Abstract
BACKGROUND: The intrinsic heterogeneity and invasiveness of diffuse gliomas complicate accurate prognosis. Existing approaches are largely constrained by subtype specificity or limited analytical dimensions. To address this gap, a multi- dimension-based prognostic framework encompassing the full glioma spectrum was developed, accompanied by an analysis of the associated immune microenvironment. METHODS: A total of 3,323 glioma samples from the SEER (n = 2181), CGGA (n = 807), and TCGA (n = 335) datasets were integrated. Differentially expressed genes were screened using the limma package, and a Lasso-Cox-based prognostic signature (Glioma-GDPM) was established. Clinical variables such as age, grade, and IDH mutation status were harmonized through propensity score matching to construct a multi-omics prognostic model (Glioma-GCDPM). GSEA, CIBERSORT-based immune infiltration analysis, and TIDE scoring were used to investigate the biological characteristics of different risk subgroups. RESULTS: Eleven key prognostic genes (such as PRAMEF2 and FADS1) and four clinical factors (age, tumor grade, IDH mutation, and 1p/19q codeletion) were identified. Glioma-GCDPM demonstrated favorable predictive ability in both the internal test cohort (AUC 0.81-0.86) and external validation sets (AUC 0.59-0.83). High-risk tumors exhibited greater invasiveness, with significant enrichment in cell cycle and proliferation-associated pathways. Additionally, a suppressed immune microenvironment was observed, reflected by elevated M2 macrophage infiltration and increased T cell dysfunction scores. CONCLUSION: The multi-omics model established in this study enables precise stratification of prognostic risk in diffuse glioma patients and reveals immunosuppressive features in high-risk individuals, providing a new basis for personalized treatment strategies.