Histopathological patterns of melanoma metastases in sentinel lymph nodes

AIMS: Sentinel lymph node biopsy (SLNB) is an important component in the staging and treatment of cutaneous melanoma (CM). The medical literature provides only limited information regarding melanoma sentinel lymph node (SLN) histology. This report details the specific histological patterns of melanoma metastases in sentinel lymph nodes (SLNs) and highlights some key factors in evaluating SLNs for melanoma. METHODS: From 281 SLNB cases between June 1998 and May 2002, 79 consecutive cases of SLN biopsies positive for metastases from CM were retrospectively reviewed. The important characteristics of the SLNs and the metastatic foci are described. RESULTS: The median size of positive SLNs was 17 mm (range, 5-38). SLNs had a median of two metastatic foci (range, 1-11), with the largest foci being a median of 1.1 mm in size (range, 0.05-24). S-100 and HMB-45 staining was positive in 100% and 92% of the detected metastatic foci, respectively. The metastatic melanoma cells were epithelioid, spindled, and mixed in 86%, 5%, and 9% of cases. Metastatic foci were most often (86%) found in the subcapsular region of the SLN. Benign naevic cells were found coexisting in 14% of positive SLNs. CONCLUSIONS: Staining for S100 is more sensitive than HMB-45 (100% v 92%), but HMB-45 staining helped to distinguish benign naevic cells from melanoma. The subcapsular region was crucial in SLN evaluation, because it contained the metastases in 86% of cases. Evaluation of the subcapsular space should not be compromised by cautery artefacts or incomplete excision of the SLN.     

Can Melan-A replace S-100 and HMB-45 in the evaluation of sentinel lymph nodes from patients with malignant melanoma?

The sentinel lymph node (SLN) biopsy has become an increasingly important procedure used in the primary staging of malignant melanoma. However, micrometastases in a lymph node can be easily missed on routine H&E-stained sections. Therefore, S-100 and HMB-45 IHC stains are standardly performed on grossly negative SLNs for detection of metastatic melanoma. Each of these IHC markers, however, is not ideal. The authors investigated whether the newer IHC marker Melan-A would improve the detection of metastatic melanoma in SLN biopsies. Forty lymph nodes previously diagnosed with metastatic melanoma were retrospectively evaluated for S-100, HMB-45, and Melan-A expression. In addition, 42 SLN biopsies for metastatic melanoma detection were prospectively collected and evaluated for S-100, HMB-45, and Melan-A expression. All lymph nodes with metastatic melanoma from the retrospective study demonstrated S-100 reactivity. Five of the lymph nodes with metastatic melanoma from the retrospective study failed to express either HMB-45 or Melan-A, all of which displayed a desmoplastic morphology. One of the metastases positive for S-100 and HMB-45 failed to show reactivity with Melan-A (3%). The prospective study found 10 lymph nodes from 42 cases to be positive for metastatic melanoma, which were positive for S-100 (100%). Nine of the involved lymph nodes were positive for HMB-45(90%), and nine were positive for Melan-A (90%). Melan-A, although very specific, cannot replace the use of S-100 and HMB-45 for the detection of metastatic melanoma in SLNs. It can, however, substitute for HMB-45 with equally good results.     

