Desmoplastic melanoma (DM) is a spindle cell melanoma that represents less than 4% of all cutaneous melanomas (Quinn et al., 1998). Because of its spindle cell morphology and its frequent lack of clinical and histologic features associated with conventional melanoma, DM is often not identified (Mccarthy et al., 2004). Diagnosis of DM by immunohistochemistry (IHC) may be difficult because primary and metastatic lesions are often negative for relatively specific melanocyte biomarkers such as HMB-45 (human melanoma black-45/gp100) and MART-1 (melanoma antigen recognized by T-cells). DM usually expresses the S-100p protein, but this protein can also be detected in Langerhans cells, dendritic cells, macrophages, Schwann cells, and a wide range of tumors, including benign and malignant nerve sheath tumors, some carcinomas, and myoepithelial tumors. The lack of a DM-specific IHC stain confounds diagnosis, particularly when DM recurs within scars (where distinction of the spindle-shaped DM cells from reactive fibroblasts can be extremely difficult due to their morphologic similarities and the presence of S-100p in some scars) or in the regional lymph nodes (LN). Other proposed IHC markers of DM, such as microphthalmia transcription factor (Mitf) (Busam et al., 2001; Granter et al., 2001) and procollagen 1, show promise, but their sensitivity has not been verified.
A large study from the Melanoma Institute Australia challenged the assumption that DM has a worse prognosis than other melanomas (Quinn et al., 1998). In this study of 280 patients, there was no difference in survival between patients with desmoplastic versus nondesmoplastic melanoma, but DM was associated with a lower rate of regional LN metastasis and a higher rate of local recurrence.
High molecular weight–melanoma-associated antigen (HMW–MAA), also known as the melanoma chondroitin sulfate proteoglycan, is expressed in >85% of primary and metastatic melanoma lesions, with limited inter-lesional and intralesional heterogeneity (Campoli et al., 2004). HMW–MAA is a membrane-bound chondroitin sulfate proteoglycan found in the melanoma cell membrane; by contrast, MART-1 co-localizes and forms a complex with the P100 form of gp100 (antigen of HMB-45) in early subcellular compartments of melanocytic cells. HMW–MAA can be used as a detection biomarker for cutaneous metastatic melanoma (Goto et al., 2008). The present study examines HMW–MAA as a potential biomarker for DM. The expression level of HMW–MAA in primary and metastatic DM lesions was assessed and compared to that of MART-1 and HMB-45 by utilizing IHC and qRT-PCR (See Supporting Information section).
Immunohistochemistry with HMW–MAA monoclonal antibodies (mAb) or MART-1 mAbs was optimized for paraffin-embedded archival tissues (PEAT) in our previous study (Goto et al., 2008). IHC with HMB-45 mAb was assessed using PEAT specimens of primary melanoma and metastatic melanoma in lung, LN and skin. Each assay was used to analyze 40 DM primaries, 23 DM metastases, nine tumor-negative LNs, and 13 tumor-negative skin specimens (see Appendix S1). As shown in Figure 1, HMW–MAA -specific mAbs stained the membrane and (to a lesser degree) the cytoplasm of melanoma cells. Table S1A,B summarize the HMW–MAA expression level in primary and metastatic DM tumors, by showing the percentage of stained cells and the staining intensity.
In primary DM, staining intensity was stronger for HMW–MAA than for MART-1 or HMB-45 (Table S2A, P < 0.0001 and P < 0.0001, respectively). Of the 40 DM primaries, 38 (95%) stained for HMW–MAA but only one (3%) stained for MART-1 and only one (3%) stained for HMB-45 (Table S2A). In primary DM, the percentage of stained cells was significantly higher for HMW–MAA than for MART-1 or HMB-45 (Table S2B, P < 0.0001 and P < 0.0001, respectively). Of the 40 primary lesions, 35 (87%) showed more than 50% staining for HMW–MAA, whereas none of 40 primaries stained for MART-1 and only one (3%) stained for HMB-45.
In metastatic DM, staining intensity was stronger for HMW–MAA than for MART-1 or HMB-45 (Table S2C, P = 0.007 and P < 0.0001, respectively). Of the 23 DM metastases, 20 (87%) stained for HMW–MAA, whereas four (18%) stained for MART-1 and two (9%) stained for HMB-45 (Table S2C). The frequency of staining was significantly higher for HMW–MAA than for MART-1 or HMB-45 (Table S2D, P < 0.0001 and P < 0.0001, respectively). Of the 23 DM metastases, 14 (61%) showed more than 50% staining for HMW–MAA; none of the metastases stained for MART-1 or HMB-45. The HMW–MAA mAb cocktail stained 12 of 14 (86%) nodal metastases (Table S1B) but did not stain any tumor-negative LN. The intensity of nodal staining was significantly stronger for HMW–MAA (12 of 14) than for MART-1 (3 of 14) or HMB-45 (1 of 14) (P = 0.007 and P = 0.0023, respectively). These results indicate that IHC staining of primary and metastatic DM was significantly more sensitive with HMW–MAA-specific mAbs than with MART-1-specific mAb or HMB-45-specific mAb.
