- Email: [email protected]
Immunohistochemical Detection of Melanoma-specific Antigens in Spontaneous Canine Melanoma
J. Comp. Path. 2002, Vol. 127, 162±168 doi:10.1053/jcpa.2002.0576, available online at http://www.idealibrary.com on
Immunohistochemical Detection of Melanoma-specific Antigens in Spontaneous Canine Melanoma S. S. Sulaimon, B. E. Kitchell and E. J. Ehrhart* Department of Veterinary Clinical Medicine and *Veterinary Diagnostic Laboratory, College of Veterinary Medicine, University of Illinois, 2001 South Lincoln Avenue, Urbana, IL 61801, USA Twenty-five formalin-fixed, paraffin wax-embedded canine melanomas were examined immunohistochemically by an immunoperoxidase method to assess their reactivity with three human melanoma-specific monoclonal antibodies (HMB-45, MEL-1, NK1/C3). HMB-45 and MEL-1 reacted with 22/25 (88%) and 18/25 (72%) of canine melanomas, respectively, but only after microwave antigen retrieval and pretreatment with potassium permanganate and oxalic acid (KMnO4/OA). Positive reactivity to HMB-45 and MEL-1 was as follows: oral melanomas, 13/16 and 9/16, respectively; cutaneous melanomas, 8/8 and 8/8; melanoma of the digit, 1/1 and 1/1; all pigmented melanomas, 16/18 and 14/18; all amelanotic tumours, 6/7 and 4/7. HMB-45 immunolabelling was characterizedbyadiffusegranularcytoplasmicpatternwithinthetumourcells,andMEL-1labellingbyacytoplasmic pattern with sporadic nuclear localization. In most tumours the labelling was homogeneous, but some showed a multifocal distribution. Generally, a higher percentage of canine melanomas was labelled by HMB-45 than by MEL-1. NK1/C3 failed to label any of seven melanomas tested, regardless of KMnO4/OA-pretreatment. Of16non-melanocytictumours(specificitycontrols), 15showed nosignficant reactivity withHMB-45,theexception being one of three plasmacytomas. Immunolabelling by MEL-1 on non-melanocytic tumours was not pursued, due to the poor sensitivity of this antibody as compared with that of HMB-45.
# 2002 Elsevier Science Ltd. All rights reserved.
Introduction Canine malignant melanoma, which is the most common oral malignancy in dogs, usually arises in the lips and gingiva but may also occur in the palatine, labial or buccal mucosa (Brodey, 1960). Canine oral melanomas are uniformly malignant and have no more than a fair to poor prognosis as compared with cutaneous melanomas, which are mostly benign (Bostock, 1979). Digital (nail bed) melanomas in dogs are rare and have a high metastatic rate (Marino et al., 1995). Melanoma is the most common intraocular tumour in dogs (Ryan and Diters, 1984). Canine malignant melanoma is a life-threatening disease, with up to 80% of affected dogs developing metastasis (Todoroff and Brodey, 1979). Because of Correspondence to: E. J. Ehrhart. 0021±9975/02/$ ± see front matter
the poor prognosis associated with this tumour, an accurate method of identification is needed. It may be difficult to distinguish amelanotic melanoma from other anaplastic tumours. Traditional histochemical stains such as Fontana-Masson or other silver stains frequently fail to identify amelanotic melanomas because melanosomes are often absent, are few in number, or are immature and lack pigment. Immunohistochemistry has proved invaluable in the diagnosis of neoplastic disease. The routine immunomarkers used to identify canine melanomas include antibodies directed against S-100 protein, vimentin, and neuron specific enolase. Positive immunolabelling for these proteins, in combination with the absence of cytokeratin immunoreactivity, supports the diagnosis of melanoma. Unfortunately, these markers lack specificity for melanoma and react with many other tumours. The S-100 protein, which was initially # 2002 Elsevier Science Ltd. All rights reserved.
Melanoma-specific Antigens in Canine Melanomas
introduced as a marker specific for human neural tissues, has since been discovered in non-neural cells (e.g., dendritic cells, cells of the salivary and sweat glands, astrocytes, oligodendrocytes, Schwann cells, Langerhans cells, adipocytes, chondrocytes and myoepithelial cells) and tumours that arise from them (Nakajima et al., 1982a, b; Vanstapel et al., 1986). Thus, although S-100 protein is the most sensitive melanoma immunomarker in common use, its specificity is low. The panel of immunohistochemical markers used in human medicine to diagnose melanoma often includes the monoclonal antibodies (mAbs), HMB45, MEL 1 and NK1/C3. HMB-45 has proved useful in distinguishing amelanotic or scantly pigmented melanoma cells from other malignant tumour cells (Gown et al., 1986). In veterinary pathology, HMB-45 has been reported to lack cross-reactivity with canine melanoma tissues (Berrington et al., 1994). MEL-1 recognizes a major cell-surface antigen, ganglioside 3 (GD 3), which is expressed in higher concentrations in melanoma cells than in normal cells (Hamilton et al., 1993). NK1/C3 recognizes a heterogeneous glycoprotein located mainly on the inner side of membranes of cytoplasmic vesicles in human melanoma cells (Vennegoor et al., 1985). The aim of this study was to evaluate the crossreactivity, sensitivity and specificity of human melanoma immunomarkers (HMB-45, MEL-1 [R-24] and NK1/C3), with the ultimate goal of improving the diagnosis of canine melanoma. Materials and Methods Tumours Formalin-fixed paraffin wax-embedded blocks from 25 canine malignant melanomas (16 oral, eight cutaneous and one digital) and 16 non-melanocytic tumours (three plasmacytomas, three fibrosarcomas, three squamous cell carcinomas, one malignant histiocytosis, three basal cell tumours and three poorly differentiated carcinomas) were selected from the pathology archives of the Veterinary Diagnostic Laboratory, College of Veterinary Medicine at the University of Illinois, Urbana-Champaign. These tissues were submissions acquired between 1996 and 1998. The selection of these tumours for immunohistochemical analysis was based on the available clinical information, including history, tumour location, and morphological characteristics, and a haematoxylin and eosin (HE)-stained section for confirmation of the histological diagnosis. All HE-stained slides were reviewed by one pathologist (EJE). For immunolabelling, sections (4±5 mm) were cut from paraffin wax
163
blocks of tissues previously fixed in 4% phosphatebuffered formaldehyde. Sections were mounted on positively charged poly-lysine-coated glass slides and dried for 45 min in a 60 C oven. Tissue sections were dewaxed with xylene and rehydrated through graded alcohol baths. Endogenous peroxidase activity was quenched by incubating sections in hydrogen peroxide 3% in methanol for 30 min at room temperature. Slides were rinsed in Tris-buffered saline pH 76 (TBS; Trizma HCl 96 mM, Trizma base 29 mM, NaCl 347 mM) for 5 min. All slides were subjected to microwave antigen retrieval before immunolabelling. For this purpose, sections were placed in citrate buffer solution, pH 64 (citric acid 2 mM, sodium citrate 89 mM) and heated in a microwave oven (800 watts) at high power until the covering solution was brought to the boil. The microwave was subsequently set at a power level of 4 for 10 min. The slides were then cooled in citrate buffer solution at room temperature for 30 min. Bleaching The presence of melanin in pigmented melanoma makes it difficult to discern any immunoperoxidase diaminobenzidine labelling pattern. Bleaching of heavy accumulations of melanin with potassium permanganate is routinely used to decolorize melanin pigment. Sections subjected to microwave antigen retrieval from each block were divided into two categories: those pretreated with potassium permanganate (KMnO4) and oxalic acid (OA), and those not pretreated. Pretreatment consisted of immersing the slides in a 025% KMnO4 bath for 30 min at room temperature, followed by a running water rinse for 5 min, immersion in 5% OA for at least 2±5 min or until the section turned colourless, and a final TBS rinse. Immunohistochemistry (IHC) After the antigen retrieval and bleaching steps, IHC was performed by a biotin-streptavidin immunoperoxidase technique. All sections were labelled in three independent experimental runs. To assess the effect of KMnO4/OA pretreatment after microwave antigen retrieval, pretreated and non-pretreated sections were run simultaneously and incubated either with the primary antibody or with an irrelevant but isotypespecific antibody as a negative control for non-specific binding. The primary murine mAbs used were: HMB-45 (Signet Laboratories, Dedham, MA, USA), an IgG1 mouse monoclonal antibody; MEL-1 (SK-MEL-28
