Immunohistochemical Profiling of Conjunctival Melanocytic Intraepithelial Lesions, Including SOX10, HMB45, Ki67, and P16.

Abstract

To determine the usefulness of melan-A, SOX10, HMB45, and p16 immunohistochemical stains in the distinction between the low-grade and high-grade conjunctival melanocytic intraepithelial lesions, either independently or as components of an immunohistochemical panel. Retrospective observational case series. Institutional pathology records between 2014 and 2018 were searched for all patients with conjunctival melanocytic intraepithelial lesions. Biopsies without supporting clinical history or tissue available for review and immunohistochemical analysis were excluded. Clinical, histopathologic, and immunohistochemical (p16, SOX10, HMB45, and Ki-67) findings were recorded. Thirty-one patients underwent 47 biopsies for conjunctival melanocytic lesions between 2014 and 2018. Pathologic diagnoses were low-grade conjunctival melanocytic intraepithelial lesion (n = 18, 38%) and high-grade conjunctival melanocytic intraepithelial lesion/melanoma in situ (n = 29, 62%). The addition of melan-A and SOX10 immunohistochemical stains resulted in an upgrade of conjunctival melanocytic intraepithelial lesion from low-grade to high-grade in 2 (4%) of 47 cases. The addition of melan-A and SOX10 immunohistochemical stains did not downgrade any of the histomorphologically high-grade lesions. In a clinical-pathologic multivariable model, the parameters most predictive of high-grade melanocytic intraepithelial lesion/melanoma in situ were involvement of the caruncle (odds ratio [OR] = 19, confidence interval [CI] 1.6-212; P = .02] and p16 cytoplasmic H-score >30 (OR = 81, CI 2.7 to >999; P = .01) CONCLUSION: Although the stains for melanocytic markers melan-A and SOX10 facilitate assessment of melanocytic intraepithelial lesions, the current immunohistochemical panels have limited value in distinction between the low-grade and high-grade intraepithelial melanocytic proliferations and need to be used judiciously.