Rapid immunohistochemistry of sentinel lymph nodes for metastatic melanoma

Sentinel lymph node (SLN) biopsy is performed on patients with malignant melanoma (MM) to assess the need for selective complete lymphadenectomy. Melanoma metastasis to regional lymph nodes is an important prognostic indicator in patients with MM. This study assesses the sensitivity and specificity of rapid immunohistochemistry (RIHC) in intraoperative delineation of melanoma metastasis to SLN. RIHC for S-100 protein, HMB45, and a melanoma marker cocktail (melan A, HMB45, and tyrosinase) was performed on 71 SLNs obtained from 28 patients with MM. Frozen sections (6 micro thick) on plus slides were fixed for 2 to 3 minutes in cold acetone and then stored at -70 degrees C. The EnVision kit (Dako, Carpinteria, CA) for rapid immunohistochemistry (RIHC) on frozen tissue sections was used, and the staining technique took 19 minutes. Together with preparation of the frozen sections and fixation in acetone, immunostained slides were available in approximately 25 minutes. Of the 71 SNLs examined, 7 showed melanoma metastasis in permanent sections. RIHC of frozen sections detected metastatic melanoma in 6 SLNs, with a sensitivity of 86% for HMB45 and 71% for S-100 protein and the melanoma cocktail and a specificity of 97% for HMB45 and 100% for S-100 and the melanoma cocktail. We conclude that RIHC for HMB45, S-100 protein, and the melanoma cocktail may help detect melanoma metastasis in SLN intraoperatively, leading to total lymph node dissection and obviating the need for 2 surgical procedures. Section folds and background stain can make interpretation difficult. Intraoperative time constraints require a more rapid technique. A recent consensus group has discouraged frozen-section examination of SLN.     

Fine-needle aspiration biopsy of malignant melanoma: a cytologic and immunocytochemical analysis

The immunoreactivity of alcohol-fixed cell blocks from 15 fine-needle aspiration (FNA) specimens of malignant melanoma was investigated using monoclonal antibodies to keratin and vimentin intermediate filaments, melanoma cytoplasmic antigen (HMB-45), and S-100, as well as polyclonal antibodies to S-100. The results were compared with the immunoprofiles obtained using formalin-fixed surgical specimens from 10 of the same patients. In all cases, immunostaining for keratin was negative and immunostaining for vimentin was positive. Immunostaining for HMB-45 was positive in 13/15 aspirates and in 9/10 surgical specimens. Immunostaining for S-100 protein showed the greatest variability in staining, with 5/15 fine needle aspiration biopsies and 9/10 surgical specimens being positive using the polyclonal antibody and only 1/15 FNA specimens and 7/10 surgical specimens being positive using the monoclonal S-100 reagent. Our findings indicate that immunocytochemical studies can be very useful as an adjunct in the FNA diagnosis of melanoma. Also included in our series is an unusual variant of malignant melanoma, the so-called signet ring melanoma. Given the location of the anal verge, the use of immunocytochemical markers was essential in establishing the correct diagnosis in this case. While S-100 protein is of limited value as a marker of melanoma in alcohol-fixed FNA specimens, a definitive diagnosis of malignant melanoma can be made using a panel of antibodies including keratin, vimentin, and HMB-45.     

PNL2, a new monoclonal antibody directed against a fixative-resistant melanocyte antigen.

We report the production of a new monoclonal antibody, PNL2, directed against a fixative resistant melanocyte antigen. The analysis of PNL2 immunostaining on a broad range of normal or malignant human tissues and on various melanocytic lesions revealed its high specificity. PNL2 gave a strong cytoplasmic staining of skin and oral mucosae melanocytes, and staining of granulocytes when used at high concentration. PNL2 stained all intra-epidermal nevi irrespective of their histologic type, but common intradermal nevi and the dermal component of compound nevi were largely non-reactive as only scattered nevus cells in the papillary dermis were labeled. PNL2 labeled more than 70% of the neoplastic cells in all primary melanomas irrespective of their histologic type. However, PNL2 did not label desmoplastic melanomas. All metastatic melanomas were also stained but the percentage of labeled cells was occasionally lower than the primary tumor. PNL2, as anti-Melan A and HMB-45 antibodies, stained most of the clear cell sarcoma cells, and a few cells in angiomyolipomas and lymphangioleiomyomatosis. None of the other non-melanocytic lesions tested were labeled. Proteomic approaches showed that the immunoaffinity purified PNL2-binding complexes isolated from melanoma cell lines comprise at least TAP1, Clathrin 17 and prealbumin proteins, but not the gp100 recognized by HMB-45. In conclusion, this new monoclonal antibody, PNL2, is directed against a new fixative resistant melanocyte associated antigen. This antigen is chemically resistant and thus allows immunostaining after melanin bleaching or decalcification. We also demonstrate that it is different from Melan A and from gp100, even if PNL2 and HMB-45 staining patterns are sometimes similar.     

