebaceous differentiation were associated with immunoreactivity for BGP and EMA. In addition, amorphous cellular debris secreted into ducts and cystic cavities was diffusely labelled (Fig. 7, A). The staining density in sebaceoma varied according to the degree of differentiation (Fig. 7, B). In sebaceous carcinoma immunoreactivity of the anti-BGP and anti-EMA antibodies was not confined to vacuolated cells. Moreover, in large areas a
Journal of the American Academy of Dermatology Volume 34, Number 5, Part 1
Metze et al.
741
diffuse cytoplasmic staining of undifferentiated epithelia could be observed (Fig. 8). A comparable staining pattern was found in the samples obtained from patients with Muir-Torre syndrome. Neoplasms with a clear cell constituency (Table II) A panel of neoplasms characterized by clear cells that were sometimes difficult to differentiate from sebocytes was either not labeled for BGP or, in addition to BGP, coexpressed other CEA glycoproteins. Beyond the different staining pattern, cells with a clear cytoplasm devoid of vacuoles were consistentlyunreactive. In clear-cell neoplasms staining for BGP was confined to glandlike structures, the luminal surface of ductal lumina, and cystic spaces, respectively (Fig. 9). Controls Preincubation of the antibodies with CEA, liver or spleen extracts, and human breast milk resulted in blocking or strong diminution of the immunohistochemical staining by the respective antibodies. Omission oftheprimary antibodies, isotype-matched antibodies, and the other control experiments applied for immunoelectron microscopy yielded no staining. In the double-immunostainingprocedure, mismatch of the secondary antibodies ruled out cross-reactivity of the detection system. DISCUSSION CEA has been widely accepted as a marker for sweat gland differentiation.6-9 However, our study demonstrates that sebaceous and sweat glands share the expression of BGP, a glycoprotein of the CEA gene family. Whereas sebaceous glands show reactivity for BGP only, normal sweat glands concurrently coexpress BGP, CEA, and NCA-90. The monoclonal antibodies applied in this study have been well characterized immunochemically.27-29, 31, 32Furthermore, theirimmunohistologic reactivity pattern has been confirmed at transcriptional level in various tissues. 37 Therefore our results seem to reliably reflect the appearance of individual CEA glycoproteins in cutaneous glands. The variable cross-reactivity of many commercially available antibodies with glycoproteins of the CEA family (personal observation) may account for the conflicting results as observed in sebaceous glands.11, la, 16, 17 In normal sebaceous glands of all body sites, BGP
Fig. 6. In sebaceous adenoma reactivity for BGP is reduced. (Original magnification x40.)
was strictly confined to mature sebocytes containing lipid vacuoles. Accordingly, cords of immature sebocytes that emanate from the base of the inftmdibulum and subsequently surrotmd the hair follicle in the form of a "mantle" are not reactive unless they contain vacuolated sebocytes. The "mantle" is discussed to represent the involution or the resting phase of a sebaceous gland cycle. 19,38 In addition, in fetal skin labeling for BGP was not detectable before complete maturation of the sebaceous glands. These findings confirm that differentiation of sebocytes is invariably associated with expression of this CEA glycoprotein. In sebaceous glands staining for BGP and EMA resulted in an identical distribution pattern. Only sebaceous glands of fetal skin expressed EMA in an earlier stage of development. EMA is a highly glycosylated protein fast described as a constituent of human milk fat globules involved in lipid secretion.39 EMA is detectable in various glandular tissues and neoplasms with epithelial differentiation. 18, 36 It appears reasonable that BGP and EMA play a role in sebaceous secretion. At ultrastmctural level BGP and EMA could be detected in small vesicles closely related to the Golgi zone that are known to contribute to the production of lipids.4°,41 The diffuse distribution of BGP within terminally differentiated sebocytes most likely can be attributed to the cellular disintegration during holocrine secretion. The expression of BGP was highly conserved in reactive, hamartomatous, and neoplastic proliferations with sebaceous differentiation. Dermatofibromas have the capacity to induce epidermal hyperplasia along with folliculosebaceous rudiments that
742
Metze et al.
