Molecular dissection of glioblastoma: Identification of prognostic hub genes through integrated bioinformatics analysis

INTRODUCTION

Glioblastoma multiforme (GBM) is among the deadliest primary brain tumor in adults, characterized by extensive cellular heterogeneity, diffuse infiltration, and high resistance to conventional therapies.^[1,2] Despite progress in surgical removal, radiation therapy, and chemotherapy, the median survival for patients with GBM remains limited to around 12–15 months.^[3] The poor prognosis is primarily attributed to the highly invasive nature of GBM cells, their ability to evade immune surveillance, as well as the presence of complex molecular alterations that drive tumor progression.^[4]

Over the past decade, large-scale initiatives such as the cancer genome atlas (TCGA) have provided deep insights into GBM biology through integrative transcriptomic and proteogenomic profiling.^[5,6] These efforts have led to the identification of numerous oncogenes, tumor suppressors, and signaling pathways implicated in GBM pathogenesis, including EGFR amplification, PTEN loss, and dysregulation of the PI3K/AKT/mTOR pathway.^[7] However, the molecular complexity and intratumoral heterogeneity of GBM require more integrative, multi-dimensional approaches to identify robust regulatory “hub genes” that not only orchestrate tumorigenesis but also modulate the immune microenvironment and influence clinical outcomes.

Previous studies have typically focused on single dimensions such as gene expression profiles or mutational landscape.^[8] A multi-dimensional integrative analysis combining transcriptomic, epigenetic, immunological, and genomic data could offer a more holistic view of GBM biology, leading to the discovery of more robust biomarkers and therapeutic targets.

In present study, we hypothesized that core hub genes regulating GBM pathogenesis could be identified through a comprehensive bioinformatic approach, integrating multiple layers of molecular information. We systematically screened GBM-associated genes from GeneCards and DisGeNet databases, followed by protein-protein interaction (PPI) network construction, functional enrichment analysis, and multi-platform validation including gene expression, promoter methylation, survival analysis, immune infiltration correlations, and genetic alteration profiling using publicly available resources such as STRING, UALCAN, GEPIA2, TISIDB, and cBioPortal.

Here, we report the identification of four core hub genes – ACTB, CASP3, ERBB2, and GAPDH – significantly associated with GBM progression and prognosis. These findings offer novel insights into the molecular mechanisms underlying GBM and highlight potential targets for diagnostic, prognostic, and therapeutic development.

MATERIAL AND METHODS

RESULTS

Identification of target gene

To identify potential key genes involved in GBM pathogenesis, we systematically queried two comprehensive databases, that is, GeneCards and DisGeNet.^[14,15] GeneCards predicted possible association of 1237 genes while DisGeNet predicted possible association with 662 genes. Genes obtained from both of these sets were subjected to Venn diagram analysis. It revealed 464 genes common to both resources [Figure 1], which were selected for subsequent network analysis and functional characterization.

Close

Figure 1:

Identification of common glioblastoma multiforme (GBM)-associated genes from multiple databases. Venn diagram depicting the overlap of GBM-related genes retrieved from GeneCards and DisGeNET databases, highlighting the set of genes commonly identified by both sources.

Export to PPT

Construction of PPI network and identification of hub genes

Using the STRING database, a PPI network was constructed for the shortlisted 464 common genes [Figure 2a].^[16] This results in the formation of a complex network comprising 464 nodes interconnected by 19,011 edges [Figure 2a]. The network demonstrated a high clustering coefficient of 0.616 with an enrichment P < 1.0E-16, indicating significant functional connectivity among the selected genes. To identify the most influential nodes within this network, we employed three established centrality measures using CytoNCA: Betweenness centrality [Figure 2b], closeness centrality [Figure 2c], and degree centrality [Figure 2d].^[17] Top 25 genes each were selected from each centrality. These 25 genes were further subject to Venn diagram analysis. Twenty-two common genes were selected and named hub genes [Figure 2e]. These 22 hub genes were subjected to further PPI network analysis [Figure 2f], revealing a tightly interconnected sub-network with 231 edges and an average node degree of 21 (PPI enrichment P = 0.0158).

