Search

Pseudomyogenic hemangioendothelioma (PHE) is an uncommon neoplasm with propensity for local recurrence. The tumor mimics epithelioid hemangioendothelioma and epithelioid sarcoma, representing a possible diagnostic pitfall. We investigated the clinicopathological, immunohistochemical, and fluorescence in situ hybridization features of PHEs. Eight cases of PHE were retrieved from our pathology archives. The clinical and outcome information was available in 6 patients. In 6 cases, the tumors were located in the lower limb, whereas the upper limb was involved in 2 cases. Three patients had exclusively bone involvement, and 2 patients had cutaneous tumor only. The tumor was multifocal in 5 cases, of which 2 patients had simultaneous bone and soft tissue involvement. Immunohistochemical stains revealed all tumors to be positive for AE1/AE3 (8/8), ERG (7/7), and FLI-1 (7/7) with preserved INI1 (7/7) expression. Most of the tumors were variably positive for CD31 (7/8), CAM5.2 (2/3), and SMA (4/6). t(7;19)(q22;q13) fusion pattern was detected in 3 cases (5 tumors) with cutaneous and multifocal PHE. In 3 cases, a translocation pattern was detected in a few nuclei but did not meet our laboratory cutoff value. MYC amplification was not detected in any of the 5 cases examined, aiding in ruling out the possibility of angiosarcoma. Four patients underwent excision, and 2 with multifocal tumor required below-knee amputation. Six patients with outcome information were alive and free of disease after a median follow-up of 35 months. We conclude that ancillary techniques can be helpful in differentiating this unusual indolent tumor from potentially aggressive epithelioid hemangioendothelioma, epithelioid sarcoma, and epithelioid angiosarcoma. Long-term follow-up is warranted.

The loss of INI1 (SMARCB1) expression, caused by SMARCB1 (INI1, SNF5L4, BAF47) inactivation, frequently occurs in epithelioid sarcoma (ES) and could aid in confirming the diagnosis. Except for INI1, the expression of switch in mating type/sucrose nonfermentation complex members in ES has never been examined. In this study, the expression of key subunits of this complex-INI1, BRG1 (SMARCA4), BRM (SNF2L2, SMARCA2), PBRM1 (hPB1, BAF180), and BAF155 (SMARCC1)-was analyzed in 23 ES cases: 15 conventional and 8 proximal type. All of the cases were reviewed and reclassified by hematoxylin-eosin staining and immunostaining for cytokeratin AE1/3, epithelial membrane antigen, CD34, vimentin, and INI1 expression. Of the 23 ES cases, 19 (82.6%) showed a loss of PBRM1, and 18 (78.3%), a loss of INI1. In most cases (17, 73.9%), loss of INI1 and PBRM1 expression was observed. The pattern of PBRM1 expression was similar to that of INI1, that is, not correlated with changes in cellular morphology. The concurrent loss of BRM, PBRM1, and INI1expression was detected in 2 cases with pure rhabdoid tumor features. The frequent observation of concurrent loss of INI1 and PBRM1 suggests that certain switch in mating type/sucrose nonfermentation complex components might act synergistically in the pathogenesis of ES by unknown mechanisms and that these components could provide new targets for therapy. The usefulness of PBRM1 as a biomarker of ES and its mechanism in ES require further investigation. Loss of BRM in ES with pure rhabdoid features suggests that BRM might be involved in the underlying mechanisms of this type of ES.

