Comprehensive Characterization of the Immune Microenvironment Based on Nested Resampling Machine Learning Framework Identifies TRAF3 Interacting Protein 3 as a Promising Regulator to Improve the Resistance to Immunotherapy in Glioma.

Abstract

Diffuse glioma, the most prevalent and malignant intracranial tumor, presents a formidable challenge due to its immunosuppressive microenvironment, which complicates conventional therapeutic approaches. This study conducted a comprehensive prognostic meta-analysis involving 2,968 patients with diffuse glioma and established a comprehensive machine learning framework with nested resampling of 18 machine learning algorithms, and developed the Immune Glioma Survival Signature (IGLoS). This signature, comprising CCL19, ICOSLG, IL11, PTGES, TNFAIP3, and TRAF3IP3, has been demonstrated to predict survival outcomes across a range of cancers and to correlate with tumor progression at the level of multi-omics. It is noteworthy that the IGLoS score enables precise patient stratification for personalized cancer treatments and elucidates pivotal resistance mechanisms to immunotherapy. Furthermore, siRNA screening has underscored the critical role of TRAF3IP3 in modulating PDL1 expression and immune pathways, with implications on the ERK pathway and NFATC2 involvement. Through single-cell analysis of published and in-house datasets, TRAF3IP3 exhibited selective enrichment in NPC-like and MES-like tumor cells, and showed a dual functionality in mediating T-Cell Exhaustion. Targeting TRAF3IP3 emerges as a promising avenue to combat immunotherapy resistance, particularly in glioma, thus paving the way for precision medicine.