79 Diagnostic Utility of DNA Methylation Profiling in CNS Tumors: A Comparative Study of FFPE and Fresh-Frozen Tissues

Introduction DNA methylation–based classification has become an essential adjunct to histopathology in central nervous system (CNS) tumor diagnostics. Although both fresh-frozen and formalin-fixed paraffin-embedded (FFPE) tissues are used for methylation profiling, comparative real-world data on their diagnostic performance remain limited. Methods We analysed 179 CNS tissue samples (162 tumors, 17 non-neoplastic controls), comprising 42 FFPE and 137 fresh-frozen specimens, all with ≥50% tumor cellularity. Genome-wide DNA methylation profiling was performed using Illumina Infinium EPIC/EPIC v2.0 arrays. Classification was carried out using the DKFZ Brain Tumor Classifier (versions 11b–12.8) and independently validated with the NIH Methylscape classifier. Classifier outputs were correlated with histopathological diagnoses and categorised as confirmed, refined, changed, or unclassified. Results Overall classification success was higher in FFPE samples compared with fresh-frozen tissue (98% vs. 92%). High-confidence classification (calibrated score >0.9) was achieved in 62% of FFPE and 74% of fresh-frozen cases. Despite this, discordant or misleading classifications were predominantly observed in fresh-frozen samples (8% of discordant cases), largely attributable to sampling bias and low effective tumor content. FFPE samples demonstrated superior diagnostic refinement (67%) and a lower rate of unsatisfactory or misleading results (2%). Methylation profiling resulted in diagnostic confirmation or refinement in the majority of cases and led to a revised diagnosis in a clinically significant subset. Conclusion DNA methylation profiling is a robust diagnostic adjunct for CNS tumors using both FFPE and fresh-frozen tissue. In routine practice, FFPE specimens show more consistent performance, likely due to histologically guided sampling and reduced tissue heterogeneity, supporting their reliable use in methylation-based diagnostics.