The final diagnosis, based on H&E and immunohistochemical stains was malignant melanoma, in-situ and invasive, Clark's level at least IV; Breslow's thickness at least 1.5 mm. Radial growth phase was present, identified as lentigo maligna type. Vertical growth phase was also present, identified as spindle cell and nevoid type. Pigmentation in the neoplasm was mild. Surface ulcer was not identified. Lymphoid response at the base was non brisk (mild). Slide section margins were involved by melanoma.
TNM stage (AJCC 2002) = Pt2a+ Nx Mx.
Comment: Most of this melanoma was composed of invasive (dermal) component, with features of spindle cell/desmoplastic melanoma.
Desmoplastic melanoma, (DM) is a relatively rare variant of malignant melanoma that represents <4% of melanomas seen at the Sydney Melanoma Unit and <2% of the melanomas seen at Memorial Sloan-Kettering Cancer Center in NYC. It was first described by Conley et al., in 1971. While there still seems to be uncertainty regarding the precursor cell of DM, a number of various "immunostain cocktails" have been developed in order to shed more light on the histogenesis of this tumor. It has been noted that DM is often seen in association with a junctional component such as lentigo maligna, as well as conventional epithelioid and spindle cell melanoma.
The most common locations for this tumor are the head and neck. In a study of 28 patients with DM at Mayo, the average age was 65 yo, (range 43-90, the median age approximately 10 years later than that for conventional melanomas) and the gender was male predominant (61%). Histologically, all tumors demonstrated dense stromal desmoplasia surrounding hyperchromatic spindled cells as well as perineural invasion in 82%. A perivascular lymphoproliferation was also a common finding in these tumors.
The paramount challenge in correctly diagnosing a desmoplastic melanoma is in accurately distinguishing it from other tumors possessing similar immunoreactivity, namely the S100 + / HMB45 - tumors such as peripheral nerve sheath tumors (MPNST, neurofibromas, schwannomas) which may also exhibit epithelioid or spindle cell morphology. In fact, Xu and colleagues (2002), suggested that desmoplastic melanoma might share pathophysiologic features with nerve sheath tumors and that the immunophenotype might be closer to peripheral nerve sheath tumors than other types of melanoma.
While S100 marks DM positively in > 90% of cases, this nonspecific stain is also known to mark a variety of other cells, including melanocytes, Langerhans cells, some eccrine cells and apocrine glandular cells, nerves, muscles, Schwann cells, myoepithelial cells, chondrocytes and histiocytes. One of the difficulties in diagnosing this lesion has been due to the sometimes hypocellular and amelanotic nature of the tumor. For this reason, once correctly diagnosed using the proper panel of stains and morphology, the S100 stain has been found to be quite useful as described by Eng and colleagues (2000), in determining tumor thickness, deep and peripheral margins and extent of neural invasion.
HMB45, on the other hand, which is a monoclonal antibody to group 100 proteins, has more specificity but less sensitivity. Whereas the majority of conventional melanomas exhibit strong staining with HMB45, in a study of 28 cases of DM done by Kay, et al., only 7% were HMB45+ and other studies reveal predominantly HMB45-negative DM tumors.
Melan-A, a melanocytic differentiation antigen, shows a similar staining pattern to S100 for melanocytes, only more specific than S100. Melan-A gives strong and diffuse staining of all the benign melanocytic lesions and most melanomas with less sensitivity for metastatic, spindle cell and desmoplastic melanomas.
Tyrosinase, an enzyme essential in early stages of melanin biosynthesis, is extremely sensitive to most melanomas, with little or no reactivity in DM. In the study by Kay, et al., out of 28 cases of DM, only 4 cases stained positive. A study of two cases of DM by Clarkson (2001) revealed positive staining of all cells with S100 and no staining with tyrosinase, Melan-A or HMB45, however, as you can see, the study was limited to only two cases of DM.