164
S. S. Sulaimon et al.
melanoma cell line; Signet Laboratories), an IgG3 monoclonal antibody; and NK1/C3 (BioGenex, San Ramon, CA, USA), an IgG1 mouse monoclonal antibody. A range of dilutions was tested for each primary antibody, the dilution (1 in 10) giving distinct immunolabelling with the least background being chosen as the working dilution for all three antibodies. The diluent was phosphate-buffered saline plus a carrier protein in 09% sodium azide (BioGenex). Negative control antibodies consisted of a purified fraction of IgG1 (Biosource International, Carmarillo, CA, USA) used at 100 mg/ml for HMB-45 and NK1/C3, and a rat antimouse IgG3 (Biodesign, Kennenbunk, ME, USA) used at 01 mg/01 ml for MEL-1. A commercially available labelling kit, consisting of a prediluted biotinylated goat antimouse IgG secondary antibody and a peroxidase-conjugated streptavidin Super Sensitive label (BioGenex), was used to label the primary antibodies. A basic immunolabelling procedure was followed. KMnO4/OA-pretreated and non-pretreated sections were incubated with a universal blocking reagent (Power Block1 ; BioGenex) for 10 min at room temperature, followed by a TBS rinse. Pretreated and non-pretreated slides were then incubated overnight (18 h) with the primary (HMB 45, MEL-1 or NK1/ C3) or irrelevant isotype-specific antibodies at room temperature in a humidity chamber. After incubation and a 3-min TBS rinse, all sections were incubated at room temperature with the biotinylated secondary antibody for 30 min, followed by application of the peroxidase-conjugated streptavidin label. Sections were finally incubated in 3,30 -diaminobenzidine chromogen (DAB) for 5 min, counterstained lightly in Mayer's haematoxylin, dehydrated, cleared and mounted. The streptavidin-peroxidase immunohistochemical technique was used to detect reactivity of HMB 45,
MEL 1 and NK1/C3 monoclonal antibodies with canine melanoma samples. The result was considered positive when immunolabelling by HMB 45, MEL-1 or NK1/C3 was restricted to the neoplasm and when no labelling was detected in the absence of primary antibody, or when irrelevant isotype specific antibody was substituted. Of the three markers, one (HMB-45) was further evaluated against 16 non-melanocytic canine tumours to determine its specificity. Results Canine Melanomas (n 25) The mAbs HMB-45 and MEL-1 reacted with 22/25 (88%) and 18/25 (72%) of the canine melanomas, respectively, but only after microwave antigen retrieval and KMnO4/OA-pretreatment (Table 1). Positive reactivity to HMB-45 and MEL-1 was shown as follows: oral melanomas, 13/16 and 9/16, respectively; cutaneous melanomas, 8/8 and 8/8; melanoma of the digit, 1/1 and 1/1; all pigmented melanomas, 16/18 and 14/18; all amelanotic tumours, 6/7 and 4/7 (Table 1). Non-pretreated sections incubated with the appropriate irrelevant isotype-specific antibody, or HMB-45 (Fig. 1a), or MEL-1 (Fig. 2a) showed no immunoreactivity. Nonspecific staining of apocrine and sebaceous glands was observed. This was probably due to nonspecific ionic binding of the antibody to secretory products. KMnO4/OA-pretreated sections did not show significant immunolabelling when the appropriate irrelevant antibodies were substituted for the primary antibody; however, positive labelling with HMB-45 mAb on pretreated sections revealed a diffuse granular cytoplasmic pattern within the tumour cells (Fig. 1b, c). MEL-1 reactivity was characterized by a cytoplasmic distribution with sporadic
Table 1 Immunohistochemical labelling of melanomas for presence of melanin and tumor location Melanomas
Oral (n 16) Pigmented (n 11) Amelanotic (n 5) Cutaneous (n 8) Pigmented (n 6) Amelanotic (n 2) Digit, pigmented (n 1) All (n 25) Pigmented (n 18) Amelanotic (n 7) N/A, not applicable.
Number of positive melanomas/number tested, with stated mAb HMB-45
MEL-1
NK1/C3
13/16 (81%) 9/11 4/5 8/8 6/6 2/2 1/1 22/25 (88%) 16/18 (89%) 6/7 (86%)
9/16 (56%) 7/11 2/5 8/8 6/6 2/2 1/1 18/25 (72%) 14/18 (78%) 4/7 (57%)
0/3 N/A N/A 0/4 N/A N/A N/A 0/7 (0%) ... ...
Melanoma-specific Antigens in Canine Melanomas
165
Fig. 1a±c. Canine oral and cutaneous amelanotic melanomas examined immunohistochemically with HMB-45. (a) No immunolabelling of oral melanoma without KMnO4/OA-pretreatment. (b) Immunolabelling (diffuse, cytoplasmic) of oral melanoma occurs after KMnO4/OA-pretreatment. (c) Labelling (diffuse, cytoplasmic) of cutaneous melanoma occurs after KMnO4/ OA-pretreatment. V, vessel. Immunoperoxidase/DAB, with Mayer's haematoxylin counterstain. Bar, 40 mm.
Fig. 2a, b. Canine amelanotic melanoma examined immunohistochemically with MEL-1. (a) No immunolabelling without KMnO4/OA-pretreatment. (b) Immunolabelling (cytoplasmic and nuclear) occurs after KMnO4/OA-pretreatment. Immunoperoxidase/DAB, with Mayer's haematoxylin counterstain. Bar, 40 mm.
166
S. S. Sulaimon et al.
nuclear labelling (Fig. 2b). Distribution of the label within most tumours was homogeneous; some tumours, however, had a multifocal pattern. HMB45 also labelled resident melanocytes within the overlying epidermis. Three out of 25 and 7/25 of the melanomas showed no reactivity with either HMB-45 or MEL-1, respectively. In positive melanomas, however, the percentages of cells labelled by HMB-45 and MEL-1 were as follows: ,25%, 2/25 and 1/25 melanomas, respectively; 25±75%, 15/25 and 16/25; .75%, 5/25 and 1/25. Generally, HMB-45 immunolabelled a higher percentage of canine melanoma cells than did MEL-1. The NK1/C3 antibody did not label any of the initial seven melanoma tissues tested, irrespective of KMnO4/OA-pretreatment, and was not used further. Non-melanocytic Canine Tumours Immunolabelling of 16 non-melanocytic tumours was performed to determine the specificity of HMB-45. Fifteen of sixteen such tumours showed no significant reactivity with HMB-45, labelling being consistently absent in fibrosarcomas, squamous cell carcinomas, poorly differentiated carcinomas, basal cell tumours, and a malignant histiocytosis. One of three plasmacytomas reacted positively. This plasmacytoma showed homogeneous labelling which was moderate, granular and cytoplasmic. Background labelling, observed as a light brown film across the entire section, was evident in three fibrosarcomas and one squamous cell carcinoma. Immunolabelling of non-melanocytic tumours with MEL-1 was not pursued, due to the poor sensitivity of this antibody for detection of canine melanomas as compared with that of HMB-45.
