DNA Methylation Profiling of Salivary Gland Tumors Supports and Expands Conventional Classification
Philipp Jurmeister,Maximilian Leitheiser,Alexander Arnold,Emma Payá Capilla,Liliana H Mochmann,Yauheniya Zhdanovic,Konstanze Schleich,Nina Jung,Edgar Calderon Chimal,Andreas Jung,Jörg Kumbrink,Patrick Harter,Niklas Prenißl,Sefer Elezkurtaj,Luka Brcic,Nikolaus Deigendesch,Stephan Frank,Jürgen Hench,Sebastian Försch,Gerben Breimer,Ilse van Engen van Grunsven,Gerben Lassche,Carla van Herpen,Fang Zhou,Matija Snuderl,Abbas Agaimy,Klaus-Robert Müller,Andreas von Deimling,David Capper,Frederick Klauschen,Stephan Ihrler
DOI: https://doi.org/10.1016/j.modpat.2024.100625
2024-09-25
Abstract:Tumors of the major and minor salivary glands histologically encompass a diverse and partly overlapping spectrum of frequent diagnostically challenging neoplasms. Despite recent advances in molecular testing and the identification of tumor-specific mutations or gene fusions, there is an unmet need to identify additional diagnostic biomarkers for entities lacking specific alterations. In this study, we collected a comprehensive cohort of 363 cases encompassing 20 different salivary gland tumor entities and explored the potential of DNA methylation to classify these tumors. We were able to show that most entities show specific epigenetic signatures and present a machine learning algorithm that achieved a mean balanced accuracy of 0.991. Of note, we showed that cribriform adenocarcinoma is epigenetically distinct from classical polymorphous adenocarcinoma, which could support risk stratification of these tumors. Myoepithelioma and pleomorphic adenoma form a uniform epigenetic class, supporting the theory of a single entity with a broad but continuous morphologic spectrum. Furthermore, we identified a histomorphologically heterogeneous but epigenetically distinct class that could represent a novel tumor entity. In conclusion, our study provides a comprehensive resource of the DNA methylation landscape of salivary gland tumors. Our data provide novel insight into disputed entities and show the potential of DNA methylation to identify new tumor classes. Furthermore, in future, our machine learning classifier could support the histopathologic diagnosis of salivary gland tumors.