To investigate the potential of HMW–MAA as a molecular biomarker to detect DM, we assessed HMW–MAA mRNA by qRT-PCR level in both PEAT primary and metastatic DM. MART-1 mRNA and GAPDH mRNA were used as controls. The qRT-PCR assay for each biomarker was optimized in PEAT specimens. The range of relative mRNA copies for DM was 0–37.4 for HMW–MAA and 0–6.39 for MART-1. The relative mRNA copy number for HMW–MAA was higher in 40 DM primaries than in tumor-free skin specimens (median 3.35 versus 1.45), and significantly higher in 23 DM LN metastases than in tumor-free LNs (median 0.47 versus 0.013; P = 0.019, Figure 2A,C). HMW–MAA positivity was 25 (63%) of 40 DM primaries and 16 (70%) of 23 DM metastases expressed HMW–MAA mRNA. Assessment of the same PEAT specimens showed that MART-1 was expressed in 9 (23%) of 40 primaries and 5 (22%) of 23 metastases (P = 0.0029 and P = 0.0023, respectively). Relative MART-1 copy number was not significantly different between 40 DM primaries and tumor-free skin specimens, or between 23 DM metastases and tumor-free LNs (P = 0.15, Figure 2B,D). Moreover, HMW–MAA mRNA was expressed in 8 (57%) of 14 nodal metastases, whereas MART-1 mRNA was expressed in 3 (21%) of 14 nodal metastases. HMB-45 (gp100) was not assessed by qRT-PCR as many types of normal cells express this pigment biomarker (Hoon et al., unpublished). These results indicate that HMW– MAA is more sensitive than MART-1 for qRT-PCR assessment of DM.
As most DM primaries are slow growing and almost half lack pigmentation, melanoma may not be considered in the clinical differential diagnosis of cutaneous tumors (Mccarthy et al., 2004). Histopathological features are also subtle. DM stains for S-100p, but S-100p also can be detected in non-neoplastic cells and in a wide range of tumors (Longacre et al., 1996). Melanocytic differentiation biomarkers such as MART-1 and HMB-45 are commonly used for detection of both primary and metastatic melanoma, including LN metastasis, but have limited value in DM. More sensitive and specific IHC biomarkers of DM are needed to avoid misdiagnosis of primary DM, reduce the risk of local persistence or recurrent DM, and prevent understaging of metastatic DM. A cocktail of HMW–MAA-specific mAbs was more sensitive than HMB-45 mAb or MART-1 mAb for IHC DM assessment. HMW–MAA was more sensitive than MART-1 for qRT-PCR of DM primary and metastasis.
Recently, several groups have reported results of sentinel lymph node (SLN) biopsy in patients with primary DM (Gyorki et al., 2003; Pawlik et al., 2006; Thelmo et al., 2001). These studies found a very low incidence of regional LN metastases (0–12.1% but usually below 4%), as compared with approximately 21% in patients with other forms of melanoma (Morton et al., 2006). The lower rate of LN positivity in DM reflects the biology of the tumor, which metastasizes less often via lymphatic routes than non-DM; however, this does not decrease the importance of detecting metastatic DM in LNs. S-100p antibody stains DM cells but may also stain non-neoplastic cells, such as cells of histiocytic derivation, which may be numerous in LNs. More sensitive and specific IHC biomarkers would be useful in the detection of metastatic DM cells in LNs. In our study, the HMW–MAA-specific mAb cocktail stained DM cells in 86% of nodal metastases; only 21% of nodal metastases were stained by either MART-1 mAb or HMB-45 mAb. We believe that assessment of SLN specimens using HMW–MAA-specific mAbs may improve the identification of metastatic DM in regional LNs.
Although reports have suggested that Mitf-specific mAb D5 is highly sensitive and specific for DM (Busam et al., 2001; Koch et al., 2001), not all investigators agree (Granter et al., 2001). We therefore performed IHC analysis with Mitf-specific mAb D5 (Neomarker, Fremont, CA, USA) in seven of the LN metastases from the present study. Only three specimens (43%) stained with Mitf-specific mAb; all three were among the five (71%) that stained with HMW–MAA mAbs. Interestingly, MART-1 and HMB-45 Abs stained only one specimen, which was negative for HMW–MAA and Mitf. Combining these two IHC biomarkers might optimize the sensitivity for detection and identification of DM more accurately.
In summary, HMW–MAA is more sensitive than HMB-45 and MART-1 for IHC diagnosis of primary DM. Moreover, the HMW–MAA mAb cocktail is valuable for detecting occult DM metastases, especially in regional tumor-draining LNs. The study indicates that HMW–MAA is more sensitive than MART-1 as an mRNA biomarker for primary DM. Because HMW–MAA mRNA was detected in metastatic DM that did not express MART-1 mRNA, combining these biomarkers for qRT-PCR assay might increase the sensitivity of mRNA assessment for high-risk DM that may metastasize to LNs. Thus HMW–MAA has potential value as a component of a multimarker probe for qRT-PCR assessment of DM metastases.