Discussion The distinction of melanomas from a variety of poorly differentiated tumours is a dilemma that faces diagnosticpathologists.Intracytoplasmicmelanin,adistinct marker for melanocytic tumours, is not synthesized in amelanotic subtypes. Problems in diagnosis may arise when melanin is confused with other pigments or when cells are extremely pleomorphic. The diverse cytological features of melanoma cells can be attributed to common embryological roots shared with neural and epithelial cells (Moulton, 1990). The melanocytic phenotype in amelanotic melanoma is difficult to confirm without ultrastructural identification of premelanosomes or melanosomes. In a diagnostic setting, however, electron microscopy is often unavailable.
Immunohistochemistry has become a cost-effective method when the diagnostic pathologist is confronted with poorly differentiated tumours. Monoclonal antibodies which have aided the immunohistochemical diagnosis of human and canine melanomas include: S-100, HMB-18, HMB-45, HMB-50, R24 (MEL-1), HMSA-1, HMSA-5, NK1/C3, NK1/beteb, EPI-3, 225285, Melan A, IBF9, cytokeratin, vimentin, and neuron specific enolase (Burnier et al., 1991; Oliver and Wolfe, 1992; Berrington et al., 1994; Fernando et al., 1994; Mottolese et al., 1994). Traditionally, antibodies against the S-100 protein have been the immunomarkers of choice, but S-100 protein has been identified in human neural and non-neural cells such as melanocytes, adipocytes, myoepithelial cells, Schwann cells, skeletal muscle cells, chondrocytes, eccrine and apocrine gland cells, and Langerhans cells (Nakajima et al., 1982a, b). Tumours arising from any of these cell types may react with anti-S-100 antibodies, thereby preventing their use as reagents specific for melanomas (OrdonÄez et al., 1988; Burnier et al., 1991). HMB-45 is generally considered to be very specific but inferior in sensitivity to S-100 antibodies for the diagnosis of human melanomas (Skeleton et al., 1991; Blessing et al., 1998). However, Wick et al. (1988) and OrdonÄez et al. (1988) reported 93% (62/67) and 91% (32/35) sensitivity for HMB-45, respectively. A few mAbs against human melanoma-associated antigens cross-react with canine melanomas. These include the HMSA-1, HMSA-5, and Melan A antibodies (Berrington et al., 1994; Ramos-Vara et al., 2000). IBF9, which recognizes a cell membrane molecule, is the only reported mAb to have been generated directly against canine melanomas (Oliver and Wolfe, 1992). IBF9 reacted positively with 63% (24/38) of canine melanomas tested. However, 12 of 40 (30%) non-melanocytic tumours reacted with IBF9, including cutaneous lymphosarcomas, squamous cell carcinomas, basal cell tumours, epulides and mammary gland tumours. This study demonstrated specific immunoreactivity of canine melanomas with HMB-45 and MEL-1 in formalin-fixed, paraffin wax-embedded tissue after microwave antigen retrieval and KMnO4/OApretreatment. The value of such pretreatment is threefold: it bleaches melanin pigment to facilitate better detection of immunolabelling; it retrieves the antigen masked by formalin fixation; and it blocks endogenous peroxidase activity (Alexander et al., 1986). The combination of microwave antigen retrieval and KMnO4/OA-pretreatment caused some distortion of cellular and nuclear features. HMB-45 was tested on 25 canine melanomas and 16 non-melanocytic tumours to evaluate its sensitivity, specificity and
Melanoma-specific Antigens in Canine Melanomas
labelling pattern. HMB-45 labelled 22/25 (88%) of the melanomas and 1/16 (6%) of the non-melanocytic tumours, suggesting that it was both sensitive and specific for canine melanomas. Immunolabelling by HMB-45 had a diffuse granular intracytoplasmic pattern similar to that reported in human melanoma. Pigmented tumours of this series showed increased labelling intensity. Melanocytes in the epidermis were reactive with HMB-45. Smoller et al. (1989) observed that melanocytes in the epidermis overlying some human melanomas were reactive with HMB-45 and hypothesized that this was due to stimulation of the epidermal melanocytes by extrinsic factors produced by tumour cells. HMB-45 appeared to be specific for canine melanomas, with 15/16 of the non-melanocytic tumours lacking any reactivity. One of three plasmacytomas, however, showed HMB-45 immunoreactivity. Positive HMB-45 labelling has been reported in human plasmacytomas, breast and endometrial adenocarcinomas, colonic and rectal adenocarcinomas, immunoblastic lymphomas, olfactory neuroblastomas, renal angiomyolipomas, and basal cell carcinomas (Bonetti et al., 1989, 1991; Smoller et al., 1989; Kornstein and Franco, 1990; Friedman and Tatum, 1991; Hancock et al., 1991; Pea et al., 1991; Herrera et al., 1992). Cross-reactivity has been attributed to antigenically related glycoprotein epitopes in these neoplasms. Alternatively, expression of oncofetal antigens containing sialic acids in an immunoreactive epitope similar to that recognized by HMB-45 may lead to cross-reactivity (Bonetti et al., 1991; Hancock et al., 1991). The MEL-1 mAb detects the cell surface disialoganglioside, GD3. Gangliosides are glycolipid molecules inserted into the cell membrane of melanoma cells and cancer cells originating from the neuroectoderm. Immunoreactivity with this mAb has been reported only in frozen sections and appears to be lost after formalin-fixation. In this study, MEL-1 immunolabelled 72% (18/25) of formalin-fixed canine melanoma samples after microwave antigen retrieval and KMnO4/OA-pretreatment, suggesting that these procedures expose epitopes masked by formalin fixation-induced protein cross-linkage. MEL-1 appears to be less sensitive than HMB-45 for labelling canine melanomas. The mAb NK1/C3 recognizes a disulphidedependent glycoprotein which is located predominantly on the inner side of the membranes of cytoplasmic vesicles. NK1/C3 did not label any of seven melanocytic tumours and was not examined further. Possible reasons for the lack of immunolabelling include the following: lack of expression of the glycoprotein recognized by NK1/C3; expression of
167
antigens occurred but was below the level needed for detection; the antigens were masked or destroyed during formalin-fixation and tissue processing. In human melanomas, lack of labelling with NK1/C3 has been associated with high metastatic potential (Vennengoor et al., 1985). In conclusion, this study demonstrated immunolabelling with HMB-45 and MEL-1 in formalin-fixed canine melanoma tissues after microwave antigen retrieval and KMnO4/OA-pretreatment. HMB-45 immunolabelling appeared both sensitive and specific, with rare cross-reactivity in plasmacytomas. In contrast, MEL-1 appeared less sensitive in identifying canine melanomas. References Alexander, R. A., Hiscott, P. S., Hart, R. I. and Grieson, I. (1986). Effect of melanin bleaching on immunoperoxidase with reference to ocular tissues and lesions. Medical Laboratory Science, 43, 121±127. Berrington, A., Jimbow, K. and Haines, D. M. (1994). Immunohistochemical detection of melanomaassociated antigens on formalin-fixed paraffin embedded canine tumours. Veterinary Pathology, 31, 455±461. Blessing, K., Sanders, S. A., and Grant, J. J. H. (1998). Comparison of immunohistochemical staining of novel antibody melan-A with S-100 protein and HMB-45 in malignant melanoma and melanoma variants. Histopathology, 32, 139±146. Bonetti, F., Colombari, R., Manfrin, E., Zamboni, G., Martignoni, G., Mombello, A. and Chilosi, M. (1989). Breast