The absence of INI1 (SMARCB1, hSNF5, BAF47) immunohistochemical reactivity is a central feature of malignant rhabdoid tumor, renal medullary carcinoma, and epithelioid sarcoma. We characterized INI1 immunoreactivity in synovial sarcoma (49 cases) in comparison with its closest histologic mimics (68 cases). We observed a unique pattern of decreased INI1 immunoreactivity with a high specificity (100%) and sensitivity (86%) for synovial sarcoma and particular sensitivity for poorly differentiated subtypes of synovial sarcoma (94%; 16/17 cases). Decreased INI1 immunoreactivity was not seen in any of the other lesions we examined, including 14 cases of Ewing sarcoma and 22 cases of malignant peripheral nerve sheath tumor. Furthermore, decreased INI1 immunoreactivity is distinct from the complete absence of INI1 immunoreactivity seen in malignant rhabdoid tumor or other INI1-negative neoplasms. We propose that this distinct INI1 immunohistochemical pattern serves as a useful diagnostic tool to provide preliminary results before molecular test results are available, especially in cases of poorly differentiated synovial sarcoma and in cases where limited material precludes confirmatory molecular studies. Awareness of this unique pattern is critical to avoid misinterpreting decreased INI1 immunoreactivity as a complete absence of INI1 and, consequently, misdiagnosing synovial sarcoma as an INI1-negative neoplasm.

Summary Epithelioid sarcoma (ES) is a malignant mesenchymal neoplasm with some morphologic or immunophenotypic evidence of epithelial differentiation. The "classic" subtype occurs in younger patients, often in distal extremities as compared with the "proximal" type. Tumors of the proximal type primarily arising in solid organs are rare with only few case reports in the literature. We report 2 cases of primary ES in the kidney of a 27-year-old woman and the adrenal gland of a 73-year-old man. Clinical examination and imaging, including computed tomography and positron-emission tomography, did not reveal tumor elsewhere in both cases. Histologic features were those of ES, proximal type with epithelioid/rhabdoid phenotype. Immunohistochemical study in both cases showed strong, diffuse expression of epithelial markers, CD34, and CD31. Nuclear expression of SMARCB1 protein was lost, but fluorescence in situ hybridization analysis was negative for SMARCB1 deletion. We believe that these are the first reports of primary kidney and adrenal gland ES.

Cardiac ankyrin repeat protein (CARP) is highly expressed in cardiac muscles and detectable in normal skeletal muscles. Arpp, a close homolog of CARP, has been demonstrated to be useful for distinguishing rhabdomyosarcoma from other malignant tumors. However, the CARP distributions among malignant tumors have been poorly investigated. Here, we analyzed the comprehensive expression of CARP in malignant tumors and evaluated its potential use for rhabdomyosarcoma diagnosis. A total of 159 malignant tumors, including 34 rhabdomyosarcomas, 85 non-rhabdomyosarcomas, and 40 carcinomas, were immunohistochemically analyzed for CARP expression. Cytoplasmic expression of CARP was detected in 29 (85%) of 34 rhabdomyosarcomas. The immunoreactivity was observed in both small cells with little differentiation and differentiated tumor cells with abundant eosinophilic cytoplasm. In contrast, focal immunoreactivity for CARP was only observed in 5 (4%) of 125 non-rhabdomyosarcomas, comprising 2 malignant fibrous histiocytomas, 1 angiosarcoma, 1 epithelioid sarcoma, and 1 squamous cell carcinoma of the lung. Comparative analysis of the CARP expression profiles with those of myogenic markers in rhabdomyosarcomas revealed that myogenin (88%) and desmin (88%) exhibited the best sensitivity, followed by CARP (85%), MyoD (82%), muscle-specific actin (79%), and myoglobin (65%). MyoD (96%) and myoglobin (96%) had the best specificity, followed by CARP (95%), myogenin (95%), desmin (89%), and muscle-specific actin (86%). Our results indicate that CARP is a sensitive and specific marker for rhabdomyosarcoma and that it will be useful for the differential diagnosis of rhabdomyosarcoma.

