Unusual variants of malignant melanoma

Abstract

Melanoma is as diverse in its presentation as it can be in its behavior. This contribution will review the clinical and histologic features of some of the more unusual variants of malignant melanoma that clinicians and pathologists are likely to encounter. There will be a disproportionate focus on the histopathology and prognosis because the clinical presentation of these lesions is in most cases, elusive, and in some cases, frankly deceptive. The discussion will include desmoplastic melanoma, nevoid melanoma, spitzoid melanoma, angiotropic melanoma, and malignant blue nevus.

Introduction

Malignant melanoma is, without question, one of the most heterogeneous and complex human neoplastic systems. Thus, it is not surprising that one encounters on a regular basis melanomas that are difficult to categorize as one of the major “clinicopathologic variants.”¹¹ If there is difficulty in establishing whether an individual lesion is a conventional melanoma, even greater difficulty clearly exists in establishing whether unusual variants of melanoma can be delineated from the objective evidence. In practical terms, the recognition and description of additional legitimate variants of melanoma must have some particular biologic significance or relevance, and there must be objective criteria for their recognition and distinction from other presentations of melanoma. It is evident that this has not been achieved for many entities and much more basic research is needed.

The description of unusual presentations of melanoma has been based on a variety of disparate phenotypic characteristics:

• epidermal surface configuration, such as verrucous, polypoid surface topography;

• stromal alterations such as desmoplasia and mucin deposition;

• morphologic resemblance or mimicry of benign melanocytic neoplasms such as “nevoid” melanoma, “spitzoid” melanoma, blue nevus-like melanoma (malignant blue nevus), resemblance to plexiform or deep-penetrating nevus;

• peculiar cytologic characteristics such as small cell, spindle cell, epithelioid cell, balloon cell, signet ring cell, rhabdoid cell, and clear cell variants of melanoma;

• other morphologic properties such as neurotropism, neural differentiation, and angiotropism, and

• pigmentary characteristics, including amelanotic, pigment-synthesizing, or “animal-type” melanoma.

It is obvious that there is overlap among a number of the latter categories.

This contribution outlines the clinical and histologic features of some of the more unusual variants of malignant melanoma that clinicians and pathologists are likely to encounter at some point in their professional careers. There will be a disproportionate focus on the histopathology and prognosis, because the clinical presentation of these lesions is in most cases, elusive, and in some cases, frankly deceptive when compared with conventional lesions of malignant melanoma. The discussion will include desmoplastic melanoma, nevoid melanoma, spitzoid melanoma, angiotropic melanoma, and malignant blue nevus.

Desmoplastic malignant melanoma (DM), like nevoid melanoma, harbors, as perhaps its greatest significance, its accurate identification as malignant melanoma. The issue is compounded because the lesion is often amelanotic and highly infiltrative, with a propensity for perineural spread, which may contribute to its hallmark capacity for local recurrence, a feature that distinguishes it from other types of melanoma.

The demographics of DM have been characterized by a predominance in men in the 1960s and 1970, with most reported lesions involving sun-damaged areas of the head and neck, although not at the exclusion of other body sites, including acral and mucosal surfaces.2, 3, 4 They present as nodules or plaques, sometimes depressed, that are very often amelanotic. In those that demonstrate pigmentation, however, it is often in the form of a conventional melanoma (often so-called lentigo maligna) component, which happens to be congruent on a histologic level because these lesions often represent the invasive component of melanomas involving chronically sun-exposed skin. (For purposes of this discussion, the historical terms “lentigo maligna” and “lentigo maligna melanoma” are used to represent lentiginous in situ and invasive melanomas arising in chronically sun-damaged skin). Desmoplastic melanoma can also arise de novo, however, and it is this situation in which it poses the greatest diagnostic challenge. The lesion may also appear scar-like clinically, which is appropriate given the histologic profile.

The histologic hallmark of DM is that of an ill-defined spindle cell neoplasm of varying concern with regard to density and cytologic atypia that often demonstrates a highly infiltrative pattern of growth amidst sclerotic collagen fibers. In fact, the collagen density on scanning magnification resembles a scar, with the embedded cellularity being reminiscent of an accompanying fibroblastic component (Figure 1).

Other variants of DM may demonstrate a more conspicuous spindle cell proliferation in which the degree of sclerosis is less ubiquitous throughout the lesion (Figure 2). The spindle cells may be present as single, markedly atypical, infiltrating cells (Figure 3) or as a predominant arrangement of fascicles that sweep through dense collagen bundles in a manner reminiscent of a neural or smooth muscle proliferation (Figure 4). There is nuclear hyperchromasia and contour irregularity of the spindle cell forms (Figure 5), often with the added histologic feature of lymphoid aggregates and solar elastosis.

In fact, these latter two features in the setting of a sclerotic lesion, even with minimal or almost no cytologic atypia and inconspicuous mitotic activity, are often instrumental in arriving at the correct diagnosis. This is particularly important in those cases that are deceptively benign in appearance. The mitotic count in most tumors is relatively modest, adding to the diagnostic confusion. Of the utmost importance is that the dermal–epidermal junction is always examined in such cases for a subtle lentiginous melanocytic component possibly representative of an overlying atypical lentiginous melanocytic proliferation or melanoma in situ/lentigo maligna (Figure 6).