A comparison of melanin bleaching and azure blue counterstaining in the immunohistochemical diagnosis of malignant melanoma.

Distinguishing heavily pigmented melanocytes from melanophages on routine hematoxylin and eosin slides can be difficult. Melanin bleaching with potassium permanganate solution is a traditional means of removing melanin from tissues and can be used before immunohistochemical staining to remove any pigment that might be confused with the brown chromogen diaminobenzidine. Azure B stains melanin granules green-blue, easily contrasts with diaminobenzidine, and may be used as a counterstain on unbleached sections after immunohistochemical staining. To our knowledge, studies comparing melanin bleaching with azure B counterstaining in the immunohistochemical evaluation of malignant melanomas have not been performed. Paraffin sections from 33 heavily pigmented malignant melanomas were bleached with a 3.0-g/L potassium permanganate solution, immunohistochemically stained for S-100 and HMB-45, and counterstained with hematoxylin. Unbleached sections were similarly stained for S-100 and HMB-45 and counterstained with azure B. To establish optimal permanganate concentrations, a variable number of sections were bleached with lower permanganate concentrations ranging from 0.125 to 2.5 g/L. S-100 antigenicity was preserved at all permanganate concentrations, whereas HMB-45 antigenicity was abolished at concentrations of 0.5 g/L and greater. At permanganate concentrations from 0.125 to 0.5 g/L, both antigenicities were preserved; however, melanin was incompletely removed. Complications of bleaching included tissue damage and loss of cytologic detail. Positive immunohistochemical staining was observed in azure B counterstained sections. Azure B stained melanin greenblue and was easily distinguished from the brown diaminobenzidine chromogen, regardless of the antibody tested. Neither tissue damage nor loss of cytologic detail was observed. We conclude that the use of azure B counterstaining is superior to permanganate bleaching in the histologic evaluation of heavily pigmented cutaneous malignant melanomas.     

Immunohistochemical study of melanocytic differentiation antigens in cutaneous malignant melanoma. A comparison of six commercial antibodies and one non-commercial antibody in nodular melanoma, superficially spreading melanoma and lentigo maligna melanoma

In certain primary and metastatic malignant melanomas diagnostic problems may arise due to their cytologic features and/or absence of synthesis of melanin. As the "classic" combination of S-100 protein and HMB-45 may occasionally fail to stain cells of malignant melanoma, we have tested a series of commercially accessible antibodies which were so far not compared by other authors in the three most frequent subtypes of this tumor. In surgical specimens from 104 cutaneous malignant melanomas (40 nodular melanomas, 46 superficially spreading malignant melanomas and 18 lentigo maligna melanomas) the staining intensity and the proportion of neoplastic cells stained with antibodies to S-100 protein, HMB-45, NKI/C3, NKI/beteb, MART 1 (Melan A), KBA 62 and Mitf was semiquantitatively analysed. The use of this group of antibodies against melanoma-associated antigens revealed it to be a favourable supplement for the bioptical or cytological diagnosis of malignant melanoma in case the traditional/conventional combination of S-100 protein and HMB-45 antibody fails. According to the authors' experience the antibody against KBA 62 has shown to be the most effective antibody followed by the antibodies against MART-1 (Melan A) and NKI/C3.     

Immunohistochemical diagnosis of sinonasal melanoma, carcinoma, and neuroblastoma with monoclonal antibodies HMB-45 and anti-synaptophysin

The efficacy of two new monoclonal antibodies with cell lineage-restricted reactivity (HMB-45 [melanocytes] and anti-synaptophysin [neuroepithelial cells]) was compared with that of "traditional" antibody panels in the delineation of malignant melanoma (MM) of the sinonasal region, nasopharyngeal carcinoma (NPC), and olfactory neuroblastoma (ONBL). HMB-45 recognized all of eight melanomas and stained one of five neuroblastomas, but failed to label any of 12 cases of NPC. All examples of ONBL were stained by anti-synaptophysin; other tumors were nonreactive with this reagent. A panel of antibodies to cytokeratin, vimentin, epithelial membrane antigen, and S100 protein was also effective in discriminating between MM, NPC and ONBL. These results suggest that HMB-45 and anti-synaptophysin are comparable in utility to more extended antibody panels in the diagnosis of sinonasal malignancies, but only if used in combination with one another.     

