Evaluation of Spindle Cell Tumors VALDA N. KAYE, MD PAUL E. SWANSON, MD
enign and malignant neoplasms of skin that are composed entirely or predominantly of spindled cells are rather uncommon, but they generally constitute a significant component of the dermatopathologist’s practice, when ordinary squamous cell carcinomas, basal cell carcinomas, and melanocytic lesions are excluded. Differences among the various benign entities in this histologic spectrum are often subtle. Nonetheless, apart from those instances in which the diagnostic alternatives also include a malignant neoplasm, the clinical cliff erences that attend these lesions are relatively small. In contrast, the evaluation of malignant cutaneous spindle cell neoplasms is often an arduous task, because histologic features are not always sufficient to distinguish among the several clinically distinct tumors that comprise this diagnostic group, including: the fibrohistiocytic “malignancies” (atypical fibroxanthoma, dermatofibrosarcoma protuberans, and malignant fibrous histiocytoma); sarcomas (angiosarcoma, leiomyosarcoma, malignant peripheral nerve sheath tumor); spindle cell malignant melanoma; and spindle cell squamous cell carcinoma (Table 1). Despite confusing histologic similarities, most of these tumors differ significantly when immunohistochemical or electron microscopic features are considered. The latter approach may be useful in situations where light microscopic techniques are inconclusive, but access to these services is often limited. In contrast, the former technique is readily available to the practicing pathologist, using standard peroxidase-antiperoxidase (PAP) or avidin-biotin-peroxidase complex (ABC) methodology and commercially-available antibodies (Table 2). Further, the result-
B
From the Departments of Laboratory Medicine and Pathology and Dermatology (VNKJ, University of Minnesota School ofMedicine, Minneapolis, Minnesota; and Department of Pathology (PES), Washington University School of Medicine, St. Louis, Missouri. Address correspondence to: Paul E. Swanson, M.D., Division of Surgical Pathology, Barnes Hospital, 1 Barnes Hospital Plaza, St. Louis MO 63120.
0 2991 by Elsevier Science Publishing
Co., Inc.
l
0738-081x/91/$3.50
ant tumoral immunophenotypes have practical applications in most clinical situations (Table 3). Thus, it is this analytic approach that is emphasized in the ensuing discussion.
Fibrohistiocytic
Tumors
As a group, the benign and malignant fibrohistiocytic tumors are histologically diverse, but as their generic name implies, they are related by a common pattern of differentiation. For practical purposes, these tumors are proliferations of fibroblasts or facultative myofibroblasts, but in virtually all settings, they are accompanied by variable numbers of histiocytes.‘J From a diagnostic point of view, the most important of the fibrohistiocytic lesions are those malignant or borderline tumors in which a spindle cell population may predominate: dermatofibrosarcoma protuberans (DFSP) and malignant fibrous histiocytoma (MFH). The first of these is a clinically indolent lesion, usually affecting adults, that presents as a firm plaque or protuberant tumoral nodule on the truck or proximal extremity. Histologically, DFSP is composed of cytologically uniform cells arranged in a tightly interweaving “storiform” pattern. Mitotic activity is often observed, but atypical division figures do not predominate.’ Stromal myxoid changes may be conspicuous in some examples, a feature that may cause some confusion with peripheral nerve sheath tumors3 There is no obvious connection between the tumor and the overlying epidermis, although neoplastic elements may abut the surface, or may cause considerable epidermal thinning and ulceration. Lateral dermal and subcuticular borders of DFSP are often ill-defined, or infiltrated by tumoral cells arranged in a loosely woven lace-like pattern. Nevertheless, deep subcutaneous extension is not typical of this neoplasm. Excision of DFSP is generally curative, but local recurrence may occur.’ Malignant fibrous histiocytoma (MFH) contrasts histologically with DFSP in several respects, most notably the
217
218
KAYE
Clinics in Dermatology
AND SWANSON
1991;9:217-226 Table 1. Malignant or Borderline Spindle Cell Neoplasms
in Skin
Fibrohistiocytic: Malignant Fibrous Histiocytoma Dermatofibrosarcoma Protuberans Atypical Fibroxanthoma Other Mesenchymal: Malignant Peripheral Nerve Sheath Tumor Leiomyosarcoma Spindle Cell Hemangioendothelioma Angiosarcoma Kaposi’s Sarcoma Spindle Cell or Desmoplastic Malignant Melanoma Soindle Cell Souamous Cell Carcinoma
presence of considerable cellular pleomorphism, multinucleated tumor giant cells, and aggressively infiltrative deep soft-tissue margins. lj4f5 Compared to that seen in DFSP, mitotic activity is increased in MFH, and atypical mitotic forms may be evident. By convention, MFH is not a tumor of dermis; rather, it is a primary lesion of subcutis and deep soft tissues. Histologically similar lesions centered on the dermis are more properly categorized as atypical fibroxanthoma (AFX).1,6-9 AFX, as with other dermal lesions that resemble deep soft-tissue malignancies, is considerably more quiescent clinically, with a relatively
Table 2. lmmunoreaxents
small risk for recurrence or metastasis following conservative surgical management. 6,7Importantly, a dermal lesion showing extensive subcuticilar or fascial spread, tumoral necrosis, or vascular invasion, is at greater risk for aggressive behavior, and should be treated as MFH.’ In general, the pleomorphic elements in AFX and MFH separate these lesions from DFSP; nevertheless, such features do not always clearly distinguish fibrohistiocytic lesions from other spindled neoplasms of skin.1,5J0 In this context, electron microscopy, as alluded to above, may be of some utility, because, (myo)fibroblastic elements in these lesions are usually admixed with histiocyte-like cells.1,2*J Immunohistochemistry contributes to the resolution of these difficult cases largely by excluding other diagnostic alternatives. In other words, the elementary immunophenotype of fibrohistiocytic tumors-using the antibodies depicted in Table 3 -includes vimentin reactivity, but very little else.2*5,9J1-14 The extent of myofibroblastic differentiation exhibited by these tumors may dictate the presence and intensity of muscle-specific actin reactivity, but in most cases of AFX and MFH, staining with antibodies to this actin isoformis at best focal. DFSP, on the other hand, rarely if ever shows actin positivity.14 What then of the histiocytic elements? A great deal of attention has been paid to the presence of proteolytic
Commonly Used in the Evaluation
of Spindle Cell Tumors Source
Dilution
Becton-Dickinson Triton Biosciences Hybritech BioGenex BioGenex Enzo Becton-Dickinson Enzo Dako Dako
1:50 1:75 1:lOO 1:4000 1:2 1:20000 1:40 1: 16000 1:800 1:80
Dako Dako Dako
1:600 1:800 1: 2000
Vector
1: 2000
Reagent Monoclonal Antibodies: Anti-cytokeratin CAM 5.2 MAK6 AEl/AE3 Anti-vimentin (V9) Anti-desmin Anti-muscle-specific actin (HHF35) Leu7 HMB45 Anti-epithelial membrane antigen (E29) Anti-factor VIII-related antigen Polyclonal Antibodies anti-S100 protein anti-alpha-1-antichymotrypsin anti-Ulex europaeus I
Lectins: Ulex europaeus I Sources: Becton Dickinson-Mountain View, CA. Triton Biosciences-Alameda, CA. Hybritech-Alameda, CA. BioGenex Laboratories-San Rnmon, CA. Dako Laboratories-Carpinteria, CA. Vector Laboratories-Burlingme, CA. Enzo Biochemicals-New York, NY.
Clinics in Dermatology 1991:9:227-226
KAYE AND SWANSON EVALUATION
OF SPINDLE
219
CELL TUMORS
Table 3. ldealized lmmunophenotypes of Spindle Cell Tumors CK
EMA
HMB
SIOO
L7
F8
UEA
ACT
+/-*
0
0
0
0
0
0
0
+
0
0
0
0
0
+
+
0
0
+/-
+ + + +
+ 0 0 0
+ 0 0 0
+/0 0 0
+/0 0 0
0 0 0 +/-
+/0 0 +
+/0 0 +/-
0 +/+/0
0 + +/0
+/0 +/+/-
+/-
0
0
+
+
0
0
0
0
+/-
+/-
Neoplasm
VIM
DES
Fibrohistiocytic Tumors Malignant Peripheral Nerve Sheath Tumor Leiomyosarcoma Angiosarcarcoma Kaposi’s Sarcoma Spindle Cell Malignant Melanoma Spindle Cell Squamous Cell Carcinoma
+
0
+
MSA
Abbreviations: VIM = uimentin; DES = desmin; MSA = muscle-specific actin; CK = cytokeratin; EMA = epitheliul membrane antigen; HMB =antigen; SloO= SIOOprotein; L7 = Law 7 antigen; F8 = factor VIII-related antigen; LEA = Ulex europaeus I agglutinin; ACT = alpha-lantichymotrypsin. l “-t/-“denotes renctivity in 20-50% ofcases. Zfmore than 50% ore typically positivefor the stated determinant, “+“is used; iffewer than 20% stain, the diagnostic group is considered to be negutiue in this idealized scheme. Most cases designated “0” are rarely, if ever, reactive for that marker.
HMB45
enzymes, such as Cathepsin B and alpha-l-antichymotrypsin, as markers of histiocytic differentiation.1*5J3-1* Whereas subsequent analyses have demonstrated the relative frequency of such substances in fibrohistiocytic lesions, they have also revealed their lack of specificity for this pattern of cellular specialization.19 Indeed, virtually all spindle cell tumors of skin occasionally stain with antibodies to these enzymes, as may a variety of malignant tumors in other sites.13 Recently, Cerio and colleagues have studied the distribution of factor XIIIa in cutaneous neoplasms. Notably, most examples of MFH and AFX are labeled with antibodies to this factor.20 However, recent evidence identifies factor XIIIa as a sensitive marker for neurofibromas as we11.21 Nevertheless, the presence of this substance together with the aforementioned enzymes, in the absence of the lineage-selective markers yet to be discussed, may Figure 1. Fibrohistiocytic Neoplasms. Spindled pleomorphic cells in superficial malignant fibrous are reactive for vimentin (4). Histiocytic elements fibroxanthoma stain for alpha-1-antichymotypsin
and histiocytoma in atypical (B).
provide presumptive evidence that the spindled neoplasm under investigation is a “fibrohistiocytic” tumor.
