J Clin Pathol 199 1;44:543-548 543 Troublesome tumours 1: Adnexal tumours of the J Clin Pathol: first published as 10.1136/jcp.44.7.543 on 1 July 1991. Downloaded from D Cotton

Introduction these are very unusual,6 and the confusion due Most adnexal tumours are benign and, if com- to the term "" being used for a pletely excised, cause no further concern. It different, malignant, tumour of other sites may therefore be thought that there is little causes considerable difficulty. Again, duct dif- need for further subclassification. The major ferentiation is CEA positive, but the bulk of arguments for considering them further can tumour cells in all these tumours (poromas, be summarised as follows: (1) if you are not , and ) are CEA sure what it is, it may be something else; (2) negative. All of the above mentioned tumours clinical associations with specific subtypes will have features reminiscent of the sweat not become apparent if the lesions are never on electron microscopical examination and subtyped; and (3) there is academic and obses- they stain variably positive with middle sional satisfaction to be derived from weight cytokeratin antibodies such as PKKI meticulously identifying lesions as accurately and are negative for CAM5 2, S100, epithelial as possible. membrane antigen (EMA) and human milk Given these justifications I will comment on fat globule 1 (HMFG 1). what I consider to be useful and interesting Poroma, , and cylindroma are aspects of certain adnexal tumours. The first all derived from the outer cells of the duct and division is into tumours showing affinity with behave as benign "epitheliomas" or eccrine and those showing affinity with "basalomas" as these terms are variously used the pilosebaceous system. The question of by Lever7 and Pinkus.8 They are the whether tumours are derived from a given equivalent of those basal cell structure or from some primordial and, at which arise from the pilosebaceous unit. least, pluripotential stem cell is unresolved, Their largely benign nature is probably due to probably unresolvable, and is, anyway, of the fact that, unlike the follicle, the little practical concern; the important point is eccrine does not actively cycle and that tumours do resemble normal structures Hair proliferate. follicle epitheliomas are http://jcp.bmj.com/ and that this enables us to classify them and, much more akin to epidermal epitheliomas as our experience of them accumulates, to because the also proliferates and predict their behaviour. regenerates actively, and such lesions con- stitute the common basal cell carcinomas. Malignant transformation of these eccrine tumours tumours is indicated by cellular pleomor- sweat The eccrine gland consists of several phism, invasion, and high mitotic rate; ab- on September 30, 2021 by guest. Protected copyright. morphological and functional regions and normal mitotic figures are generally highly there is at least one recognisable tumour for suggestive of . each of these (fig 1). The outer cells of the The cells lining the inner aspect of the coiled intraepidermal duct give rise to the eccrine duct and those of the secretory coil are eccrine poroma. This tumour may nest within concerned with active secretion/resorption the epidermis as hidroacanthoma simplex, or processes and their benign are it may become detached from the epidermis adenomas-the so-called hidradenomas. These and occur as an intradermal nodule, termed a are more prone to malignant transformation dermal duct tumour; actual ducts are rare but and occur as simple termed are positive with carcinoembryonic antigen hidradenocarcinomas. Because the cells lining (CEA) antibody when they occur. Malignant the duct have different functions and express poromas have been clearly described and their different markers from those forming the behaviour is relatively unpredictable.' It has secretory coil, the hidradenomas are rather been claimed that the clear cell of variable in form and in their marker expres- Degos has a similar histogenesis,2 but the sion. Those derived from the duct show gen- evidence for this is poor and the lesion is eral staining with CEA antibody in most of Department of probably reactive rather than neoplastic.3 4 their cells; those from the secretory coil show Pathology, University The intradermal straight duct gives rise to CAM 5-2 and S100 positivity but no CEA of Sheffield Medical eccrine School, Beech Hill spiradenomas, and again ductular dif- antibody positivity.' The intercellular can- Road, Sheffield ferentiation is rare but CEA positive when it aliculi of the secretory coil express HMFG1 S10 2RY occurs. Highly vascular and giant forms5 and EMA positivity, and small spidery areas D Cotton occur which can give some diagnostic dif- corresponding to this can be seen to stain with Correspondence to: ficulty but malignant transformation is again these markers in many Most Dr D Cotton hidradenomas. rare. The coiled duct seems to be the origin of hidradenomas show a mixture of differentia- Accepted for publication 21 November 1990 eccrine cylindromas; malignant variants of tion towards both duct lining cells and 544 Cotton

