The Implementation of DNA Methylation Profiling into a Multistep Diagnostic Process in Pediatric Neuropathology: A 2-Year Real-World Experience by the French Neuropathology Network.

Melanie Pages,Claude-Alain Maurage,David Meyronet,On Behalf Of The Renoclip-Loc Network ,Audrey Rousseau,Guillaume Gauchotte,Dominique Figarella-Branger,Catherine Godfraind,Karima Mokhtari,Emmanuelle Uro-Coste,Carole Colin,Karen Silva,Pascale Varlet

doi:10.3390/cancers13061377

Abstract

Simple SummarySupported by the “easy-to-use” and free DKFZ central nervous system (CNS) tumor classification web tool, DNA methylation profiling is a method changing the routine diagnostic practice in neuro-oncology. This work depicts a real-world practice experience by the French neuropathology network of incorporating DNA methylation profiling into the diagnostic process of challenging pediatric CNS tumors. After two rounds of histopathological review by neuropathology experts—including morphology, neuroimaging, immunohistochemistry, panel sequencing and FISH—62 tumors still presenting diagnostic uncertainty were selected for DNA methylation profiling. Using the DKFZ “classifier” and combining all additional information obtained from DNA methylation array, we observed significant diagnostic refinements and amendments. DNA methylation was successful in a significant number of cases (71%) despite the complex specificities of the cohort. Our study evaluates how DNA methylation testing would impact diagnosis and presents illustrative and representative cases.DNA methylation profiling has recently emerged as a powerful tool to help establish diagnosis in neuro-oncology. Here we present our national diagnostic strategy as the French neuropathology network (RENOCLIP-LOC) and our current approach of integrating DNA methylation profiling into our multistep diagnostic process for challenging pediatric CNS tumors. The tumors with diagnostic uncertainty were prospectively selected for DNA methylation after two rounds of review by neuropathology experts. We first integrated the classifier score into the histopathological findings. Subsequent analyses using t-SNE (t-Distributed Stochastic Neighbor Embedding) representation were performed. An additional step consisted of analyzing copy-number variation data (CNV). Finally, we combined all data to establish diagnoses and evaluated the impact of DNA methylation profiling on diagnostic and grading changes that would affect patient management. Over two years, 62 pediatric tumors were profiled. (1) Integrating the classifier score to the histopathological findings impacted the diagnosis in 33 cases (53%). (2) t-SNE analysis provided arguments for diagnosis in 26/35 cases with calibrated scores <0.84 (74.3%). (3) CNV investigations also evidenced alterations used for diagnosis and prognostication. (4) A diagnosis was finally established for 44 tumors (71%). Our results support the use of DNA methylation for challenging pediatric tumors. We demonstrated how additional methylation-based analyses complement the classifier score to support conventional histopathological diagnosis.

Highlights

The current WHO classification of CNS tumors, which encompasses more than 150 histologically and/or molecularly defined entities, mirrors the wide range of distinct groups and subgroups of CNS tumors that have recently emerged, in pediatrics [1]
Prior to reporting our experience of incorporating DNA methylation profiling in our diagnostic process, we briefly present the organization of the French neuropathology network (RENOCLIP-LOC Réseau de Neuro-Oncologie Clinico-pathologique pour les cancers rares du SNC) whose primary purpose is to jointly analyze difficult adult and pediatric CNS tumors, facilitate access to molecular biology platforms and harmonize the histopathological diagnosis of CNS tumors
DNA methylation profiling is only proposed as a last resort after a collegial decision is made by the national experts when consensus on the diagnosis has not been achieved because of conflicting morphological and/or molecular findings, non-informative molecular testing or other confusing aspects

Summary

Introduction

The current WHO classification of CNS (central nervous system) tumors, which encompasses more than 150 histologically and/or molecularly defined entities, mirrors the wide range of distinct groups and subgroups of CNS tumors that have recently emerged, in pediatrics [1]. The recent molecular findings have led to refining the classification by introducing newly described entities as well as subdividing morphologybased entities into molecular subgroups, resulting in challenges for routine diagnostic practice [1,2,3,4,5,6,7,8]. In parallel to these findings, the recent technological advances made have encouraged the integration of molecular testing, alongside morphological examination for minimizing inter-observer variability in histopathological diagnosis. The Heidelberg group made this approach applicable in a routine diagnostic setting by developing a free web tool accessible to everyone everywhere that provides an analysis report based on their

Methods

Results

Conclusion