Spindle Cell Squamous Carcinoma Although the spindled variant of squamous carcinoma is uncommon, it poses a diagnostic challenge that leaves most of us wishing it never occured at all. The clinical appearance of spindle cell squamous carcinoma (SCSC) may resemble that of any other form of squamous carcinoma, with ulceration, excoriation and hyperkeratosis, but its ability to “drop off” the epidermis makes it possible for SCSC to present as a lesion with little obvious epidermal damage. Histologically, SCSC arises from epidermis, and appears to stream into superficial dermis. In lesions with epidermal ulceration, however, the association between epidermis and tumor may be lost, and the distinction between SCSC and other spindle neoplasms is often problematic. 10,22-24Nonetheless, the recognition of SCSC is important, as local recurrence and metatasis are relatively uncommon, compared to histologically similar melanocytic and fibrohistiocytic malignancies.1*23 The electron microscopic appearance of SCSC is often definitive, because true epithelial features are generally preserved. As in the more typical examples of squamous cell carcinoma, the cells of SCSC contain bundles of tonofilaments that insert into desmosomes or desmosomelike cell junctions.22-25 Immunohistochemistry is usually reliable as well, and for practical purposes obviates the need for electron microscopy. The immunophenotype of SCSC is reproducible in most instances, and is distinct among spindle cell tumors of skin: reactivity for both cytokeratin (CK) epithelial membrane antigen (EMA) is readily demonstrated. 11-14~17,26*27 The latter determinant, although somewhat less commonly expressed than CK, is virtually specific for epithelial differentiation in cutane-
220
KAYE AND SWANSON
Clinics in Dermatology 1991;9:22 7- 226
or acral lentiginous melanoma, and thus are often the anticipated sequalae of previously documented pigmented lesions.28*29 Examples of SMM that arise de novo are less readily recognized, especially in cases where pigmentation is sparse. As in SCSC, SMM may have an obvious association with the overlying epidermis, but in such cases, a lentiginous melanotic lesion is usually present.*s The cells of SMM are cytologically atypical, and may be conspicuously amelanotic. They exhibit a diffuse pattern of dermal or subcuticular spread, and generally are not arranged in fascicles or a distinct storiform pattern. Mitotic activity is present at all levels of the lesion, in contrast to the benign spindle-cell melanocytic lesions. Stroma is often quite fibrous or desmoplastic,28*30 and striking neurotropism may be observed. *s~*~The latter trait, among others, has formed the basis for comparisons of SMM to tumors of peripheral nerves. The validity of this comparison may be argued, but such a debate exceeds the scope of this discussion. As in SCSC, the electron microscopy of SMM is often diagnostic. The cardinal feature is the premelanosome; despite the lack of obvious pigment in many examples,
Figure 2. Spindle Cell Squamous Carcinoma. Cytokeratin reactivity is usually conspicuous in the majority of neoplastic cells.
ous neoplasms, as only rare examples of smooth muscle or peripheral nerve sheath tumors are EMA-reactive, whereas melanocytic lesions are only anecdotally positive. What leads to confusion in cases of SCSC is the variable presence of vimentin,12-*4 particularly when CK or EMA are only focally present.i4 In such cases, careful histologic and immunohistochemical evaluation of all available material may be necessary. Unfortunately, as with the other neoplasms discussed herein, not all examples can be diagnosed with certainty.
Spindle Cell (Desmoplastic) Malignant Melanoma The phenotypic complexity and diagnostic evaluation of melanocytic lesions are covered in greater detail elsewhere in this monograph. Hence, the investigation of only the overtly malignant spindle-cell melanocytic lesions will be described here. Clinical clues are often instrumental in the accurate recognition of spindled malignant melanoma (SMM). Most examples occur in association with lentigo maligna
Figure 3. Spindle Cell Malignant Melanoma. spindled cells closely appose epidermis.
SlOO-reactive
Clinics in Dermatology 1991;9:217- 226 these organelles may occasionally be abundant. As most dermatopathologists know, however, there are as many cases where premelanosomes defy detection. The utility of immunohistochemistry is well documented, particularly in efforts to separate SMM from SCSC. Unlike the latter tumor, SMM does not express either CK or EMA,‘*J4 but instead contains SlOO protein.*1J3J4,32,33 However, the latter substance is found in several other cell types, among them chondrocytes, adipocytes, Schwann cells, and melanocytes. Hence, although SlOO protein is a sensitive marker for melanocytic differentiation, it is by no means specific. Peripheral nerve sheath tumors are typically reactive for SlOO, whereas occasional examples of dermal leiomyosarcoma and SCSC may also contain this protein.13J4 Other markers have been described that are selective for neoplastic melanocytes; the most important of these is the HMB-45 antigen. In most forms of malignant melanoma, this determinant proves to be a specific tumor marker. Unfortunately, few examples of SMM are immunoreactive for this substance.32*34*35 Hence, the diagnosis of SMM in difficult cases relies on the clinical, histologic, and immunohistochemical exclusion of epithelial, myogenie, and neurilemmal differentiation.
Peripheral Nerve Sheath Tumors The most commonly encountered peripheral nerve sheath tumor in skin is the neurofibroma, a tumor that may occupy either the dermis or subcutis in virtually any anatomic location. Histologically, the uniformity of the spindle cells, together with “wavy“ nuclear and plasmalemma1 profiles, are compelling evidence of nerve sheath differentiation.’ They may be variably cellular, and may be dominated by a myxoid stroma. However, in cutaneous lesions, cellular pleomorphism is generally slight or absent, and mitotic activity is uncommon. Loose, infihrative dermal margins may be seen in some instances, and not all such tumors are easily distinguished from benign fibrohistiocytic neoplasms.’ Schwannomas (neurilemmomas) may also arise in superficial soft tissues; unlike neurofibromas, they are more variegated in their histologic appearance. Classically, schwannomas are composed of two dominant patterns, the first consisting of spindle cells showing nuclear palisading or neurotactoid differentiation (Antoni A), the second containing similar cells that are more loosely dispersed in a myxoid or degenerating stroma.’ Cystic change and hemorrhage may be conspicuous in these latter Antoni B foci. As in neurofibroma, schwannomas are not mitotically active, and except in tumors with extensive degeneration, nuclear pleomorphism and cytologic atypia are lacking.’
KAYE AND SWANSON EVALUATION OF SPINDLE CELL TUMORS
221
Figure 4. Malignant Peripheral Nerve Sheath Tumor. Both SlOOprotein (A) and Leu 7 antigen @) are detected in the majority of these neoplasms.
Among the benign nerve sheath tumors of skin, it is perhaps the neurothekeoma that evokes the greatest concern for malignancy. 1,36Composed of irregular nodules or lobules separated by distinct fibrous (perineurial?) septae, neurothekeoma may assume a plexiform growth pattern similar to that of both plexiform fibrous histiocytoma3’ and plexiform neurofibroma.’ However, in contrast to both of these alternatives, the cells of neurothekeoma are arranged in small clusters or whorls; cells may be ovoid or spindled, and varying amounts of myxoid stroma may be present. 1~36The term “dermal nerve sheath myxoma” has been applied to cases where the latter features predominate.38 Cytologically, neurothekeoma also differs from other benign nerve sheath tumors, in that nuclear atypia may be focally pronounced, and mitotic activity may be observed. Nonetheless, stromal or capsular invasion, perineurial spread, and vascular permeation are not features of this tumor. There has been some debate regarding the true nature of neurothekeoma, with some authors arguing for a perineurial, as opposed to a schwannian pattern of differentiation.1,36,3* In our experience, however, the ultrastructural and immunohistochemical features of these lesions are similar to those of other nerve sheath tumors, as will be described. In contrast to the relative frequency of benign nerve sheath tumors in skin, malignant peripheral nerve sheath tumors (MPNST) are quite uncommon.1*5 We have analyzed two such neoplasms,39 and a small series has recently been reported. 4o Unlike the benign lesions in which they may arise, MPNST are cellular neoplasms, composed of cytologically atypical and mitotically active spindled, pleomorphic, or epithelioid elements. Often, the predominant pattern is that of interweaving cell bun-
222
KAYE AND SWANSON
dles or a storiform pattern, not unlike that seen in MFH or fibrosarcoma. Fewer than one-third of all cutaneous and deep soft-tissue MPNST exhibit the histologic features of benign schwannoma. lp5One characteristic of MPNST not shared by histologically similar neoplasms is a propensity for divergent chondro-osseous, rhabdomyoblastic, endothelial, or glandular differentiation.’ However, this phenomenon has not characterized cutaneous MPNST and will not be emphasized here. Despite the lack of distinctive histologic features in many cases, MPNST shares with benign nerve sheath neoplasms the ultrastructural features of Schwann cell differentiation.’ In particular, these tumors, to varying degrees, may elaborate numerous cell processes that interdigitate, and form desmosome-like cell junctions at points of apposition. In benign tumors, these processes may encircle bundles of collagen in a manner reminiscent of the mesaxon of the normal Schwann cell. Basal lamina is typically present, although it may be absent in some MPNST. Perineurial elements consist of fibroblast-like cells with delicate cell processes that are often intimately associated with mast cells, The latter cells are often present in abundance in benign and malignant nerve sheath tumors, although this feature alone is of little diagnostic significance. In neurifibromas, cells with perineurial features may be abundant, but otherwise typical fibroblasts and Schwann cells are more commonly seen. In schwannoma, perineurial and fibroblastic elements are sparse or absent.’ Immunohistochemically, MPNST and benign nerve sheath tumors share the expression of SlOO protein, as well as two other substances associated with Schwann cell differentiation: Leu 7 antigen and myelin basic protein.13J4,41,42 Leu 7 is an antigenic determinant found on a variety of proteins, including a myelin-associated glycoprotein that has cell adhesion properties. Myelin basic protein is a well-characterized constituent of the myelin sheath. Although its presence has been questioned in cutaneous MPNST, we have found it in over one-third of deeply situated examples. 14,42As noted previously, S 100 is also expressed by melanocytic lesions, but the combination of SlOO protein and Leu 7 or myelin basic protein reactivity in a non-epithelial spindle cell tumor is compelling evidence of nerve sheath differentiation.42 Vimentin is uniformly expressed by nerve sheath lesions,‘* and the expression of glial fibrillary acidic protein in occasional benign and malignant PNST appears to be a unique trait among spindle cell neoplasms of skin.43-45 Among benign tumors, neurofibromas are somewhat less likely to express either SlOO or Leu 7, but their presence may be of particular importance in the occasional instance where neurotization of a dermal nevus has imparted a spindled appearance.14*46 In the latter context,
Clinics in Dermatology 1991;9:217-226 neither SlOO nor Leu 7 antigen is typically detected.46 The potential relevence of factor XIIIa staining to this differential diagnosis is discussed elsewhere.21 Finally, there have been occasional reports of both cytokeratin” and epithelial membrane antigen*a reactivity in nerve sheath tumors. The former, we believe, is spurious, but may nonetheless complicate the immunohistochemical diagnosis of MPNST. In contrast, the presence of EMA is an expected finding in some cases, as perineurial cells are normally reactive for this substance.49
Smooth Muscle Tumors Benign smooth muscle tumors of the dermis and subcutis are lesions that often present as solitary, firm, domeshaped nodules without epidermal ulceration, These lesions are uncommon among all age groups, and probably represent the least common group of benign mesenchyma1 neoplasms in skin.’ Histologically, leiomyomas are composed of fascicles of cytologically uniform fusiform or ovoid cells with abundant eosinophilic cytoplasm, and centrally located “cigar-shaped’ nuclei. Mitotic activity is not conspicuous, and cellular pleomorphism is minimal. These fascicles may form nodules or lobules that are related to or centered on pilosebaceous units, or may form angiomyomatous nodules in the superficial subcutis. Excision of such lesions is generally curative.’ Leiomyosarcomas (LMS) uncommonly arise in the dermis or subcutaneous tissue, and are distinguished from benign smooth muscle tumors by their larger size, their tendency to infiltrate adjacent dermis or subcutis, and their propensity for increased cytologic atypia (which may be quite pronounced), mitotic activity, and tumoral necrosis.1,5,50,51 The second of these as a marker for malignancy has been challenged in extracutaneous smooth muscle neoplasms, but the general consensus remains that superficial lesions with 2 or more mitotic figures per high-power field should be considered malignant.1~50~51 On rare occasion, vascular invasion may be seen; this feature is more typical of subcutaneous-based lesions, and portends a less favorable outcome.1*50 As Fields and Helwig have emphasized, the clinical behavior of superficial LMS depends in large part on the location of the tumor in skin.50 Lesions centered on, and largely limited to, the dermis rarely recur or metastasize, whereas deep dermal or subcuticular lesions more commonly show such behavior. Metastasis is rare in either setting when adequate surgical excision is performed.1~50~51 Although tumors dominated by epithelioid or polygonal cells have proven to be smooth muscle in nature, most examples of LMS are spindled in appearance.1~5~50~51 The architectural arrangement of these lesions, the lack of
Clinics in Dermatology 1991;9:217- 226
KAYE AND SWANSON
223
EVALUATION OF SPINDLE CELL TUMORS
protein and Leu 7 antigen is the expected phenotype of MPNST,42 and is a highly unlikely outcome in LMS. Leiomyosarcoma, like most mesenchymal neoplasms, contains vimentin, and occasional examples may express cytokeratins as well .54,55Nonetheless, the reproducibility of a myogenic (actin- and desmin-positive) phenotype in LMS provides a reliable contrast to spindle cell squamous carcinoma and other cytokeratin-containing neoplasms.
