Construction and Multi-dimensional Validation of a Lactylation-Related Signature for Glioblastoma Multiforme Prognostic and Therapeutic Purposes.

Abstract

Glioblastoma multiforme, one of the most malignant types of brain tumor, heavily relies on glycolytic pathways and is significantly influenced by immune infiltration and its surrounding microenvironment. Growing evidence implies that increase in glycolysis can lead to lactate accumulation, which further contributed to histone lactylation, playing a crucial role in tumor development, maintenance, and therapeutic response. This study explores the prognostic and therapeutic potential of lactylation-related genes in glioblastoma multiforme. Using single-cell (GSE162631) and bulk transcriptome datasets (TCGA, CGGA, and GSE16011), we identified lactylation-related genes through ssGSEA and WGCNA. Moreover, a machine learning framework, incorporating 10 algorithms and 101 combinations, was used to establish an eight-gene lactylation-related signature (POLDIP3, MMP14, MDK, KDELR2, GSTK1, DEDD2, CD151, and BRI3) with robust predictive accuracy for patient survival. A nomogram with lactylation-related signature integration was developed as a quantitative prognostic instrument for clinical use. Moreover, patients classified by lactylation-related signature risk scores showed distinct immune status, tumor mutation burden, immunotherapy response, and drug sensitivity. The expression of those lactylation-related genes was further validated by quantitative PCR and functional experiment in normal and GBM cell lines. Overall, this study establishes a lactylation-related signature with significant potential for glioblastoma multiforme prognostic prediction, targeted prevention, and individualized therapy.