carcinoma with positive results for melanoma marker (HMB-45): HMB-45 immunoreactivity in normal and neoplastic breast. American Journal of Clinical Pathology, 92, 491±495. Bonetti, F., Pea, M., Martignoni, G., Mombello, A., Colombari, R., Zamboni, G., Scarpa, A., Piubello, Q., Bacchi, C. E. and Gown, A. M. (1991). False positive immunolabelling of normal epithelium and carcinomas with ascites fluid preparations of anti-melanoma antibody HMB-45. American Journal of Clinical Pathology, 95, 454±459. Bostock, D. E. (1979). Prognosis after surgical excision of canine melanomas. Veterinary Pathology, 16, 32±40. Brodey R. S. (1960). Clinical and pathologic study of 130 neoplasms of the mouth and pharynx in the dog. American Journal of Veterinary Research, 21, 787±818. Burnier, M. N., McLean, I. W. and Gamel, J. W. (1991). Immunohistochemical evaluation of uveal melanocytic tumours. Expression of HMB-45, S-100 protein and neuron-specific enolase. Cancer, 68, 809±814. Fernando, S. S. E., Johnson, S. and Mate, J. (1994). Immunohistochemical analysis of cutaneous malignant melanoma: comparison of S-100 protein, HMB-45 monoclonal antibody and NK1/C3 monoclonal antibody. Pathology, 26, 16±19. Friedman, H. D. and Tatum, A. (1991). HMB-45 detection in adenocarcinomas. Archives of Pathology and Laboratory Medicine, 115, 826±830.
168
S. S. Sulaimon et al.
Gown, A. M., Vogel, A. M., Hoak, D., Gough, F. and McNutt, M. A. (1986). Monoclonal antibodies specific for melanocytic tumours distinguish subpopulations of melanocytes. American Journal of Pathology, 123, 195±203. Hamilton, W. B., Helling, F., Lloyd, K. O. and Livingston, P. O. (1993). Ganglioside expression on human malignant melanoma assessed by quantitative immune thin-layer chromatography. International Journal of Cancer, 53, 566±573. Hancock, C., Allen, B. C. and Herrera, G. A. (1991). HMB-45 detection in adenocarcinomas. Archives of Pathology and Laboratory Medicine, 115, 886±890. Herrera, G. A., Hancock, C. and Allen, B. C. (1992). Specificity of antibody HMB-45. Archives of Pathology and Laboratory Medicine, 116, 900±901. Kornstein, M. J. and Franco, A. P. (1990). Specificity of HMB-45. Archives of Pathology and Laboratory Medicine, 114, 450. Marino, D. J., Matthiesen, D. T., Stefanacci, J. D. and Moroff, S. D. (1995). Evaluation of dogs with digit masses: 117 cases (1981±1991). Journal of the American Veterinary Medical Association, 207, 726±728. Mottolese M., Venturo, I., Benevolo, F., Di Fillippo, F., Lopez, M., Bigotti, A. and Natali, P. G. (1994). Immunocytochemical diagnosis of amelanotic metastatic melanoma using monoclonal antibodies HMB-45 and EP1-3. Melanoma Research, 4, 53±58. Moulton, J. E. (1990). Tumours in Domestic Animals, 3rd Edit., University of California Press, Berkeley, pp. 75±87. Nakajima, T., Watanabe, S., Sato, Y., Kameya, T., Shimosato, Y. and Ishihara, K. (1982a). Immunohistochemical demonstration of S-100 protein in malignant melanoma and pigmented nevus, and its diagnostic applications. Cancer, 50, 912±918. Nakajima, T., Watanabe, S., Sato, Y., Kameya, T., Shimosato, Y. and Ishihara, K. (1982b). An immunoperoxidase study of S-100 protein distribution in normal and neoplastic tissues. American Journal of Surgical Pathology, 6, 715±727. Oliver, J. L. and Wolfe, L. G. (1992). Antigen expression in canine tissues, recognized by monoclonal antibodies generated against canine melanoma cells. American Journal of Veterinary Research, 53, 123±128. OrdonÄez, N. G., Xiaolong, J. and Hickey, R. C. (1988). Comparison of HMB-45 monoclonal antibody and S-100 protein in the immunohistochemical diagnosis of melanoma. American Journal of Clinical Pathology, 90, 385±390.
Pea, M., Bonetti, F., Zamboni, G., Martignoni, G., Riva, M., Colombari, R., Mombello, A., Bonzanini, M., Scarpa, A. and Ghimenton, C. (1991). Melanocytemarker-HMB-45 is regularly expressed in angiomyolipoma of the kidney. Pathology, 23, 185±188. Ramos-Vara, J. A., Beissenherz, M. E., Miller, M. A., Johnson, G. C., Pace, L. W., Fard, A. and Kottler, S. J. (2000). Retrospective study of 338 canine oral melanomas with clinical, histologic, and immunohistochemical review of 129 cases. Veterinary Pathology, 37, 597±608. Ryan, A. M. and Diters R. W. (1984). Clinical and pathologic features of canine ocular melanomas. Journal of the American Veterinary Medical Association, 184, 60±67. Skeleton, H. G., Smith, K. J., Barrett, T. L., Lupton, G. P. and Graham, J. H. (1991). HMB-45 staining in benign and malignant melanocytic lesions. A reflection of cellular activation. American Journal of Dermatopathology, 13, 543±550. Smoller, B. R., McNutt, N. S. and Hsu, A. (1989). HMB45 recognizes stimulated melanocytes. Journal of Cutaneous Pathology, 16, 49±53. Todoroff, R. J. and Brodey, R. S. (1979). Oral and pharyngeal neoplasia in the dog. A retrospective survey of 361 cases. Journal of the American Veterinary Medical Association, 175, 567±571. Vanstapel, M. U., Gatter, K., De Wolf-Peeters, C., Mason, D. and Doesmet, V. (1986). New sites of human S-100 immunoreactivity detected with monoclonal antibodies. American Journal of Clinical Pathology, 85, 160±168. Vennengoor, C., Calafat, J., Hageman, P. H., Van Buitenen, F., Janssen, H., Kolk, A. and Rumke, P. H. (1985). Biochemical characterization and cellular localization of a formalin-resistant melanomaassociated antigen reacting with monoclonal antibody NK1/C3. International Journal of Cancer, 35, 287±295. Wick, M. R., Swanson, P. E. and Rocamora, A. (1988). Recognition of malignant melanoma by monoclonal antibody HMB-45. An immunohistochemical study of 200 paraffin-embedded cutaneous tumours. Journal of Cutaneous Pathology, 15, 201±207. Received; June 27th; 2001 Accepted; April 29th; 2002
Immunohistochemical Detection of Melanoma-specific Antigens in Spontaneous Canine Melanoma S. S. Sulaimon, B. E. Kitchell and E. J. Ehrhart* Department of Veterinary Clinical Medicine and *Veterinary Diagnostic Laboratory, College of Veterinary Medicine, University of Illinois, 2001 South Lincoln Avenue, Urbana, IL 61801, USA Twenty-five formalin-fixed, paraffin wax-embedded canine melanomas were examined immunohistochemically by an immunoperoxidase method to assess their reactivity with three human melanoma-specific monoclonal antibodies (HMB-45, MEL-1, NK1/C3). HMB-45 and MEL-1 reacted with 22/25 (88%) and 18/25 (72%) of canine melanomas, respectively, but only after microwave antigen retrieval and pretreatment with potassium permanganate and oxalic acid (KMnO4/OA). Positive reactivity to HMB-45 and MEL-1 was as follows: oral melanomas, 13/16 and 9/16, respectively; cutaneous melanomas, 8/8 and 8/8; melanoma of the digit, 1/1 and 1/1; all pigmented melanomas, 16/18 and 14/18; all amelanotic tumours, 6/7 and 4/7. HMB-45 immunolabelling was characterizedbyadiffusegranularcytoplasmicpatternwithinthetumourcells,andMEL-1labellingbyacytoplasmic pattern with sporadic nuclear localization. In most tumours the labelling was homogeneous, but some showed a multifocal distribution. Generally, a higher percentage of canine melanomas was labelled by HMB-45 than by MEL-1. NK1/C3 failed to label any of seven melanomas tested, regardless of KMnO4/OA-pretreatment. Of16non-melanocytictumours(specificitycontrols), 15showed nosignficant reactivity withHMB-45,theexception being one of three plasmacytomas. Immunolabelling by MEL-1 on non-melanocytic tumours was not pursued, due to the poor sensitivity of this antibody as compared with that of HMB-45.