Vasogenic Neoplasms
A Figure 5. Leiomyosarcoma. Unlike other malignant spindled tumors of skin, leiomyosarcoma may be diffusely reactive for desmin (A) and muscle-specific actin 0. overt neurotropism, and their cytologic features often are sufficient for their accurate diagnosis, but as in all other settings described in this review, the occasional LMS presents a diagnostic challenge. Electron microscopy is particularly useful, as the contractile nature of these cells is readily demonstrated by this technique. Typically, cells are elongate, their cytoplasm containing little rough endoplasmic reticulum (in contrast to the (myo)fibroblastic elements of fibrohistiocytic lesions, in which such organelles are abundant), but instead is filled by bundles of microfilaments interrupted sporadically by dense bodies. Plasmalemmal plaques are evident, as is pinocytosis. The relative abundance of these structures is in part dictated by the degree of differentiation exhibited by the tumor.1*51 Immunohistochemistry is equally efficient at detecting myogenic differentiation, although there are potential pitfalls in interpretation. As expected, most smooth muscle tumors are immunoreactive for muscle-specific actin1*14~51and the intermediate filament desmin.12-14~51~52 Desmin is virtually always present in LMS, in our experience, but has been found less often by other investigators.52,53 Staining for desmin must be remembered in any evaluation of spindle cell tumors, because, as already described, fibrohistiocytic neoplasms may also occasionally exhibit focal staining for muscle-specific actin. Additionally, rare cases of MPNST may contain either actin or desmin.” Conversely, cutaneous LMS may contain SlOO protein-like substances, or the Leu 7 antigen.13J4*51 We must emphasize, however, that the potential confusion caused by this immunophenotypic overlap between LMS and MPNST is the exception rather than the rule, because the diffuse coexpression of actin and desmin, as seen in well over 50% of LMS in our experience,51 is not a trait revealed by MPNST. Similarly, the coexpression of SlOO
The evaluation of vascular tumors in general has been emphasized elsewhere in this monograph. The salient conclusions that can be applied to an evaluation of spindle cell tumors are: 1) regardless of histologic appearance, tumors of endothelium are virtually always reactive with the UIex europaeus I agglutinin (UEA), although the substance (H antigen) recognized by this lectin is not exclusively expressed by endothelial tumors; 2) the endothelial-specific marker von Willebrand factor (vWF: factor VIII-related antigen) is identified in only half of benign and fewer than 20% of malignant endothelial neoplasms. Similarly, ultrastructural evaluation may reveal the characteristic organelles of endothelium, including the Weibel-Palade body, and abundant pinocytotic vesicles, but the majority of angiosarcomas do generally contain these in numbers sufficient for their easy recognition. How do these observations bear upon the various spindle cell endothelial tumors of skin? Among malignant spindle cell tumors, only the spindle cell angiosarcoma an unusual morphologic variant-exhibits overt vasogenesis, or forms intracytoplasmic lumens that contain red blood cells.1,5 In difficult cases, reliance on clinical clues (the predilection for the head and neck region; the bruise-like or angiomatous clinical appearance1s5) may be augmented by immunohistochemistry, in that staining with UEA, or less likely vWF, is accompanied by vimentin reactivity,12J4 but not by staining for epithelial, myogenie, or “neural” determinants.13J4 There have been sporadic reports of cytokeratin-like reactivity in vascular tumors, particularly those with an epithelioid morphology, 56,57but such findings are of uncertain significance at this time. As in MPNST, we consider cytokeratin immunostaining of vascular neoplasms to be a spurious phenomenon, one which is not reproduced in our experience. Nonetheless, attention to this possibility in one’s own practice may help to avoid unnecessary misinterpretation. The application of these techniques to the so-called spindle cell hemangioendothelioma (SCH),5*,59 a lesion peculiar to the skin of the extremities and trunk, has been detailed elsewhere. This tumor is sufficiently distinct histologically (with Kaposi’s sarcoma-like spindled areas
224
KAYE AND SWANSON
Clinics in Dermatology 1991;9:217-226
admixed with cavernous vascular spaces) to allow accurate diagnosis. 58,59Kaposi’s sarcoma, on the other hand, presents considerable problems in some settings, both for its spindled morphology in advanced stages, and for its reluctance in many laboratories to stain for either vWF or UEA.14 When predominantly a spindle cell tumor, Kaposi’s sarcoma may be recognized by the following features: tightly cellular clusters and fascicles; slit-like spaces between tumor cells that are occasionally filled with red blood cells; intracytoplasmic eosinophilic globules that in most cases represent phagolysosomes containing red cell fragments; hemorrhage or hemosiderin pigment; a lymphoplasmacytic inflammatory infiltrate; and dilated vascular channels at the periphery of the lesion.1*60-62 The ultrastructural bated,
phenotype
as endothelial,
of Kaposi’s
pericytic,
tures have been identified.’ features are clinical: location
sarcoma
is still de-
and myofibroblastic
fea-
To date, the most reliable on the extremities in elderly
clearly of greatest clinical concern, focus of additional study.
and should be the
References 1. Enzinger FM, Weiss SW. Soft Tissue Tumors, Second Edition, St. Louis, C.V. Mosby, 1988. 2. Martorelli M, Calabuig C, Peydro-Olaya A, et al. Fibroblast and myofibroblast participation in malignant fibrous histiocytoma (MFH) of bone. Ultrastructural study of eight cases with immunohistochemical support. Path01 Res Pratt 1989;184:582-90. 3. Frierson HF, Cooper PH. Myxoid variant of dermatofibrosarcoma protuberans. Am J Surg Path01 1983;7:445-50. 4. Enzinger FM. Atypical fibroxanthoma and malignant flbrous histiocytoma. Am J Dermatopathol 1979;1:185-92. 5. Manivel JC, Dehner LP, Wick MR. Nonvascular sarcomas of the skin. In Wick MR (ed), Pathology of Unusual Malignant Cutaneous Tumors. New York, Marcel Dekker, 1985, pp 211-79.
individuals; the epidemic association with HIV infection; and the presence of multiple synchronous and me-
6. Helwig EB. Atypical 59:664-67.
tachronous
7. Helwig EB, May D. Atypical fibroxanthoma of the skin with metastasis. Cancer 1986;57:368-76.
cutaneous,
in immunosuppressed
lymphnodal,
and visceral
lesions
patients.60*62
fibroxanthoma.
Tex J Med 1963;
8. Lanigan S, Gilkes J. Spectrum of atypical fibroxanthoma of the skin. J Royal Sot Med 1984;77:27-30.
Summary The most important aspect of any evaluation of spindle cell tumors in the skin or superficial soft tissues is the
clinical examination, as a great deal can be learned from the location, appearance, and size of the tumor in question. As recounted in this chapter, the histologic features of these tumors may also be distinctive; however, in some instances, histologic examination alone is insufficient for diagnosis. In such cases, electron microscopy holds considerable promise, but the technique is too dependent upon both the availability of adequately preserved tissues and access to the technique itself. As a result, immunohistochemistry remains the favored approach to most problematic lesions. In our experience, at least 90% of histologically enigmatic tumors will exhibit a characteristic immunophenotype, the remainder usually being indeterminant for a specific pattern of differentiation. The latter outcome is often the result of improper tissue preservation, but may also reflect the primitive nature of some neoplasms. Fortunately, the least common outcome is an ambiguous or “mixed-lineage” phenotype, in which neither one of two or more patterns of differentiation is resolved with certainty. The most common settings in which these problems arise are the separation of MPNST from LMS, and the recognition of melanocytic lesions as distinct from tumors of peripheral nerve sheath. The latter is
9. Ricci A Jr, Cartun RW, Zakowski MF. Atypical fibroxanthoma. A study of 14 cases emphasizing the presence of Langerhan’s histiocytes with implications for differential diagnosis by antibody panels. Am J Surg Path01 1988; 12:591-98. 10 Kuwano H, Hashimoto H, Enjoji M. Atypical fibroxanthoma distinguishable from spindle cell carcinoma in sarcoma-like skin lesions. Cancer 1985;55: 172 - 80.