Close

Figure 2:

Protein-protein interaction (PPI) network construction and hub gene identification in glioblastoma multiforme (GBM). (a) PPI network generated using the STRING database for 464 common GBM-associated genes, consisting of 464 nodes and 19,011 edges (clustering coefficient =0.616; PPI enrichment P < 1.0E-16). (b-d) Ranking of the top 25 genes based on (b) betweenness centrality, (c) closeness centrality, and (d) degree centrality using CytoNCA in Cytoscape. (e) Venn diagram illustrating the intersection of top-ranked genes across all three centrality measures, identifying 22 putative hub genes. (f) Refined PPI network of the 22 hub genes, comprising 231 edges and an average node degree of 21 (PPI enrichment P = 0.0158).

Export to PPT

Functional enrichment analysis of hub genes

To elucidate the biological significance of the hub genes, a comprehensive functional enrichment analysis was performed utilizing GO and KEGG databases.^[18,19] The analysis revealed that these genes are involved in regulating diverse biological activities encompassing MF, BP, and CC.

KEGG pathway analysis further mapped these genes to established cancer-related pathways [Figure 3a], underscoring their involvement in key oncogenic processes. The KEGG enrichment highlighted the participation of these genes in critical pathways associated with cancer regulation. In addition, biological process enrichment analysis using GO terms [Figure 3b] demonstrated significant overrepresentation in processes essential to the pathophysiology of GBM, including cellular responses to stimuli, signal transduction, and cell communication networks.

Close

Figure 3:

Functional enrichment analysis of prospective hub genes in glioblastoma multiforme (GBM) (a) Mapping of hub genes to canonical cancer-related signaling pathways, highlighting their involvement in key oncogenic mechanisms. (b) Gene ontology (GO) biological process enrichment analysis, showing significant enrichment in pathways related to signal transduction, immune modulation, and apoptosis. (c) Circos plot illustrating the integrative functional annotation of hub genes across multiple databases, including GO terms, kyoto encyclopedia of genes and genomes pathways, and WikiPathways, emphasizing their multifunctional roles in GBM pathogenesis.

Export to PPT

To visualize the complex interrelationships between hub genes and their associated functional pathways, a Circos diagram was constructed [Figure 3c], integrating enrichment data from WikiPathways, KEGG, and GO analyses. This visualization revealed that genes such as TP53, STAT3, CD44, CTNNB1, and AKT1 exhibited extensive interactions with a wide array of pathways, suggesting their pivotal roles as central regulators within the network. Furthermore, the chord visualization emphasized that multiple genes participate concurrently in shared pathways, reflecting overlapping biological functions. For instance, TP53, BCL2, and CASP3 demonstrated strong associations with apoptosis and cell survival pathways, whereas interleukin (IL)-6, tumor necrosis factor (TNF), and STAT3 were predominantly linked to immune signaling and inflammatory response processes. The color-coded mapping of pathways and GO terms [Figure 3c] facilitated the identification of functional clusters, with cancer-related pathways (e.g. hsa05205) and immune-related processes (e.g. GO:0001775) being the most prominently represented among the selected genes. Collectively, these findings underscore the multifunctional nature of key oncogenes and tumor suppressor genes, highlighting their potential as critical therapeutic targets and biomarkers in cancer and immunological research.

Expression analysis and promoter methylation profiling of hub genes

Using UALCAN to analyze the TCGA data, we examined the mRNA expression patterns of 22 hub genes in GBM compared with normal brain tissue [Figure 4a].^[20] Fourteen genes showed statistically significant differences in expression (P < 0.05), 13 of which (ACTB, AKT1, CASP3, CD44, CTNNB1, EGFR, ERBB2, GAPDH, HIF1A, KRAS, MYC, STAT3, TGFB1, and TP53) showed up-regulation, and only PTEN showed significant down-regulation in the GBM samples [Figure 4b].