The eccrine sweat gland '.-.lf i Hidroacanthoma simplex Intraepidermal duct C.inPoroma

:: 4 ~~~~~Dermal duct tumour J Clin Pathol: first published as 10.1136/jcp.44.7.543 on 1 July 1991. Downloaded from Spiradenoma Intradermal duct straight Hdaeoa

Secretory coil () Figure 1 Variousfunctional regions of the eccrine sweat gland and the tumours derivedfrom them.

secretory coil cells sometimes recapitulating the normal gland with "ductal" areas nearer the epidermis and "secretory coil" areas deeper in the skin. In some hidradenomas the outer cells differentiate to myoepithelial cells and either produce or induce large volumes of stromal mucins. This compresses the aden- Hair papilla omatous part of the lesion and produces the Figure 2 Vertical section ofhairfollicle showing the characteristic mixed tumour of skin or chon- various morphological elements. droid syringoma. The stroma differs from nor- mal in being mucopolysaccharide-rich epithelial interactions peculiar to the hair folli- and is the counterpart of the "private stroma" cle and which can be recognised in distorted seen around the normal sweat gland. All form within some tumours. As with most varieties of hidradenomas can be malignant adnexal tumours, the main area of concern is and the criteria are those for adenocarcinomas distinguishing between benign and malignant of any site. Whatever their nature, no eccrine lesions; unfortunately there are no reliable tumour, unless it is very large and very old, morphological correlates to help in the mar- ever shows calcification; this is the hallmark of ginal cases. pilosebaceous tumours. Tumours ofthe hair follicle show differentia- tion towards many of the normal follicular elements and are generally named accordingly.

Pilosebaceous tumours It is much more difficult to confirm these http://jcp.bmj.com/ While all skin adnexal structures develop in a morphological impressions with techniques similar manner from embryonic epidermal such as immunocytochemistry and electron anlagen, the convention has been to separate microscopy than is the case with sweat gland eccrine sweat glands from pilosebaceous com- tumours. The reason why immunocyto- plexes and to view the latter as a somewhat chemistry is less useful in classifying hair integrated entity. This seems logical as the follicle tumours than eccrine sweat gland biological functions of apocrine and sebaceous tumours is that the different layers of the hair on September 30, 2021 by guest. Protected copyright. glands seem to be related to maintaining the follicle represent metabolic variations on the "condition" of hair and the whole complex is theme of keratinisation and there is consider- morphologically closely related. When it able overlap in specificity to the available cyto- comes to adnexal tumours, however, there are markers. In eccrine sweat glands dif- clearly some which are apocrine, some which ferent areas of the gland express markers that are sebaceous, and a large number which are show almost no overlapping specificities (CEA, pilar, notwithstanding the fact that there are HMFG1, S100, CAM5-2, PKK1) (table 1). also many The themselves The various benign hair follicle tumours hybrids. hybrids have been reviewed and classified most satisfac- support this classification as I see many sweat in and little has gland tumours with apparently mixed sweat torily by Headington 1976, gland histogenesis but relatively few that also been added since that time.'0 Malignant hair show any pilosebaceous differentiation. follicle tumours have been recently reviewed tumours show by Wick and Coffin, although they comment Similarly, pilar may apocrine ... been and sebaceous areas, but very rarely do they that, "... only a handful have show eccrine With these documented."." differentiation. Table 2 is modified from Headington's paper reservations in mind I will review some exam- of hair ples of pilosebaceous tumours under the and shows the histogenic classification various subheadings of pilar, apocrine, and follicle tumours based on morphology. This classification is based not only on the perceived sebaceous neoplasms. morphological similarities between the tumours and the normal components ofthe hair HAIR FOLLICLE (PILAR) TUMOURS follicle, but also on current ideas relating to the The hair follicle is a complex structure and at various hair "germs" (stem cells at various least seven layers can be distinguished within levels of commitment) that are believed to be the wall (fig 2). Added to this there are dermal/ involved in the normal development and Adnexal tumours of the skin 545