11 Eusebi V, Ceccarelli C, Piscioli F, et al. Spindle-cell tumors of the skin of debatable origin: an immunocytochemical study. J Path01 1984;144:189-99.
12. Miettinen M, Lehto V-P, Virtanen I. Antibodies to intermediate filament proteins: The differential diagnosis of cutaneous tumors. Arch Dermatol 1985;121:736-41. 13. Silvis NG, Swanson PE, Manivel JC, et al. Spindle-cell and pleomorphic neoplasms of the skin. A clinicopathologic and immunohistochemical study of 30 cases, with emphasis on “atypical fibroxanthomas.” Am J Dermatopathol 1988;10:9- 19. 14. Wick MR, Kaye VN. The role of diagnostic immunohistochemistry in dermatology. Semin Dermatol 1986;5:34658. 15. du Boula y CEH. Demonstration of alpha- 1 -antitrypsin and alpha-l-antichymotrypsin in fibrous histiocytomas using the immunoperoxidase technique. Am J Surg Path01 1982;6:559-63. 16. Kindblom L, Jacobsen GK, Jacobsen M. Immunohistochemica1 investigations of tumors of supposed fibroblastic-histiocytic origin. Hum Path01 1982;13:834-40.
KAYE AND SWANSON
Clinics in Dermatology
1991;9:217-226
EVALUATION
225
OF SPINDLE CELL TUMORS
17. Harmon JB. Atypical fibroxanthoma and spindle cell squamous carcinomas of the head and neck: An immunohistochemical study. Lab Invest 1983;48:33A.
34. Gown AM, Vogel AM, Hoak D, et al. Monoclonal antibodies specific for melanocytic tumors distinguish subpopulations of melanocytes. Am J Path01 1986;123:195-203.
18. Nakanishi S, Hizawa K. Enzyme histochemical observation of fibrohistiocytic tumors. Acta Path01 Jpn 1984;34:100316.
35. Wick MR, Swanson PE, Rocamora A. Recognition of malignant melanoma by monoclonal antibody HMB45. J Cutan Path01 1988;15:201-07.
19. Leader M, Pate1 J, Collins M, Henry K: Anti-alpha-l-antichymotrypsin staining of 194 sarcomas, 38 carcinomas, and 17 malignant melanomas: Its lack of specificity as a tumor marker. Am J Surg Path01 1987;11:133-39.
36. Gallager RL, Helwig EB. Neurothekeoma: A benign cutaneous tumor of nerve sheath origin. Am J Clin Path01 1980;74:759-64.
20. Cerio R, Spaull J, Oliver GF, Wilson Jones E. A study of factor XIIIa and MAC387 immunolabeling in normal and pathological skin. Am J Dermatopathol 1990;12:221-33. 21. Gray MH, Smoller BR, McNutt NS, Hsu A. Immunohistochemical demonstration of factor XIIIa expression in neurofibromas. Arch Dermatol 199;126:427-76. 22. Battifora H. Spindle-cell carcinoma: Ultrastructural evidence of squamous origin and collagen production by the tumor cells. Cancer 1976;37:2275-82. 23. Evans HL, Smith JL. Spindle cell squamous carcinomas and sarcoma-like tumors of the skin: a comparative study of 38 cases. Cancer 1980;45:2687-97. 24. Woyke S, Domagala W, Olszewski W, Korabiec M. Pseudosarcoma of the skin: An electron microscopic study and comparison with the fine structure of the spindle-cell variant of squamous carcinoma. Cancer 1974;33:970-80. 25. Feldman I’S, Barr RJ. Ultrastructure of spindle-cell squamous carcinoma. J Cutan Path01 1976;3:17-24. 26. Penneys NS, Nadji M, Ziegels-Weissman J, Morales AR. Prekeratin in spindle cell tumours of the skin. Arch Dermato1 1983;119:476-79. 27. Pinkus GS, Kurtin PJ. Epithelial membrane antigen-A diagnostic discriminant in surgical pathology. Hum Path01 1985;16:929-40. 28. Jain S, Allen PW. Desmoplastic malignant melanoma and its variants: A study of 45 cases. Am J Surg Path01 1989;13:358-73. 29. Kossard S, Doherty E, Murray E. Neurotropic melanoma: A variant of desmoplastic melanoma. Arch Dermatol 1987;123:907-12. 30. From L, Hanna W, Kahn HJ, Baumal R. Origin of the desmoplasia in desmoplastic malignant melanoma. Hum Path01 1983;14:1072-80. 31. Suster S, Amazon K, Rosen LB, Ollague JM. Malignant epithelioid schwannoma of the skin. A low-grade neurotropic malignant melanoma? Am J Dermatopathol 1989;11:338-44. 32. Ordonez NG, Ji XL, Hickey RC. Comparison of HMB45 monoclonal antibody and SlOO protein in the immunohistochemical diagnosis of melanoma. Am J Clin Path01 1988;90:385-90. 33. Springall DR, Gu J, Cocchia D, et al. The value of SlOO immunostaining as a diagnostic tool in human malignant melanomas. Virchows Arch (Path01 Anat) 1983;400:33144.
37. Enzinger FM, Zhang R. Plexiform fibrohistiocytic tumor presenting in chiidren and young adults. An analysis of 65 cases. Am J Surg Path01 1988;12:818-26. 38. Pulitzer DR, Reed RJ. Nerve sheath myxoma. Am J Dermatopathol 1985;7:409-15. 39. George E, Swanson PE, Wick MR. Malignant peripheral nerve sheath tumors of the skin. Am J Dermatopathol 1989;11:213-21. 40. Dabski C, Reiman HM Jr, Muller SA. Neurofibrosarcoma of skin and subcutaneous tissue. Mayo Clin Proc 1990; 65:164-72. recogni41. Perentes E, Rubinstein L. Immunohistochemical tion of human nerve sheath tumors by anti-Leu 7(HNK-1) monoclonal antibody. Acta Neuropathol 1985;68:319-26. 42. Swanson PE, Manivel JC, Wick MR. Immunoreactivity for Leu 7 in neurofibrosarcoma and other spindle-cell soft tissue sarcomas. Am J Path01 1987;126:546-60. 43. Gould V, Moll R, Moll I, et al. The intermediate filament complement of the spectrum of nerve sheath neoplasms. Lab Invest 1986;55:463 - 71. 44. Kawahara E, Oda Y, Ooi A, et al. Expression of glial fibrillary acidic protein (GFAP) in peripheral nerve sheath tumors. Am J Surg Path01 1988;12:115-21. 45. Stanton C, Perentes E, Collins V, Rubinstein L. GFA protein reactivity in nerve sheath tumors: A polyvalent and monoclonal antibody study. J Neuropathol Exp Neurol 1987; 46:634-41. 46. Gray MH, Smoller BR, NcNutt NS, Hsu A. Neurofibromas and neurotized melanocytic nevi are immunohistochemitally distinct neoplasms. Am J Dermatopathol 1990;12: 234-41. of peripheral nerve 47. Miettinen M. Immunohistochemistry sheath sarcomas (abstr). Lab Invest 1990;62:67A. 48. Swanson PE, Manivel JC, Scheithauer BW, Wick MR. Epithelial membrane antigen reactivity in mesenchymal neoplasms: An immunohistochemical study of 306 soft tissue sarcomas. Surg Path01 1989;2:313-22. 49. Theaker JM, Gatter KC, Puddle J. Epithelial membrane antigen expression by the perineurium of peripheral nerve and in peripheral nerve tumours. Histopathol 1988; 13:171-79. 50. Fields JP, Helwig EB. Leiomyosarcoma of the skin and subcutaneous tissue. Cancer 1981;47:156-69. 51. Swanson PE, Stanley NW, Scheithauer BW, Wick MR. Primary cutaneous leiomyosarcoma: A histological and im-
226
KAYE AND SWANSON
Clinics in Dermatology 2991;9:22 7- 226
munohistochemical study of 9 cases, with ultrastructural correlation. J Cutan Path01 1988;15:129-41. 52. Azumi N, Ben-Ezra J, Battifora H. Immunophenotypic diagnosis of leiomyosarcomas and rhabdomyosarcomas with monoclonal antibodies to muscle-specific actin and desmin in formalin-fixed tissue. Mod Path01 1988;1:46974. 53. Leader M, Collins M, Pate1 J, Henry K. Desmin: Its value as a marker of muscle derived tumours using a commercial antibody. Virchows Arch (Path01 Anat) 1987;411:345-49. 54. Gown AM, Boyd HC, Chang Y, et al. Smooth muscle cells can express cytokeratins to “simple” epithelium. Immunocytochemical and biochemical studies in vitro and in vivo. Am J Path01 1988;132:223-32.
57. Eusebi V, Carcangiu ML, Dina R, Rosai J. Keratin-positive epithelioid angiosarcoma of thyroid. A report of four cases. Am J Surg Path01 1990;14:737-47. 58. Scott GA, Rosai J. Spindle cell hemangioendothelioma. Report of seven additional cases of a recently described vascular neoplasm. Am J Dermatopathol 1988;10:281-88. 59. Weiss SW, Enzinger FM. Spindle-cell hemangioendothelioma: A low-grade angiosarcoma resembling a cavernous hemangioma and Kaposi’s sarcoma. Am J Surg Path01 1986;10:521-30. 60. Jones RR, Pault J, Spay C, Wilson-Jones E. The histogenesis of Kaposi’s sarcoma in patients with and without AIDS. J Clin Path01 1986;39:742-49.
55. Miettinen M. Immunoreactivity for cytokeratin and epitheha1 membrane antigen in leiomyosarcoma. Arch Path01 Lab Med 1988;112:637-40.
61. Kao GF, Johnson FB, Sulica VI. The nature of hyaline (eosinophilic) globules and vascular slits of Kaposi’s sarcoma. Am J Dermatopathol 1990;12:256-67.
56. Gray MH, Rosenberg AE, Dickersin GR, Bhan AK. Cytokeratin expression in epithelioid vascular neoplasms. Hum Path01 1990;21:212-17.