# 2002 Elsevier Science Ltd. All rights reserved.
Introduction Canine malignant melanoma, which is the most common oral malignancy in dogs, usually arises in the lips and gingiva but may also occur in the palatine, labial or buccal mucosa (Brodey, 1960). Canine oral melanomas are uniformly malignant and have no more than a fair to poor prognosis as compared with cutaneous melanomas, which are mostly benign (Bostock, 1979). Digital (nail bed) melanomas in dogs are rare and have a high metastatic rate (Marino et al., 1995). Melanoma is the most common intraocular tumour in dogs (Ryan and Diters, 1984). Canine malignant melanoma is a life-threatening disease, with up to 80% of affected dogs developing metastasis (Todoroff and Brodey, 1979). Because of Correspondence to: E. J. Ehrhart. 0021±9975/02/$ ± see front matter
the poor prognosis associated with this tumour, an accurate method of identification is needed. It may be difficult to distinguish amelanotic melanoma from other anaplastic tumours. Traditional histochemical stains such as Fontana-Masson or other silver stains frequently fail to identify amelanotic melanomas because melanosomes are often absent, are few in number, or are immature and lack pigment. Immunohistochemistry has proved invaluable in the diagnosis of neoplastic disease. The routine immunomarkers used to identify canine melanomas include antibodies directed against S-100 protein, vimentin, and neuron specific enolase. Positive immunolabelling for these proteins, in combination with the absence of cytokeratin immunoreactivity, supports the diagnosis of melanoma. Unfortunately, these markers lack specificity for melanoma and react with many other tumours. The S-100 protein, which was initially # 2002 Elsevier Science Ltd. All rights reserved.
Melanoma-specific Antigens in Canine Melanomas
introduced as a marker specific for human neural tissues, has since been discovered in non-neural cells (e.g., dendritic cells, cells of the salivary and sweat glands, astrocytes, oligodendrocytes, Schwann cells, Langerhans cells, adipocytes, chondrocytes and myoepithelial cells) and tumours that arise from them (Nakajima et al., 1982a, b; Vanstapel et al., 1986). Thus, although S-100 protein is the most sensitive melanoma immunomarker in common use, its specificity is low. The panel of immunohistochemical markers used in human medicine to diagnose melanoma often includes the monoclonal antibodies (mAbs), HMB45, MEL 1 and NK1/C3. HMB-45 has proved useful in distinguishing amelanotic or scantly pigmented melanoma cells from other malignant tumour cells (Gown et al., 1986). In veterinary pathology, HMB-45 has been reported to lack cross-reactivity with canine melanoma tissues (Berrington et al., 1994). MEL-1 recognizes a major cell-surface antigen, ganglioside 3 (GD 3), which is expressed in higher concentrations in melanoma cells than in normal cells (Hamilton et al., 1993). NK1/C3 recognizes a heterogeneous glycoprotein located mainly on the inner side of membranes of cytoplasmic vesicles in human melanoma cells (Vennegoor et al., 1985). The aim of this study was to evaluate the crossreactivity, sensitivity and specificity of human melanoma immunomarkers (HMB-45, MEL-1 [R-24] and NK1/C3), with the ultimate goal of improving the diagnosis of canine melanoma. Materials and Methods Tumours Formalin-fixed paraffin wax-embedded blocks from 25 canine malignant melanomas (16 oral, eight cutaneous and one digital) and 16 non-melanocytic tumours (three plasmacytomas, three fibrosarcomas, three squamous cell carcinomas, one malignant histiocytosis, three basal cell tumours and three poorly differentiated carcinomas) were selected from the pathology archives of the Veterinary Diagnostic Laboratory, College of Veterinary Medicine at the University of Illinois, Urbana-Champaign. These tissues were submissions acquired between 1996 and 1998. The selection of these tumours for immunohistochemical analysis was based on the available clinical information, including history, tumour location, and morphological characteristics, and a haematoxylin and eosin (HE)-stained section for confirmation of the histological diagnosis. All HE-stained slides were reviewed by one pathologist (EJE). For immunolabelling, sections (4±5 mm) were cut from paraffin wax
163
blocks of tissues previously fixed in 4% phosphatebuffered formaldehyde. Sections were mounted on positively charged poly-lysine-coated glass slides and dried for 45 min in a 60 C oven. Tissue sections were dewaxed with xylene and rehydrated through graded alcohol baths. Endogenous peroxidase activity was quenched by incubating sections in hydrogen peroxide 3% in methanol for 30 min at room temperature. Slides were rinsed in Tris-buffered saline pH 76 (TBS; Trizma HCl 96 mM, Trizma base 29 mM, NaCl 347 mM) for 5 min. All slides were subjected to microwave antigen retrieval before immunolabelling. For this purpose, sections were placed in citrate buffer solution, pH 64 (citric acid 2 mM, sodium citrate 89 mM) and heated in a microwave oven (800 watts) at high power until the covering solution was brought to the boil. The microwave was subsequently set at a power level of 4 for 10 min. The slides were then cooled in citrate buffer solution at room temperature for 30 min. Bleaching The presence of melanin in pigmented melanoma makes it difficult to discern any immunoperoxidase diaminobenzidine labelling pattern. Bleaching of heavy accumulations of melanin with potassium permanganate is routinely used to decolorize melanin pigment. Sections subjected to microwave antigen retrieval from each block were divided into two categories: those pretreated with potassium permanganate (KMnO4) and oxalic acid (OA), and those not pretreated. Pretreatment consisted of immersing the slides in a 025% KMnO4 bath for 30 min at room temperature, followed by a running water rinse for 5 min, immersion in 5% OA for at least 2±5 min or until the section turned colourless, and a final TBS rinse. Immunohistochemistry (IHC) After the antigen retrieval and bleaching steps, IHC was performed by a biotin-streptavidin immunoperoxidase technique. All sections were labelled in three independent experimental runs. To assess the effect of KMnO4/OA pretreatment after microwave antigen retrieval, pretreated and non-pretreated sections were run simultaneously and incubated either with the primary antibody or with an irrelevant but isotypespecific antibody as a negative control for non-specific binding. The primary murine mAbs used were: HMB-45 (Signet Laboratories, Dedham, MA, USA), an IgG1 mouse monoclonal antibody; MEL-1 (SK-MEL-28
