Integrating Multi-Omics Analysis for Enhanced Diagnosis and Treatment of Glioblastoma: A Comprehensive Data-Driven Approach

Simple Summary The most prevalent and lethal primary brain tumor, glioblastoma multiforme (GBM), exhibits fast growth and widespread invasion and has a poor prognosis. The recurrence and mortality rates of GBM patients are still significant due to the intricacy of their molecular process. Therefore, screening GBM biomarkers is urgently required to demonstrate the therapy impact and enhance the prognosis. The findings of this study revealed 11 genes (UBC, HDAC1, CTNNB1, TRIM28, CSNK2A1, RBBP4, TP53, APP, DAB1, PINK1, and RELN), five miRNAs (has-mir-221-3p, hsa-mir-30a-5p, hsa-mir-15a-5p, has-mir-130a-3p, and hsa-let-7b-5p), six metabolites (HDL, N6-acetyl-L-lysine, cholesterol, formate, N, N-dimethylglycine/xylose, and X2. piperidinone), and 15 distinct signaling pathways that are essential for the development of GBM disease. The top genes, miRNAs, and metabolite signatures identified in this study may be used to develop early diagnosis procedures and construct individualized therapeutic approaches to GBM. Abstract The most aggressive primary malignant brain tumor in adults is glioblastoma (GBM), which has poor overall survival (OS). There is a high relapse rate among patients with GBM despite maximally safe surgery, radiation therapy, temozolomide (TMZ), and aggressive treatment. Hence, there is an urgent and unmet clinical need for new approaches to managing GBM. The current study identified modules (MYC, EGFR, PIK3CA, SUZ12, and SPRK2) involved in GBM disease through the NeDRex plugin. Furthermore, hub genes were identified in a comprehensive interaction network containing 7560 proteins related to GBM disease and 3860 proteins associated with signaling pathways involved in GBM. By integrating the results of the analyses mentioned above and again performing centrality analysis, eleven key genes involved in GBM disease were identified. ProteomicsDB and Gliovis databases were used for determining the gene expression in normal and tumor brain tissue. The NetworkAnalyst and the mGWAS-Explorer tools identified miRNAs, SNPs, and metabolites associated with these 11 genes. Moreover, a literature review of recent studies revealed other lists of metabolites related to GBM disease. The enrichment analysis of identified genes, miRNAs, and metabolites associated with GBM disease was performed using ExpressAnalyst, miEAA, and MetaboAnalyst tools. Further investigation of metabolite roles in GBM was performed using pathway, joint pathway, and network analyses. The results of this study allowed us to identify 11 genes (UBC, HDAC1, CTNNB1, TRIM28, CSNK2A1, RBBP4, TP53, APP, DAB1, PINK1, and RELN), five miRNAs (hsa-mir-221-3p, hsa-mir-30a-5p, hsa-mir-15a-5p, hsa-mir-130a-3p, and hsa-let-7b-5p), six metabolites (HDL, N6-acetyl-L-lysine, cholesterol, formate, N, N-dimethylglycine/xylose, and X2. piperidinone) and 15 distinct signaling pathways that play an indispensable role in GBM disease development. The identified top genes, miRNAs, and metabolite signatures can be targeted to establish early diagnostic methods and plan personalized GBM treatment strategies.


Introduction
Glioblastoma (GBM) is the most common high-grade primary malignant brain tumor with a poor prognosis [1,2]. It is urgently necessary to develop new therapeutic strategies for GBM with approved treatments due to its poor survival rates [3,4]. Following decades of basic science investment in GBM, innovative clinical trials currently utilize improved genetic and epigenetic profiling to treat the disease [5,6]. Proneural, neural, classical, and mesenchymal are the four molecular subtypes of GBM [7]. The complex genetic In our study, we did not use statistical methods for integrative analysis. This approach is appropriate when integrating different types of data together and then analyzing them. As part of the integration process, a sequential approach was first taken to analyze the protein network of different strategies. A regulatory network was constructed from the obtained proteins to identify important miRNAs. Following that, metabolic networks were analyzed separately and in conjunction with the important proteins identified.
Additionally, the type of enrichment method used is specified in the web server method. Since the approach used in this study was not one of machine learning, integrative statistical methods, variable selection methods (supervised, unsupervised, and semi-supervised), and Bayesian variable selection were not used. Moreover, stability analysis of the identified omics features in the current study was not included. Stability and validity are discussed to assess the reliability of the selected set of features and variables. One possible measure of stability which can be used in research is the measurement of the consistency of the obtained features across different data sets and platforms. To select features, one must consider whether the model is supervised, unsupervised, or semisupervised. Since this study did not use supervised, unsupervised, or semi-supervised methods, the selection of features did not occur, so the stability was checked [15].