ERG is immunoexpressed in vascular endothelial tumors, blastic extramedullary myeloid tumors, and tumors with ERG-involved translocation, such as prostate carcinoma or Ewing sarcoma. Recently, ERG immunoexpression was reported in an epithelioid sarcoma, which is a SMARCB1/INI1-deficient tumor, although epithelioid sarcoma is not associated with chromosomal translocations involving ERG and is categorized as a tumor with uncertain differentiation. SALL4 is essential for a proliferation and stabilization of embryonic stem cells. It was reported that SALL4 expression may aid in distinguishing epithelioid sarcoma from malignant rhabdoid tumor. We analyzed the frequency of ERG and SALL4 expressions in 80 SMARCB1/INI1-deficient tumors, including 45 epithelioid sarcomas (conventional-type, 24; proximal-type, 20), 17 malignant rhabdoid tumors, 5 atypical teratoid/rhabdoid tumors, 6 undifferentiated/unclassified sarcomas, 5 myoepithelial tumors, and 4 extraskeletal myxoid chondrosarcomas. We found that ERG expression was present in 18 of the epithelioid sarcomas (41%), including 13 conventional-type (54%) and 5 proximal-type (25%), whereas all 17 of the malignant rhabdoid tumors exhibited negative immunoreactivity. One atypical teratoid/rhabdoid tumor (20%), 1 myoepithelial carcinoma (20%), 1 undifferentiated/unclassified sarcoma (17%), and no extraskeletal myxoid chondrosarcomas (0%) also showed ERG expression. SALL4 expression was recognized in 5 epithelioid sarcomas (11%), 12 malignant rhabdoid tumors (71%), 2 atypical teradoid/rhabdoid tumors (40%), 4 undifferentiated/unclassified sarcomas (67%), 1 myoepithelial tumor (20%), and none of the extraskeletal myxoid chondrosarcomas (0%). Therefore, the evaluation of ERG and SALL4 immunoexpressions may be a useful diagnostic tool to distinguish epithelioid sarcoma, especially proximal type, from malignant rhabdoid tumor.

Low-affinity nerve growth factor receptor (p75) is a member of the tumor necrosis factor receptor family. It may modulate the binding of nerve growth factor (NGF) to the functional high-affinity receptor tyrosine kinase (trk) A. NGF is thought to be responsible for growth, apoptosis, and function of the nervous system. The presence of this receptor (p75) was determined in a large group of neural and nonneural tumors and fetal and adult tissues. One thousand one hundred fifty tumors were analyzed with monoclonal antibody for p75, along with selected normal fetal and adult tissues. Immunoreactivity for p75 was present in adult pericytes, perivascular fibroblasts, basal cells of several types of epithelia, perineurial cells, and dendritic reticulum cells. Additionally, a wide zone of subepithelial mesenchyme and skeletal muscle were positive in the first-trimester fetus, but were diminished or negative in the adult. Consistently positive nonneural mesenchymal tumors included dermatofibrosarcoma protuberans (DFSP), embryonal and alveolar rhabdomyosarcoma, synovial sarcoma, and spindle cell hemangio(endotheli)oma. Schwann cell tumors, ganglioneuroma, granular cell tumor, and malignant peripheral nerve sheath tumor (MPNST) were also p75 positive. Mesenchymal nonneural tumors that were variably positive (32% to 69%) for p75 included fibrosarcoma variants, solitary fibrous tumor, hemangiopericytoma, spindle cell lipoma, Ewing's sarcoma, mesenchymal chondrosarcoma, and malignant melanoma. Nervous system tumors such as paragangliomas, neuroblastoma, meningioma, and perineurioma and nonneural mesenchymal tumors, including extraskeletal osteosarcoma, benign fibrous histiocytomas, fibromas, alveolar soft part sarcoma, epithelioid sarcoma, smooth muscle and gastrointestinal stromal tumors, and angiosarcomas, were almost always negative for p75. Epithelial tumors that were consistently positive included mixed tumor and adenoid cystic carcinoma, whereas mesothelioma, adenocarcinomas, and most squamous cell carcinomas were negative. p75 is not a specific marker for nerve sheath tumors. It is present in a variety of other mesenchymal tumors including synovial sarcoma and in CD34-positive tumors such as DFSP, spindle cell lipoma, and hemangiopericytoma. The presence of p75 in nonneural tumors such as DFSP and rhabdomyosarcoma mimic its presence in early fetal mesenchyme and skeletal muscle, suggesting oncofetal expression in these tumors. p75 may be useful to distinguish DFSP from benign fibrous histiocytoma.