In a paper by Xu, et al., MAGE-1, a member of the melanoma antigen encoding gene family, was found to exhibit protein expression in malignant melanomas. Unfortunately, other non-melanocytic malignant tumors can also express MAGE, as well. For this reason, it has been recommended that a panel of these immunostains be used to determine the source and extent of tumor in these types of lesions.
Skelton, et al., from the AFIP reported on 128 patients in a large study group, 59% of whom had DM in the head/neck region. Typically, it had been noted that desmoplastic melanoma demonstrated a better prognosis than other melanomas. The exception to this observation would be those DMs located in the head/neck. While they reported a 61% 5 year survival rate for DMs invading to a depth of >4mm, these same tumors when located in the head/neck worsened the 5-year survival rate to 38%. It has been postulated that the reason for this is that the thin layer of the dermis in this area allows faster access to the bloodstream and lymphatics, as well as the fact that cosmetically, this region may not allow for the clinician to achieve such wide surgical margins as necessary. They identified other poor prognostic factors, to include positivity for tyrosinase, melan-A and SMA staining, as well as tumor size and tumor thickness.
Most DM tumors have been found to be Clark level IV or V, leading Kay, et al., to state that a Breslow measurement of tumor thickness would have more clinical importance than Clark level in this instance.
In a study by Busam and colleagues, 92 patients with either "pure" or "combined" DM from a single institution were studied between 1980 and 2002. Their purpose was to identify clinical and pathologic parameters that could be used as significant variables for disease-free survival. They found that it was important clinically to distinguish "pure DMs", where desmoplasia was prominent throughout the entire tumor (constituting > 90% of the invasive melanoma) from "combined DMs", in which desmoplasia is only a partial, usually minor component (densely cellular tumor foci without stromal fibrosis constituted > 10% of the entire tumor). They found through Kaplan-Meier curves and hazard ratios for prognostic parameters that pure forms were associated with longer disease-specific survival. Histologically, in the "combined" DMs, the non-desmoplastic dense cellular aggregates had more atypical nuclei, higher mitotic index and overall higher cellular density compared with "pure" DM. Also, more "combined" tumors than pure DMs were associated with ulceration, (24% vs. 16%.) Of four clinical and histologic parameters identified by statistical analysis: Clark level (P=0.005), subtype of DM (P=0.01), tumor mitotic rate (P=0.01), and tumor thickness (P=0.02), the effect of subtype had the strongest influence on survival in patients with deeply infiltrating (Clarks V) melanomas. Patients with a combined DM had a 3.5 times greater risk of an event such as death of disease or metastatic disease than those with pure DM. Regarding the controversy over the prognosis of DM, in Conley's studies, 5 of 7 patients had local recurrences, 3 had lymph node mets and 4 died of disease. This led the authors to conclude that DMs may be associated with a worse clinical outcome than conventional melanomas. In 1988, Egbert, et al, had 12 of 23 patients with DM have local recurrences and 3 of those die of disease, leading him to share Conley's conclusion. There was, however, never an adjustment for tumor thickness in these studies. Also in 1988, Walsh et al, after adjusting for tumor thickness, suggested that DM might be associated with a more favorable outcome. They reported the high incidence of local recurrences, but also noted longer survival in patients with DM compared to patients with conventional melanomas of similar thickness. However since Walsh's 1988 study, the waters have again been muddied, with a large study of 280 patients from the Sydney Melanoma Unit. Here, they did not detect a difference in survival rates between patients with DM vs. conventional melanoma. This might be, in part due to the fact that the patients studied in the SMU series had melanomas < 4mm in thickness, whereas in the Walsh studies, patients had tumor thicknesses of > 4mm. Busam postulates that the favorable prognostic impact may only be significant in those patients with thicker and more deeply infiltrating tumors, (i.e., Clarks level IV-V, and > 4mm in thickness).
