The assessment and diagnosis of melanocytic lesions can present a formidable challenge to even a seasoned pathologist, which is especially true when dealing with the subset of nevi occurring at special sites—where baseline variations inherent to particular locations on the body can preclude the use of features routinely used to diagnose malignancy elsewhere. These so-called special-site nevi previously have been described in the literature along with suggested criteria for differentiating malignant lesions from their benign counterparts.1 Locations generally considered to be special sites include the acral skin, anogenital region, breast, ear, and flexural regions.1,2
When evaluating non–special-site melanocytic lesions, general characteristics associated with a malignant diagnosis include confluence or pagetoid spread of melanocytes, nuclear pleomorphism, cytologic atypia, and irregular architecture3; however, these features can be compatible with a benign diagnosis in special-site nevi depending on their extent and the site in question. Although they can be atypical, special-site nevi tend to have the bulk of their architectural distortion and cytologic atypia in the center of the lesion as opposed to the edges.1 If a given lesion is from a special site but lacks this reassuring feature, special care should be taken to rule out malignancy.
Preferentially expressed antigen in melanoma (PRAME) is an antigen first identified in tumor-reactive T-cell populations in patients with malignant melanoma. It is the product of an oncogene that frequently is overexpressed in melanomas, lung squamous cell carcinomas, sarcomas, and acute leukemias.4 It functions as an antagonist of the retinoic acid signaling pathway, which normally serves to induce further cell differentiation, senescence, or apoptosis.5 PRAME inhibits retinoid signaling by forming a complex with both the ligand-bound retinoic acid holoreceptor and the polycomb protein EZH2, which blocks retinoid-dependent gene expression by encouraging chromatin condensation at the RARβ promoter site5; therefore, expressing PRAME allows lesional cells a substantial growth advantage.
PRAME expression has been extensively characterized in non–special-site nevi and has filled the need for a rather specific marker of melanoma.6-10 Although PRAME has been studied in acral nevi,11 the expression pattern in nevi of special sites has yet to be elucidated. Herein, we present a dataset characterizing PRAME expression in these challenging lesions.
Methods
We performed a retrospective case review at the University of Virginia (Charlottesville, Virginia) and collected a panel of 36 special-site nevi that previously were diagnosed as benign by a trained dermatopathologist from January 2020 through December 2022. Special-site nevi were identified using a natural language filter for the following terms: acral, palm, sole, ear, auricular, lip, axilla, armpit, breast, groin, labia, vulva, umbilicus, and penis. This study was approved by the University of Virginia institutional review board.
The original hematoxylin and eosin slides used for primary diagnosis were re-examined to verify the prior diagnosis of benign nevus at a special site. We performed a detailed microscopic examination of all benign nevi in our cohort to determine the frequency of various characteristics at each special site. Sections were prepared from the formalin-fixed and paraffin-embedded tissue blocks and stained with a commercial PRAME antibody (#219650 [Abcam] at a 1:50 dilution) and counterstain. A trained dermatopathologist (S.S.R.) examined the stained sections and recorded the percentage of tumor cells with nuclear PRAME staining. We reported our results using previously established criteria for scoring PRAME immunohistochemistry7: 0 for no expression, 1+ for 1% to 25% expression, 2+ for 26% to 50% expression, 3+ for 51% to 75% expression, and 4+ for diffuse or 76% to 100% expression. Only strong clonal expression within a population of cells was graded.
Data handling and statistical testing were performed using the R Project for Statistical Computing (https://www.r-project.org/). Significance testing was performed using the Fisher exact test. Plot construction was performed using ggplot2 (https://ggplot2.tidyverse.org/).