61. Rusczczk Z, Mayer DA, Silva A, Organos CE. Kaposi’s sarcoma in AIDS. Multicentric angioneoplasia in early skin lesions. Am J Dermatopathol 1987;9:388-98.
enign and malignant neoplasms of skin that are composed entirely or predominantly of spindled cells are rather uncommon, but they generally constitute a significant component of the dermatopathologist’s practice, when ordinary squamous cell carcinomas, basal cell carcinomas, and melanocytic lesions are excluded. Differences among the various benign entities in this histologic spectrum are often subtle. Nonetheless, apart from those instances in which the diagnostic alternatives also include a malignant neoplasm, the clinical cliff erences that attend these lesions are relatively small. In contrast, the evaluation of malignant cutaneous spindle cell neoplasms is often an arduous task, because histologic features are not always sufficient to distinguish among the several clinically distinct tumors that comprise this diagnostic group, including: the fibrohistiocytic “malignancies” (atypical fibroxanthoma, dermatofibrosarcoma protuberans, and malignant fibrous histiocytoma); sarcomas (angiosarcoma, leiomyosarcoma, malignant peripheral nerve sheath tumor); spindle cell malignant melanoma; and spindle cell squamous cell carcinoma (Table 1). Despite confusing histologic similarities, most of these tumors differ significantly when immunohistochemical or electron microscopic features are considered. The latter approach may be useful in situations where light microscopic techniques are inconclusive, but access to these services is often limited. In contrast, the former technique is readily available to the practicing pathologist, using standard peroxidase-antiperoxidase (PAP) or avidin-biotin-peroxidase complex (ABC) methodology and commercially-available antibodies (Table 2). Further, the result-
B
From the Departments of Laboratory Medicine and Pathology and Dermatology (VNKJ, University of Minnesota School ofMedicine, Minneapolis, Minnesota; and Department of Pathology (PES), Washington University School of Medicine, St. Louis, Missouri. Address correspondence to: Paul E. Swanson, M.D., Division of Surgical Pathology, Barnes Hospital, 1 Barnes Hospital Plaza, St. Louis MO 63120.
0 2991 by Elsevier Science Publishing
Co., Inc.
l
0738-081x/91/$3.50
ant tumoral immunophenotypes have practical applications in most clinical situations (Table 3). Thus, it is this analytic approach that is emphasized in the ensuing discussion.
Fibrohistiocytic
Tumors
As a group, the benign and malignant fibrohistiocytic tumors are histologically diverse, but as their generic name implies, they are related by a common pattern of differentiation. For practical purposes, these tumors are proliferations of fibroblasts or facultative myofibroblasts, but in virtually all settings, they are accompanied by variable numbers of histiocytes.‘J From a diagnostic point of view, the most important of the fibrohistiocytic lesions are those malignant or borderline tumors in which a spindle cell population may predominate: dermatofibrosarcoma protuberans (DFSP) and malignant fibrous histiocytoma (MFH). The first of these is a clinically indolent lesion, usually affecting adults, that presents as a firm plaque or protuberant tumoral nodule on the truck or proximal extremity. Histologically, DFSP is composed of cytologically uniform cells arranged in a tightly interweaving “storiform” pattern. Mitotic activity is often observed, but atypical division figures do not predominate.’ Stromal myxoid changes may be conspicuous in some examples, a feature that may cause some confusion with peripheral nerve sheath tumors3 There is no obvious connection between the tumor and the overlying epidermis, although neoplastic elements may abut the surface, or may cause considerable epidermal thinning and ulceration. Lateral dermal and subcuticular borders of DFSP are often ill-defined, or infiltrated by tumoral cells arranged in a loosely woven lace-like pattern. Nevertheless, deep subcutaneous extension is not typical of this neoplasm. Excision of DFSP is generally curative, but local recurrence may occur.’ Malignant fibrous histiocytoma (MFH) contrasts histologically with DFSP in several respects, most notably the
217
218
KAYE
Clinics in Dermatology
AND SWANSON
1991;9:217-226 Table 1. Malignant or Borderline Spindle Cell Neoplasms
in Skin
Fibrohistiocytic: Malignant Fibrous Histiocytoma Dermatofibrosarcoma Protuberans Atypical Fibroxanthoma Other Mesenchymal: Malignant Peripheral Nerve Sheath Tumor Leiomyosarcoma Spindle Cell Hemangioendothelioma Angiosarcoma Kaposi’s Sarcoma Spindle Cell or Desmoplastic Malignant Melanoma Soindle Cell Souamous Cell Carcinoma
presence of considerable cellular pleomorphism, multinucleated tumor giant cells, and aggressively infiltrative deep soft-tissue margins. lj4f5 Compared to that seen in DFSP, mitotic activity is increased in MFH, and atypical mitotic forms may be evident. By convention, MFH is not a tumor of dermis; rather, it is a primary lesion of subcutis and deep soft tissues. Histologically similar lesions centered on the dermis are more properly categorized as atypical fibroxanthoma (AFX).1,6-9 AFX, as with other dermal lesions that resemble deep soft-tissue malignancies, is considerably more quiescent clinically, with a relatively
Table 2. lmmunoreaxents
small risk for recurrence or metastasis following conservative surgical management. 6,7Importantly, a dermal lesion showing extensive subcuticilar or fascial spread, tumoral necrosis, or vascular invasion, is at greater risk for aggressive behavior, and should be treated as MFH.’ In general, the pleomorphic elements in AFX and MFH separate these lesions from DFSP; nevertheless, such features do not always clearly distinguish fibrohistiocytic lesions from other spindled neoplasms of skin.1,5J0 In this context, electron microscopy, as alluded to above, may be of some utility, because, (myo)fibroblastic elements in these lesions are usually admixed with histiocyte-like cells.1,2*J Immunohistochemistry contributes to the resolution of these difficult cases largely by excluding other diagnostic alternatives. In other words, the elementary immunophenotype of fibrohistiocytic tumors-using the antibodies depicted in Table 3 -includes vimentin reactivity, but very little else.2*5,9J1-14 The extent of myofibroblastic differentiation exhibited by these tumors may dictate the presence and intensity of muscle-specific actin reactivity, but in most cases of AFX and MFH, staining with antibodies to this actin isoformis at best focal. DFSP, on the other hand, rarely if ever shows actin positivity.14 What then of the histiocytic elements? A great deal of attention has been paid to the presence of proteolytic
Commonly Used in the Evaluation
of Spindle Cell Tumors Source
Dilution
Becton-Dickinson Triton Biosciences Hybritech BioGenex BioGenex Enzo Becton-Dickinson Enzo Dako Dako
1:50 1:75 1:lOO 1:4000 1:2 1:20000 1:40 1: 16000 1:800 1:80
Dako Dako Dako
1:600 1:800 1: 2000
Vector
1: 2000
Reagent Monoclonal Antibodies: Anti-cytokeratin CAM 5.2 MAK6 AEl/AE3 Anti-vimentin (V9) Anti-desmin Anti-muscle-specific actin (HHF35) Leu7 HMB45 Anti-epithelial membrane antigen (E29) Anti-factor VIII-related antigen Polyclonal Antibodies anti-S100 protein anti-alpha-1-antichymotrypsin anti-Ulex europaeus I
Lectins: Ulex europaeus I Sources: Becton Dickinson-Mountain View, CA. Triton Biosciences-Alameda, CA. Hybritech-Alameda, CA. BioGenex Laboratories-San Rnmon, CA. Dako Laboratories-Carpinteria, CA. Vector Laboratories-Burlingme, CA. Enzo Biochemicals-New York, NY.
Clinics in Dermatology 1991:9:227-226
KAYE AND SWANSON EVALUATION
OF SPINDLE
219
CELL TUMORS
Table 3. ldealized lmmunophenotypes of Spindle Cell Tumors CK
EMA
HMB
SIOO
L7
F8
UEA
ACT
+/-*
0
0
0
0
0
0
0
+
0
0
0
0
0
+
+
0
0
+/-
+ + + +
+ 0 0 0
+ 0 0 0
+/0 0 0
+/0 0 0
0 0 0 +/-
+/0 0 +
+/0 0 +/-
0 +/+/0
0 + +/0
+/0 +/+/-
+/-
0
0
+
+
0
0
0
0
+/-
+/-
Neoplasm
VIM
DES
Fibrohistiocytic Tumors Malignant Peripheral Nerve Sheath Tumor Leiomyosarcoma Angiosarcarcoma Kaposi’s Sarcoma Spindle Cell Malignant Melanoma Spindle Cell Squamous Cell Carcinoma
+
0
+
MSA
Abbreviations: VIM = uimentin; DES = desmin; MSA = muscle-specific actin; CK = cytokeratin; EMA = epitheliul membrane antigen; HMB =antigen; SloO= SIOOprotein; L7 = Law 7 antigen; F8 = factor VIII-related antigen; LEA = Ulex europaeus I agglutinin; ACT = alpha-lantichymotrypsin. l “-t/-“denotes renctivity in 20-50% ofcases. Zfmore than 50% ore typically positivefor the stated determinant, “+“is used; iffewer than 20% stain, the diagnostic group is considered to be negutiue in this idealized scheme. Most cases designated “0” are rarely, if ever, reactive for that marker.
HMB45
enzymes, such as Cathepsin B and alpha-l-antichymotrypsin, as markers of histiocytic differentiation.1*5J3-1* Whereas subsequent analyses have demonstrated the relative frequency of such substances in fibrohistiocytic lesions, they have also revealed their lack of specificity for this pattern of cellular specialization.19 Indeed, virtually all spindle cell tumors of skin occasionally stain with antibodies to these enzymes, as may a variety of malignant tumors in other sites.13 Recently, Cerio and colleagues have studied the distribution of factor XIIIa in cutaneous neoplasms. Notably, most examples of MFH and AFX are labeled with antibodies to this factor.20 However, recent evidence identifies factor XIIIa as a sensitive marker for neurofibromas as we11.21 Nevertheless, the presence of this substance together with the aforementioned enzymes, in the absence of the lineage-selective markers yet to be discussed, may Figure 1. Fibrohistiocytic Neoplasms. Spindled pleomorphic cells in superficial malignant fibrous are reactive for vimentin (4). Histiocytic elements fibroxanthoma stain for alpha-1-antichymotypsin
and histiocytoma in atypical (B).
provide presumptive evidence that the spindled neoplasm under investigation is a “fibrohistiocytic” tumor.