In a second study by Busam, he emphasizes the importance of distinguishing DM from conventional melanomas. Underlying this is the fact that the clinical presentation of the two entities is so very different. Unlike conventional melanoma, most DMs are rarely diagnosed at early stage, and are >4mm at time of diagnosis. The majority of DMs are Clarks level IV or V at the time of diagnosis and 18-20% are ulcerated. A subset of DMs are neurotropic and exhibit prominent infiltration of peripheral nerves, with some evidence that these tumors are more aggressive than those without neurotropism, leading to DNM as a variant of DM, but data is conflicting. He goes on to point out that the histologic differential diagnosis of DM vs. other sclerosing lesions can be daunting at times. He recommends a panel of immunohistochemical stains including S100, 34BE12, SMA, desmin, gp100, Melan-A, tyrosinase, microphthalmia transcription factor and factor XIIIa to distinguish DM from entities such as sclerosing nevi, dermatofibroma, neurofibroma, fibrosarcoma, desmoplastic leiomyosarcoma, sclerosing sarcomatoid squamous cell carcinoma and MPNST. Importantly, evidence indicates that lymph node mets are much lower at presentation in patients with DM than in patients with conventional melanoma. DMs need to be completely excised, and due to the high incidence of nerve involvement, a wide excision is recommended, if aesthetically possible. As sentinel lymph node involvement is much less common in DM that conventional melanomas, the low yield of positive results would thus imply limited, if any prognostic value as far as risk stratification for patients with DM. For that reason, there is no potential benefit for this procedure, according to Busam, and it is no longer recommended in patients with pure DM. As for radiation therapy, while conventional melanomas have been shown to be radiation-resistant, there was a study at the University of California showing that post-op adjuvant radiation decreased the incidence of local recurrences, and was therefore recommended by the authors. The need for wide and deep margins, the response to adjuvant radiation, as well as the low frequency of mets to sentinel lymph nodes at presentation of DM, underscore the importance in distinction between DM and conventional melanoma as separate subtypes, each requiring different management.
In a paper by Jukic and Spencer, discussing a 71 year-old man with clear cell sarcoma of tendons and aponeuroses (CCS), they underline the importance of differentiating this entity from malignant melanoma (MM), Pecoma, follicular dendritic and interdigitating cell tumors (IDCT/FDCT), and malignant peripheral nerve sheath tumors (MPNST). They go on to elaborate on how the distinction can be made between first CCS, MM and MPNST, all of which are derived from neural crest, and all of which show S100 positivity. First and perhaps most important, CCS has been shown recently to have a characteristic translocation, t(12;22)(q13;q13), resulting in a fusion gene, EWS/ATF12, and is apparently pathognomonic for these tumors. Additionally, CCS usually presents in young adults with a slight female predominance and was described as a slow-growing, painful nodule of the (often) lower extremities, usually arising in deep tissues around fascia, tendons and aponeuroses. Over the course of years, it may spread to regional lymph nodes, lungs or bone. The 5-year survival rate is estimated at 48-67%, lower if widely metastatic at diagnosis. The three can usually be separated immunohistochemically by a panel of stains. To name a few, MPNST can be distinguished from MM and CCS, as it is negative for HMB45, Melan-A, and Tyrosinase, while the other two are positive. Likewise, MM and CCS can be distinguished by GFAP, as MM is usually weakly positive and CCS is negative. CCS is also positive for synaptophysin, while MM is negative and CCS can be positive for chromogranin, and MM is not. Pecomas do not express S100 reactivity. IDCTs/FDCTs are both negative for Melan-A and Tyrosinase, while CCS is positive for both, as are Pecomas. Other stains that were compared were NSE, CD56, MITF-1, CD68, Vimentin, EMA, CD117, SMA, Desmin, HHF35, CD35 and CD21. The authors go on to point out that misdiagnosis is a danger to the dermatopathologist regarding these tumors, and early recognition of lesions is paramount. They stress the importance of being aware of the number of differential diagnoses associated with these types of lesions and the wide array of techniques at one's disposal for identifying the correct entity.
Contributed by Allene Gagliano, MD, Leena Lourduraj, MD and Drazen Jukic, MD, PhD