Malignant melanoma with a myxoid stroma: a diagnostic pitfall on fine-needle aspiration biopsy

Malignant melanoma (MM), both primary and metastatic, may be associated with a prominent myxoid stromal reaction causing diagnostic confusion on fine-needle aspiration biopsy (FNAB), most often with sarcomas that demonstrate a myxoid stroma, particularly malignant peripheral nerve sheath tumor (MPNST). We present a case of a 32-yr-old man with no past medical history who presented with a unilateral neck mass clinically suspicious for lymphoma. FNAB produced a specimen composed of large sheets of anaplastic cells encased in a myxoid stroma that was S100 and vimentin-positive but HMB-45-negative. A diagnosis of MPNST was made. Excision demonstrated a metastatic MM of unknown primary, with a prominent myxoid stromal reaction. A repeat HMB-45 was again negative. Electron microscopy demonstrated intracytoplasmic melanasomes and cisternae of rough endoplasmic reticulum with intracisternal parallel tubules, confirming the diagnosis. Although HMB-45 is typically negative in both tumors, S100 should be strongly positive in myxoid MM and only focal in MPNST.     

Primary mucosal malignant melanoma: an immunohistochemical study of 12 cases with comparison to cutaneous and metastatic melanomas

Immunohistochemical analysis of 40 formalin-fixed, paraffin-embedded malignant melanomas (12 primary mucosal, 16 primary cutaneous, and 12 metastatic cutaneous) was performed to study the possible differences in immunostaining profiles according to location. The majority of melanomas were reactive with a polyclonal antibody to S100 protein (P-S100; 85%), a monoclonal melanoma-specific antibody (HMB-45; 88%), and a monoclonal antibody to vimentin (90%), and there were no differences in staining profiles for these antibodies by anatomic location. In contrast, while 13 of 16 cutaneous melanomas (81%) and ten of 12 metastatic melanomas (83%) were reactive with a monoclonal antibody to S100 protein (MoAb-079), only five of 12 mucosal tumors (42%) showed positive staining for MoAb-079. Similarly, 14 cutaneous melanomas (88%) and 11 metastatic melanomas (92%) showed positive staining for neuron specific enolase (NSE), while only four mucosal melanomas (33%) were NSE-positive. Of the 40 melanomas, all but two were reactive with either P-S100, MoAb-079, or HMB-45. These findings suggest that MoAb-079 and NSE may be less sensitive markers than P-S100 and HMB-45 for routinely processed mucosal melanomas as compared with cutaneous and metastatic tumors.     

A primary amelanotic melanoma of the vagina, diagnosed by immunohistochemical staining with HMB-45, which recurred as a pigmented melanoma

Usually, malignant melanoma is readily diagnosed by the presence of melanin granules. Although amelanotic melanoma contains a few melanin granules, it is often difficult to differentiate from non-epithelial malignant tumours. This report describes a case of amelanotic melanoma of the vagina, which was originally suspected to be a non-epithelial malignant tumour, but was subsequently correctly diagnosed by immunohistochemical staining with the HMB-45 antibody and for the S-100 protein. A light grey tumour with superficial ulceration was located in the upper third of the vagina. The patient was treated with irradiation followed by chemotherapy. Subsequently, the tumour disappeared and cytology was negative; thus, she achieved complete remission. However, 20 months after complete remission, the tumour recurred locally: the site had a grossly black appearance, which was pathognomonic for a malignant melanoma. Thus, HMB-45 and S-100 protein immunohistochemistry confirmed the diagnosis of amelanotic melanoma.     

Immunohistochemical characterization of atypical fibroxanthoma and dermatofibrosarcoma protuberans.

