Melanoma, a highly aggressive type of skin cancer originating from pigment-producing cells called melanocytes, can be a source of diagnostic confusion for pathologists. While primary melanomas are easily recognizable, metastatic melanomas can dedifferentiate and exhibit varied and misleading cyto-architectural characteristics, loss of immunohistochemical markers or present with aberrant morpho-immunophenotypes, leading to diagnostic confusion with non-melanocytic neoplasms. To improve diagnostic accuracy, clinical, histopathological, and genotypic criteria have been defined for dedifferentiated melanoma. These criteria include melanoma history, presence of a minimally differentiated clone, emergence of a non-specific histology, detection of a melanoma-associated mutation and the absence of an alternative primary tumor source.   

Hench et al interrogated clinical, histopathologic and molecular features, including methylation signature (MS) and copy number profile (CNP), in a retrospective analysis of 78 dedifferentiated melanoma (DedM) tissue samples from 61 patients retrieved from the European Organization for Research and Treatment of Cancer (EORTC). Tissue samples were collected from metastatic DedM, and when available, matched primary cutaneous melanomas along with clinical data including age, gender and treatment status. The H&E-stained slides were scanned using Aperio AT2 scanners. Each digitized slide was imported into the HALO Link image management system for remote review by three pathologists who were experts in melanoma and soft tissue pathology located across Europe. A diagnosis of ‘genuine DedM’ was given if the tumor met all the above criteria as well as testing negative for all routinely used melanoma immunohistochemical markers.  

Most patients (60/61) had a metastatic DedM, with the most common histological pattern being unclassified pleomorphic and spindle cell morphology similar to undifferentiated pleomorphic sarcoma (UPS)-like histologic features (21/60). Small round cell sarcoma-like histologic features with admixed medium-sized epithelioid cells was the second most common pattern (20/60). The authors observed large rhabdoid/epithelioid cell morphology in 12/60 cases, spindle cell sarcoma-like patterns resembling low-grade fibromyxoid sarcoma in 11/60 cases, small round cells admixed with spindled cells in 9/60 cases, myxoid/myxofibrosarcoma-like patterns in 3/60 cases and one case each of chondroblastic/osteogenic sarcoma and adenocarcinoma-like patterns. Foci of pleomorphic rhabdomyosarcoma-like patterns were observed in 2/60 cases with UPS-like morphology. One case with spindle-cell sarcoma-like features showed a focus of osteogenic features.  

Analysis of specimens from 20 of the 61 patients resulted in interpretable data for 16 patients. Hench et al found retained melanoma-like MS in seven tissue samples while a non-melanoma-like MS was observed in 13 tissue samples. In two patients with multiple specimens analyzed, some of the samples preserved cutaneous melanoma MS while other specimens exhibited an epigenetic shift towards mesenchymal/sarcoma-like profile, matching the observed histological features. In these two patients, CNP was largely identical across all analyzed specimens, in line with their common clonal origin, despite significant modification of their epigenome.  

Hench et al found that loss of MS occurs in DedM to the degree that it becomes impossible to epigenetically identify them as melanomas. They also found that the epigenetic features that were present in DedM could lead to misdiagnosis in a large proportion of cases, typically resulting in misdiagnosis as osteosarcoma or malignant peripheral nerve sheath tumor. They also found that in most cases of DedM, CNP was preserved despite loss of the melanoma MS. They concluded that clinical features, mutation profile, CNP and MS should all be utilized to correctly identify cases of DedM, which is an important identification to make as many patients with DedM are responsive to systemic therapy.