Abstract The WHO 2021 classification of oligodendroglioma grade 2 (G2) and grade 3 (G3), is separated by histopathology alone and does not universally reflect survival outcome. DNA methylation-based profiling also fails to separate oligodendroglioma grades. Our aim was to define transcriptome-based molecular subgroups that predict survival of oligodendroglioma (IDH-mutant and 1p/19q-codeleted) patients. Clinical and whole-transcriptome data for oligodendroglioma were obtained from TCGA (n=132), CGGA (n=53), and the Exsegen Genomics Research Biobank (n=16) (CTRI/2021/09/036861); while that of control brain samples (n=9) were obtained from CPTAC. Overall survival (OS) was significantly different in TCGA samples between WHO CNS G2 and G3 oligodendrogliomas (log-rank test, p=0.0001); but, not in CGGA (p=0.1833) or Exsegen (p>0.06) samples. Genes with highly variable expression (Median Absolute Deviation, i.e. MAD>1; 1645 genes) across TCGA oligodendroglioma and CPTAC control brain samples were used to exclude 9 oligodendroglioma samples with Pearson’s correlation>0.75 with control brain samples. Consensus clustering (K-means and spectral) of remaining 124 TCGA oligodendroglioma samples identified three significantly different (SigClust, p<0.05) transcriptomic classes using 1347 highly variable genes (MAD>1) in these samples. Principal component analysis (PCA) on expression of 1347 genes in TCGA samples revealed partial overlap of G2 and G3 tumours; but transcriptomic classes with distinct OS could be identified at both extremes of survival. Gene expression signatures for each transcriptomic class were developed using core TCGA samples from each class. For further validation, gene signatures were used to predict the transcriptomic classes of CGGA and Exsegen samples based on highest cosine similarity between sample gene expression and gene signature of each transcriptomic class. OS was significantly different between the transcriptomic classes in TCGA samples and predicted ones in CGGA samples. Transcriptome profiling separated oligodendrogliomas into distinct classes which can predict survival and potentially augment grading and refine prognostication and treatment options.
Read full abstract