Diagnostic utility of PRAME, p53 and 5-hmC immunostaining for distinguishing melanomas from naevi, neurofibromas, scars and other histological mimics.

High expression of PRAME (PReferentially expressed Antigen in MElanoma) and p53 (a proposed marker of desmoplastic melanoma) and low expression of 5-hydroxymethylcytosine (5-hmC) have each been reported in melanoma. However, their combined diagnostic utility for distinguishing melanomas, including uncommon variants, from histological mimics is unknown. This study sought to determine the utility of PRAME, p53 and 5-hmC immunostains for diagnosing melanocytic tumours. A total of 333 cutaneous melanocytic tumours (melanoma n=280, naevi n=53), 20 cutaneous neurofibromas, and 15 scars were evaluated using multiplex immunofluorescence (n=313) or single-plex chromogenic immunohistochemical staining (n=55). Immunostaining for PRAME and 5-hmC were each scored using a previously described semiquantitative scale 0 (absent) to 4+ (diffuse). p53 was scored using a previously described immunoreactive score (range 0-300). PRAME expression was significantly higher in melanomas than in naevi (p<0.0001), with the lowest PRAME expression found in low-grade desmoplastic melanomas compared to the other melanoma subtypes. In non-desmoplastic melanomas, 38% showed 4+ staining (>75% positive tumour cells) and 70% showed 3+ or 4+(>50% positive tumour cells). Conversely, 96% of naevi showed 0, 1+ or 2+ expression. 5-hmC expression was significantly lower in melanomas than in naevi (p<0.0001). However, acral melanomas were not significantly associated with loss of 5-hmC expression (p=0.84). Compared with using PRAME in isolation, combining PRAME and 5-hmC scores increased sensitivity (64%-84%) for detecting melanoma. With respect to desmoplastic melanoma compared to scar or neurofibroma, strong PRAME or p53 staining were almost exclusively found in high-grade desmoplastic melanomas; low-grade desmoplastic melanomas, neurofibromas and scars were negative. 5-hmC was not useful in distinguishing desmoplastic melanomas from neurofibromas or scars. Our data support the use of PRAME as a highly specific ancillary investigation in the diagnosis of melanoma, however PRAME should be considered 'positive' if there is 3+ or 4+ staining (rather than the widely recommended 4+ threshold). 5-hmC, PRAME and p53 appear to have a limited role in the diagnosis of low-grade desmoplastic melanomas.