Antibodies to S-100 protein have been used widely as markers of malignant melanoma, despite abundant evidence that they are non-specific for this neoplasm. Hence, alternatives to these reagents are desirable in diagnostic dermatopathology. We evaluated the characteristics of a new monoclonal antibody (HMB-45) which does have putative specificity for melanoma, and compared it with a polyclonal anti-S-100 reagent in immunohistochemical staining of 67 melanomas of the skin and 133 non-melanomatous cutaneous neoplasms. All specimens were formalin-fixed and paraffin-embedded, and were studied with the avidin-biotin-peroxidase complex technique. HMB-45 labelled 62 of 67 melanomas, while anti-S-100 recognized all tumors of this type. On the other hand, S-100 also was expressed by 15 of the non-melanocytic neoplasms, all 133 of which were HMB-45-negative. The only cases of melanoma that were missed by the latter reagent were of the spindle-cell type. Hence, HMB-45 was 100% specific and 93% sensitive, relative to a diagnosis of malignant melanoma in paraffin sections. Epithelioid and small-cell neoplasms are reliably recognized by this antibody, but it would appear that spindle-cell melanomas must be detected by other immunohistochemical means.

A review was conducted of 335 malignant melanomas to identify variant morphologic patterns that might be confused with other tumors. In all, 27 predominantly amelanotic neoplasms with unusual histologic features were selected for additional study. These included nine with an adenoid or pseudopapillary pattern, seven small cell neoplasms, five with prominent myxoid stroma, four with a hemangiopericytoma-like appearance, and two composed of neoplastic cells with a signet-ring configuration. A diagnosis of melanoma was confirmed in all cases by Fontana-Masson strains for melanin pigment, electron microscopic examination, or the results of immunohistochemical analyses for cytokeratin, vimentin, S-100 protein, and the HMB-45 antigen. One tumor was associated with a congenital hairy melanocytic nevus, five were vulvovaginal lesions, four arose in the sinonasal tract, and one occurred in the rectum. Four of the specified microscopic patterns were observed in both primary and secondary neoplasms; the two signet-ring cell melanomas were recurrent lesions. The authors conclude that malignant melanomas may assume the histologic guise of adenocarcinomas, small cell carcinomas, and sarcomas, in a variety of tissue sites. Special studies designed to detect melanocytic differentiation are therefore appropriate in diverse differential diagnostic settings.

The human hematopoietic progenitor cell antigen (CD34) is a cell surface protein expressed by human hematopoietic progenitor cells, vascular endothelium, and many mesenchymal tumors. Sections from six samples of normal skin and from 41 epithelial tumors of the skin were studied. Immunostaining of epithelial cells from the external root sheath below the attachment of the arrector pili muscle and above the matrix cells was noted in normal samples. Tumors derived from or differentiated toward cells of the outer sheath, especially trichilemmomas, were immunostained with QBEND/10 (anti-CD34 antibody), whereas other epithelial tumors studied were negative. CD34 could serve as a marker of outer sheath cell derivation and may well be of value in the distinction between trichilemmomas and other lesions with similar histopathological features.

In order to determine whether or not phenotypic differences existed between reactive angioendotheliomatosis (RAE) and malignant angioendotheliomatosis (MAE), we studied the histological and immunohistochemical features of 4 and 8 cases of these lesions, respectively. Antibodies to leukocyte common antigen (LCA), specialized B- and T-lymphocytic determinants, Factor VIII-related antigen (FVIIIRAG), blood group isoantigens A, B, and H (BGI), epithelial antigens, vimentin, and actin; and Ulex europaeus I lectin (UEL) were utilized. Cutaneous lesions in all cases of MAE were part of a disseminated, fatal, intravascular cellular proliferation, with highly atypical cytological features. Because one of the patients in this group had cardiac valvular vegetations at autopsy, this case had been reported previously as representative of RAE. However, the latter example, as well as all others of MAE, stained strongly for LCA, B-cell antigens, and vimentin in tumor cells. FVIIIRAG was seen focally in 6 cases, in cells entrapped in platelet-fibrin thrombi; however, UEL binding and reactivity for BGI were uniformly absent. Conversely, RAE was typified by a cytologically-bland intravascular proliferation, with actin-positive, perivascular, pericytic cuffs. All 4 patients in this group had cutaneous involvement only, and the lesions tended to be self-resolving. One had pulmonary tuberculosis, but evidence for an underlying infection was absent in the remainder of RAE cases. Immunohistologically, RAE displayed universal reactivity for FVIII-RAG, BGI, UEL, and vimentin, and negativity for LCA in intravascular cells. Neither MAE nor RAE showed the presence of epithelial determinants. These data indicate that MAE and RAE are clinicopathologically distinct entities, showing lymphoid and endothelial features, respectively. Because of the phenotypic properties of the former condition, it would appear advisable to substitute the term "intravascular lymphomatosis" for "malignant angioendotheliomatosis".

No abstract