Disease Module Identification
This study found GBM-related disease modules using the NeDRex plugin version 1.0.0 (https://nedrex.net/ (accessed on 13 January 2022)) implemented in the Cytoscape platform (version 3.7.2). This investigation used two different NeDRex algorithms: MuST (Multi-Steiner trees) and DIAMOnD (DIseAse MOdule Detection). To extract a connected subnetwork engaged in the disease pathways, MuST combines a variety of approximate Steiner tree calculations that are not all unique [16,17]. Based on the idea that the connectivity importance for known disease proteins is highly distinctive, the DIA-MOnD algorithm determines the disease module surrounding a set of known disease genes or proteins (seeds) [18]. The genes collected from various algorithms were each given special consideration. A summary of the data sources used in this study is presented in Table 1. Under the heading "Set of genes obtained from the two algorithms of MuST and DIAMOnD are listed and used in the study" (122 and 305 genes, respectively) (Supplementary Tables S1 and S2), the genes collected from various algorithms were each given special consideration. A schematic figure of the bioinformatics approaches designed in this study is given in Figure 1. Table 1. Data sources for GBM multi-omics analysis.

Database Name Type of Data Purpose
The Steiner tree calculations that are not all unique [16,17]. Based on the idea that the connectivity importance for known disease proteins is highly distinctive, the DIAMOnD algorithm determines the disease module surrounding a set of known disease genes or proteins (seeds) [18]. The genes collected from various algorithms were each given special consideration. A summary of the data sources used in this study is presented in Table 1. Under the heading "Set of genes obtained from the two algorithms of MuST and DIA-MOnD are listed and used in the study" (122 and 305 genes, respectively) (Supplementary  Tables S1 and S2), the genes collected from various algorithms were each given special consideration. A schematic figure of the bioinformatics approaches designed in this study is given in Figure 1. An overview of the bioinformatics approaches used in this study. We created four gene lists from several different study levels. By merging these gene lists and performing various analyses, 11 genes and five key miRNAs were identified.

Network Reconstruction of GBM-Related Signaling Pathways
A third analysis level was also established to learn about GBM disease and determine the essential proteins. Two techniques were established: (i) a literature review on genes associated with epithelial-mesenchymal transition (EMT), cytoskeleton remodeling, autophagy, secretory autophagy, and metabolism; and (ii) compiling the genes from 78 distinct signaling pathways. Table 2 lists the names of these pathways. The data from steps i and ii were combined in the following step to create a different list of 3860 genes (Supplementary Table S6). In the next step, centrality analysis was per- formed similarly to the previous step (Supplementary Table S7), and 20 proteins with the highest scores from each centrality were selected and combined. After processing 3860 initial entries, 35 essential proteins were discovered and selected for further analysis (Supplementary Table S8).  The final gene list and gene network for GBM disease were developed by combining four earlier study methodologies: MuST algorithm (122 genes), DIAMOnD algorithm (305 genes), glioblastoma-related protein-protein interaction network (48 genes), and the network analysis of signaling pathways associated with GBM (35 genes). In the end, 351 genes (Supplementary Table S9) were integrated, and a network was reconstructed using the GeneMANIA plugin and Cytoscape software (version 3.7.2). Similar to the previous steps, we performed a centrality analysis and selected five proteins with the highest scores from each centrality (Supplementary Table S10). Using the results of this stage, 11 essential genes (Table 3) related to GBM disease were identified, and a miRNAgene regulatory network was drawn via NetworkAnalyst (https://www.networkanalyst. ca/ (accessed on 10 August 2020)) [26,27] and miRTarBase v8.0 [28]. An analysis of miRNA centrality was conducted using NetworkAnalyst based on degree and betweenness. Our findings indicated that five miRNAs are essential based on the characteristics of the network based on integration results (Table 4).

