Invasion/metastasis-related differentially methylated genes predict prognosis in diffuse gliomas

This study aimed to identify invasion/metastasis-related differentially methylated genes (DMGs) with prognostic relevance in diffuse gliomas and to develop a DNA methylation-based risk model for prognostic stratification.

Invasion/metastasis-related DMGs were identified by integrating gene expression and DNA methylation profiles from public glioma datasets. Prognostic DMGs were screened using univariate Cox regression and log-rank tests, and a risk model was constructed and evaluated by survival analysis and time-dependent receiver operating characteristic curves. Independent validation was performed in external cohorts. Functional enrichment, immune infiltration, mutation, and clinical correlation analyses were conducted. Key genes were functionally validated in T98G and U251 glioma cells.

Twelve prognostic DMGs were identified, and two marker genes, ERRFI1 and MYO1G, were selected to construct the risk model. The model showed prognostic value in the training cohort and retained predictive performance across independent validation cohorts. Enrichment analyses linked the risk signature to adaptive immune and inflammatory responses. High-risk patients had worse overall survival (OS), altered immune infiltration, and increased immune checkpoint expression. Knockdown of ERRFI1 or MYO1G inhibited proliferation, invasion, and migration in T98G and U251 cells. Preliminary Western blot analysis suggested that ERRFI1 knockdown altered EGFR/MAPK/ERK pathway phosphorylation.

This study identifies an invasion/metastasis-related DNA methylation signature based on ERRFI1 and MYO1G that may provide complementary prognostic information and biological insight for diffuse gliomas.