Spindle Cell Squamous Carcinoma Although the spindled variant of squamous carcinoma is uncommon, it poses a diagnostic challenge that leaves most of us wishing it never occured at all. The clinical appearance of spindle cell squamous carcinoma (SCSC) may resemble that of any other form of squamous carcinoma, with ulceration, excoriation and hyperkeratosis, but its ability to “drop off” the epidermis makes it possible for SCSC to present as a lesion with little obvious epidermal damage. Histologically, SCSC arises from epidermis, and appears to stream into superficial dermis. In lesions with epidermal ulceration, however, the association between epidermis and tumor may be lost, and the distinction between SCSC and other spindle neoplasms is often problematic. 10,22-24Nonetheless, the recognition of SCSC is important, as local recurrence and metatasis are relatively uncommon, compared to histologically similar melanocytic and fibrohistiocytic malignancies.1*23 The electron microscopic appearance of SCSC is often definitive, because true epithelial features are generally preserved. As in the more typical examples of squamous cell carcinoma, the cells of SCSC contain bundles of tonofilaments that insert into desmosomes or desmosomelike cell junctions.22-25 Immunohistochemistry is usually reliable as well, and for practical purposes obviates the need for electron microscopy. The immunophenotype of SCSC is reproducible in most instances, and is distinct among spindle cell tumors of skin: reactivity for both cytokeratin (CK) epithelial membrane antigen (EMA) is readily demonstrated. 11-14~17,26*27 The latter determinant, although somewhat less commonly expressed than CK, is virtually specific for epithelial differentiation in cutane-
220
KAYE AND SWANSON
Clinics in Dermatology 1991;9:22 7- 226
or acral lentiginous melanoma, and thus are often the anticipated sequalae of previously documented pigmented lesions.28*29 Examples of SMM that arise de novo are less readily recognized, especially in cases where pigmentation is sparse. As in SCSC, SMM may have an obvious association with the overlying epidermis, but in such cases, a lentiginous melanotic lesion is usually present.*s The cells of SMM are cytologically atypical, and may be conspicuously amelanotic. They exhibit a diffuse pattern of dermal or subcuticular spread, and generally are not arranged in fascicles or a distinct storiform pattern. Mitotic activity is present at all levels of the lesion, in contrast to the benign spindle-cell melanocytic lesions. Stroma is often quite fibrous or desmoplastic,28*30 and striking neurotropism may be observed. *s~*~The latter trait, among others, has formed the basis for comparisons of SMM to tumors of peripheral nerves. The validity of this comparison may be argued, but such a debate exceeds the scope of this discussion. As in SCSC, the electron microscopy of SMM is often diagnostic. The cardinal feature is the premelanosome; despite the lack of obvious pigment in many examples,
Figure 2. Spindle Cell Squamous Carcinoma. Cytokeratin reactivity is usually conspicuous in the majority of neoplastic cells.
ous neoplasms, as only rare examples of smooth muscle or peripheral nerve sheath tumors are EMA-reactive, whereas melanocytic lesions are only anecdotally positive. What leads to confusion in cases of SCSC is the variable presence of vimentin,12-*4 particularly when CK or EMA are only focally present.i4 In such cases, careful histologic and immunohistochemical evaluation of all available material may be necessary. Unfortunately, as with the other neoplasms discussed herein, not all examples can be diagnosed with certainty.
Spindle Cell (Desmoplastic) Malignant Melanoma The phenotypic complexity and diagnostic evaluation of melanocytic lesions are covered in greater detail elsewhere in this monograph. Hence, the investigation of only the overtly malignant spindle-cell melanocytic lesions will be described here. Clinical clues are often instrumental in the accurate recognition of spindled malignant melanoma (SMM). Most examples occur in association with lentigo maligna
Figure 3. Spindle Cell Malignant Melanoma. spindled cells closely appose epidermis.
SlOO-reactive
Clinics in Dermatology 1991;9:217- 226 these organelles may occasionally be abundant. As most dermatopathologists know, however, there are as many cases where premelanosomes defy detection. The utility of immunohistochemistry is well documented, particularly in efforts to separate SMM from SCSC. Unlike the latter tumor, SMM does not express either CK or EMA,‘*J4 but instead contains SlOO protein.*1J3J4,32,33 However, the latter substance is found in several other cell types, among them chondrocytes, adipocytes, Schwann cells, and melanocytes. Hence, although SlOO protein is a sensitive marker for melanocytic differentiation, it is by no means specific. Peripheral nerve sheath tumors are typically reactive for SlOO, whereas occasional examples of dermal leiomyosarcoma and SCSC may also contain this protein.13J4 Other markers have been described that are selective for neoplastic melanocytes; the most important of these is the HMB-45 antigen. In most forms of malignant melanoma, this determinant proves to be a specific tumor marker. Unfortunately, few examples of SMM are immunoreactive for this substance.32*34*35 Hence, the diagnosis of SMM in difficult cases relies on the clinical, histologic, and immunohistochemical exclusion of epithelial, myogenie, and neurilemmal differentiation.
Peripheral Nerve Sheath Tumors The most commonly encountered peripheral nerve sheath tumor in skin is the neurofibroma, a tumor that may occupy either the dermis or subcutis in virtually any anatomic location. Histologically, the uniformity of the spindle cells, together with “wavy“ nuclear and plasmalemma1 profiles, are compelling evidence of nerve sheath differentiation.’ They may be variably cellular, and may be dominated by a myxoid stroma. However, in cutaneous lesions, cellular pleomorphism is generally slight or absent, and mitotic activity is uncommon. Loose, infihrative dermal margins may be seen in some instances, and not all such tumors are easily distinguished from benign fibrohistiocytic neoplasms.’ Schwannomas (neurilemmomas) may also arise in superficial soft tissues; unlike neurofibromas, they are more variegated in their histologic appearance. Classically, schwannomas are composed of two dominant patterns, the first consisting of spindle cells showing nuclear palisading or neurotactoid differentiation (Antoni A), the second containing similar cells that are more loosely dispersed in a myxoid or degenerating stroma.’ Cystic change and hemorrhage may be conspicuous in these latter Antoni B foci. As in neurofibroma, schwannomas are not mitotically active, and except in tumors with extensive degeneration, nuclear pleomorphism and cytologic atypia are lacking.’
KAYE AND SWANSON EVALUATION OF SPINDLE CELL TUMORS
221
Figure 4. Malignant Peripheral Nerve Sheath Tumor. Both SlOOprotein (A) and Leu 7 antigen @) are detected in the majority of these neoplasms.
Among the benign nerve sheath tumors of skin, it is perhaps the neurothekeoma that evokes the greatest concern for malignancy. 1,36Composed of irregular nodules or lobules separated by distinct fibrous (perineurial?) septae, neurothekeoma may assume a plexiform growth pattern similar to that of both plexiform fibrous histiocytoma3’ and plexiform neurofibroma.’ However, in contrast to both of these alternatives, the cells of neurothekeoma are arranged in small clusters or whorls; cells may be ovoid or spindled, and varying amounts of myxoid stroma may be present. 1~36The term “dermal nerve sheath myxoma” has been applied to cases where the latter features predominate.38 Cytologically, neurothekeoma also differs from other benign nerve sheath tumors, in that nuclear atypia may be focally pronounced, and mitotic activity may be observed. Nonetheless, stromal or capsular invasion, perineurial spread, and vascular permeation are not features of this tumor. There has been some debate regarding the true nature of neurothekeoma, with some authors arguing for a perineurial, as opposed to a schwannian pattern of differentiation.1,36,3* In our experience, however, the ultrastructural and immunohistochemical features of these lesions are similar to those of other nerve sheath tumors, as will be described. In contrast to the relative frequency of benign nerve sheath tumors in skin, malignant peripheral nerve sheath tumors (MPNST) are quite uncommon.1*5 We have analyzed two such neoplasms,39 and a small series has recently been reported. 4o Unlike the benign lesions in which they may arise, MPNST are cellular neoplasms, composed of cytologically atypical and mitotically active spindled, pleomorphic, or epithelioid elements. Often, the predominant pattern is that of interweaving cell bun-
222
KAYE AND SWANSON
dles or a storiform pattern, not unlike that seen in MFH or fibrosarcoma. Fewer than one-third of all cutaneous and deep soft-tissue MPNST exhibit the histologic features of benign schwannoma. lp5One characteristic of MPNST not shared by histologically similar neoplasms is a propensity for divergent chondro-osseous, rhabdomyoblastic, endothelial, or glandular differentiation.’ However, this phenomenon has not characterized cutaneous MPNST and will not be emphasized here. Despite the lack of distinctive histologic features in many cases, MPNST shares with benign nerve sheath neoplasms the ultrastructural features of Schwann cell differentiation.’ In particular, these tumors, to varying degrees, may elaborate numerous cell processes that interdigitate, and form desmosome-like cell junctions at points of apposition. In benign tumors, these processes may encircle bundles of collagen in a manner reminiscent of the mesaxon of the normal Schwann cell. Basal lamina is typically present, although it may be absent in some MPNST. Perineurial elements consist of fibroblast-like cells with delicate cell processes that are often intimately associated with mast cells, The latter cells are often present in abundance in benign and malignant nerve sheath tumors, although this feature alone is of little diagnostic significance. In neurifibromas, cells with perineurial features may be abundant, but otherwise typical fibroblasts and Schwann cells are more commonly seen. In schwannoma, perineurial and fibroblastic elements are sparse or absent.’ Immunohistochemically, MPNST and benign nerve sheath tumors share the expression of SlOO protein, as well as two other substances associated with Schwann cell differentiation: Leu 7 antigen and myelin basic protein.13J4,41,42 Leu 7 is an antigenic determinant found on a variety of proteins, including a myelin-associated glycoprotein that has cell adhesion properties. Myelin basic protein is a well-characterized constituent of the myelin sheath. Although its presence has been questioned in cutaneous MPNST, we have found it in over one-third of deeply situated examples. 14,42As noted previously, S 100 is also expressed by melanocytic lesions, but the combination of SlOO protein and Leu 7 or myelin basic protein reactivity in a non-epithelial spindle cell tumor is compelling evidence of nerve sheath differentiation.42 Vimentin is uniformly expressed by nerve sheath lesions,‘* and the expression of glial fibrillary acidic protein in occasional benign and malignant PNST appears to be a unique trait among spindle cell neoplasms of skin.43-45 Among benign tumors, neurofibromas are somewhat less likely to express either SlOO or Leu 7, but their presence may be of particular importance in the occasional instance where neurotization of a dermal nevus has imparted a spindled appearance.14*46 In the latter context,
Clinics in Dermatology 1991;9:217-226 neither SlOO nor Leu 7 antigen is typically detected.46 The potential relevence of factor XIIIa staining to this differential diagnosis is discussed elsewhere.21 Finally, there have been occasional reports of both cytokeratin” and epithelial membrane antigen*a reactivity in nerve sheath tumors. The former, we believe, is spurious, but may nonetheless complicate the immunohistochemical diagnosis of MPNST. In contrast, the presence of EMA is an expected finding in some cases, as perineurial cells are normally reactive for this substance.49
Smooth Muscle Tumors Benign smooth muscle tumors of the dermis and subcutis are lesions that often present as solitary, firm, domeshaped nodules without epidermal ulceration, These lesions are uncommon among all age groups, and probably represent the least common group of benign mesenchyma1 neoplasms in skin.’ Histologically, leiomyomas are composed of fascicles of cytologically uniform fusiform or ovoid cells with abundant eosinophilic cytoplasm, and centrally located “cigar-shaped’ nuclei. Mitotic activity is not conspicuous, and cellular pleomorphism is minimal. These fascicles may form nodules or lobules that are related to or centered on pilosebaceous units, or may form angiomyomatous nodules in the superficial subcutis. Excision of such lesions is generally curative.’ Leiomyosarcomas (LMS) uncommonly arise in the dermis or subcutaneous tissue, and are distinguished from benign smooth muscle tumors by their larger size, their tendency to infiltrate adjacent dermis or subcutis, and their propensity for increased cytologic atypia (which may be quite pronounced), mitotic activity, and tumoral necrosis.1,5,50,51 The second of these as a marker for malignancy has been challenged in extracutaneous smooth muscle neoplasms, but the general consensus remains that superficial lesions with 2 or more mitotic figures per high-power field should be considered malignant.1~50~51 On rare occasion, vascular invasion may be seen; this feature is more typical of subcutaneous-based lesions, and portends a less favorable outcome.1*50 As Fields and Helwig have emphasized, the clinical behavior of superficial LMS depends in large part on the location of the tumor in skin.50 Lesions centered on, and largely limited to, the dermis rarely recur or metastasize, whereas deep dermal or subcuticular lesions more commonly show such behavior. Metastasis is rare in either setting when adequate surgical excision is performed.1~50~51 Although tumors dominated by epithelioid or polygonal cells have proven to be smooth muscle in nature, most examples of LMS are spindled in appearance.1~5~50~51 The architectural arrangement of these lesions, the lack of
Clinics in Dermatology 1991;9:217- 226
KAYE AND SWANSON
223
EVALUATION OF SPINDLE CELL TUMORS
protein and Leu 7 antigen is the expected phenotype of MPNST,42 and is a highly unlikely outcome in LMS. Leiomyosarcoma, like most mesenchymal neoplasms, contains vimentin, and occasional examples may express cytokeratins as well .54,55Nonetheless, the reproducibility of a myogenic (actin- and desmin-positive) phenotype in LMS provides a reliable contrast to spindle cell squamous carcinoma and other cytokeratin-containing neoplasms.
