Molecular Subtypes in Glioblastoma Multiforme: Integrated Analysis Using Agilent GeneSpring and Mass Profiler Professional Multi-Omics Software

Significance of the Topic

Glioblastoma multiforme (GBM) is the most aggressive primary brain tumor in adults, with a median survival under 15 months despite standard treatment. Molecular subtyping of GBM has revealed distinct biological subgroups that differ in prognosis and therapeutic response. Integrating multi-omics data—including mRNA, miRNA, protein expression, and copy number aberrations—enables a more comprehensive view of tumor biology. This integrated approach can uncover key regulatory mechanisms, identify robust biomarkers, and guide personalized therapies in neuro-oncology.

Objectives and Study Overview

This study applied Agilent GeneSpring and Mass Profiler Professional (MPP) software to perform an integrated analysis of TCGA GBM datasets—including mRNA, miRNA, and copy number variations (CNVs)—together with an independent label-free proteomics dataset. The goals were to:

• Reproduce known GBM molecular subtypes (Proneural, Classical, Mesenchymal, Neural) via mRNA expression clustering and PCA.
• Correlate subtype classification with CNVs in key oncogenes and tumor suppressors (EGFR, PDGFRA, TP53, NF1, PTEN, CDKN2A, FGFR2).
• Identify miRNA regulators of subtype-specific gene signatures involved in neurogenesis and p53 signaling.
• Combine TCGA mRNA signatures with proteomics data to define a compact core signature differentiating tumor subtypes.

Methodology and Instrumentation

mRNA and miRNA data for 173 and 534 GBM samples, respectively, were downloaded from TCGA. GeneSpring 13.0 imported unified single-color expression values; CNVs were imported as metadata. Hierarchical clustering (Euclidean distance, Ward’s linkage), PCA, and Pearson correlation confirmed subtype structure and evaluated batch effects and sample quality parameters. miRNA–mRNA pairwise correlations were computed using Pearson similarity within GeneSpring, focusing on 25 genes involved in nervous system development and 38 differentially expressed miRNAs (p≤0.05).

For proteomics, ten GBM tumor and ten epilepsy control tissue specimens were processed by in-gel trypsin digestion with internal markers, separated by SDS-PAGE, and analyzed in triplicate using an Agilent HPLC-Chip/6550 iFunnel Q-TOF MS. Data were searched against UniProt at 1% FDR with Spectrum Mill, then exported to MPP for differential expression analysis (t-test p≤0.05, fold change ≥2).

Main Results and Discussion

• mRNA clustering and PCA recapitulated the four established GBM subtypes with minimal batch effects and no bias from sample quality metrics.
• CNV analysis revealed subtype-specific patterns: Classical tumors exhibited EGFR amplification and CDKN2A deletion; Proneural tumors were enriched for PDGFRA amplification and TP53 deletion; Mesenchymal tumors showed frequent NF1 hemizygous loss; FGFR2 and PTEN co-deletion suggested a single-event aberration on chromosome 10.
• MGMT promoter methylation did not align strictly with expression subtypes, indicating additional regulatory influences.
• GO enrichment identified a Proneural gene cluster involved in neurogenesis. Correlation analysis highlighted miR-21, miR-34a, miR-155, miR-221/222, and others as potential regulators of signature genes OLIG2, SOX11, NKX2-2, and ASCL1, suggesting miRNA-mediated repression of p53 signaling in Proneural tumors.
• Proteomics analysis yielded 14,187 protein groups, of which 587 proteins distinguished tumor from control samples and suggested tumor subgroups. Metadata did not explain subgrouping by age, sex, location, or MGMT status.
• Integration of the 587 proteins with 840 TCGA subtype genes identified a core signature of 54 overlapping gene–protein entities. These 54 molecules effectively classified both TCGA samples and the independent proteomics cohort into known subtypes.

Benefits and Practical Applications

• The multi-omics workflow in GeneSpring/MPP provides a unified platform for subtype validation, biomarker discovery, and pathway analysis in GBM.
• A concise 54-member signature may serve as a robust panel for diagnostic or prognostic assays, minimizing technical complexity while retaining classification power.
• miRNA–mRNA correlation insights offer potential targets for therapeutic modulation of signaling pathways specific to GBM subtypes.

Future Trends and Potential Applications

• Expansion to single-cell multi-omics could resolve intratumoral heterogeneity and refine subtype boundaries.
• Integration with epigenomic and spatial transcriptomic data may uncover microenvironment interactions and identify novel therapeutic vulnerabilities.
• Translation of the 54-gene/protein signature into multiplexed clinical assays (e.g., targeted proteomics or digital PCR) could enable real-time patient stratification.

Conclusion

This integrated multi-omics analysis confirms and extends GBM molecular subtypes, uncovers coordinated CNV events, and identifies miRNA regulators of subtype-specific networks. The distilled 54-marker core signature demonstrates powerful classification across transcriptomic and proteomic platforms. The GeneSpring/MPP framework facilitates comprehensive biomarker discovery and underscores the value of data integration for precision oncology in GBM.

References

1. Cancer Genome Atlas Research Network. Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature 2008.
2. Verhaak et al. Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma. Cancer Cell 2010.
3. Phillips et al. Molecular subclasses of high-grade glioma predict prognosis and resemble neurogenesis stages. Cancer Cell 2006.
4. Noushmehr et al. Identification of a CpG island methylator phenotype that defines a distinct subgroup of glioma. Cancer Cell 2010.
5. Silber et al. miR-34a repression in proneural malignant gliomas upregulates PDGFRA and promotes tumorigenesis. PLoS One 2012.
6. Ligon et al. Olig2-regulated lineage-restricted pathway controls replication competence in neural stem cells and malignant glioma. Neuron 2007.