Abstract
BACKGROUND: Genetic and epigenetic profiles are critical in managing brainstem gliomas (BSG), whose heterogeneity is far beyond the realm of the Diffuse midline glioma, H3K27 altered. Cerebrospinal fluid (CSF) circulating tumor DNA (ctDNA)-based liquid biopsy provides minimally-invasive strategies to acquire molecular information for brain tumors, whereas there is a deficiency in techniques for co-detection of genetic and epigenetic alterations due to the limited yield of ctDNA. This study aims to develop a reliable minimally-invasive approach to simultaneously detect the mutation and methylation profiles in the CSF ctDNA of BSGs, thereby enhancing diagnostic accuracy, prognostic capability, and monitoring potential. METHODS: A cohort of 80 BSG cases with 138 CSF samples and 71 tissues was retrospectively established. Public tissue methylation profiles (N = 1016) were used for the development of H3K27M and IDH mutation-specific assay. The mutation and methylation co-detection classifier (BSGdiag) was trained and tested in tissue cohorts and further validated in CSF samples. CSF Methylation Risk Score (MRS) was defined and used for prognostication and monitoring. RESULTS: The methylation assay demonstrated robust three-class (H3K27M-mut, IDH-mut and double-wildtype) classification with microAUC values of 1.00, 0.973, and 0.813 across public datasets, tissue cohorts, and CSF samples, respectively. BSGdiag achieved a sensitivity of 95.6%, specificity of 83.3%, and AUC of 0.949 for the H3K27M subtype, and a microAUC of 0.990 for the three-class classification in CSFs. MRS-stratified CSF methylation risk group was an independent prognostic factor (HR = 2.61, 95% CI: 1.09-6.25, P = 0.032). Methylation information in CSF remained even with clinical, radiological and CSF genetic indications of no disease, suggesting its utility in monitoring minimal residual disease. CONCLUSIONS: The study de novo developed the first methylation assay for robust BSG molecular subtyping and introduced a novel methodology for co-detecting CSF ctDNA mutations and methylation in BSGs. The BSGdiag enhances the utility of ctDNA by leveraging both genetic and epigenetic information. Its comprehensiveness, minimal invasiveness, robustness, and reliability make it highly promising for future clinical applications and trial designs.