Desmoplastic neurotropic melanoma. A clinicopathologic analysis of 28 cases.

BACKGROUND: Desmoplastic neurotropic melanoma (DNM) is a rare variant of malignant melanoma, the natural history and histogenesis of which still are being defined.
METHODS: The clinical and histologic features of 28 cases of DNM were studied. All published cases of DNM to date were reviewed. Paraffin sections from 26 cases were investigated with a panel of 10 tissue markers. The ultrastructural features of seven cases were evaluated.
RESULTS: A comparison of this study's findings with that of other published cases revealed many similarities regarding clinical and pathologic findings and outcome. The patients were white (15 men:13 women; mean and median age, 59 years; range, 22-83 years). Most tumors were located on the head and neck (75%) and were nonpigmented (57%). An associated intraepidermal melanocytic proliferation was identified in 85% of the patients (lentigo maligna in 56%). Histologically, the dermal tumors were composed of tapered, nonpigmented spindle cells in peripheral nerve sheath patterns resembling neuromas, schwannomas, neurofibromas, and perineurial proliferations accompanied by variable neurotropism and desmoplasia; desmoplasia was the most notable feature in most tumors. The mean depth of tumor invasion was 4.1 mm (range, 0.32-9.0 mm). Tumors with continuity between the epidermal and dermal components had a significantly thinner depth of invasion and a more extensive intraepidermal melanocytic proliferation than those tumors with a grenz zone between the two components (2.3 mm vs. 4.6 mm, P = 0.015). Mitotic activity ranged from 0/HPF in 10 cases, 1-6/high power field (HPF) in 12 cases, and to greater than 6/HPF in 4 cases. An ulcer was present in 5/27 tumors, regression in 4/27, a microsatellite in 1, and brisk and had nonbrisk tumor infiltrating lymphocytic responses in 2 and 14, respectively. Vimentin was uniformly positive and keratins AE1.3 and Cam 5.2 and Leu-7 were uniformly negative. S100 protein, also uniformly positive, had patchy reactivity in most tumors that expressed EMA (43%). Smooth muscle actin (52%), neuron-specific enolase (42%), and FXIIIa (30%) had patchy positivity. HMB-45 was reactive only in the epidermal and superficial papillary dermal component in 21% of cases. Ultrastructurally, the common features were long, often intertwining cellular processes, intercellular junctions, and discontinuous basal lamina. Melanosomes were not identified. Follow-up data available on 26/28 patients (mean, 36 months; median, 24 months; range, 5-132 months) showed 20 (70%) alive without disease, 2 alive with disease and 3 dead from disease. Seven patients had recurrent local tumor (multiple in four); four had lymph node metastases, and three had visceral metastases. Patients with recurrent disease of any type had significantly thicker tumors (5.4 mm vs. 3.4 mm, P = 0.046) and were more likely to have an ulcerated tumor (P = 0.03). Actuarial 5-year survival for tumors with greater than a 4-mm thickness was 72%, which was greater than that for other types of melanoma with greater than a 4-mm thickness.
CONCLUSIONS: Desmoplastic neurotropic melanomas are neuroectodermal tumors that usually arise from an intraepidermal melanocytic proliferation but rarely develop de novo in the dermis. Schwannian and perineurial differentiation may account for the desmoplasia and neurotropism encountered in these neoplasms. Desmoplastic neurotropic melanomas present at a more advanced stage locally and may be associated with a better survival than associated with conventional melanomas of similar depth of invasion.