ABSTRACT
Accurate pathological diagnosis is crucial for optimal management of patients with cancer. For the approximately 100 known tumour types of the central nervous system, standardization of the diagnostic process has been shown to be particularly challenging-with substantial inter-observer variability in the histopathological diagnosis of many tumour types. Here we present a comprehensive approach for the DNA methylation-based classification of central nervous system tumours across all entities and age groups, and demonstrate its application in a routine diagnostic setting. We show that the availability of this method may have a substantial impact on diagnostic precision compared to standard methods, resulting in a change of diagnosis in up to 12% of prospective cases. For broader accessibility, we have designed a free online classifier tool, the use of which does not require any additional onsite data processing. Our results provide a blueprint for the generation of machine-learning-based tumour classifiers across other cancer entities, with the potential to fundamentally transform tumour pathology.
Subject(s)
Central Nervous System Neoplasms/diagnosis , Central Nervous System Neoplasms/genetics , DNA Methylation , Adolescent , Adult , Aged , Aged, 80 and over , Central Nervous System Neoplasms/classification , Central Nervous System Neoplasms/pathology , Child , Child, Preschool , Cohort Studies , Female , Humans , Infant , Male , Middle Aged , Reproducibility of Results , Unsupervised Machine Learning , Young AdultABSTRACT
CIC and FUBP1 mutations have recently been detected in oligodendrogliomas but not in oligoastrocytomas. However, allelic losses in the regions on chromosomal arms 19q and 1p harboring CIC and FUBP1 are a common feature of both, oligodendrogliomas and oligoastrocytomas. To resolve this discrepancy, we analyzed CIC and FUBP1 mutations in a set of primary brain tumors including 18 oligodendrogliomas and 42 oligoastrocytomas. In addition, we analyzed 10 astrocytomas and 16 glioblastomas with allelic losses on 19q as well as a set of 12 medulloblastomas for CIC mutations. CIC mutations were found in 15/18 oligodendrogliomas, 14/42 oligoastrocytomas and 3/10 preselected astrocytomas. With the exception of a single case, all CIC mutations occurred in tumors with combined 1p/19q losses. In contrast to oligodendrogliomas where CIC mutations were always detected along with 1p/19q co-deletion, CIC mutations were only found in 52 % of the 1p/19q co-deleted oligoastrocytomas. FUBP1 mutations were detected in 7/61 tumors, all presenting with CIC mutations. FUBP1 mutations appear to cluster in the DNA binding domain spanning exons 5-14. CIC and FUBP1 mutations exclusively occurred in presence of either IDH1 or IDH2 mutations. Our data confirm CIC and FUBP1 mutations in oligodendrogliomas and demonstrate the presence of these mutations in oligoastrocytomas.