P.099 Integrating DNA methylation profiling in brain tumour diagnosis directly changes patient oncological care

Abstract

Background: Molecular signatures are being increasing used to classify central nervous system (CNS) tumors with incorporation into World Health Organization (WHO) classifications. A recently published genome-wide DNA methylation-based CNS tumor classifier assisted in diagnostically challenging cases. However its impact on patient care has not been reported, limiting translation to other centres. Methods: All 55 challenging CNS tumour diagnoses over three years underwent DNA methylation profiling. Tumor classification along with copy number variant (CNV) plot results were integrated with histopathological findings to determine final diagnoses and corresponding clinical impact was assessed. Results: After methylation profiling 46/55 (84%) received clinically relevant diagnostic changes, 30 (55%) with a new diagnosis or resolved differential diagnosis and 16 (29%) with clinically important molecular diagnostic or subtyping changes. WHO grade changed in 15 (27%), with two-thirds upgraded. Nine new IDH mutations in gliomas, four new molecular subtypes in medulloblastomas/ependymomas, and three false positive 1p/19q codeletions were identified. Patient care was directly changed by methylation profiling in 7/47 (15%) followed-up cases to avoid unnecessary treatment in three, insufficient treatment in three, and medically assisted death in one. Conclusions: This real-world use of methylation-based CNS tumor classification substantially impacts patient care for diagnostically challenging tumors and also avoids misdiagnosis-related uncessary resource use.