Keratin polypeptides of the nonhair type, numbered 1 through 20 in the Moll catalog, are selectively expressed in normal and neoplastic tissues. Keratin 1 (K1), the highest-molecular-weight keratin (67 kd), was generally considered specific for keratinizing squamous epithelia. However, recent studies have shown that it is an integral component of the multiprotein kininogen receptor of endothelial cells. A library of formalin-fixed, paraffin-embedded tissue samples was evaluated immunohistochemically (avidin-biotin peroxidase complex method) for K1 expression using a specific monoclonal antibody (Novocastra clone 34βB4). The study group included a wide variety of normal tissues and 541 tumors of epithelial or mesenchymal derivation. The specificity of the antibody to K1 was verified in normal epithelial tissues, where the staining was essentially limited to the epidermis and Hassal corpuscles of the thymus and focally to other squamous epithelia. Among carcinomas, it was essentially limited to keratinizing squamous carcinomas. It was also regularly found in endothelial cells of normal capillaries, veins, and arteries. Capillary, cavernous, and venous hemangiomas often had endothelia with K1 positivity. Among the malignant vascular tumors, epithelioid hemangioendotheliomas were consistently positive (8 of 8). However, angiosarcomas had more variable expression (59 of 81 were positive), with well-differentiated tumors generally having greater reactivity than poorly differentiated examples. Mesenchymal tumors with K1 expression included schwannomas (10 of 16), epithelioid sarcomas (26 of 37), and synovial sarcomas (19 of 68). In the last 2 tumor types, K1 reactivity was detected in both epithelioid and spindled neoplastic populations. In addition to its specificity for keratinizing squamous epithelia, K1 can be immunohistochemically detected in normal vascular endothelial cells and a spectrum of vascular tumors. However, its expression in poorly differentiated vascular tumors is variable, suggesting that this marker is poorly conserved in highly transformed endothelia. The unexpected K1 immunoreactivity in nonvascular soft tissue tumors, such as synovial sarcoma, epithelioid sarcoma, and schwannomas, requires further study. HUM PATHOL 32:873-879. This is a US government work. There are no restrictions on its use.

Epithelioid sarcoma (ES) is a distinctive soft tissue neoplasm with a predilection for the distal extremities of young adults. This tumor typically contains nodular aggregates of epithelioid and spindle cells with zonal necrosis. The neoplastic cells are generally reported to coexpress keratin and vimentin and are often stated to be positive for CD34. However, there is no large series with extensive immunohistochemical data, there are few data with regard to expression of different keratin subtypes, and there are no large series discussing the epithelioid sarcoma subtypes. In the current study, we immunohistochemically evaluated 88 typical and 24 variant (8 angiomatoid, 9 large cell/rhabdoid, and 7 "fibroma-like") ESs. Nearly all ESs with typical histology (94%) were positive for keratin 8 (K8), whereas 72% were positive for K19, 48% for intermediate- and high-molecular-weight keratins (34betaEH12), and 22% for K7; reactivity with the latter two antibodies was usually seen in only a minority of tumor cells. Vimentin reactivity was present in all cases, EMA in 96% of cases and muscle-specific actin and CD34 were noted in 41% and 52% of the cases, respectively. A few ESs (7%) showed focal cytoplasmic CD31 reactivity, but none exhibited a distinctive membrane staining pattern, and examples tested for FVIIIRAg were negative. The angiomatoid, fibroma-like, and large cell-rhabdoid ES variants had immunohistochemical profiles similar to the classic cases, supporting a common pathogenesis. Although not consistently expressed in ES, the presence of CD34 is helpful in distinguishing this entity from primary and metastatic carcinomas and other sarcomas such as malignant rhabdoid tumor.