164
S. S. Sulaimon et al.
melanoma cell line; Signet Laboratories), an IgG3 monoclonal antibody; and NK1/C3 (BioGenex, San Ramon, CA, USA), an IgG1 mouse monoclonal antibody. A range of dilutions was tested for each primary antibody, the dilution (1 in 10) giving distinct immunolabelling with the least background being chosen as the working dilution for all three antibodies. The diluent was phosphate-buffered saline plus a carrier protein in 09% sodium azide (BioGenex). Negative control antibodies consisted of a purified fraction of IgG1 (Biosource International, Carmarillo, CA, USA) used at 100 mg/ml for HMB-45 and NK1/C3, and a rat antimouse IgG3 (Biodesign, Kennenbunk, ME, USA) used at 01 mg/01 ml for MEL-1. A commercially available labelling kit, consisting of a prediluted biotinylated goat antimouse IgG secondary antibody and a peroxidase-conjugated streptavidin Super Sensitive label (BioGenex), was used to label the primary antibodies. A basic immunolabelling procedure was followed. KMnO4/OA-pretreated and non-pretreated sections were incubated with a universal blocking reagent (Power Block1 ; BioGenex) for 10 min at room temperature, followed by a TBS rinse. Pretreated and non-pretreated slides were then incubated overnight (18 h) with the primary (HMB 45, MEL-1 or NK1/ C3) or irrelevant isotype-specific antibodies at room temperature in a humidity chamber. After incubation and a 3-min TBS rinse, all sections were incubated at room temperature with the biotinylated secondary antibody for 30 min, followed by application of the peroxidase-conjugated streptavidin label. Sections were finally incubated in 3,30 -diaminobenzidine chromogen (DAB) for 5 min, counterstained lightly in Mayer's haematoxylin, dehydrated, cleared and mounted. The streptavidin-peroxidase immunohistochemical technique was used to detect reactivity of HMB 45,
MEL 1 and NK1/C3 monoclonal antibodies with canine melanoma samples. The result was considered positive when immunolabelling by HMB 45, MEL-1 or NK1/C3 was restricted to the neoplasm and when no labelling was detected in the absence of primary antibody, or when irrelevant isotype specific antibody was substituted. Of the three markers, one (HMB-45) was further evaluated against 16 non-melanocytic canine tumours to determine its specificity. Results Canine Melanomas (n 25) The mAbs HMB-45 and MEL-1 reacted with 22/25 (88%) and 18/25 (72%) of the canine melanomas, respectively, but only after microwave antigen retrieval and KMnO4/OA-pretreatment (Table 1). Positive reactivity to HMB-45 and MEL-1 was shown as follows: oral melanomas, 13/16 and 9/16, respectively; cutaneous melanomas, 8/8 and 8/8; melanoma of the digit, 1/1 and 1/1; all pigmented melanomas, 16/18 and 14/18; all amelanotic tumours, 6/7 and 4/7 (Table 1). Non-pretreated sections incubated with the appropriate irrelevant isotype-specific antibody, or HMB-45 (Fig. 1a), or MEL-1 (Fig. 2a) showed no immunoreactivity. Nonspecific staining of apocrine and sebaceous glands was observed. This was probably due to nonspecific ionic binding of the antibody to secretory products. KMnO4/OA-pretreated sections did not show significant immunolabelling when the appropriate irrelevant antibodies were substituted for the primary antibody; however, positive labelling with HMB-45 mAb on pretreated sections revealed a diffuse granular cytoplasmic pattern within the tumour cells (Fig. 1b, c). MEL-1 reactivity was characterized by a cytoplasmic distribution with sporadic
Table 1 Immunohistochemical labelling of melanomas for presence of melanin and tumor location Melanomas
Oral (n 16) Pigmented (n 11) Amelanotic (n 5) Cutaneous (n 8) Pigmented (n 6) Amelanotic (n 2) Digit, pigmented (n 1) All (n 25) Pigmented (n 18) Amelanotic (n 7) N/A, not applicable.
Number of positive melanomas/number tested, with stated mAb HMB-45
MEL-1
NK1/C3
13/16 (81%) 9/11 4/5 8/8 6/6 2/2 1/1 22/25 (88%) 16/18 (89%) 6/7 (86%)
9/16 (56%) 7/11 2/5 8/8 6/6 2/2 1/1 18/25 (72%) 14/18 (78%) 4/7 (57%)
0/3 N/A N/A 0/4 N/A N/A N/A 0/7 (0%) ... ...
Melanoma-specific Antigens in Canine Melanomas
165
Fig. 1a±c. Canine oral and cutaneous amelanotic melanomas examined immunohistochemically with HMB-45. (a) No immunolabelling of oral melanoma without KMnO4/OA-pretreatment. (b) Immunolabelling (diffuse, cytoplasmic) of oral melanoma occurs after KMnO4/OA-pretreatment. (c) Labelling (diffuse, cytoplasmic) of cutaneous melanoma occurs after KMnO4/ OA-pretreatment. V, vessel. Immunoperoxidase/DAB, with Mayer's haematoxylin counterstain. Bar, 40 mm.
Fig. 2a, b. Canine amelanotic melanoma examined immunohistochemically with MEL-1. (a) No immunolabelling without KMnO4/OA-pretreatment. (b) Immunolabelling (cytoplasmic and nuclear) occurs after KMnO4/OA-pretreatment. Immunoperoxidase/DAB, with Mayer's haematoxylin counterstain. Bar, 40 mm.
166
S. S. Sulaimon et al.
nuclear labelling (Fig. 2b). Distribution of the label within most tumours was homogeneous; some tumours, however, had a multifocal pattern. HMB45 also labelled resident melanocytes within the overlying epidermis. Three out of 25 and 7/25 of the melanomas showed no reactivity with either HMB-45 or MEL-1, respectively. In positive melanomas, however, the percentages of cells labelled by HMB-45 and MEL-1 were as follows: ,25%, 2/25 and 1/25 melanomas, respectively; 25±75%, 15/25 and 16/25; .75%, 5/25 and 1/25. Generally, HMB-45 immunolabelled a higher percentage of canine melanoma cells than did MEL-1. The NK1/C3 antibody did not label any of the initial seven melanoma tissues tested, irrespective of KMnO4/OA-pretreatment, and was not used further. Non-melanocytic Canine Tumours Immunolabelling of 16 non-melanocytic tumours was performed to determine the specificity of HMB-45. Fifteen of sixteen such tumours showed no significant reactivity with HMB-45, labelling being consistently absent in fibrosarcomas, squamous cell carcinomas, poorly differentiated carcinomas, basal cell tumours, and a malignant histiocytosis. One of three plasmacytomas reacted positively. This plasmacytoma showed homogeneous labelling which was moderate, granular and cytoplasmic. Background labelling, observed as a light brown film across the entire section, was evident in three fibrosarcomas and one squamous cell carcinoma. Immunolabelling of non-melanocytic tumours with MEL-1 was not pursued, due to the poor sensitivity of this antibody for detection of canine melanomas as compared with that of HMB-45.