The immunohistochemical profile in most DM lesions is characterized by S100 (Figure 7), p75 neurotrophin receptor, and vimentin positivity. Smooth muscle actin positivity may be present in many cases. Human melanoma black-45 (HMB-45) and Melan-A (Figure 8) stains are inconsistent at best, and cases are completely negative with regard to the invasive spindle cell component.

As a practical aside, caution must be exercised with the S100 stain in biopsy and reexcision specimens of DM, because almost invariably some degree of S100 staining is present within the native dermis and in the setting of nonmelanocytic proliferations. These may be indicative of dermal antigen-presenting cells. Such cells have also been documented in scars (Figure 9), where the nature of these cells have been postulated by different researchers as being representative of Langerhans cells,⁵ fibroblasts,⁶ and Schwann cells or regenerating nerve twigs.⁷ It is important to note that such cells do normally exist in the dermis, and they represent a common pitfall that may contribute to an erroneous diagnosis of DM, particularly in the setting of scars that mimic DM and also in the setting of reexcision specimens, where they may be misinterpreted as being representative of residual DM.

The pathogenesis of DM remains unresolved. Some investigators, by way of immunohistochemical and ultrastructural analysis, have suggested that bidirectional differentiation is involved in these tumors.⁸

The histologic differential diagnosis of DM includes scar, desmoplastic nevus, desmoplastic Spitz nevus, sclerosing blue nevus, dermatofibroma, dermatofibrosarcoma protuberans, neural tumors, sarcomas, spindle cell squamous cell carcinoma, and atypical fibroxanthoma. Perhaps the most important initial step in distinguishing a DM is an examination of the dermal–epidermal junction for the presence of a suspicious junctional melanocytic component and the lesion for the presence of melanin within tumor cells.

Desmoplastic melanoma can quite incredibly mimic a scar. It is important to beware of “scars” that demonstrate (1) conspicuous cellularity, (2) solar elastosis, (3) lymphoid aggregates, (4) retained adnexal structures, and (5) a lack of horizontal fibrosis, vertically oriented vessels, and erythrocyte extravasation. Greater scrutiny should be applied to the cytomorphology in such cases with regard to plump, spindled cells with nuclear hyperchromasia and enhanced mitotic activity.

Desmoplastic nevi are usually well circumscribed and show features of benign nevomelanocytes only embedded in an inordinately sclerotic stroma (Fig. 10, Fig. 11). Aside from this sclerotic stroma, there is little architectural and cytomorphologic resemblance to DM in most instances.

Desmoplastic Spitz nevi often impart the clinical and histologic appearance of a dermatofibroma. Examination of the cytomorphology may raise some concern, given the plumpness of the cells that is typical of spitzoid lesions. There is not, however, a dominant spindled cytomorphology with wide dermal involvement. The lesions are usually relatively well circumscribed, and the cells are usually identifiable as being spitzoid (Fig. 12, Fig. 13). There is perhaps more chance for confusion with an atypical Spitz tumor rather than with a DM.

Once again, sclerosing blue nevus resembles DM only by way of the dermal sclerosis. Dermal sclerosis is a relatively consistent feature of common blue nevi and sclerosing blue nevi. The presence of relatively bland epithelioid and fusiform cells with delicate to prominent cytoplasmic melanization dominates the histologic appearance (Fig. 14, Fig. 15).

Conventional and cellular dermatofibromas have been known to mimic DM. Some dermatofibromas indeed show a plump, spindled cytomorphology with, in some cases, significant cytologic atypia and enhanced mitotic activity. There is usually evidence, however, of peripheral collagen trapping, no junctional melanocytic component, and the cytomorphology in most is relatively bland. In addition, the heterogeneity of the cellular populace, in the form of multinucleate and xanthomatized histiocytes and a mixed inflammatory cell element, helps to distinguish dermatofibroma from DM. This applies even to those cases of cellular dermatofibroma and atypical fibrous histiocytoma that show dense dermal cellularity and bizarre cells. Dermatofibroma variants can nearly always be distinguished from DM on routine staining.

Dermatofibrosarcoma protuberans demonstrates a highly cellular and diffuse dermal proliferation of spindled cells that involves the dermis and subcutis. The primary arrangement is storiform in most cases, and the cytomorphology is characteristically monotonous; in fact, it is more monotonous and homogeneous than dermatofibromas. The spindled nuclei are delicate and betray their malignant nature largely by way of architecture, rather than cytomorphology, in the form of diffuse involvement of the dermis and subcutis with entrapment of subcutaneous adipocytes. Diagnostic confirmation is achieved with positive staining for CD34, and lesional negativity for S100 allows for histologic distinction from DM.