Close

Figure 4:

Differential expression analysis of hub genes in glioblastoma multiforme (GBM) using the cancer genome atlas data. (a) Box plot illustrating the expression profiles of the 22 prospective hub genes in normal brain tissues versus GBM samples, highlighting overall expression trends. (b) Bar graph displaying transcript per million values for the 14 hub genes that were significantly differentially expressed between normal brain (blue) and GBM (red) tissues, underscoring their potential diagnostic and prognostic relevance.

Export to PPT

In particular, EGFR showed the highest up-regulation with a 13.5-fold increase in expression compared to normal brain tissue, which is consistent with its well-known role in the pathogenesis of GBM. These genes were further looked for promoter methylation on UALCAN. The β value reflects the extent of promoter methylation, ranging from 0 (completely unmethylated) to 1 (fully methylated) [Figure 5]. Thresholds for interpretation vary, with values between 0.7 and 0.5 typically indicating hypermethylation, while values from 0.3 to 0.25 suggest hypomethylation.^[9,10] Among the 22 hubs, IL1B and TNF showed significant change in their promoter methylation. In normal brain tissues, IL1B exhibited high promoter methylation (β-value ≈0.87), which was reduced to 0.60 in GBM samples (P < 0.001), indicating hypomethylation. In contrast, TNF displayed the opposite pattern, with promoter β-values increasing from 0.55 in normal tissues to 0.87 in GBM (P < 0.001), reflecting hypermethylation and likely transcriptional silencing. The divergent methylation profiles of these cytokines highlight a dual regulatory program in GBM: IL1B hypomethylation may facilitate pro-tumorigenic inflammation, whereas TNF hypermethylation may attenuate anti-tumor immunity.

Close

Figure 5:

Deoxyribonucleic acid methylation analysis of hub genes in glioblastoma multiforme (GBM). Beta value distribution plots illustrating promoter methylation levels of the 22 prospective hub genes in normal brain tissues versus GBM samples, revealing distinct epigenetic alterations potentially contributing to gene dysregulation in glioblastoma.

Export to PPT

Hub genes and GBM survival

To assess the prognostic significance of the 22 hub genes, we performed Kaplan–Meier survival analysis using GEPIA2 [Figure 6].^[21] This prognostic significance was used to shortlist the core-hub genes for the further study. Only genes which have HR 1.5 or above with significant P value for HR and significant P value log ranks were selected. Four genes – ACTB, CASP3, ERBB2, and GAPDH – fulfilled these criteria, demonstrating strong associations between their elevated expression and poor overall survival in GBM patients. Based on their prognostic significance, these four genes were designated as core hub genes and selected for further validation and mechanistic exploration.

Close

Figure 6:

Survival analysis of hub genes in glioblastoma multiforme (GBM) patients. Kaplan–Meier survival curves generated using GEPIA2, illustrating the association between high expression (red) versus low expression (blue) of hub genes and overall survival in GBM patients. Elevated expression of specific hub genes correlates with significantly reduced survival, indicating their potential prognostic value.

Export to PPT

Protein expression validation of core hub genes

To validate our findings at the protein level, we analyzed the immunohistochemical staining patterns of the four core hub genes using the Human Protein Atlas database [Figure 7].^[22] Interestingly, we observed discordance between mRNA and protein expression for ACTB and GAPDH, which showed decreased protein levels in GBM tissues despite elevated transcript levels. This discrepancy suggests potential post-transcriptional regulatory mechanisms affecting these genes in GBM. In contrast, CASP3 and ERBB2 exhibited concordant upregulation at both mRNA and protein levels, with notably increased immunostaining intensity in GBM samples compared to normal brain tissue. These data were further confirmed with IHC images.

Close

Figure 7:

Protein expression validation of core hub genes in glioblastoma multiforme (GBM). Representative immunohistochemical staining images from the Human Protein Atlas depicting the differential protein expression of ACTB, CASP3, ERBB2, and GAPDH in normal brain tissue (n) compared to GBM tumor samples (t). The staining patterns support transcriptomic findings and highlight protein-level dysregulation in glioblastoma.