Table I Immunocytochemical staining patterns in eccrine sweat glands and their pore analogous to the origin of the eccrine tumours poroma"3; this histogenesis seems unlikely Staining in sweat glands Staining in tumours because all of the morphological evidence Antibody indicates that the follicular pore is normal PKKI Outer cells of duct including basal Patchy but most cells of spiradenoma, layers of epidermis cylindroma epidermis unlike the eccrine pore which is J Clin Pathol: first published as 10.1136/jcp.44.7.543 on 1 July 1991. Downloaded from CAM5.2 Some cells of secretory coil (non Some cells of hidradenoma and highly specialised and even continues into S100 positive) chondroid syringoma The CEA Inner lining cells of all regions of Any ductal cells in poroma, structured keratin on the palm and sole. duct and the inner cuticle spiradenoma and cylindroma. Some lesion consists of an endophytic mass of of the clefts in chondroid syringoma in HMFGI Intercellular canaliculi Intercellular canaliculi in hidradenoma squamous eddies identical with those seen and chondroid syringoma as spidery irritated basal cell papillomas. The lesion has areas between cells no Si1O Some cells of secretory coil (non Some areas of hidradenoma and no keratin tunnels, however, and pseudo- CAM 5.2 positive) chondroid syringoma horn cysts, which serves to distinguish it, Some myoepithelial cells Outer cells of chondroid syringoma and cells in the stroma although some authors still think that inverted follicular keratoses are all irritated basal cell papillomas or .'4 growth cycles of the hair. Most familiar hair Desmoplastic are benign follicle tumours are either "" or hamartomas with no major tendency to malig- tumours of or hair sheath, the nancy and are often confused with morphea- majority ofthe other tumours in the table are so like basal cell carcinomas and therefore treated rare that the only realistic way to recognise inappropriately. Takei et al " published a list of them is to have seen one before in a slide set 26 features that they believe differentiate the (table 2). two lesions.'5 The major features characteristic The hallmark of those hair sheath tumours of the benign lesion include (i) symmetry; (ii) that keratinise is pilar keratinisation, the form small rather than large cellular aggregates; (iii) of keratinisation commonly seen in pilar cysts. little ;ariation in size and no bizarre forms to It lacks a keratohyalin granular layer at light the aggregates; (iv) no clefting between microscopy, although characteristic tricho- aggregates and stroma; (v) compact collagen hyalin granules can be seen at electron micro- rims around aggregates; (vi) focal hair follicle scopy. The granules can not be seen on light differentiation; (vii) possibly ghost cells and microscopy. The keratinisation surface is calcification. Mitoses are rare. rather wavy and the keratin is amorphous Pilomatrixomas (derived from hair matrix) rather than layered. It is this type of keratin are readily recognised by the triad of basaloid that can undergo calcification; normal epider- cells, ghost cells, and calcification, but a more mal keratin never calcifies. Hair sheath tumours aggressive variant has also been described may also show ghost cells, more commonly which shows a proliferation edge, invasion of seen in pilomatixomas." fat and skeletal muscle, and even vascular Unfortunately, hair follicle tumours may be invasion, although distant metastases have not

mixed and typical external hair sheath features been reported. Even in the usual form of http://jcp.bmj.com/ may be found in tumours that are predomi- pilomatrixomas local recurrence after excision nantly of some other derivation. occurs in about 3% of cases. Inverted follicular is a benign Proliferating pilar tumours show the features tumour that is said to arise from the follicular of pilar keratinisation but proliferate exten- sively, invade locally, and recur following Table 2 Histogenic classification of tumours and cysts of are often to be hairfollicles'0 excision. They thought ruptured

