A DIABLO–Elastic Net Cox Framework Reveals Prognostic Biomarkers and a Plasticity, Guidance, and Invasion Axis Driving Glioblastoma Progression
Abdullah Al-Sultan, Mai Al-RashidiBackground
Glioblastoma (GBM) is the most common and lethal primary malignant brain tumour in adults, characterised by marked molecular heterogeneity, therapeutic resistance and a median overall survival of approximately 15 months. Improved prognostic biomarkers and a deeper understanding of the molecular mechanisms driving disease progression remain critical unmet needs.
Methods
A novel framework combining Data Integration Analysis and Biomarker discovery using Latent cOmponents (DIABLO) with Elastic Net Cox regression was applied to transcriptomic, proteomic, phosphoproteomic and miRNA data from the CPTAC GBM cohort. Prognostic performance was evaluated using Kaplan–Meier analysis, Harrell’s concordance index (C-index) and time-dependent ROC analysis. External validation was performed in four independent cohorts (TCGA GBM, GDC-TCGA GBM, CGGA-693 and CGGA-325), followed by benchmarking against published prognostic signatures.
Results
Multi-omics integration revealed a coordinated plasticity, guidance and invasion axis underlying GBM progression. The model identified a six-gene prognostic signature comprising CHI3L1, DRAXIN, HOXC9, EPHB2, SPP1 and ADAMTS4. In the discovery cohort, the signature significantly stratified patients by survival (
Conclusions
The proposed framework integrates multi-omics biomarker discovery with mechanistic characterization of GBM progression and identifies a reproducible prognostic signature with potential utility for patient risk stratification and clinical management.