Atypical fibroxanthoma (AFX) and dermatofibrosarcoma protuberans (DFSP) have generated undue interest regarding their histogenesis, biological behavior, and differentiation from other forms of spindle cell tumors of the skin, including spindle cell squamous carcinomas and desmoplastic melanomas. To identify characteristic immunophenotypes, 12 AFXs and 15 DFSPs were examined with a panel of antibodies against cytokeratin; vimentin; desmin; proteolytic enzymes (alpha-1-antitrypsin and alpha-1-antichymotrypsin); melanoma-associated antigens defined by HMB-45, HMB-50, and NKI/C3; muscle-specific actin (HHF-35); and S-100 protein. The staining patterns of these two tumors were nearly identical. All cases tested negative for cytokeratin, desmin, and S-100 protein and strongly positive for vimentin. Six (50%) AFXs and 12 (80%) DFSPs tested focally positive for muscle-specific actin. None of the cases were reactive with melanoma antibodies HMB-45 and HMB-50; NKI/C3 strongly stained 26 of 27 tumors. Compared to HMB-45 and HMB-50, NKI/C3 cross-reacted with nonmelanocytic neoplasms. Two AFXs stained for alpha-1-antitrypsin and alpha-1-antichymotrypsin. This study confirms (1) the immunophenotypic similarity of AFX and DFSP, (2) the presence of myofibroblastic differentiation in both tumors, as reflected by HHF-35 staining, and (3) that AFX and DFSP are easily distinguished from spindle cell squamous carcinoma and desmoplastic melanoma by the absence of cytokeratin, HMB-45, and HMB-50 staining.     

HMB-45 detection in adenocarcinomas.

Several studies have suggested that HMB-45 is a specific marker for melanoma, presumably due to its ability to detect a glycoprotein that is present in premelanosomes. The present study was conducted to evaluate whether HMB-45 is an absolutely specific antigenic determinant for melanoma and the role that testing with this antibody has in the differential diagnostic workup of amelanotic melanoma vs adenocarcinoma. Formaldehyde solution-fixed, paraffin-embedded tissue samples from 52 adenocarcinomas (primary or metastatic) and five melanomas (two primary and three metastatic) were immunostained with the use of a commercially available monoclonal antibody (MoAb), ie, HMB-45 (Enzo), a polyclonal antibody to S100 protein, a wide-spectrum keratin polyclonal antibody, and a keratin MoAb, ie, AE1/AE3. Approximately 10% (ie, 9.6%) of the adenocarcinomas (five cases) expressed HMB-45 with varied intensity and distribution. Positive primary tumors (n = 3) included one each from the breast, colon, and kidney; positive metastatic tumors (n = 2) included one each from the breast and endometrium. Fifty-two percent of the adenocarcinomas were positive for S100 protein. One renal carcinoma was negative for both keratins when tested with the AE1/AE3 MoAb and polyclonal antibody (Dako). This was the only adenocarcinoma that was negative when the keratin polyclonal antibody (Dako) was used. All but one additional adenocarcinoma demonstrated keratin expression when the AE1/AE3 MoAb was used for testing. This study showed that HMB-45 is not absolutely specific for melanoma. HMB-45 may react with some adenocarcinomas, at least when tested with the commercially available MoAb (Enzo). This fact, in conjunction with aberrant keratin expression by some melanomas and S100 protein expression by adenocarcinomas and other neoplasms other than melanomas, should be considered when antibody panels are evaluated in the workup of poorly differentiated tumors. However, HMB-45 appears to be the most specific marker that is available at the present time for supporting a diagnosis of melanoma.     

CEA immunoreactivity in metastatic malignant melanoma.