Discussion The distinction of melanomas from a variety of poorly differentiated tumours is a dilemma that faces diagnosticpathologists.Intracytoplasmicmelanin,adistinct marker for melanocytic tumours, is not synthesized in amelanotic subtypes. Problems in diagnosis may arise when melanin is confused with other pigments or when cells are extremely pleomorphic. The diverse cytological features of melanoma cells can be attributed to common embryological roots shared with neural and epithelial cells (Moulton, 1990). The melanocytic phenotype in amelanotic melanoma is difficult to confirm without ultrastructural identification of premelanosomes or melanosomes. In a diagnostic setting, however, electron microscopy is often unavailable.
Immunohistochemistry has become a cost-effective method when the diagnostic pathologist is confronted with poorly differentiated tumours. Monoclonal antibodies which have aided the immunohistochemical diagnosis of human and canine melanomas include: S-100, HMB-18, HMB-45, HMB-50, R24 (MEL-1), HMSA-1, HMSA-5, NK1/C3, NK1/beteb, EPI-3, 225285, Melan A, IBF9, cytokeratin, vimentin, and neuron specific enolase (Burnier et al., 1991; Oliver and Wolfe, 1992; Berrington et al., 1994; Fernando et al., 1994; Mottolese et al., 1994). Traditionally, antibodies against the S-100 protein have been the immunomarkers of choice, but S-100 protein has been identified in human neural and non-neural cells such as melanocytes, adipocytes, myoepithelial cells, Schwann cells, skeletal muscle cells, chondrocytes, eccrine and apocrine gland cells, and Langerhans cells (Nakajima et al., 1982a, b). Tumours arising from any of these cell types may react with anti-S-100 antibodies, thereby preventing their use as reagents specific for melanomas (OrdonÄez et al., 1988; Burnier et al., 1991). HMB-45 is generally considered to be very specific but inferior in sensitivity to S-100 antibodies for the diagnosis of human melanomas (Skeleton et al., 1991; Blessing et al., 1998). However, Wick et al. (1988) and OrdonÄez et al. (1988) reported 93% (62/67) and 91% (32/35) sensitivity for HMB-45, respectively. A few mAbs against human melanoma-associated antigens cross-react with canine melanomas. These include the HMSA-1, HMSA-5, and Melan A antibodies (Berrington et al., 1994; Ramos-Vara et al., 2000). IBF9, which recognizes a cell membrane molecule, is the only reported mAb to have been generated directly against canine melanomas (Oliver and Wolfe, 1992). IBF9 reacted positively with 63% (24/38) of canine melanomas tested. However, 12 of 40 (30%) non-melanocytic tumours reacted with IBF9, including cutaneous lymphosarcomas, squamous cell carcinomas, basal cell tumours, epulides and mammary gland tumours. This study demonstrated specific immunoreactivity of canine melanomas with HMB-45 and MEL-1 in formalin-fixed, paraffin wax-embedded tissue after microwave antigen retrieval and KMnO4/OApretreatment. The value of such pretreatment is threefold: it bleaches melanin pigment to facilitate better detection of immunolabelling; it retrieves the antigen masked by formalin fixation; and it blocks endogenous peroxidase activity (Alexander et al., 1986). The combination of microwave antigen retrieval and KMnO4/OA-pretreatment caused some distortion of cellular and nuclear features. HMB-45 was tested on 25 canine melanomas and 16 non-melanocytic tumours to evaluate its sensitivity, specificity and
Melanoma-specific Antigens in Canine Melanomas
labelling pattern. HMB-45 labelled 22/25 (88%) of the melanomas and 1/16 (6%) of the non-melanocytic tumours, suggesting that it was both sensitive and specific for canine melanomas. Immunolabelling by HMB-45 had a diffuse granular intracytoplasmic pattern similar to that reported in human melanoma. Pigmented tumours of this series showed increased labelling intensity. Melanocytes in the epidermis were reactive with HMB-45. Smoller et al. (1989) observed that melanocytes in the epidermis overlying some human melanomas were reactive with HMB-45 and hypothesized that this was due to stimulation of the epidermal melanocytes by extrinsic factors produced by tumour cells. HMB-45 appeared to be specific for canine melanomas, with 15/16 of the non-melanocytic tumours lacking any reactivity. One of three plasmacytomas, however, showed HMB-45 immunoreactivity. Positive HMB-45 labelling has been reported in human plasmacytomas, breast and endometrial adenocarcinomas, colonic and rectal adenocarcinomas, immunoblastic lymphomas, olfactory neuroblastomas, renal angiomyolipomas, and basal cell carcinomas (Bonetti et al., 1989, 1991; Smoller et al., 1989; Kornstein and Franco, 1990; Friedman and Tatum, 1991; Hancock et al., 1991; Pea et al., 1991; Herrera et al., 1992). Cross-reactivity has been attributed to antigenically related glycoprotein epitopes in these neoplasms. Alternatively, expression of oncofetal antigens containing sialic acids in an immunoreactive epitope similar to that recognized by HMB-45 may lead to cross-reactivity (Bonetti et al., 1991; Hancock et al., 1991). The MEL-1 mAb detects the cell surface disialoganglioside, GD3. Gangliosides are glycolipid molecules inserted into the cell membrane of melanoma cells and cancer cells originating from the neuroectoderm. Immunoreactivity with this mAb has been reported only in frozen sections and appears to be lost after formalin-fixation. In this study, MEL-1 immunolabelled 72% (18/25) of formalin-fixed canine melanoma samples after microwave antigen retrieval and KMnO4/OA-pretreatment, suggesting that these procedures expose epitopes masked by formalin fixation-induced protein cross-linkage. MEL-1 appears to be less sensitive than HMB-45 for labelling canine melanomas. The mAb NK1/C3 recognizes a disulphidedependent glycoprotein which is located predominantly on the inner side of the membranes of cytoplasmic vesicles. NK1/C3 did not label any of seven melanocytic tumours and was not examined further. Possible reasons for the lack of immunolabelling include the following: lack of expression of the glycoprotein recognized by NK1/C3; expression of
167