Journal of the AmericanAcademyof Dermatology May 1996
Fig. 7. In sebaceoma aggregations of cells in all stages of sebaceous maturation and cystic stmc~xes are labeled for BGP (A). Reactivity in sebaceoma with low differentiation is confined to scattered sebocytes (B). (Original magnification; A, MOO; B, x60.)
Fig. 8. Sebaceous carcinoma shows diffuse staining of nonvacuolated ceils. (Original magnification x400.)
Fig. 9. Clear ceils as seen in solid-cystic hidradenoma are not reactive for BGP. Note staining of granulocytes in blood vessel. (Original magnification x200.)
showed staining for EMA and BGP. 19 Furthermore, in follicular neoplasms vacuolated epithelia consistent with sebaceous differentiation were labeled by the antibodies recognizing BGP and EMA. 2° All cases of neoplasms with sebaceous differentiation revealed an invariable reactivity for BGP and EMA that was strictly confined to vacuolated epithelia and secreted debris, sparing undifferentiated cells or ductal epithelia. Alternatively, a diffuse cytoplasmic labeling of undifferentiated epithelia in sebaceous carcinoma may be related to malignant
transformation. The consistent reactivity pattern of the antibody panel as summarized in Table I and the control experiments appear to exclude nonspecific staining because of cross-reactivity. Because there may be difficulties in the distinction of various neoplasms with a clear-cell constituency from sebaceous gland neoplasms, their staining patterns were compared. 19,21"24 However, sweat gland neoplasms, neoplastic proliferations ofkeratinocytes, cutaneous metastases, dermatofibromas, and melanocytic neoplasms containing clear cells could be
Journal of the American Academy of Dermatology Volume 34, Number 5, Part 1
clearly differentiated by their immunohistochemical profile (Table ID. In addition, clear cells devoid of vacuoles consistently lacked reactivity for BGP as opposed to sebocytes characterized by a vacuolated cytoplasm. For practical purposes, the accurate diagnosis of adnexal neoplasms with sebaceous differentiation is important because they serve as cutaneous markers for the Muir-Torre syndrome.42,43 This hereditary syndrome requires recognition because the patients are at high risk for multiple gastrointestinal and other malignancies.44, 45 In conclusion, the detection of BGP in sebaceous glands disqualifies the CEA family to be a monospecific marker for sweat gland differentiation. However, sweat and sebaceous glands are characterized by their differential expression of CEA glycoproteins. Accordingly, adnexal neoplasms can be distinguished by a panel of well-defined CEA antibodies. The monoclonal antibodies 47 and 192 were provided by F. Buchegger, University of Lausanne. The excellent technical assistance of U. Keller, C. Focke, and A. Wissel is gratefully acknowledged. REFERENCES
1. Thompson J, Zimmermann W. The carcinoembryonic antigen gene family:structure, expression and evolution. Tumour Biol 1988;9:63-83. 2. Thomas P, Toth CA, Saini KS, et al. The structure, metabolism, and function of the carcinoembryonic antigen gene family. Biochem Biophys Acta 1990; 1032:177-89. 3. Thompson JA, Grtmert F, Zimmermann W. Carcinoembryonic antigen gene family: molecular biology and clinical perspectives. J Clin Lab Anal 1991;5:344-66. 4. Beauchemin N, Turbide C, Afar D, et al. A mouse analogue of the human carcinoembryonic antigen. Cancer Res 1989; 49:2017-21. 5. Katz AM, Rosenthal D, Sander DN. Cell adhesion molecules: structure, function, and implication in a variety of cutaneous and other pathologic conditions. Int J Dermatol 1991 ;30:153-60. 6. Eto H, Nishiyama S. Sweat gland cells. In: Hiroaki U, Hideo Y, eds. A colour arias of dermato-immunohistocytology. Tokyo: Wolfe Medical Publications, 1989:40-1. 7. Penneys NS, Nadji M, McKinney EC. Carcinoembryonic antigen present in human eccrine sweat. J AM ACADDERMATOL 1981;4:401-3. 8. Cotton DWK. Immunohistochemical staining of normal sweat glands. Br J Dermatol 1986;114:441-9. 9. Graham RM, McKee PH, Chapman DV, et al. Intercellular canaliculi in eccrine sweat glands: an immunoperoxidase study. Br J Dermatol 1985;112:397-403. 10. Penneys NS. Immunohistochemistry of adnexal neoplasms. J Cutan Pathol 1984;11:357-64. 11. Kariniemi AL, Forsman LM, Wahlstr/Sm T, et al. Expression of differentiation antigens in benign sweat gland tumours. Br J Dermatol 1984; 111:175-82.