pilar cysts that are proliferating as a result of on September 30, 2021 by guest. Protected copyright. Hamartomas of hair germ the rupture, and a typically violent dermal Localised with limited morphodifferentiation reaction to extruded keratin can often be seen in Hamartomas of the sebaceous complex the surroundings which tends to support this. Localised with advanced morphodifferentiation But their behaviour is that of a low grade Basal cell squamous showing primarily pilar Congenital vellus hamartoma as Tumours of hair germ keratinisation and they should be treated Epithelial trichogenic tumors without induction such. 16 can Epithelial trichogenic tumors with induction Three distinctive tumours of hair sheath Trichoblastic fibroma be recognised and as they are relatively com- Trichogenic trichoblastoma Mesenchymal trichogenic tumors mon their differentiating features deserve brief Trichogenic myxoma comment. Tumours of hair matrix Trichomatricoma () The of Winer" is a keratin Carcinoma of hair matrix (matrical carcinoma) plugged proliferation of a hair follicle-like Tumours of extemal hair sheath (tricholemma) Tricholemmal cyst (pilar cyst) structure with a small degree of lobulated Proliferating tricholemmal tumour proliferation around the edges. The mode of Tricholemmal keratosis Tricholemmoma keratinisation is epidermal with a granular Tricholemmal carcinoma layer suggesting that the lesion arises at the Tumours of perifollicular mesenchyme level of the infundibulum. The degree of Perifollicular fibroma proliferation is mild and but for the lack of any Tumours of the intraepidermal follicle and infundibulum Inverted follicular keratosis (acrptrichoma) deeper hair follicle elements it strongly resem- Tumour of follicular infundibulum bles a comedome. Dilated pore of Winer Tumours of miscellaneous type The pilar sheath acanthoma is related to the dilated pore of Winer'7 but is generally deeper Trichoma and much more proliferative, often with a Adapted from Headington JT. Am J Pathol 76;85:480-505. pseudo-epitheliomatous appearance. It has a 546 Cotton

central keratin filled pit and keratinisation is microscopy is in fact degeneration or clear cell again epidermal in type. The deeper cells often change and does not stain with these markers show glycogen-rich, clear cell change. (Cotton DWK, Parsons MA, Lee WR, un- consists of multiple lobules published observations). On the whole, those

of epithelial cells with the appearance of outer tumours that look like basal cell carcinomas J Clin Pathol: first published as 10.1136/jcp.44.7.543 on 1 July 1991. Downloaded from root sheath generally showing pronounced behave like them and those that look like clear cell change. The lesion is bounded by a squamous carcinomas behave like them, with thickened basement membrane and the outer the possible caveat about eyelid tumours layer of cells are palisaded. There is generally mentioned above. Sebaceous carcinomas, again no central keratin plug. A clinical association particularly those of the eyelid, may show with various thyroid, gastrointestinal, soft tis- pagetoid spread and enter the differential diag- sue and reproductive organ tumours occurs as nosis of Paget's disease, malignant , Cowden's disease. and hidroacanthoma simplex, as well as the differs histopathologically various Borst-Jadassohn lesions. from these lesions by virtue of its growth pattern which gives the impression ofarising in CERUMINOUS TUMOURS several adjacent hair follicles. The typical Occasionally, tumours of adnexal appearance glassy cells should not be confused with the may arise in the external auditory meatus and glycogen-rich clear cells of the above lesion; may be locally very aggressive in spite of glassy cells have a glassy cytoplasm, clear cells apparently benign histology, eroding temporal have an empty appearance because most of the bone, and eventually metastasising. They give glycogen is lost during routine processing. confusing staining patterns with strong cyto- These lesions also have an involuting clinical keratin staining with CAM5.2 and with course whereas the pore of Winer, the pilar apocrine markers such as EMA and HMFG1 sheath acanthoma, and the trichilemmoma and the main clue to their diagnosis is their require surgical excision. site (Cotton DWK, Hird PM, unpublished The other tumours of hair follicle origin, observations). including those interesting lesions that show neoplastic versions of the normal epithelial/ CYSTS stromal interactions that are the induction Numerous types of cyst occur within the skin, phase of the normal hair cycle, are too rare to either as solitary lesions, as cysts occurring in justify detailed description. conjunction with other lesions, or as multiple cystic structures as an entity.18 Some of these APOCRINE TUMOURS seem to arise due to mechanical blockage of Apocrine dilatation is often seen in organoid ducts, others because of cystic transformation naevi of the epidermis and adnexa but these within a more solid tumour, while many seem may be passive dilatations due to blockage of to form as cysts and to persist in that form. the outlet ducts. are common Benign cysts and Small miliarial type cysts may form from http://jcp.bmj.com/ are the exact analogues ofeccrine cysts. It is not damaged eccrine ducts following blistering at all uncommon to see cysts with walls show- eruptions, but true neoplastic cysts of similar ing both apocrine and eccrine differentiation appearance such as can arise with and this may reflect the fact that there are some no history of trauma; whatever their origin, sweat glands which show similarly mixed these cysts show typical luminal staining with features. There are also numerous benign CEA antibody.