Immunohistochemical techniques may aid in the diagnosis of poorly differentiated metastatic tumors. Anti-carcinoembryonic antigen (CEA) antibodies have been used in the identification of epithelial neoplasms. However, recent unpublished data report CEA reactivity in malignant melanoma and melanoma cell lines. We studied 28 cases of known metastatic malignant melanoma with an antibody panel for CEA (polyclonal and monoclonal), AE1:3, S-100, and HMB-45. Reactivity for CEA (polyclonal) was seen in 15 of 28 (53%) cases: nine exhibited strong diffuse positivity, five moderate focal positivity, and one globular cytoplasmic staining. Focal reactivity for cytokeratin (AE1:3) was seen in three of 28 cases. HMB-45 staining was present in 23 of 28 (82%, including strong positivity in the cytokeratin-reactive cases). Staining for S-100 protein was strong in all cases. No staining was seen for CEA (monoclonal). CEA immunoreactivity is seen in a significant number of metastatic malignant melanoma cases. This may be due to CEA expression by tumor cells, or crossreactivity of the polyclonal antibody with substances such as nonspecific crossreacting antigen (NCA) that share antigenic sites with CEA. These findings emphasize the need for care in interpreting immunohistochemical results. Immunohistochemical evaluation of CEA should not be made alone, but only as part of a diagnostic antibody panel.     

Sclerosing hemangioma of the lung in a young woman with cutaneous melanoma: the role of electron microscopy in preventing an erroneous diagnosis of metastasis

A young woman with a melanoma of the left forearm was found to have a right lung mass. This was initially interpreted as metastatic melanoma on the basis of clinical, radiographic, and light microscopic features, together with positive staining of tumor cells with antibody HMB-45. Electron microscopic examination performed for confirmation of the diagnosis revealed no evidence of melanocytic differentiation. Instead, there were features suggestive of the alternative diagnosis of sclerosing hemangioma (SH). This diagnosis was confirmed with additional immunocytochemical stains. To the authors' knowledge this is the first report of HMB-45 positivity in SH. This case illustrates a potentially disastrous diagnostic pitfall in interpreting lung tumors in patients with melanoma, and the vital role of electron microscopy in resolving conflicting and/or misleading immunocytochemical results.     

Renal angiomyolipoma: further immunophenotypic characterization of an expanding morphologic spectrum

BACKGROUND: Renal angiomyolipoma is a benign tumor histologically characterized by proliferation of spindle cells, epithelioid cells, and adipocytic cells in concert with many thick-walled blood vessels. To add further diagnostic confusion, an epithelioid cell-predominant variant of renal angiomyolipoma has recently been described. HMB-45 immunoreactivity correlates with ultrastructural striated organelles that closely resemble premelanosomes, although no evidence of melanogenesis has been documented in this tumor. OBJECTIVE: To further characterize the immunophenotypic and ultrastructural profile of renal angiomyolipoma based on phenotypic cell type (epithelioid, spindle, and adipocytic cell). DESIGN: Formalin-fixed, paraffin-embedded tissues from 27 renal angiomyolipomas and 8 renal cell carcinomas were immunostained with monoclonal antibodies to the melanoma-associated antigens HMB-45, HMB-50, NKI/C3 (CD63), and tyrosinase; the smooth muscle-related antigens calponin and muscle-specific actin (HHF-35); S100; and cytokeratin (CK). All renal angiomyolipomas were also immunostained with a polyclonal antibody to renin. Ultrastructural examination was performed on 9 selected cases. RESULTS: All renal angiomyolipomas stained positive for HMB-45, HMB-50, NKI/C3, muscle-specific actin (HHF-35), and calponin. Overall, HMB-45, HMB-50, and NKI/C3 preferentially stained the epithelioid cells. Tyrosinase staining was present in 50% of the renal angiomyolipomas with adequate tissue for staining (12 of 24 cases); positive staining and intensity paralleled HMB-45, HMB-50, and NKI/C3. Muscle-specific actin (HHF-35) and calponin preferentially stained the spindle cells. The adipocytic cells stained positive for both melanoma-associated antigens and smooth muscle antigens. Epithelioid cells, spindle cells, and adipocytic cells were CK, S100, and renin negative. Ultrastructural findings paralleled immunohistochemical staining patterns. Premelanosome-like organelles and electron dense granules were more readily detected in the epithelioid cells within the tumor, whereas ultrastructural characteristics of smooth muscle cells were more easily found in the spindle cells. All renal cell carcinomas stained positive for CK, NKI/C3 staining was variable, and all were negative for HMB-45, HMB-50, smooth muscle actin (HHF-35), and calponin. CONCLUSION: In renal angiomyolipoma, the epithelioid and spindle cells have preferential staining patterns for melanoma-associated antigens versus smooth muscle antigens, respectively. Positivity in renal angiomyolipoma for HMB-50, NKI/C3, and tyrosinase, in addition to HMB-45, provides evidence for the presence of different melanoma-associated gene products. Immunophenotypic overlap of the 3 histologically distinct renal angiomyolipoma cell populations suggests a common cell line, supporting a unitarian concept for renal angiomyolipoma. Ultrastructural characteristics of the 3 renal angiomyolipoma cell phenotypes parallel the immunophenotype, giving further support to a common cell line. Our study lends further credence to the perivascular epithelioid cell concept as proposed by Bonetti and colleagues.     

