Abstract
BACKGROUND: Recent advances in genetics and epigenetics research have underscored the heterogeneity of pediatric brain tumors, characterized by overlapping morphological and molecular features across various tumor types. This complexity presents substantial challenges for precision diagnosis in routine diagnostics. In this study, we collected a cohort of pediatric high-grade brain tumors that remain unresolved after next-generation sequencing (NGS). Our goal is try to refine diagnosis by DNA methylation profiling, thereby evaluating the utility of this new diagnostic tool against unresolved and challenging cases in standard clinical practice. METHODS: In this retrospective analysis, we selected 10 cases from local pathology archives based on the following criteria: (1) age at diagnosis< 18 years; (2) brain tumor with high-grade evidence; (3) NGS was employed but failed to provide a definitive diagnosis or not aligned to WHO CNS tumor classification. All cases were subjected to the Illumina Infinium MethylationEPIC v2.0 BeadChip, DNA methylation profiling was conducted by using an customized pipeline alongside the DKFZ brain tumor classifier. NGS was previously done by utilizing a DNA-panel targeting over 500 genes associated with tumor-related genetic variations. The collection of tumor samples and clinical data was processed in accordance with standards approved by the local ethical committees. RESULT: All cases were supratentorial high-grade tumors, aged 1 to 16 years (n=10). Utilizing DNA methylation profiling, definitive diagnoses were established in 60% of cases: two cases of embryonal tumor with multilayered rosettes (ETMR), one pediatric diffuse high-grade glioma MYCN subtype, one pleomorphic xanthoastrocytoma (PXA), one medulloblastoma non-WNT/non-SHH subgroup 7, and one alveolar rhabdomyosarcoma. The remaining cases (n=4) neither yielded convincing calibrated scores from the DKFZ brain tumor classifier (v12.8) nor provided valuable insights from methylation based t-SNE clustering or copy-number profiling analysis. These 4 cases includes: a pediatric high grade glioma (pHGG) harboring CDK12, DICER1, JAK3, NF1, NTRK2, RB1 and SPEN mutation; a pHGG harboring PIK3CA, NF1 and TP53 mutation; a malignant embryonal tumor in the pineal gland harboring ATRX, BCL2, BLM, CYSLTER2, GPR101, HIF1A and SHOC2 mutation; and a malignant embryonal tumor with TERT, SMARCA4 mutation and EML4-ALK fusion. CONCLUSION: In confronting the diagnostic complexities of pediatric high-grade brain tumors, DNA methylation profiling may provide more critical evidence that can address and unravel intricate clinical diagnostic challenges. SUPPORT/DISCLOSURE: This research was supported by Natural Science Foundation of Chongqing (No.CSTB2023NSCQ-BHX0105).