Epithelioid sarcomas are soft tissue tumors with an indolent, but potentially aggressive, clinical behavior. Distinction from other benign and malignant entities may be a diagnostic dilemma. In this study, we evaluate the presence of loss of heterozygosity (LOH) of chromosome 22q in tumor DNA from 13 epithelioid sarcomas, four epithelioid angiosarcomas, and two epithelioid hemangioendotheliomas, and investigate its possible role in diagnosis. LOH was detected in 6 of 10 (60%) of the informative epithelioid sarcomas. No allele loss was detected in the informative vascular tumors, three angiosarcomas, and two hemangioendotheliomas. Chromosome 22q carries the locus of a tumor suppressor gene, the neurofibromatosis 2 (NF2) gene, which has been shown to be lost or mutated in some NF2-related tumors, sporadic meningiomas, and vestibular schwannomas, as well as a few other tumors. Our data suggest that a region of chromosome 22q may be the locus of a tumor suppressor gene involved in the tumorigenesis of these neoplasms. Genetic alterations of yetunknown tumor suppressor genes in this region, or even the NF2 tumor suppressor gene, may play a role in epithelioid sarcomas tumorigenesis. The fact that LOH was only detected in epithelioid sarcomas and not in the vascular tumors studied suggests a possible role for this marker in diagnosis. Hum Pathol 29:604–608. This is a US government work. There are no restrictions on its use.

Loss of SMARCB1/INI1 protein expression is considered useful for confirming a histologic diagnosis of malignant rhabdoid tumor. However, loss of SMARCB1/INI1 protein expression has recently been reported in other tumors as well, including a few cases of epithelioid sarcoma. In addition, the histopathologic differences between proximal-type epithelioid sarcoma and malignant rhabdoid tumor have not been conclusively defined. We analyzed SMARCB1/INI1 protein expression in 54 epithelioid sarcoma (proximal-type, 25; distal-type, 29) and examined alterations of the SMARCB1/INI1 gene in the cases lacking protein expression. We found that 19 (76.0%) proximal-type epithelioid sarcoma and 27 (93.1%) distal-type epithelioid sarcoma showed loss of SMARCB1/INI1 protein expression. Analysis of 39 cases with loss of protein expression revealed 4 cases (10.3%) with SMARCB1/INI1 gene alterations at the DNA level (homozygous deletion, 2; 1- or 2-bp deletion, 2) that could have induced the loss of gene products, and all 4 of these were proximal-type epithelioid sarcoma. Epithelioid sarcoma was thus associated with a high frequency of loss of SMARCB1/INI1 protein expression similar to that in malignant rhabdoid tumor. However, the frequency of SMARCB1/INI1 gene alteration at the DNA level in proximal-type epithelioid sarcoma was significantly lower than that in malignant rhabdoid tumor. In addition, the prognosis of patients with malignant rhabdoid tumor is significantly worse than that of patients with proximal-type epithelioid sarcoma (P = .001). Therefore, proximal-type epithelioid sarcoma and malignant rhabdoid tumor are suggested to be distinctive tumors with respect to the mechanism of the loss of SMARCB1/INI1 protein expression. Analysis of alterations in the SMARCB1/INI1 gene may thus be a useful diagnostic tool to distinguish proximal-type epithelioid sarcoma from malignant rhabdoid tumor.