Metze et aL
743
12. Penneys NS, Nadji M, Morales A. Carcinoembryonic antigen in benign sweat gland tumors. Arch Dennatol 1982; 118:225-7. 13. Penneys NS, Nadji M, Ziegels-Weissmann J, et al. Carcinoembryonic antigen in sweat gland carcinomas. Cancer 1982;50:1608-1 l. 14. Maiorana A, Nigrisoli E, Papotti M. Immunohistochemical markers of sweat gland tumors. J Cutan Pathol 1986; 13:187-96. 15. Metze D, Bhardwaj R, Amann U, et al. Glyeoproteins of the carcinoembryonic antigen (CEA) family are expressed in sweat and sebaceous glands of human fetal and adult skin. J Invest Dermatol 1996;106:64-9. 15a. Metze D, Gmnert F, Neumaier M, et al. Neoplasms with sweat gland differentiation express various glycoproteins of the carcinoembryonic antigen (CEA) family. J Cutan Pathol 1996;23:1-11. 16. Heyderman E, Graham RM, Chapman DV, et al. Epithelial markers in primary skin cancer: an immunoperoxidase study of the distribution of epithelial membrane antigen (EMA) and carcinoembryonic antigen (CEA) in 65 primary skin carcinomas. Histopathology 1984;8:423-34. 17. Friedman KJ, Boudreau S, Farmer ER. Superficial epithelioma with sebaceous differentiation. J Cutan Pathol 1987; 14:193-7. 18. Sloane JP, Ormerod MG. Distribution of epithelial membrane antigen in normal and neoplastic tissues and its value in diagnostic minor pathology. Cancer 1981;47:1786-95. 19. Steffen CH, Ackerman AB, eds. Neoplasms with sebaceous differentiation. Philadelphia: Lea & Febiger, 1994. 20. Ackerman AB, de Viragh PA, Chongchitnant N, eds. Neoplasms with follicular differentiation. Philadelphia: Lea & Febiger, 1993. 21. Weedon D. Turnouts of cutaneous appendages. In: Symmers WSC, ed. Systemic pathology: the skin; vol 9. Edinburgh: Churchill-Livingstone, 1992:817-72. 22. Brownstein MH, Shapiro L. The pilosebaceous minors. Int J Dermatol 1977;16:340-52. 23. Prioleau PG, Santa Cruz DJ. Sebaceous gland neoplasia. J Cutan Pathol 1984; 11:396-414. 24. Troy JL, Ackerman AB. Sebaceoma: a distinctive benign neoplasm of adnexal epithefium differentiating towards sebaceous cells. Am J Dermatopathol 1984;6:7-13. 25. Wick MR, Goellner JR, Wolfe JT, et al. Adnexal carcinomas of the skin: 1I. extraocular sebaceous carcinomas. Cancer 1985;56:1!63-72. 26. Nelson BR, Hamlet KR, Gillard M, et al. Sebaceous carcinoma. J AM ACAO DF2,MATOL1995;33:1-15. 27. Gmnert F, Abu Harfeil N, Schwarz K, et al. Two CEA and three NCA species, although distinguishable by monoclonal antibodies, have nearly identical peptide patterns. Int J Cancer 1985;36:357-62. 28. Daniel S, Nagel G, Johnson JB, et al. Determination of the specificities of monoclonal antibodies recognizing members of the CEA family using a panel of transfectants. Int J Cancer 1993;55:303-10. 29. Drzeniek Z, Lamerz R, Fenger U, et al. Identifcafion of membrane antigens in granulocytes and colonic carcinoma cells by a monoclonal antibody specific for biliary glycoprotein, a member of the carcinoembryonic antigen family. Cancer Lett 1991;56:173-6. 30. Nap M, ten Hoor KA, Fleuren GJ. Cross-reactivity with normal antigens in commercial anti-CEA sera, used for immunohistology: the need for tissue controls and absorption. Am J Clin Pathol 1983;79:25-31. 31. Hammarstrom S, Shively JE, Paxton RJ, et al. Antigenic
744
32.