Export to PPT

Correlation between core hub gene expression and immune cell infiltration

Given the growing importance of tumor immune microenvironment in GBM progression and treatment response, we investigated potential relationships between the expression of core hub genes and the infiltration of key immune cell populations using the TISIDB database [Figure 8].^[23] Our analysis focused on three immune cell types with established roles in GBM biology: CD8⁺ cytotoxic T cells, tumor-associated macrophages (TAMs), and Tregs. Among the four core hub genes, only ACTB expression showed significant positive correlation with both macrophage (r = 0.412, P < 0.001) and Treg (r = 0.378, P < 0.001) infiltration. This finding suggests that ACTB may contribute to an immunosuppressive microenvironment in GBM, potentially limiting the effectiveness of anti-tumor immune responses and contributing to its association with poor prognosis.

Close

Figure 8:

Correlation between core hub gene expression and immune cell infiltration in glioblastoma multiforme (GBM). Scatter plots generated using the TISIDB database illustrating the association between expression levels of ACTB, CASP3, ERBB2, and GAPDH and the infiltration of key immune cell populations – including CD8⁺ cytotoxic T-cells, tumor-associated macrophages, and regulatory T-cells – in GBM samples. These correlations suggest potential roles of the core hub genes in modulating the tumor immune microenvironment.

Export to PPT

Genetic alteration analysis of core hub genes

Apart from expression, there are multiple factors which could affect the gene function and could contribute to their effect on disease prognosis. These factors may be, but not limited to mutations, deletions, and amplification in the sequences. These alterations could result in the protein folding, activity, and its function. To characterize the spectrum of genetic alterations affecting the core hub genes, we performed comprehensive genomic analysis using cBioPortal [Figure 9].^[24] Each gene exhibited a distinct pattern of alterations in GBM samples. ACTB showed both mutations and amplifications, with amplification frequencies ranging from 1.5% to 4% across different GBM cohorts. CASP3 demonstrated a broader spectrum of alterations, including amplifications, mutations, and deep deletions (up to 2%). ERBB2 was predominantly affected by mutations, with frequencies reaching up to 4%, while GAPDH exhibited both mutations and amplifications at elevated frequencies. These diverse alteration patterns suggest different mechanisms of dysregulation for each core hub gene in GBM pathogenesis.

Close

Figure 9:

Genetic alteration landscape of core hub genes in glioblastoma multiforme (GBM). Oncoprint visualization generated using cBioPortal illustrating the spectrum and frequency of genetic alterations – including mutations, amplifications, and deep deletions – affecting ACTB, CASP3, ERBB2, and GAPDH across GBM patient cohorts. These alterations highlight the genomic instability and oncogenic potential of the core hub genes in glioblastoma.

Export to PPT

Multi-dimensional association analysis of core hub genes with GBM

To comprehensively evaluate the relationship between the core hub genes and GBM, we conducted multi-dimensional association analysis using cBioPortal [Figure 10]. All four genes demonstrated significant text mining associations with GBM, indicating substantial literature evidence supporting their relevance in GBM biology. In addition, CASP3 showed significant expression atlas association, providing further experimental validation of its altered expression in GBM datasets. ERBB2 exhibited both ChEMBL and pathway associations, highlighting its potential as a therapeutic target and its involvement in established GBM-related signaling pathways. Collectively, these multi-faceted associations provide robust evidence supporting the biological and clinical significance of our identified core hub genes in GBM.

Close

Figure 10:

Multi-dimensional association analysis of core hub genes with glioblastoma multiforme (GBM). Integrated analysis illustrating the associations of ACTB, CASP3, ERBB2, and GAPDH with GBM using multiple data sources, including text mining, gene expression profiles, ChEMBL drug interactions, and pathway enrichment. This comprehensive evaluation underscores the multifaceted roles of these genes in glioblastoma pathobiology and therapeutic relevance.

Export to PPT

In the present integrated multi-omics analysis, we identified ACTB, CASP3, ERBB2, and GAPDH as core regulatory genes in GBM. These genes are linked to poor prognosis, immune modulation, and distinct genomic alterations. Their consistent significance across transcriptomic, epigenetic, and proteomic layers underscores their potential as prognostic biomarkers and therapeutic targets. These findings establish a foundation for future functional validation and translational applications in GBM.

