Desmoplastic Melanomas
Desmoplastic melanomas (DM) are an uncommon subtype of melanoma included in the WHO Classification of tumors and are generally defined by the presence of neoplastic spindle cells separated from one another by dense collagenous stroma.1–5 Often, DMs present on chronically sun-damaged skin with a predilection for the head and neck region in older individuals with a median age of diagnosis being ten years later than conventional melanomas; however, cases have been reported in young patients as well as in a variety of anatomic locations.2,3,6 Tumors usually present as an indurated, ill-defined plaque or nodule with overlying pigmentation in varying shades of tan and nearly half lacking pigmentation entirely.2,4,6
These tumors can have a broad spectrum of histologic appearances ranging from traditional paucicellular scar-like morphology with only mild atypia to overtly malignant pleomorphic spindle cells.3–5 DMs often show deep invasion in the reticular dermis or subcutis diffusely replacing the fat lobules. Characteristic diagnostic clues that may be present include prominent lymphocytic aggregates at the tumor edge and neural involvement.1–3, 6 While DMs have a well-known association with lentigo maligna melanoma, up to 30% of cases do not have an identifiable melanoma in situ (MIS) component.4 Ancillary immunohistochemical stains are of limited value as virtually all DMs are distinctively negative for classic melanoma markers such as HMB45, Melan-A, and Tyrosinase. The majority label with S-100 protein and SOX10 in up to 96% and 92% of cases, respectively, according to one meta-review.1–5 However, these stains may show more limited labeling when compared to conventional melanomas.
Due to the heterogeneity in tumor morphology, classification criteria were set by Busam et al. in 2004 based on the percentage of invasive tumor cells displaying desmoplastic features (i.e., individual tumor cells separated from one another by fibrous stroma).1,2,6 Pure DM (pDM) is defined by 90% or greater of the invasive component being desmoplastic, while mixed DM (mDM) is defined as having a desmoplastic component accounting for less than 90% but more than 10% of the invasive component. The non-desmoplastic portion of mDM is often composed of pleomorphic spindle or fusiform cells organized in fascicle-like structures or, less commonly, non-spindled epithelioid cells. When these criteria are strictly applied, the pure desmoplastic subtypes tend to behave more akin to sarcomas with a greater risk of local recurrence, increased hematogenous spread to the lung, and rare sentinel lymph node metastasis when compared to mDM and melanoma without any desmoplastic features.1 Additionally, pDM is associated with longer disease-free survival than mixed DM and conventional melanomas of a similar thickness.1,2,6
While not codified by the WHO Classification system, the commonly termed “spindle cell melanoma” is a morphologic variant of interest. They also demonstrate a predominantly spindle cell morphology but notably lack a desmoplastic component (i.e., cells individually surrounded by collagenous matrix). These have historically been considered to be on a histologic spectrum with DM; however, given the different clinical behavior between conventional melanoma and DM when strictly defined, there is potential value in distinguishing these two entities.2 Melanomas with less than 10% of a desmoplastic component are not considered to be desmoplastic melanomas by some, but there is at least practical value in noting the presence of desmoplastic features in these lesions as it may let the pathologist evaluating a subsequent specimen know what she/he may find. The nosologic distinction between subtypes of melanoma notwithstanding, they pose a diagnostic challenge when the differential diagnosis includes other spindle cell lesions, particularly malignant peripheral nerve sheath tumor (MPNST).2,7
Malignant Peripheral Nerve Sheath Tumor
MPNST is an uncommon sarcoma composed of fascicles of spindle cells arranged in alternating hypo- and hypercellular areas with a tendency to show perivascular accentuation of tumor cells. The degree of cellular pleomorphism is often limited but may be striking, particularly in cases with heterologous differentiation.8,9 These tumors frequently arise in a large peripheral nerve trunk or a pre-existing benign nerve sheath tumor and present as an enlarging mass or fusiform enlargement of a nerve, most commonly in the trunk or extremities, followed by the head and neck region.7,8 While up to 50% of cases occur in the setting of Neurofibromatosis Type 1 (NF1), tumors may occur sporadically or in the setting of previous radiation.7 Sporadic MPNST tends to occur between 20 to 60 years of age, with patients diagnosed with NF1 presenting at a younger age.8,9
There is general agreement that the diagnosis of MPNST can be made when a lesion displays appropriate histomorphologic features and arises from a peripheral nerve or benign peripheral nerve sheath tumor. However, the diagnosis is quite challenging when these relationships are lacking or are unclear, particularly in sporadic cases. In these cases, it is broadly accepted that with the appropriate morphology, the diagnosis of a conventional MPNST can be made based on the identification of neural differentiation, ordinarily evidenced by focal or patchy S100-protein or SOX10 positivity, and or the loss of the methylation marker, H3K27me3.7,8 While the loss of the H3K27me3 is considered to be highly specific in this context, up to 50% of MPNSTs demonstrate an intact staining pattern. The utility of the H3K27me3 marker is most significant in the post-radiation setting and in high-grade lesions. Such lesions will show loss of staining in 90% of cases. In contrast, low-grade lesions will show a loss pattern in only 30% of cases.9
This diagnostic challenge is further complicated when considering epithelioid MPNST. This rare subtype is composed of plump, epithelioid cells with abundant eosinophilic cytoplasm that may be embedded in a hyalinized or myxoid extracellular matrix. It is traditionally not associated with NF1 and, uniquely, may be seen arising from conventional schwannoma. SOX10 and S100-protein IHC is likely to be strong and diffuse with a retained pattern of expression for H3K27me3.8
While differentiating MPNST from melanoma with spindle cell morphology is not a common scenario one may encounter in the diagnosis of melanocytic lesions, in our tertiary referral practice, we have seen cases where this has been an issue, particularly in sporadic cases and sites where the ‘deep’ soft tissues can be sampled with superficial methods in the clinic, as in the head and neck. Furthermore, most DM arise in the head and neck region and often have more extensive neurotropism than DM at other sites.2,4,5 Neurotropic patterns that have been described closely mimicking MPNSTs include “neural transformation,” in which the invasive component closely mimics the growth pattern of a nerve sheath tumor, as well as a rare “nerve-centered” pattern that is reserved when growth is confined entirely to the nerve or nerve sheath.2 MPNSTs and the aforementioned melanoma subtypes can be nearly impossible to distinguish on morphology alone without an associated in situ lesion. Immunohistochemical studies have been considered to be of limited diagnostic use as there is considerable overlap in the expression patterns. Others have reported on these challenges surrounding this differential.3,7,9,10,11
Prame
PRAME (PReferentially expressed Antigen in MElanoma) is an antigen expressed in cutaneous and ocular melanomas. Lezcano et al. found that diffuse nuclear labeling with PRAME was present in up to 83% of the 155 primary melanomas tested. Of the primary melanomas, 20 were classified as desmoplastic melanomas, with 35% showing labeling for PRAME, and of the mixed desmoplastic melanomas, the non-desmoplastic portion preferentially expressed PRAME.12 To our knowledge, Caldwell et al. have the only published study evaluating the diagnostic utility of PRAME immunohistochemistry (IHC) in MPNSTs compared to benign peripheral nerve sheath tumors with up to 65% of MPNSTs demonstrating PRAME labeling.13 Due to the rarity of MPNSTs and DMs, the evaluation of large cohorts is problematic, and further investigation is warranted to characterize PRAME IHC labeling patterns fully. Our goal is to assess the diagnostic utility in the PRAME IHC marker in the context of distinguishing MPNSTs from histomorphologically similar melanomas, specifically spindle cell melanomas.