GLUT-1, an erythrocyte-type glucose transporter protein expressed in juvenile hemangiomas, has recently been shown to be a sensitive marker of perineurial cells and their tumors in a small number of cases. However, GLUT-1 expression has not been systematically examined in other mesenchymal neoplasms. Prompted by a recent report of GLUT-1 expression in epithelioid sarcoma, a tumor not generally felt to show perineurial differentiation, we examined GLUT-1 expression in a wide variety of mesenchymal tumors. Sections from 247 mesenchymal tumors of a variety of histologic subtypes were retrieved from our archives and immunostained for GLUT-1 using heat-induced epitope retrieval and the DAKO ADVANCE detection system (DAKO, Carpinteria, CA). Scoring was as follows: negative (5% of cells), 1+ (5%-25% of cells), 2+ (25%-50% of cells), and 3+ (50% of cells). All benign nerve sheath tumors showed a peripheral rim of positive normal perineurial cells, with 2 neurofibromas and 3 schwannomas showing more extensive staining. Three of 4 perineuriomas showed strong GLUT-1 expression. All juvenile hemangiomas were GLUT-1 positive. GLUT-1 expression was also seen in a wide variety of benign and malignant mesenchymal tumors. However, GLUT-1 expression was absent in nonjuvenile hemangioma endothelial tumors and in almost all low-grade lesions that enter the histologic differential diagnosis of perineurial tumors, including low-grade fibromyxoid sarcoma, dermatofibrosarcoma protuberans, and myxofibrosarcoma. We conclude that GLUT-1 expression in mesenchymal tumors is by no means specific for perineurial differentiation, but may instead represent upregulation of this protein within hypoxic zones, secondary to upstream activation of proteins such as hypoxia-inducible factor 1-alpha.

Seven epithelioid and eight non-epithelioid vascular tumors were studied by the avidin-biotin-peroxidase method for the presence of endothelial- and epithelial-associated markers, using Ulex europaeus agglutinin-1 (UEA-1) lectin, and antibodies directed against factor VIII-related antigen, (FVIII-RA), vimentin, keratin, carcinoembryonic antigen, and epithelial membrane antigen. The cases included four epithelioid hemangiomas, two epithelioid hemangioendotheliomas (EHE), one epithelioid angiosarcoma (EAS), four common non-epithelioid capillary hemangiomas, and four non-epithelioid angiosarcomas. Staining for FVIII-RA, UEA-1, and vimentin were observed in all cases. The EAS showed staining for keratin in formalin-fixed, paraffin-embedded sections and in frozen sections. Staining for keratin was also observed in frozen sections of one EHE. Both keratin-positive vascular tumors were confirmed with electron microscopy. Carcinoembryonic antigen and epithelial membrane antigen stains were negative in all cases. Our results show that the epithelioid vascular tumors EHE and EAS, in addition to staining for the endothelial markers and vimentin, may also express the epithelial marker keratin. This is important since these tumors may closely resemble carcinomas by routine light microscopy. This study further underscores the importance of using a broad panel of immunohistochemical markers in the diagnostic workup of soft-tissue neoplasms.

Summary Arpp, a protein including an a nkyrin- r epeat, P EST motif, and p roline-rich region, is a recently identified protein that is exclusively expressed in striated muscles. This study comprehensively analyzed its expression among soft tissue sarcomas of various histological types and evaluated its potential use for the differential diagnosis of rhabdomyosarcoma (RMS). Formalin-fixed, paraffin-embedded tissues, including 37 RMS cases, 88 non-RMS sarcomas, and 38 carcinomas, were analyzed for Arpp expression. Arpp was detected in 33 (89.2%) of 37 RMS cases by immunohistochemistry. Western blot analysis revealed expression of Arpp in all RMS cases tested. High expression of Arpp was generally associated with morphological evidence of skeletal muscle differentiation of tumor cells. In contrast, Arpp displayed 6.3% (8/126) positivity among the non-RMS tumors. Focal or weak expression was seen in malignant fibrous histiocytoma (2/27), synovial sarcoma (1/11), Ewing sarcoma (1/5), and epithelioid sarcoma (3/5), whereas one epithelioid sarcoma displayed strong expression for Arpp. A comparative analysis of the Arpp profile with that of myogenic markers in RMS revealed that the sensitivity of Arpp (89.2%) was higher than that of myoglobin (59.6%) and comparable with that of myogenin (88.2%), MyoD (80.6%), muscle-specific actin (83.8%), and desmin (89.2%). These results suggested that Arpp is sensitive to and specific for RMS. Thus, we proposed that Arpp is a novel skeletal muscle–specific marker, which is useful for differential diagnosis of RMS.