DISCUSSION

GBM continues to pose one of the greatest challenges in the field of neuro-oncology due to its aggressive progression, extensive heterogeneity, and resistance to standard therapies, resulting in consistently poor prognosis.^[3,25] In this study, we employed a multi-dimensional integrative bioinformatics approach to identify and characterize core regulatory genes that may serve as prognostic biomarkers or therapeutic targets in GBM.

Starting from a gene pool derived from GeneCards and DisGeNET, we identified 464 common GBM-associated genes. Network-based centrality measures narrowed these down to 22 hub genes, which were subjected to functional and enrichment analyses. The use of degree, betweenness, and closeness centrality provided complementary perspectives on network influence, ensuring robust hub gene identification rather than reliance on a single topological parameter. These genes were significantly enriched in cancer-associated BP and pathways, including signal transduction, immune modulation, and apoptosis. Circos plots further revealed the multifunctional nature of these genes such as TP53, STAT3, and AKT1, underscoring their role in GBM pathophysiology.^[5,6]

Expression analysis through UALCAN demonstrated that 14 out of 22 hub genes were differentially expressed between GBM and normal brain tissues. Out of these 14, 13 genes ACTB, AKT1, CASP3, CD44, CTNNB1, EGFR, ERBB2, GAPDH, HIF1A, KRAS, MYC, STAT3, TGFB1, and TP53, genes were over expressed in GBM compared to the normal tissue. These expression patterns align with prior research demonstrating the critical roles of these genes in GBM pathogenesis. For instance, EGFR and ERBB2 are frequently amplified in GBM, STAT3, HIF1A, and TGFB1 that are key mediators of immunosuppression and hypoxia-induced adaptation in the tumor microenvironment.^[26-29] Oncogenes KRAS and MYC drive proliferation and metabolic reprogramming, while CD44 and CTNNB1 are involved in cell adhesion and invasion.^[30-32] GAPDH participates in GBM progression through non-glycolytic roles, including transcription regulation and redox signaling.^[33,34] PTEN being the only significantly downregulated gene which is consistent with its – well-established tumor suppressor role in gliomagenesis.^[7]

In promoter methylation study, only two genes – IL1B and TNF – showed significant changes in promoter methylation. In GBM, IL1B undergoes promoter hypomethylation, with β values decreasing from ~0.87 in normal tissues to ~0.60 in tumors, a change that likely de-represses its transcription and may potentiate pro-inflammatory signaling within the tumor microenvironment. In contrast, TNF exhibits significant hyper-methylation, with β values rising from ~0.55 to ~0.87 in GBM, suggesting transcriptional silencing of this key cytokine. These findings are consistent with genome-wide methylation analyses in GBM, where TNF hypermethylation is reported among genes exhibiting CpG island methylation shifts in tumor samples including GBM.^[35-37]

This coordinated yet opposing promoter methylation pattern underscores a complex regulatory program: IL1B de-repression may support tumor-associated inflammation and infiltrative behaviors, while TNF silencing could facilitate immune evasion by dampening anti-tumor cytokine signaling.^[38,39] These changes align with broader epigenetic reprogramming observed in GBM, where promoter methylation alterations are selectively targeted to genes controlling immunity and survival.^[36] The β-value cutoffs adopted in this study (0.7–0.5 for hypermethylation and 0.3–0.25 for hypomethylation) are supported by prior large-scale analyses in endometrial and glioblastoma cohorts, where similar thresholds reliably distinguished tumor-specific epigenetic shifts.^[9,10] Using these ranges ensures consistency with published standards, facilitates cross-study comparison, and strengthens the biological relevance of the identified methylation alterations in GBM. Network-based methylation studies in other diseases, such as gestational diabetes, highlight the robustness of integrative approaches in uncovering disease-specific epigenetic shifts.^[40]

