Abstract
BACKGROUND: Gliomas are common primary malignant brain tumors, with glioblastoma (GBM) being the most aggressive subtype. GBM is characterized by high recurrence rates and treatment resistance, leading to poor patient outcomes. Current prognostic models have limited predictive power, underscoring the need to elucidate underlying mechanisms and identify novel biomarkers to improve therapeutic strategies and prognostic models. METHODS: Gene expression and clinical data for GBM and LGG were obtained from the TCGA and CGGA database, while single-cell sequencing data from GSE167960 were selected from the GEO database. Molecular characteristics of gliomas were revealed through normalization, consensus clustering analysis, immune scoring, cell infiltration analysis, and pathway analysis. TUBA1B, identified as a key gene through machine learning, was incorporated into a nomogram model using multivariate Cox regression. Its functions were validated through qRT-PCR, in vitro functional assays, and mouse xenograft models. All data analyses and statistics were performed using R software. RESULTS: Consensus clustering of the TCGA glioma dataset identified two aggrephagy subtypes (C1 and C2), with C2 showing worse survival outcomes and higher immune infiltration. TUBA1B was identified as an independent prognostic marker, with high expression associated with upregulated cell cycle pathways and alterations in the immune microenvironment. TUBA1B was shown to influence glioma cell proliferation, migration, invasion, and autophagy, impacting tumor progression and treatment response through intercellular communication and metabolic pathways. CONCLUSION: The study demonstrates that high TUBA1B expression is closely associated with glioma malignancy and poor prognosis, making it a potential therapeutic target.