apocrine tumours whose morphology is suf- Some cystic lesions seem to cross the boun- on September 30, 2021 by guest. Protected copyright. ficiently characteristic to permit straight- dary between eccrine and apocrine lesions, and forward diagnosis with haematoxylin and eosin hidrocytomas may show clear evidence of both staining, which can be confirmed with types of histogenesis within the one cyst wall. antibodies such as HMFG1 and EMA, or by Combinations of pilar and epidermal cysts electron microscopical examination. are more unusual, although by no means Primary apocrine carcinoma is rare but is uncommon, and generally occur when a pilar often said to be the underlying lesion in extra- cyst retains a punctum, in which case this may mammary Paget's disease. A recent review be lined with typical epidermal epithelium. disputes this and is discussed below. Malignancy in these cystic structures is uncommon, but involvement of epidermal SEBACEOUS TUMOURS cysts by general skin processes has been These fall into two groups-those from the reported for Bowen's disease,20 mycosis general skin surface and those occurring fungoides,2' and Paget's disease of the breast.22 around the eyelids. Traditionally these have been felt to have differing properties according MERKEL CELL CARCINOMA to site, with those around the eyelids having a The Merkel cell is sparsely represented in man more aggressive behaviour than identical but may be seen as an occasional, CAM5.2 appearing tumours from other sites, although positive, cell at the base of the epidermis. It the specific survival figures for ocular and non- seems to be involved in some tactile function ocular sites do not support this in the case of because it is found in abundance at the base of sebaceous carcinomas.'8 9 Most sebaceous dif- vibrissae in animals such as the vole. The ferentiation can be shown with HMFG1 or tumours are very characteristic and at first sight EMA antibodies, but when these are used it is often suggest oat cell carcinoma. They occur on apparent that much of the vacuolation that the head and neck of the elderly usually and seems to be sebaceous differentiation at light they often cause death due to widespread Adnexal tumours of the skin 547

