The Impact of Next-generation Sequencing on Interobserver Agreement and Diagnostic Accuracy of Desmoplastic Melanocytic Neoplasms

Chen A, Sharma N, Patel P, Olivares S, Bahrami A, Barnhill RL, Blokx WAM, Bosenberg M, Busam KJ, de La Fouchardière A, Duncan LM, Elder DE, Ko JS, Landman G, Lazar AJ, Lezcano C, Lowe L, Maher N, Massi D, Messina J, Mihic-Probst D, Parker DC, Redpath M, Scolyer RA, Shea CR, Spatz A, Tron V, Xu X, Yeh I, Jung Yun S, Zembowicz A, Gerami P. Am J Surg Pathol, doi:10.1097/pas.0000000000002226.

Abstract

Next-generation sequencing (NGS) is increasingly being utilized as an ancillary tool for diagnostically challenging melanocytic neoplasms. It is incumbent upon the pathology community to perform studies assessing the benefits and limitations of these tools in specific diagnostic scenarios. One of the most challenging diagnostic scenarios faced by skin pathologists involves accurate diagnosis of desmoplastic melanocytic neoplasms (DMNs). In this study, 20 expert melanoma pathologists rendered a diagnosis on 47 DMNs based on hematoxylin and eosin sections with demographic information. After submitting their diagnosis, the experts were given the same cases, but this time with comprehensive genomic sequencing results, and asked to render a diagnosis again. Identification of desmoplastic melanoma (DM) improved by 7%, and this difference was statistically significant (P<0.05). In addition, among the 15 melanoma cases, in the pregenomic assessment, only 12 were favored to be DM by the experts, while after genomics, this improved to 14 of the cases being favored to be DM. In fact, some cases resulting in metastatic disease had a substantial increase in the number of experts recognizing them as DM after genomics. The impact of the genomic findings was less dramatic among benign and intermediate-grade desmoplastic tumors (BIDTs). Interobserver agreement also improved, with the Fleiss multirater Kappa being 0.36 before genomics to 0.4 after genomics. NGS has the potential to improve diagnostic accuracy in the assessment of desmoplastic melanocytic tumors. The degree of improvement will be most substantial among pathologists with some background and experience in bioinformatics and melanoma genetics.