33.
34.
35.
36.
37.
Journal of the American Academy of Dermatology May 1996
Metze et aL
sites in carcinoembryonic antigen. Cancer Res 1989;49: 4852-8. Gmnert F, Soubt M, Jantscheff P, et al. Specificity and epitope localization of CD 66/67 monoclonal antibodies. In: Schlossmann Sf, Boumsell L, Gilks W, et al, eds. Leukocyte typing V. White cell differentiation antigens. Oxford: Oxford Press. (in press) Heyderman E, Stmdiey I, Powell G, et al. A new monoclonal antibody to epithelial membrane antigen (EMA)E29: a comparison of its immunocytochemical reactivity with polyclonal anti-EMA antibodies and with another monoclonal antibody, HMFG-2. Br J Cancer 1985;52:35561. Lucocq JM, Roth J. Colloidal gold and colloidal silvermetallic markers for light microscopic histochemistry. In: Bullock GR, Petrusz P, eds. Techniques in immunocytochemistry; vol 3. New York: Academic Press, 1985:20336. Romano EL, Romano M. Staphylococcal protein A bound to colloidal gold: a useful reagent to label antigen-antibody sites in electron microscopy. Immunohistochemistry 1977;14:711-5. Heyderman E, Steele K, Ormerod MG. A new antigen on the epithelial membrane: its immunoperoxidase localisation in normal and neoplastic tissue. J Clin Pathol 1979; 32:35-9. Thompson J, M6ssinger S, Reichardt V, et al. A poly-
38. 39.
40.
41.
42.
43.
44.
45.
merase-chain-reaction assay for the specific identification of transcripts encoded by individual CEA gene family members. Int J Cancer 1993;55:311-9. Steffen.CH. Mantleoma. Am J Dermatopathol 1993; 15:306-10. Ormerod MG, Steele K, Westwood JI-I, et al. Epithelial membrane antigen:partialptnification assay and properties. Br J Cancer 1983;48:533-41. Strauss JS, Downing DT, Ebling FJ, et al. Sebaceous glands. In: Goldsmith LA, ed. Physiology, biochemistry, and molecular biology of the skin. Oxford: Oxford University Press, 1991:712-40. Bell M. A comparative study of the ultrastrucmre of the sebaceous glands of man and other primates. J Invest Dermatol 1974;62:132-43. Rothenberg J, Lambert WC, Vail JT Jr, et al. The Muir-Tone (Torre's) syndrome: the significance of a sop italy sebaceous tumor. J AM ACAD DERMATOL 1990;23: 638-40. Banse-Kupin L, Morales A, Barlow M. Torre's syndrome: report of two cases and review of the literature. J AM ACAD DERMATOL 1984;10:803-17. Cohen PR, Kohn SR, Kurzrock R. Association of sebaceous gland rumors and internal malignancy: the MuirTorre syndrome. Am J Med 1990;90:606-13. Schwartz RA, Torre DP. The Muir-Torre syndrome: a 25year retrospect. J AM ACAD DERMATOL1995;33:90-104.