Identification of metabolism-associated molecular classification and prognostic genes for medulloblastoma based on bioinformatics analysis.

objectiveMedulloblastoma (MB) is one of the most prevalent solid brain tumors in the pediatric population. Although hundreds of Differentially Expressed Genes (DEGs) have been identified as diagnostic molecules and therapeutic targets for Medulloblastoma (MB), the function of metabolism-associated genes in the pathophysiological mechanisms of MB remains unclear.

methodsThe Gene Expression Omnibus (GEO) provided five datasets that contain mRNA expression profiles and clinical information. In detail, 599 MB samples extracted from GSE85217 were set as the training set (n = 400) and test set (n = 199). For validation, 74 samples from GSE37418, 8 scRNA-seq samples, and 37 RNA-seq samples were used. Nonnegative matrix factorization clustering was conducted, leading to the identification of four MB subclasses (C1, C2, C3, and C4) in both the training and test sets.

resultsC1 and C3 are metabolically active, and C2 involves metabolic hypoactivity; C4 did not exhibit obvious metabolic characteristics. Next, GSE50161, GSE74195, and GSE86574 were used to identify DEGs, and a 17-gene metabolism-associated signature for prognosis prediction was established. The authors performed transcriptome and single-cell sequencing on medulloblastoma bulk specimens to further validate metabolic subclasses and the prognosis prediction model at the transcriptome and cellular level.

conclusionThe present study classifies Medulloblastoma (MB) based on metabolic signatures, supplementing existing subtype characterization from a metabolic perspective. The authors provide preliminary insights into MB's metabolic hallmarks and a potential reference for developing multimolecule-based personalized therapies and prognostic tools ‒ with the caveat that the 17-gene signature requires additional multi-center validation to confirm clinical utility.