. Histologically they are small, "blue duration of the pre-existing benign lesion in cell" tumours arranged in a trabecular pattern much the same way as with mixed salivary and are variably positive with neuroendocrine gland tumours. Generalisations are diffcult, markers, but also with CAM5.2, which oat cells however, because of the rarity of these lesions; are not.23 they are reported to represent 0-005% of all J Clin Pathol: first published as 10.1136/jcp.44.7.543 on 1 July 1991. Downloaded from skin tumours, and one person's experience of ADENOID CYSTIC AND MUCINOUS CARCINOMA them is bound to be very limited. The evidence that these rare tumours can be primary in the skin seems to be irresistible yet PROGNOSIS there are no reports indicating to what normal In one study of 10 sweat gland tumours the structure they are related. By analogy with authors describe them collectively as salivary glands we might expect them to be "hidradenocarcinomas" but subsume both related to sweat glands, yet pcrsonal experience eccrine and apocrine tumours of widely differ- with one such tumour showed no immuno- ing types under this one heading and then peroxidase staining pattern to support this. A attempt to relate histological type to prog- similar situation exists regarding mucinous nosis.6 Wicks and Coffin reviewed published carcinomas of the skin; this is very rare and reports extensively and commented on the identical to colloid carcinomas ofbreast and has large degree of disagreement to be found." definite malignant potential." They observed that it was difficult to evaluate the rate of malignant behaviour from reported PAGET' S DISEASE cases, as these had generally been reported The mammary form ofPaget's disease has been because they uncharacteristically showed convincingly shown to be due to spread of malignant behaviour. In their own series they mainly intraduct carcinoma of the breast along found metastases in 58% of sweat gland carci- larger ducts and into the epidermis. There is nomas showing ductal differentiation, but only still some argument that there are cells peculiar one third of mucinous sweat gland carcinomas to the normal nipple that can be considered to metastasised, and none of these proved fatal. be the progenitor cells of Paget's of the breast, They offered no figures for pilar tumours. but my personal view is that extrusion of The best advice regarding adnexal tumours normal breast duct cells occurs through the seems to be that the surgeon should aim for epidermis and the Paget's disease is just the complete excision with a margin of normal malignant version of this. The opposite may tissue, and that the pathologist should be apply to extra-mammary Paget's disease as it cautiously optimistic while bearing in mind the was long held to be the case that the condition fact that some undoubted, if rare, cases of arises from an underlying apocrine carcinoma. malignant behviour will arise out of very This has been strongly challenged and many benign looking adnexal tumours, and that authorities now feel that extra-mammary cytologically alarming adnexal tumours may, Paget's disease begins in the epidermis as a sort none the less, show relatively indolent http://jcp.bmj.com/ of malignant metaplasia and spreads down behaviour. ducts into the dermis, and that only a minority of cases arise by the opposite route.24 For Addendum practical purposes, if any case of extra- Since this article was written the first volume of mammary Paget's arises by spread from an Ackerman's Histologic Diagnosis of Neoplastic underlying tumour, then all cases have to be Skin Disease has been published. This volume investigated to exclude this. The differential covers Neoplasms with eccrine differentiation on September 30, 2021 by guest. Protected copyright. diagnosis always includes malignant mela- very thoroughly. It is published by Lea and , even on the nipple or vulva, and Febiger, Philadelphia 1990 (ISBN No 8121 theoretically also includes hidroacanthoma 12369). simplex and the various Borst-Jadassohn lesions, although in practice this does not represent a great diagnostic problem. 1 Ryan JF, Darley CR, Pollock DJ. Malignant eccrine poroma: report of three cases. J Clin Pathol 1986;39: 1099-104. Adnexal carcinoma 2 Lindgren AGH, Neumann E. Some evidence concerning the sweat duct origin of . Acta From what has been discussed so far it is Dermatovener 1973;53:511-4. apparent that there are two types of adnexal 3 Penneys NS, Nadji M, Ziegels-Weissman J. Clear cell acanthoma: Not of sweat gland origin. Acta Dermatovener carcinoma: those which show affinities with, 1981;61:569-70. and may be thought to derive from, benign 4 Cotton DWK, Mills PM, Stephenson TJ, Underwood JCE. On the nature of clear cell acanthomas. Br J Dermatol adnexal tumours; and those which seem to be 1987;1 17:569-74. malignant from the start. In many cases the 5 Cotton DWK, Slater DN, Rooney N, Goepel JR, Mills PM. Giant vascular eccrine spiradenomas: a report of two cases tumours are so poorly differentiated that they with histology, immunohistology and electron micro- show no recognisable features of skin adnexal scopy. Histopathology 1986;10:1093-9. 6 Dissanayake RVP, Salm R. Sweat gland carcinomas: Prog- differentiation and in this case it is impossible to nosis related to histological type. Histopathology 1980;4: place them in either category with any degree of 45-66. 7 Lever WF, Schaumburg-Lever G. Histopathology of the certainty. Some other tumours are relatively skin. 7th Ed. Philadelphia: J B Lippincott Co, 1990. well differentiated carcinomas but seem to bear 8 Pinkus H, Mehregan AH. A guide to dermatohistopathology. 3rd Ed. New York: Appleton-Century-Crofts, 1981. no relation to any recognisable benign counter- 9 Cotton DWK. An immunohistochemical study of benign part, such as of skin. eccrine sweat gland tumours. MD thesis in the University of Sheffield 1987;1-278. In general it seems that the risk of malignancy 10 Headington JT. Tumors of the hair follicle. Am J Pathol in a benign tumour is related to size and 1976;85:480-505. 548 Cotton