Survival analysis identified ACTB, CASP3, ERBB2, and GAPDH as core hub genes associated with poor prognosis. Prior studies have shown that these four core hub genes are involved in critical processes in GBM prognosis. ACTB is involved in cytoskeletal remodeling, invadopodia formation, and GBM cell invasion.^[41,42] Moreover, ACTB has been implicated in mechanotransduction pathways, suggesting it plays a critical role in how GBM cells sense and respond to extracellular matrix stiffness – a defining feature of tumor progression.^[43,44] CASP3 paradoxically exhibits non-apoptotic, pro-tumorigenic functions in GBM. Previous studies have reported that CASP3 can promote tumor repopulation following therapy-induced apoptosis and activate survival pathways such as AKT.^[45] ERBB2 (human epidermal growth factor receptor 2) has reported amplification or overexpression in ~15–20% of GBM cases.^[46]

Earlier report also found significantly elevated ERBB2 mRNA in GBM compared to lower-grade gliomas and normal brain.^[47] ERBB2 activates PI3K/AKT and MAPK pathways – promoting GBM proliferation, invasion, and therapeutic resistance – especially in PTEN-deficient tumors.^[26,48] ERBB2 also forms heterodimers with EGFR, enhancing malignancy and limiting the effectiveness of RTK-targeted therapies.^[26] GAPDH which is often dismissed as a mere housekeeping gene showed robust overexpression in GBM and was linked to poor prognosis. Recent studies reveal that GAPDH participates in the Warburg effect, DNA repair, and hypoxia response.^[33,49] GAPDH expression is known to increase under hypoxic conditions, due to its glycolytic role and potential regulation by HIF-1α, which may support tumor metabolic adaptation and therapy resistance.^[50] A novel contribution of this study is the integration of transcriptomic, epigenetic, immune infiltration, and genomic alteration analyses into a single framework. While earlier studies have implicated individual genes such as EGFR, TP53, or STAT3, our work uniquely identifies a composite core hub gene signature (ACTB, CASP3, ERBB2, and GAPDH) with multidimensional validation, providing a broader systems-level insight into GBM biology.

Protein expression and immune correlation analyses revealed unique dysregulation patterns for each. ACTB correlated with macrophage and Treg infiltration, implicating it in immune suppression and cytoskeletal remodeling.^[43,51,52] CASP3, though pro-apoptotic, exhibited pro-tumorigenic functions, consistent with reports of therapy resistance and AKT activation.^[45] ERBB2 was overexpressed in ~15–20% of GBM cases and associated with poor outcomes, acting through PI3K/AKT and MAPK pathways, particularly in PTEN-deficient tumors.^[26,46-48] GAPDH was robustly overexpressed and linked to hypoxia response and metabolic adaptation through HIF-1α signaling.^[49,50]

Interestingly, our protein validation revealed interesting discrepancies. CASP3 and ERBB2 showed consistent mRNA and protein expression patterns, while ACTB and GAPDH exhibited lower protein levels despite elevated transcript abundance. We assume that this could be due to post-transcriptional regulation, such as miRNA targeting or protein degradation. However, genetic alteration analysis highlighted that ACTB and GAPDH exhibited frequent amplifications, whereas CASP3 showed mutations and deep deletions. These differences imply varied oncogenic pressures and selection mechanisms acting on each gene in GBM evolution.^[5]

The immune microenvironment of GBM is characterized by a complex and often paradoxical interplay between immune effector and suppressor cells. Among the multitude of immune populations present in the tumor milieu, CD8⁺ cytotoxic T lymphocytes, TAMs, and Tregs emerge as the most biologically relevant and consistently observed infiltrates across multiple studies.^[11-13] We focused on CD8⁺ T cells, TAMs, and Tregs as these represent the most biologically relevant immune subsets in GBM. CD8⁺ T cells, though central to anti-tumor immunity, often become functionally exhausted under programmed death (PD-1)/PD-ligand 1 (PD-L1) signaling and TAM-derived cytokines.^[11] TAMs, which comprise up to half of the tumor mass, adopt an M2 phenotype that promotes angiogenesis, recruits Tregs, and suppresses effector T-cells, thereby fostering an immunosuppressive niche.^[8-10] Tregs further reinforce this cycle by dampening CD8⁺ T-cell responses through IL-10 and TGF-β, and their enrichment is strongly associated with poor prognosis. Together, these three cell types form a critical immunological triad that shapes GBM progression and therapeutic resistance. This triad represents both the immune attack and immune evasion arms of the GBM immune landscape. CD8⁺ T-cells are the primary antitumor effector cells responsible for recognizing and killing malignant cells. Their presence within GBM tissue has been correlated with improved prognosis and longer survival.^[53]