Vasogenic Neoplasms
A Figure 5. Leiomyosarcoma. Unlike other malignant spindled tumors of skin, leiomyosarcoma may be diffusely reactive for desmin (A) and muscle-specific actin 0. overt neurotropism, and their cytologic features often are sufficient for their accurate diagnosis, but as in all other settings described in this review, the occasional LMS presents a diagnostic challenge. Electron microscopy is particularly useful, as the contractile nature of these cells is readily demonstrated by this technique. Typically, cells are elongate, their cytoplasm containing little rough endoplasmic reticulum (in contrast to the (myo)fibroblastic elements of fibrohistiocytic lesions, in which such organelles are abundant), but instead is filled by bundles of microfilaments interrupted sporadically by dense bodies. Plasmalemmal plaques are evident, as is pinocytosis. The relative abundance of these structures is in part dictated by the degree of differentiation exhibited by the tumor.1*51 Immunohistochemistry is equally efficient at detecting myogenic differentiation, although there are potential pitfalls in interpretation. As expected, most smooth muscle tumors are immunoreactive for muscle-specific actin1*14~51and the intermediate filament desmin.12-14~51~52 Desmin is virtually always present in LMS, in our experience, but has been found less often by other investigators.52,53 Staining for desmin must be remembered in any evaluation of spindle cell tumors, because, as already described, fibrohistiocytic neoplasms may also occasionally exhibit focal staining for muscle-specific actin. Additionally, rare cases of MPNST may contain either actin or desmin.” Conversely, cutaneous LMS may contain SlOO protein-like substances, or the Leu 7 antigen.13J4*51 We must emphasize, however, that the potential confusion caused by this immunophenotypic overlap between LMS and MPNST is the exception rather than the rule, because the diffuse coexpression of actin and desmin, as seen in well over 50% of LMS in our experience,51 is not a trait revealed by MPNST. Similarly, the coexpression of SlOO
The evaluation of vascular tumors in general has been emphasized elsewhere in this monograph. The salient conclusions that can be applied to an evaluation of spindle cell tumors are: 1) regardless of histologic appearance, tumors of endothelium are virtually always reactive with the UIex europaeus I agglutinin (UEA), although the substance (H antigen) recognized by this lectin is not exclusively expressed by endothelial tumors; 2) the endothelial-specific marker von Willebrand factor (vWF: factor VIII-related antigen) is identified in only half of benign and fewer than 20% of malignant endothelial neoplasms. Similarly, ultrastructural evaluation may reveal the characteristic organelles of endothelium, including the Weibel-Palade body, and abundant pinocytotic vesicles, but the majority of angiosarcomas do generally contain these in numbers sufficient for their easy recognition. How do these observations bear upon the various spindle cell endothelial tumors of skin? Among malignant spindle cell tumors, only the spindle cell angiosarcoma an unusual morphologic variant-exhibits overt vasogenesis, or forms intracytoplasmic lumens that contain red blood cells.1,5 In difficult cases, reliance on clinical clues (the predilection for the head and neck region; the bruise-like or angiomatous clinical appearance1s5) may be augmented by immunohistochemistry, in that staining with UEA, or less likely vWF, is accompanied by vimentin reactivity,12J4 but not by staining for epithelial, myogenie, or “neural” determinants.13J4 There have been sporadic reports of cytokeratin-like reactivity in vascular tumors, particularly those with an epithelioid morphology, 56,57but such findings are of uncertain significance at this time. As in MPNST, we consider cytokeratin immunostaining of vascular neoplasms to be a spurious phenomenon, one which is not reproduced in our experience. Nonetheless, attention to this possibility in one’s own practice may help to avoid unnecessary misinterpretation. The application of these techniques to the so-called spindle cell hemangioendothelioma (SCH),5*,59 a lesion peculiar to the skin of the extremities and trunk, has been detailed elsewhere. This tumor is sufficiently distinct histologically (with Kaposi’s sarcoma-like spindled areas
224
KAYE AND SWANSON
Clinics in Dermatology 1991;9:217-226
admixed with cavernous vascular spaces) to allow accurate diagnosis. 58,59Kaposi’s sarcoma, on the other hand, presents considerable problems in some settings, both for its spindled morphology in advanced stages, and for its reluctance in many laboratories to stain for either vWF or UEA.14 When predominantly a spindle cell tumor, Kaposi’s sarcoma may be recognized by the following features: tightly cellular clusters and fascicles; slit-like spaces between tumor cells that are occasionally filled with red blood cells; intracytoplasmic eosinophilic globules that in most cases represent phagolysosomes containing red cell fragments; hemorrhage or hemosiderin pigment; a lymphoplasmacytic inflammatory infiltrate; and dilated vascular channels at the periphery of the lesion.1*60-62 The ultrastructural bated,
phenotype
as endothelial,
of Kaposi’s
pericytic,
tures have been identified.’ features are clinical: location
sarcoma
is still de-
and myofibroblastic
fea-
To date, the most reliable on the extremities in elderly
clearly of greatest clinical concern, focus of additional study.
and should be the
References 1. Enzinger FM, Weiss SW. Soft Tissue Tumors, Second Edition, St. Louis, C.V. Mosby, 1988. 2. Martorelli M, Calabuig C, Peydro-Olaya A, et al. Fibroblast and myofibroblast participation in malignant fibrous histiocytoma (MFH) of bone. Ultrastructural study of eight cases with immunohistochemical support. Path01 Res Pratt 1989;184:582-90. 3. Frierson HF, Cooper PH. Myxoid variant of dermatofibrosarcoma protuberans. Am J Surg Path01 1983;7:445-50. 4. Enzinger FM. Atypical fibroxanthoma and malignant flbrous histiocytoma. Am J Dermatopathol 1979;1:185-92. 5. Manivel JC, Dehner LP, Wick MR. Nonvascular sarcomas of the skin. In Wick MR (ed), Pathology of Unusual Malignant Cutaneous Tumors. New York, Marcel Dekker, 1985, pp 211-79.
individuals; the epidemic association with HIV infection; and the presence of multiple synchronous and me-
6. Helwig EB. Atypical 59:664-67.
tachronous
7. Helwig EB, May D. Atypical fibroxanthoma of the skin with metastasis. Cancer 1986;57:368-76.
cutaneous,
in immunosuppressed
lymphnodal,
and visceral
lesions
patients.60*62
fibroxanthoma.
Tex J Med 1963;
8. Lanigan S, Gilkes J. Spectrum of atypical fibroxanthoma of the skin. J Royal Sot Med 1984;77:27-30.
Summary The most important aspect of any evaluation of spindle cell tumors in the skin or superficial soft tissues is the
clinical examination, as a great deal can be learned from the location, appearance, and size of the tumor in question. As recounted in this chapter, the histologic features of these tumors may also be distinctive; however, in some instances, histologic examination alone is insufficient for diagnosis. In such cases, electron microscopy holds considerable promise, but the technique is too dependent upon both the availability of adequately preserved tissues and access to the technique itself. As a result, immunohistochemistry remains the favored approach to most problematic lesions. In our experience, at least 90% of histologically enigmatic tumors will exhibit a characteristic immunophenotype, the remainder usually being indeterminant for a specific pattern of differentiation. The latter outcome is often the result of improper tissue preservation, but may also reflect the primitive nature of some neoplasms. Fortunately, the least common outcome is an ambiguous or “mixed-lineage” phenotype, in which neither one of two or more patterns of differentiation is resolved with certainty. The most common settings in which these problems arise are the separation of MPNST from LMS, and the recognition of melanocytic lesions as distinct from tumors of peripheral nerve sheath. The latter is
9. Ricci A Jr, Cartun RW, Zakowski MF. Atypical fibroxanthoma. A study of 14 cases emphasizing the presence of Langerhan’s histiocytes with implications for differential diagnosis by antibody panels. Am J Surg Path01 1988; 12:591-98. 10 Kuwano H, Hashimoto H, Enjoji M. Atypical fibroxanthoma distinguishable from spindle cell carcinoma in sarcoma-like skin lesions. Cancer 1985;55: 172 - 80.
