The current glioma classification could be optimized to cover such a separate and individualized prognosis ranging from a few months to over ten years. Considering its highly conserved role and potential in therapies, autophagy might be a promising element to be incorporated as a refinement for improved survival prognostication. The expression and RNA-seq data of 881 glioma patients from the Gene Expression Omnibus and The Cancer Genome Atlas were included, mapped with autophagy-related genes. Weighted gene coexpression network analysis and Cox regression analysis were used for the autophagy signature establishment, which composed of MUL1, NPC1, and TRIM13. Validations were represented by Kaplan-Meier plots and receiver operating curves (ROC). Cluster analysis suggested the IDH1 mutant involved in the favorable prognosis of the signature clusters. The signature was also immune-related shown by the Gene Ontology analysis and the Gene Set Enrichment Analysis. The high signature risk group held a higher ESTIMATE score (p = 2.6e - 11) and stromal score (p = 1.8e - 10). CD276 significantly correlated with the signature (r = 0.51, p < 0.05). The final nomogram integrated with the autophagy signature, IDH1 mutation, and pathological grade was built with accuracy and discrimination (1-year survival AUC = 0.812, 5-year survival AUC = 0.822, and 10-year survival AUC = 0.834). Its prognostic value and clinical utility were well-defined by the superiority in the comparisons with the current World Health Organization glioma classification in ROC (p < 0.05) and decision curve analysis. The autophagy signature-based IDH1 mutation and grade nomogram refined glioma classification for a more individualized and clinically applicable survival estimation and inspired potential autophagy-related therapies.
Copyright © 2020 Yuxiang Fan et al.