Neural proliferations such as malignant peripheral nerve sheath tumor (MPNST) pose a significant challenge, for in the absence of a junctional melanocytic component, histologic and immunohistochemical overlap exists with S100. A potentially useful discriminating feature in this setting is the patchy positivity of S100 in MPNST as opposed to the diffuse lesional positivity for S100 in most cases of DM. Some MPNSTs may also be associated with a background of neurofibroma. Malignant peripheral nerve sheath tumor arising in the skin is a distinctly uncommon phenomenon, but distinguishing between these two conditions can pose a serious diagnostic challenge.

Sarcomas such as fibrosarcoma and leiomyosarcoma can show significant histologic overlap with DM, but immunohistochemistry allows consistent differentiation from DM.

Spindle cell squamous cell carcinoma also shows considerable histologic overlap with DM. Scrutiny of the dermal–epidermal junction can be instrumental in such cases should either keratinocyte atypia or a junctional melanocytic proliferation be identified. If this does not prove useful, immunohistochemistry is a reliable arbitrator in these cases.

Atypical fibroxanthoma can show spindle cell features. Once again, the immunohistochemical profile serves to resolve the diagnostic confusion.

The biologic behavior of DM has been a subject of some controversy, largely due to the relatively low number of cases and the increasing recognition that the natural history of this lesion appears to differ from that of conventional melanomas. First, diagnostic delay, by virtue of the subtlety of the clinical and histologic findings, contributes significantly to the advanced Breslow depth that this lesion often demonstrates at diagnosis. Even in those that are biopsied, limited lesional sampling may sometimes preclude the comprehensive histologic analysis that is required to diagnose these lesions. A recent large review of the literature found the mean Breslow depths were 2.0 to 6.5 mm.⁹ Most cases present with at least a Clark level IV.

Local recurrence, more so than distant metastasis, distinguishes this tumor from other types of melanoma. Perineural invasion and the ill-defined nature of these lesions likely play a significant role in this regard. Distant spread is thought to be less common despite the advanced tumor thickness that is characteristic of DM.¹⁰ One study of 129 patients with a mean tumor thickness of 4.2 mm reported recurrence in 51%.¹¹ A review of the literature encompassing 703 patients documented a 26.2% recurrence rate, a 7.1% nodal metastatic rate, and a 19.8% rate of systemic metastasis.¹ A more recent study of nodal metastasis in DM reported one nodal metastasis in a cohort of 18 who underwent lymph node biopsy.¹⁰

The prognosis of DM is generally regarded as being more favorable than that of conventional melanomas for a given tumor thickness. A study of 129 patients with a mean tumor thickness of 4.2 mm reported survival rates of 76% at 5 years and 64% at 10 years.¹¹ A study at the Sydney Melanoma Unit¹² that evaluated 280 patients with a mean tumor thickness of 2.5 mm reported a 5-year survival of 75%. The neurotropic variety of DM was present in 90 of these 280 patients. The researchers failed to demonstrate that perineural invasion affected rates of survival in DM. They further reported fewer lymph node metastasis in DM compared with those associated with conventional melanomas.¹² Another report indicated that DM lesions greater than 4 mm in thickness demonstrate a 5-year survival of 60% compared with 40% in non-DMs.¹³ These findings contrast with those of a study of 89 patients that reported survival rates that were similar to those of non-DMs of similar thickness.¹⁴

A 2005 study engaged in a nuanced examination of the histologic features for their prognostic contribution.¹⁵ Cases of DM were subdivided into “pure” and “mixed” subtypes, being respectively representative of lesions with prominent stromal desmoplasia and those in which the desmoplastic component was simply an element of what was otherwise a more conventional melanoma. Those patients classified as pure DM showed a significantly lower incidence of lymph node metastasis and a lower 5-year melanoma-specific mortality compared with the mixed DM patients. The pure DM and conventional melanoma patients had the same melanoma-specific mortality rate despite a threefold higher median tumor depth in the former. Another interesting finding was that lymph node metastases in the mixed DM patients were characterized by the nondesmoplastic component of the primary mixed DM lesions.¹⁵ It may, therefore, be reasonable to subclassify these lesions as some suggest¹⁵ and designate the pure forms of DM as “spindle cell” melanomas.

Reasonable evidence therefore exists to support the notion that DMs maintain a prognostic advantage compared with conventional lesions of malignant melanoma. The question that is perhaps more intriguing, however, relates to the pathobiology of this neoplasm. Specifically, should a lesion that deviates on a morphologic, immunohistochemical, and behavioral basis from melanoma continued to be classified as a melanoma versus a “primitive melanocytic or neurocristic spindle cell sarcoma” (one can recall the use of the historical term “melanosarcoma”)? Perhaps future studies will seek to address this issue.

Desmoplastic malignant melanoma carries a high risk for local recurrence that is directly related to its frequent amelanotic appearance, often advanced stage at the time of diagnosis, Breslow thickness exceeding 4.0 mm, and neurotropic and angiotropic properties, as mentioned. Resection of the DM with clear surgical margins as early as possible in its development is thus crucial to its successful clinical management. Although not systematically studied, most surgeons recommend a minimum clearance of at least 1 cm. As already discussed, there is increasing evidence that sentinel lymph node (SLN) biopsy may not be indicated for “pure” variants of DM because of their low incidence of regional lymph node metastases.