An epithelioid sarcoma of the perineum of a 60-year-old man with widespread metastases produced leukocytosis, myeloid hyperplasia of the bone marrow, and splenomegaly. High titers of granulocyte colony-stimulating factor (G-CSF) were found in the patient's serum and primary culture medium of the tumor tissue. The tumor tissue extract contained m-RNA for G-CSF in large quantities, proving that the tumor was the source of this cytokine.

Glycoproteins isolated from human milk fat globule membranes, designated epithelial membrane antigen (EMA), have been detected immunohistochemically in most nonneoplastic epithelia and are potentially a highly effective marker for establishing the epithelial nature of neoplastic cells. With commercially available monoclonal antibodies and an indirect immunoperoxidase technique, EMA localization was evaluated in paraffin-embedded tissues from a wide variety of neoplasms (320 specimens). Adenocarcinomas from various primary sites (breast, lung, colon, stomach, pancreas, gallbladder, prostate, endocrine glands, ovary, kidney, thyroid) were immunoreactive for EMA in 88 of 97 cases (91 per cent). Cytoplasmic and apical luminal membrane staining were the most common patterns of immunoreactivity, with peripheral membrane staining or other patterns also seen in some neoplasms. Squamous cell (13 of 13 cases) and transitional cell (12 of 12 cases) carcinomas, small cell anaplastic carcinomas (12 of 12 cases), and mesotheliomas (six of six cases) were also uniformly EMA-positive. Malignant lymphomas of the Hodgkin's (15 cases) and non-Hodgkin's types (74 cases), except for the true histiocytic lymphomas and occasional T-cell lymphomas, were nonreactive for EMA. Neoplastic and nonneoplastic plasma cells showed variable EMA positivity. Endocrine neoplasms (17 cases), including carcinoid tumors, medullary carcinoma of thyroid, adrenocortical carcinomas and pheochromocytomas, and germ cell tumors (eight cases, embryonal carcinoma and seminoma), and a wide variety of soft tissue tumors (27 cases) generally lacked immunoreactivity for EMA; the exceptions to this finding were synovial sarcomas and an epithelioid sarcoma. Malignant melanomas (eight cases) were typically nonreactive. Based on the observations in this large series of neoplasms, EMA is an excellent marker of epithelial differentiation, appears to be highly reliable for discriminating between poorly differentiated carcinomas and malignant lymphomas, and is especially helpful in characterizing small cell anaplastic carcinomas. Epithelial membrane antigen immunoreactivity is well preserved in paraffin sections of routinely processed tissues, facilitating application of this technique in diagnostic surgical pathology.

This is a report of nine cases of epitheloid sarcoma, a term introduced recently to designate a special variant of malignant soft tissue tumor generally involving tendons, aponeuroses, and more superficial tissues of the extremities. The histologic features of this tumor, which consists of a mixture of polygonal epithelioid cells and elongated fibroblastic cells with necrotizing foci, have often misled the pathologist into diagnoses of necrotizing granuloma, malignant melanoma, angiosarcoma, or metastatic carcinoma. The tumor is definitely malignant, is difficult to eradicate by local surgical excision, and often metastasizes to lymph nodes and via the blood. The histogenesis of this tumor is still controversial, but some light and electron microscopic features suggest a possible synovial origin.

Seven epithelioid sarcomas were examined by light and electron microscopy and by immunohistochemical methods for the presence of vimentin and the epithelial markers keratin and epithelial membrane antigen. All the tumors displayed vimentin and both epithelial markers. Electron microscopy showed a spectrum of cellular differentiation from primitive fibrohistiocytic cells to epithelial-type cells with junctions, microvilli, and tonofilaments. Spindle cells showed myofibroblastic and fibroblastic differentiation. Epithelioid sarcoma appears to be a tumor of primitive cells with the potential for mesenchymal and epithelial differentiation, like some embryonic tumors.