11 Eusebi V, Ceccarelli C, Piscioli F, et al. Spindle-cell tumors of the skin of debatable origin: an immunocytochemical study. J Path01 1984;144:189-99.
12. Miettinen M, Lehto V-P, Virtanen I. Antibodies to intermediate filament proteins: The differential diagnosis of cutaneous tumors. Arch Dermatol 1985;121:736-41. 13. Silvis NG, Swanson PE, Manivel JC, et al. Spindle-cell and pleomorphic neoplasms of the skin. A clinicopathologic and immunohistochemical study of 30 cases, with emphasis on “atypical fibroxanthomas.” Am J Dermatopathol 1988;10:9- 19. 14. Wick MR, Kaye VN. The role of diagnostic immunohistochemistry in dermatology. Semin Dermatol 1986;5:34658. 15. du Boula y CEH. Demonstration of alpha- 1 -antitrypsin and alpha-l-antichymotrypsin in fibrous histiocytomas using the immunoperoxidase technique. Am J Surg Path01 1982;6:559-63. 16. Kindblom L, Jacobsen GK, Jacobsen M. Immunohistochemica1 investigations of tumors of supposed fibroblastic-histiocytic origin. Hum Path01 1982;13:834-40.
KAYE AND SWANSON
Clinics in Dermatology
1991;9:217-226
EVALUATION
225
OF SPINDLE CELL TUMORS
17. Harmon JB. Atypical fibroxanthoma and spindle cell squamous carcinomas of the head and neck: An immunohistochemical study. Lab Invest 1983;48:33A.
34. Gown AM, Vogel AM, Hoak D, et al. Monoclonal antibodies specific for melanocytic tumors distinguish subpopulations of melanocytes. Am J Path01 1986;123:195-203.
18. Nakanishi S, Hizawa K. Enzyme histochemical observation of fibrohistiocytic tumors. Acta Path01 Jpn 1984;34:100316.
35. Wick MR, Swanson PE, Rocamora A. Recognition of malignant melanoma by monoclonal antibody HMB45. J Cutan Path01 1988;15:201-07.
19. Leader M, Pate1 J, Collins M, Henry K: Anti-alpha-l-antichymotrypsin staining of 194 sarcomas, 38 carcinomas, and 17 malignant melanomas: Its lack of specificity as a tumor marker. Am J Surg Path01 1987;11:133-39.
36. Gallager RL, Helwig EB. Neurothekeoma: A benign cutaneous tumor of nerve sheath origin. Am J Clin Path01 1980;74:759-64.
20. Cerio R, Spaull J, Oliver GF, Wilson Jones E. A study of factor XIIIa and MAC387 immunolabeling in normal and pathological skin. Am J Dermatopathol 1990;12:221-33. 21. Gray MH, Smoller BR, McNutt NS, Hsu A. Immunohistochemical demonstration of factor XIIIa expression in neurofibromas. Arch Dermatol 199;126:427-76. 22. Battifora H. Spindle-cell carcinoma: Ultrastructural evidence of squamous origin and collagen production by the tumor cells. Cancer 1976;37:2275-82. 23. Evans HL, Smith JL. Spindle cell squamous carcinomas and sarcoma-like tumors of the skin: a comparative study of 38 cases. Cancer 1980;45:2687-97. 24. Woyke S, Domagala W, Olszewski W, Korabiec M. Pseudosarcoma of the skin: An electron microscopic study and comparison with the fine structure of the spindle-cell variant of squamous carcinoma. Cancer 1974;33:970-80. 25. Feldman I’S, Barr RJ. Ultrastructure of spindle-cell squamous carcinoma. J Cutan Path01 1976;3:17-24. 26. Penneys NS, Nadji M, Ziegels-Weissman J, Morales AR. Prekeratin in spindle cell tumours of the skin. Arch Dermato1 1983;119:476-79. 27. Pinkus GS, Kurtin PJ. Epithelial membrane antigen-A diagnostic discriminant in surgical pathology. Hum Path01 1985;16:929-40. 28. Jain S, Allen PW. Desmoplastic malignant melanoma and its variants: A study of 45 cases. Am J Surg Path01 1989;13:358-73. 29. Kossard S, Doherty E, Murray E. Neurotropic melanoma: A variant of desmoplastic melanoma. Arch Dermatol 1987;123:907-12. 30. From L, Hanna W, Kahn HJ, Baumal R. Origin of the desmoplasia in desmoplastic malignant melanoma. Hum Path01 1983;14:1072-80. 31. Suster S, Amazon K, Rosen LB, Ollague JM. Malignant epithelioid schwannoma of the skin. A low-grade neurotropic malignant melanoma? Am J Dermatopathol 1989;11:338-44. 32. Ordonez NG, Ji XL, Hickey RC. Comparison of HMB45 monoclonal antibody and SlOO protein in the immunohistochemical diagnosis of melanoma. Am J Clin Path01 1988;90:385-90. 33. Springall DR, Gu J, Cocchia D, et al. The value of SlOO immunostaining as a diagnostic tool in human malignant melanomas. Virchows Arch (Path01 Anat) 1983;400:33144.
37. Enzinger FM, Zhang R. Plexiform fibrohistiocytic tumor presenting in chiidren and young adults. An analysis of 65 cases. Am J Surg Path01 1988;12:818-26. 38. Pulitzer DR, Reed RJ. Nerve sheath myxoma. Am J Dermatopathol 1985;7:409-15. 39. George E, Swanson PE, Wick MR. Malignant peripheral nerve sheath tumors of the skin. Am J Dermatopathol 1989;11:213-21. 40. Dabski C, Reiman HM Jr, Muller SA. Neurofibrosarcoma of skin and subcutaneous tissue. Mayo Clin Proc 1990; 65:164-72. recogni41. Perentes E, Rubinstein L. Immunohistochemical tion of human nerve sheath tumors by anti-Leu 7(HNK-1) monoclonal antibody. Acta Neuropathol 1985;68:319-26. 42. Swanson PE, Manivel JC, Wick MR. Immunoreactivity for Leu 7 in neurofibrosarcoma and other spindle-cell soft tissue sarcomas. Am J Path01 1987;126:546-60. 43. Gould V, Moll R, Moll I, et al. The intermediate filament complement of the spectrum of nerve sheath neoplasms. Lab Invest 1986;55:463 - 71. 44. Kawahara E, Oda Y, Ooi A, et al. Expression of glial fibrillary acidic protein (GFAP) in peripheral nerve sheath tumors. Am J Surg Path01 1988;12:115-21. 45. Stanton C, Perentes E, Collins V, Rubinstein L. GFA protein reactivity in nerve sheath tumors: A polyvalent and monoclonal antibody study. J Neuropathol Exp Neurol 1987; 46:634-41. 46. Gray MH, Smoller BR, NcNutt NS, Hsu A. Neurofibromas and neurotized melanocytic nevi are immunohistochemitally distinct neoplasms. Am J Dermatopathol 1990;12: 234-41. of peripheral nerve 47. Miettinen M. Immunohistochemistry sheath sarcomas (abstr). Lab Invest 1990;62:67A. 48. Swanson PE, Manivel JC, Scheithauer BW, Wick MR. Epithelial membrane antigen reactivity in mesenchymal neoplasms: An immunohistochemical study of 306 soft tissue sarcomas. Surg Path01 1989;2:313-22. 49. Theaker JM, Gatter KC, Puddle J. Epithelial membrane antigen expression by the perineurium of peripheral nerve and in peripheral nerve tumours. Histopathol 1988; 13:171-79. 50. Fields JP, Helwig EB. Leiomyosarcoma of the skin and subcutaneous tissue. Cancer 1981;47:156-69. 51. Swanson PE, Stanley NW, Scheithauer BW, Wick MR. Primary cutaneous leiomyosarcoma: A histological and im-
226
KAYE AND SWANSON
Clinics in Dermatology 2991;9:22 7- 226
munohistochemical study of 9 cases, with ultrastructural correlation. J Cutan Path01 1988;15:129-41. 52. Azumi N, Ben-Ezra J, Battifora H. Immunophenotypic diagnosis of leiomyosarcomas and rhabdomyosarcomas with monoclonal antibodies to muscle-specific actin and desmin in formalin-fixed tissue. Mod Path01 1988;1:46974. 53. Leader M, Collins M, Pate1 J, Henry K. Desmin: Its value as a marker of muscle derived tumours using a commercial antibody. Virchows Arch (Path01 Anat) 1987;411:345-49. 54. Gown AM, Boyd HC, Chang Y, et al. Smooth muscle cells can express cytokeratins to “simple” epithelium. Immunocytochemical and biochemical studies in vitro and in vivo. Am J Path01 1988;132:223-32.
57. Eusebi V, Carcangiu ML, Dina R, Rosai J. Keratin-positive epithelioid angiosarcoma of thyroid. A report of four cases. Am J Surg Path01 1990;14:737-47. 58. Scott GA, Rosai J. Spindle cell hemangioendothelioma. Report of seven additional cases of a recently described vascular neoplasm. Am J Dermatopathol 1988;10:281-88. 59. Weiss SW, Enzinger FM. Spindle-cell hemangioendothelioma: A low-grade angiosarcoma resembling a cavernous hemangioma and Kaposi’s sarcoma. Am J Surg Path01 1986;10:521-30. 60. Jones RR, Pault J, Spay C, Wilson-Jones E. The histogenesis of Kaposi’s sarcoma in patients with and without AIDS. J Clin Path01 1986;39:742-49.
55. Miettinen M. Immunoreactivity for cytokeratin and epitheha1 membrane antigen in leiomyosarcoma. Arch Path01 Lab Med 1988;112:637-40.
61. Kao GF, Johnson FB, Sulica VI. The nature of hyaline (eosinophilic) globules and vascular slits of Kaposi’s sarcoma. Am J Dermatopathol 1990;12:256-67.
56. Gray MH, Rosenberg AE, Dickersin GR, Bhan AK. Cytokeratin expression in epithelioid vascular neoplasms. Hum Path01 1990;21:212-17.
61. Rusczczk Z, Mayer DA, Silva A, Organos CE. Kaposi’s sarcoma in AIDS. Multicentric angioneoplasia in early skin lesions. Am J Dermatopathol 1987;9:388-98.