TAMs, derived from both infiltrating monocytes and resident microglia, comprise up to 30–50% of the GBM tumor mass.^[13] While macrophages can exert tumoricidal effects in other cancers; in GBM, they are frequently polarized toward an M2-like, pro-tumorigenic phenotype, secreting immunosuppressive cytokines such as IL-10 and TGF-β, promoting angiogenesis, and supporting tumor invasion and recurrence.^[11,54]

Tregs represent a specialized subset of CD4⁺ T-cells that in GBM, often enriched in the tumor microenvironment and are directly associated with immune escape and poor patient outcomes.^[12] Their expansion is frequently driven by tumor-secreted factors and by the very TAMs that they interact with, creating a vicious cycle of immunosuppression.^[13]

The functional crosstalk between these three immune populations further underscores their collective importance in GBM pathophysiology. TAMs can promote Treg recruitment and differentiation, while simultaneously inhibiting CD8⁺ T-cell activation through PD-L1 expression and arginase activity.^[13] This dynamic interplay not only facilitates tumor immune evasion but also renders GBM largely refractory to conventional immunotherapies. Given their central roles, CD8⁺ T-cells, TAMs, and Tregs serve as robust immunological indicators of GBM progression and therapy resistance. ACTB’s association with TAMs and Tregs in the present study reinforce its role as core hub gene and immune suppression.^[55,56]

Despite the insights gained from public transcriptomic databases, this study has some important limitations. Bulk RNA-seq datasets, such as TCGA, often suffer from technical variability (e.g. differences in sequencing platforms and normalization pipelines) and intrinsic biological biases, including sample heterogeneity and tumor purity, which can confound results.^[57] Furthermore, commonly used housekeeping genes such as GAPDH and ACTB, though traditionally viewed as stable controls, exhibit variable expression across different cancers and experimental conditions, making their alteration in GBM potentially misleading if used without validation.^[58] This limitation is particularly critical, as the reliance on housekeeping genes such as ACTB and GAPDH as reference markers in cancer studies may obscure their context-dependent roles in tumor progression. Their dysregulation in GBM underscores the need for careful interpretation and suggests that future studies should avoid assuming these genes as neutral internal controls. These genes undergo context-dependent regulation in cancer, including GBM. Further, while comprehensive in silico analyses can identify candidate biomarkers and pathways, these findings remain preliminary. Functional validation through wet-lab experiments – such as quantitative reverse transcription-polymerase chain reaction, Western blotting, coimmunoprecipitation, and cell-based assays – and ultimately confirmation in clinical cohorts are essential next steps to establish biological relevance and translational potential.^[59] In addition, the relatively small number of GBM cases and its rapid clinical progression pose limitations to large-scale omics research. Invasive biopsy access is limited, and tumor heterogeneity adds another layer of complexity to single-sample analyses.^[60] Nonetheless, converging lines of evidence across multiple datasets reinforce the relevance of our identified core hub genes.

CONCLUSION

In summary, this study identified ACTB, CASP3, ERBB2, and GAPDH as four core hub genes and highlights their multifaceted roles of in GBM biology. These genes are strongly implicated in GBM ranging from disease progression, tumor invasion, immune modulation, and metabolic adaptation to apoptosis resistance. These genes do not merely function in isolation but form part of an interconnected molecular network, as evidenced by prior studies. Their consistent significance across transcriptomic, epigenetic, proteomic, and immunological dimensions underscores their potential utility as prognostic biomarkers and therapeutic targets. It is important to note that these findings are preliminary and derived from in silico analyses. Rigorous experimental validation in cell-based and clinical settings will be essential before translating these observations into diagnostic or therapeutic applications. Future in vitro and in vivo validation, along with mechanistic studies, will be critical to translating these findings into clinically actionable strategies for GBM patients.