11 Wick MR, Coffin CM. Sweat gland and pilar carcinomas. In: 18 Massa MC, Medenica M. Cutaneous adnexal tumors and Wick MR, ed. Pathology of unusual malignant cutaneous cysts: A review. Part 1-Tumors with hair follicular and tumors. New York: Marcel Dekker Inc, 1985. sebeceous glandular differentiation and cysts related to 12 LeBoit PE, Parslow TG, Choy S-H. Hair matrix differentia- different parts of the hair follicle. Pathology Annual tion. Occurrence in lesions other than pilomatricoma. 1985;Part II:189-233. Am J Dermatopathol 1987;9:399-405. 19 Wolfe JT, Wick MR, Campbell RJ. of 13 Azzopardi JG, Laurini R. Inverted follicular keratosis. the oculocutaneous adnexa and the extraocular skin. In: J Clin Pathol: first published as 10.1136/jcp.44.7.543 on 1 July 1991. Downloaded from J Clin Pathol 1975;28:465-71. Wick MR, ed. Pathology of unusual malignant cutaneous 14 Spielogel RL, Austin C, Ackerman AB. Inverted follicular tumors. New York: Marcel Dekker Inc, 1985. keratosis is not a specific keratosis but a verruca vulgaris (or 20 Shelley WB, Wood ME. Occult Bowen's disease in keratin- ) with squamous eddies. Am J Der- ous cysts. Br J Dermatol 1981;105:105-8. matol 1983;5:427-42. 21 Slater DN, Messenger AG, Rooney N. Mycosis fungoides in 15 TIakei Y, Fukushiro S, Ackerman AB. Criteria for histologic an . Histopathology 1985;9:659-62. differentiation ofdesmoplastic trichoepithelioma (scleros- 22 Stephenson TJ, Cotton DWK. Paget's disease in an epider- ing epithelial hamartoma) from morphea-like basal-cell mal cyst. Dermatologica 1987;174:186-90. carcinoma. Am J Dermatopathol 1985;7:207-21. 23 Heenan PJ, Cole JM, Spagnolo DV. Primary cutaneous 16 Arico M, LaRocca E, Noto G, Pravata G, Rodolico V. neuroendocrine carcinoma [Merkel cell tumor). An Proliferating tricholemmal tumour with lymph node adnexal epithelial . Am J Dermatopathol 1990; metastases. Br J Dermatol 1989;121:793-7. 12:7-16. 17 Klovekorn G, Klovekorn W, Plewig G, Pinkus H. Reisen- 24 Jones RE, Austin C, Ackerman AB. Extramammary Paget's pore und Haarscheidenakanthom. Klinische und histolog- disease. A critical re-examination. Am J Dermatopathol ische Diagnose. Der Hautarzt 1983;34:209-16. 1979;1:101-32. http://jcp.bmj.com/ on September 30, 2021 by guest. Protected copyright.