Anatomic Pathology / COMPARISON OF FIVE MELANOMA MARKERS IN CYTOLOGIC PREPARATIONS

We determined the sensitivity and specificity of 3 novel antibodies (microphthalmia transcription factor [Mitf], Melan-A, and tyrosinase) as markers for melanoma in cytologic preparations and compared the results with those of commonly used markers (S-100 protein [S-100] and HMB-45). We stained 72 cell blocks from 40 patients with melanoma and 32 with nonmelanocytic malignant neoplasms with antibodies against S-100, HMB-45, Mitf, Melan-A, and tyrosinase. Histologic correlation was available in more than 95% of cases. Nuclear staining for Mitf and cytoplasmic staining for S-100, HMB-45, Melan-A, and tyrosinase in more than 10% of tumor cells was considered positive. All 3 novel markers demonstrated sensitivity superior to S-100 and HMB-45. HMB-45, Melan-A, and Mitf demonstrated specificities of 97%. S-100 protein and tyrosinase were less specific. Sensitivity and specificity for the combination Mitf+/Melan-A+ were 95% and 100%, respectively, whereas they were 80% and 100%, respectively, for S-100+/HMB-45+. Mitf, Melan-A, and tyrosinase are sensitive markers for epithelioid melanoma. Mitf and Melan-A seem more specific than S-100 and tyrosinase. An antibody panel consisting of Mitf and Melan-A is superior to a panel of S-100 and HMB-45 in the diagnosis of melanoma in cytologic specimens. The cytologic diagnosis of metastatic melanoma can be challenging. Melanoma often manifests with a diverse cytologic appearance that may include a dyshesive single cell pattern or a cohesive cellular arrangement.1,2 The cell shape varies from epithelioid to spindled, or a mixture of epithelioid and spindle cell patterns might be seen.1,2 Furthermore, the cytologic features of melanoma often are shared with other poorly differentiated malignant neoplasms, including carcinomas, lymphomas, and sarcomas. In addition, metastatic melanoma can be found anywhere in the body and may manifest with a myriad of clinical signs and symptoms. Nevertheless, it is important to differentiate melanoma from nonmelanocytic malignant neoplasms because prognosis and therapy differ radically among these entities. Immunocytochemical studies often are used as an aid in the diagnosis of melanoma.3-7 The most frequently used melanocytic markers in clinical practice are S-100 protein and HMB-45. Monoclonal antibody to S-100 protein, a calcium binding F-hand protein originally isolated from the brain, is a sensitive marker that reacts with more than 90% of melanomas. However, this protein also is present in adipocytes, chondrocytes, Schwann cells, and myoepithelial cells.8,9 Tumors derived from these tissues usually retain immunoreactivity to S-100 protein.10 Thus, S-100 protein reacts with a broad range of benign and malignant neoplasms, therefore limiting its specificity as a melanocytic marker.10 In addition, certain epithelial neoplasms such as mammary carcinoma may be positive for S-100 protein.10,11 Monoclonal antibody against HMB-45 antigen recognizes melanosome-specific gp100.12 Although it is quite specific for melanocytic neoplasms, HMB-45 is less sensitive than S100 protein for identifying melanoma.13-15 Some melanomas, Anatomic Pathology / ORIGINAL ARTICLE Am J Clin Pathol 2002;118:930-936 931 931 DOI: 10.1092/EWK9LUPR6BC51GXV 931 © American Society for Clinical Pathology particularly the spindle cell and desmoplastic variants, fail to react with HMB-45.16,17 In addition, HMB-45 has been shown to react with other neural crest–derived tumors and occasionally with adenocarcinomas and other neoplasms.18 Several new monoclonal antibodies raised against melanocytic differentiation antigens have become available for routine diagnostic use. These markers, which include microphthalmia transcription factor (Mitf; clone D5), tyrosinase (clone T311), and Melan-A (or MART-1; clone A103), are thought to be specific for