antigens occurred but was below the level needed for detection; the antigens were masked or destroyed during formalin-fixation and tissue processing. In human melanomas, lack of labelling with NK1/C3 has been associated with high metastatic potential (Vennengoor et al., 1985). In conclusion, this study demonstrated immunolabelling with HMB-45 and MEL-1 in formalin-fixed canine melanoma tissues after microwave antigen retrieval and KMnO4/OA-pretreatment. HMB-45 immunolabelling appeared both sensitive and specific, with rare cross-reactivity in plasmacytomas. In contrast, MEL-1 appeared less sensitive in identifying canine melanomas. References Alexander, R. A., Hiscott, P. S., Hart, R. I. and Grieson, I. (1986). Effect of melanin bleaching on immunoperoxidase with reference to ocular tissues and lesions. Medical Laboratory Science, 43, 121±127. Berrington, A., Jimbow, K. and Haines, D. M. (1994). Immunohistochemical detection of melanomaassociated antigens on formalin-fixed paraffin embedded canine tumours. Veterinary Pathology, 31, 455±461. Blessing, K., Sanders, S. A., and Grant, J. J. H. (1998). Comparison of immunohistochemical staining of novel antibody melan-A with S-100 protein and HMB-45 in malignant melanoma and melanoma variants. Histopathology, 32, 139±146. Bonetti, F., Colombari, R., Manfrin, E., Zamboni, G., Martignoni, G., Mombello, A. and Chilosi, M. (1989). Breast carcinoma with positive results for melanoma marker (HMB-45): HMB-45 immunoreactivity in normal and neoplastic breast. American Journal of Clinical Pathology, 92, 491±495. Bonetti, F., Pea, M., Martignoni, G., Mombello, A., Colombari, R., Zamboni, G., Scarpa, A., Piubello, Q., Bacchi, C. E. and Gown, A. M. (1991). False positive immunolabelling of normal epithelium and carcinomas with ascites fluid preparations of anti-melanoma antibody HMB-45. American Journal of Clinical Pathology, 95, 454±459. Bostock, D. E. (1979). Prognosis after surgical excision of canine melanomas. Veterinary Pathology, 16, 32±40. Brodey R. S. (1960). Clinical and pathologic study of 130 neoplasms of the mouth and pharynx in the dog. American Journal of Veterinary Research, 21, 787±818. Burnier, M. N., McLean, I. W. and Gamel, J. W. (1991). Immunohistochemical evaluation of uveal melanocytic tumours. Expression of HMB-45, S-100 protein and neuron-specific enolase. Cancer, 68, 809±814. Fernando, S. S. E., Johnson, S. and Mate, J. (1994). Immunohistochemical analysis of cutaneous malignant melanoma: comparison of S-100 protein, HMB-45 monoclonal antibody and NK1/C3 monoclonal antibody. Pathology, 26, 16±19. Friedman, H. D. and Tatum, A. (1991). HMB-45 detection in adenocarcinomas. Archives of Pathology and Laboratory Medicine, 115, 826±830.
168
S. S. Sulaimon et al.
Gown, A. M., Vogel, A. M., Hoak, D., Gough, F. and McNutt, M. A. (1986). Monoclonal antibodies specific for melanocytic tumours distinguish subpopulations of melanocytes. American Journal of Pathology, 123, 195±203. Hamilton, W. B., Helling, F., Lloyd, K. O. and Livingston, P. O. (1993). Ganglioside expression on human malignant melanoma assessed by quantitative immune thin-layer chromatography. International Journal of Cancer, 53, 566±573. Hancock, C., Allen, B. C. and Herrera, G. A. (1991). HMB-45 detection in adenocarcinomas. Archives of Pathology and Laboratory Medicine, 115, 886±890. Herrera, G. A., Hancock, C. and Allen, B. C. (1992). Specificity of antibody HMB-45. Archives of Pathology and Laboratory Medicine, 116, 900±901. Kornstein, M. J. and Franco, A. P. (1990). Specificity of HMB-45. Archives of Pathology and Laboratory Medicine, 114, 450. Marino, D. J., Matthiesen, D. T., Stefanacci, J. D. and Moroff, S. D. (1995). Evaluation of dogs with digit masses: 117 cases (1981±1991). Journal of the American Veterinary Medical Association, 207, 726±728. Mottolese M., Venturo, I., Benevolo, F., Di Fillippo, F., Lopez, M., Bigotti, A. and Natali, P. G. (1994). Immunocytochemical diagnosis of amelanotic metastatic melanoma using monoclonal antibodies HMB-45 and EP1-3. Melanoma Research, 4, 53±58. Moulton, J. E. (1990). Tumours in Domestic Animals, 3rd Edit., University of California Press, Berkeley, pp. 75±87. Nakajima, T., Watanabe, S., Sato, Y., Kameya, T., Shimosato, Y. and Ishihara, K. (1982a). Immunohistochemical demonstration of S-100 protein in malignant melanoma and pigmented nevus, and its diagnostic applications. Cancer, 50, 912±918. Nakajima, T., Watanabe, S., Sato, Y., Kameya, T., Shimosato, Y. and Ishihara, K. (1982b). An immunoperoxidase study of S-100 protein distribution in normal and neoplastic tissues. American Journal of Surgical Pathology, 6, 715±727. Oliver, J. L. and Wolfe, L. G. (1992). Antigen expression in canine tissues, recognized by monoclonal antibodies generated against canine melanoma cells. American Journal of Veterinary Research, 53, 123±128. OrdonÄez, N. G., Xiaolong, J. and Hickey, R. C. (1988). Comparison of HMB-45 monoclonal antibody and S-100 protein in the immunohistochemical diagnosis of melanoma. American Journal of Clinical Pathology, 90, 385±390.
Pea, M., Bonetti, F., Zamboni, G., Martignoni, G., Riva, M., Colombari, R., Mombello, A., Bonzanini, M., Scarpa, A. and Ghimenton, C. (1991). Melanocytemarker-HMB-45 is regularly expressed in angiomyolipoma of the kidney. Pathology, 23, 185±188. Ramos-Vara, J. A., Beissenherz, M. E., Miller, M. A., Johnson, G. C., Pace, L. W., Fard, A. and Kottler, S. J. (2000). Retrospective study of 338 canine oral melanomas with clinical, histologic, and immunohistochemical review of 129 cases. Veterinary Pathology, 37, 597±608. Ryan, A. M. and Diters R. W. (1984). Clinical and pathologic features of canine ocular melanomas. Journal of the American Veterinary Medical Association, 184, 60±67. Skeleton, H. G., Smith, K. J., Barrett, T. L., Lupton, G. P. and Graham, J. H. (1991). HMB-45 staining in benign and malignant melanocytic lesions. A reflection of cellular activation. American Journal of Dermatopathology, 13, 543±550. Smoller, B. R., McNutt, N. S. and Hsu, A. (1989). HMB45 recognizes stimulated melanocytes. Journal of Cutaneous Pathology, 16, 49±53. Todoroff, R. J. and Brodey, R. S. (1979). Oral and pharyngeal neoplasia in the dog. A retrospective survey of 361 cases. Journal of the American Veterinary Medical Association, 175, 567±571. Vanstapel, M. U., Gatter, K., De Wolf-Peeters, C., Mason, D. and Doesmet, V. (1986). New sites of human S-100 immunoreactivity detected with monoclonal antibodies. American Journal of Clinical Pathology, 85, 160±168. Vennengoor, C., Calafat, J., Hageman, P. H., Van Buitenen, F., Janssen, H., Kolk, A. and Rumke, P. H. (1985). Biochemical characterization and cellular localization of a formalin-resistant melanomaassociated antigen reacting with monoclonal antibody NK1/C3. International Journal of Cancer, 35, 287±295. Wick, M. R., Swanson, P. E. and Rocamora, A. (1988). Recognition of malignant melanoma by monoclonal antibody HMB-45. An immunohistochemical study of 200 paraffin-embedded cutaneous tumours. Journal of Cutaneous Pathology, 15, 201±207. Received; June 27th; 2001 Accepted; April 29th; 2002