Expression of gp100, MART-1, tyrosinase, and S100 in paraffin-embedded primary melanomas and locoregional, lymph node, and visceral metastases: implications for diagnosis and immunotherapy. A study conducted by the EORTC Melanoma Cooperative Group

With the recent availability of novel antibodies against melanoma antigens tyrosinase and MART-1, it is important to validate their usefulness in pathology practice and in screening patients for immunotherapy treatment. In the present study conducted by the Melanoma Cooperative Group of the European Organization for Research and Treatment of Cancer (EORTC-MCG), immunohistochemical staining for gp100 (antibodies NKI-beteb and HMB-45), MART-1 (A103), tyrosinase (T311), and S100 (S100) was compared on formalin-fixed and paraffin-embedded tumour lesions from 80 patients with 130 malignant melanoma lesions, comprising 44 primary tumours, 18 locoregional metastases, 41 lymph node metastases, and 27 visceral metastases from the lung, liver, and brain. A score between 0 and 5 was allocated to each immunohistochemically stained section. These scores were evaluated in a statistical analysis. S100 was by far the most sensitive marker in all four types of lesions tested. Apart from a significantly better performance for T311 in primary melanomas compared with HMB-45, no significant differences were observed between the four remaining antigens tested. Three settings were next investigated to determine whether the expression of melanoma antigens decreases with tumour progression. First, within the primary melanomas, only NKI-beteb and A103 staining showed a nearly significant negative correlation with Clark's level of invasion and a similar tendency was observed for these antibodies with Breslow thickness. Second, when comparing primary melanoma-metastasis pairs from the same patient, lymph node metastases showed less staining with NKI-beteb, HMB-45, A103, and T311, at a level near significance. This difference was not significant when comparing the primary tumour with visceral metastases, probably due to the lower numbers of pairs. Third, regarding tumour progression from primary melanoma to locoregional, to lymph node, to visceral metastasis, a significant decrease with progression was found only for T311. The apparently stable expression of most of the melanoma antigens, and the small contribution of decreased expression in melanoma tumour progression, supports the rationale for immunotherapy based on the melanoma immunogens gp100, MART-1, and tyrosinase.     

Sclerosing Hemangioma of the Lung in a Young Woman with Cutaneous Melanoma: The Role of Electron Microscopy in Preventing an Erroneous Diagnosis of Metastasis

A young woman with a melanoma of the left forearm was found to have a right lung mass. This was initially interpreted as metastatic melanoma on the basis of clinical, radiographic, and light microscopic features, together with positive staining of tumor cells with antibody HMB-45. Electron microscopic examination performed for confirmation of the diagnosis revealed no evidence of melanocytic differentiation. Instead, there were features suggestive of the alternative diagnosis of sclerosing hemangioma (SH). This diagnosis was confirmed with additional immunocytochemical stains. To the authors' knowledge this is the first report of HMB-45 positivity in SH. This case illustrates a potentially disastrous diagnostic pitfall in interpreting lung tumors in patients with melanoma, and the vital role of electron microscopy in resolving conflicting and/or misleading immunocytochemical results.