melanocytic lesions. Mitf is a nuclear transcription factor critical for melanocyte development in the embryo and for survival of these melanocytes postnatally.19 Tyrosinase is a key enzyme involved in the early stages of melanin production and is present in cells of melanocytic lineage.20 Melan-A protein is a product of the MART-1 gene that is recognized by autologous cytotoxic T cells.21 It is a cytoplasmic protein that is expressed in mature melanocytes; however, its biologic function is unknown.22 In a previous study, Dorvault et al23 demonstrated that Mitf was a sensitive and specific marker for melanomas in cytologic specimens and might be superior to the presently used melanocytic markers, S-100 protein and HMB-45. The purpose of the present study was to compare the diagnostic usefulness of the other novel melanoma markers, namely, tyrosinase and Melan-A with Mitf and with S-100 protein and HMB-45 as markers for melanoma in cytologic preparations. Materials and Methods We retrieved 72 formalin-fixed, paraffin-embedded cell blocks from 40 patients with melanoma (including 2 cases of spindle cell melanoma) and 32 patients with nonmelanocytic malignant neoplasms (carcinoma, 25; mesothelioma, 4; lymphoma, 2; islet cell tumor, 1) from the files of the Departments of Pathology at the University of Alabama at Birmingham and New York University Medical Center, New York, NY. The primary sites of the carcinomas were as follows: ovary, 7; lung, 6; kidney, 4; breast, 3; colorectal, 3; liver, 1; and uterine cervix, 1. These cases were retained from a previous study.23 Corresponding histologic material was available for 69 cases. Three cases of melanoma were from patients with disseminated metastatic disease. Among the melanomas, specimen sources were fine-needle aspiration biopsy of the following: lymph nodes, 15; lungs, 9; liver, 6; subcutaneous nodules, 6; orbital mass, 1; and pancreas, 1; and from pleural and peritoneal fluid, 1 each. Among the nonmelanocytic malignant neoplasms, the specimen sources were fine-needle aspiration biopsy of the following: kidney, 6; liver, 6; lung, 2; and pancreas, 1; and from pleural fluid, 10; and peritoneal fluid, 7. Immunocytochemical analysis was performed using standard protocols. Briefly, 5-μm sections were cut, placed on electrostatically charged glass slides, and deparaffinized. Staining for Mitf was performed using the monoclonal antibody to Mitf (D5, undiluted, provided by D.E.F.) obtained from tissue culture supernatant. For Mitf, heat-induced epitope retrieval was accomplished by microwave heating for 10 minutes in a 0.01-mol/L concentration of citrate buffer, pH 6. Immunostaining then was accomplished manually by using a modified avidin-biotin peroxidase technique. Staining for the remaining primary antibodies was performed using the Ventana ES automated immunohistochemistry system and the Ventana DAB Detection Kit (Ventana Medical Systems, Tucson, AZ). The primary antibody used included antibodies to S-100 protein (clone S-100, Ventana Medical Systems), HMB-45 antigen (clone gp100, Ventana Medical Systems), tyrosinase (clone T311, Neomarkers, Fremont, CA), and Melan-A (clone A013, Neomarkers). All 4 antibodies were prediluted and incubated with the tissue section for 1 hour at room temperature. Counterstaining was performed with hematoxylin. Appropriate positive and negative controls were used; the latter was achieved by omitting the primary antibodies. All cases were examined independently and blindly by 3 observers (Mitf reviewed by C.C.D., K.N.W., and D.C.C.; S-100 protein and HMB-45 by C.C.D., H.Y., and D.C.C.; tyrosinase and Melan-A by M.V.S., H.Y., and D.C.C.). Nuclear staining with Mitf in more than 10% of tumor cell nuclei was considered positive, whereas the presence of cytoplasmic staining in more than 10% of the tumor cells was considered positive staining with S-100 protein, HMB-45, tyrosinase, and Melan-A.