Identification of Significant Metabolites and SNPs That Interact with Eleven Essential Genes
We identified significant SNPs and metabolites, interacting with eleven essential genes using the mGWAS-Explorer database (https://www.mgwas.ca/ (accessed on 15 July 2022)) [38], a user-friendly web-based tool that connects SNPs, metabolites, genes, and their known disease relationships using sophisticated network visual analytics. To identify the role of metabolites in glioma and GBM in greater depth, we extracted the list of essential metabolites (182 cases) (Supplementary Table S11) from recent studies [39,40]. In the next step, pathway, joint-pathway, and network analyses were performed using the MetaboAnalyst 5.0 (https://www.metaboanalyst.ca/ (accessed on 30 October 2019)) database [41,42].

The Network Obtained from the NeDRex Plugin to Identify Disease Modules
The proteins in GBM were identified by delineating the disease network using the NeDRex plugin ( Figure 2). In the following, step disease modules were determined by two distinct algorithms (MuST and DIAMOnD). The MuST algorithm identified five disease modules around the essential genes MYC, EGFR, PIK3CA, SUZ12, and SPRK2. IRAK1, PTK2, and BMI1 also represent bridging roles between MYC-EGFR, EGFR-PIK3R1, and SUZ12-SPRK2, respectively ( Figure 3). The DIAMOnD algorithm identified only one module with a high number of genes (305 genes) compared to the MuST algorithm (122 genes).

miRNA-Gene Regulatory Network Analysis
The interaction network of the eleven essential proteins identified with the corresponding miRNAs is shown in Figure 4. After the centrality analysis, it was inevitable to investigate the interaction of the five prominent miRNAs (hsa-mir-221-3p, hsa-mir-30a-5p, hsa-mir-15a-5p, hsa-mir-130a-3p, and hsa-let-7b-5p) with eleven identified genes (Figure 5). The two miRNAs (hsa-mir-221-3p and hsa-mir-30a-5p) showed a higher degree and betweenness of the centrality levels, and their targets (TP53, CTNNB1, UBC, TRIM28, and HDAC1) seem to play a more critical role in GBM than others.

miRNA-Gene Regulatory Network Analysis
The interaction network of the eleven essential proteins identified with the corresponding miRNAs is shown in Figure 4. After the centrality analysis, it was inevitable to investigate the interaction of the five prominent miRNAs (hsa-mir-221-3p, hsa-mir-30a-5p, hsa-mir-15a-5p, hsa-mir-130a-3p, and hsa-let-7b-5p) with eleven identified genes ( Figure 5). The two miRNAs (hsa-mir-221-3p and hsa-mir-30a-5p) showed a higher degree and betweenness of the centrality levels, and their targets (TP53, CTNNB1, UBC, TRIM28, and HDAC1) seem to play a more critical role in GBM than others. Figure 4. Gene regulatory network obtained from eleven identified proteins. A wide ran acting genes-miRNAs was determined.

Analyzing the Status of Identified Gene Expression in Healthy and Malignant Brain Tissue
Using microarray and RNA-Seq data, transcriptomics results from proteomicsDB revealed that all eleven identified genes were expressed in healthy brain tissue ( Figure 6). In addition, Gliovis' findings showed that, except for CSNK2A1, the remaining ten genes displayed a significant variation between the GBM and the non-tumor condition. Six genes represent an increased expression in the GBM state, whereas four genes were downregulated. Furthermore, all eleven specific genes were altered during the primary stage of the tumor ( Table 5). As a consequence of the correlation analysis, ten positive correlations were found between UBC (APP), HDAC1 (TP53, RBBP4, TRIM28, and CTNNB1), RBBP4 Figure 5. The scheme of the relationship between the five identified miRNAs and their targets. hsa-mir-221-3p and hsa-mir-30a-5p showed a higher degree and betweenness of the centrality levels.

Metabolic Pathway Analysis
The outcomes of the metabolic pathway enrichment analysis and pathway topology analysis were combined for this investigation. The results were divided into four categories according to the factors considered, as shown in Table 6.

Metabolic Pathway Analysis
The outcomes of the metabolic pathway enrichment analysis and pathway topology analysis were combined for this investigation. The results were divided into four categories according to the factors considered, as shown in Table 6. The results were analyzed and summarized based on three criteria: 1. The selection of the first five cases in each of the centralities-relative betweenness centrality (R-b C) and out-degree centrality (O-d C)-based on the highest score obtained in the impact parameter. 2. In the next step, the results were classified based on FDR, and five cases (phenylalanine, tyrosine, and tryptophan biosynthesis; synthesis and degradation of ketone bodies; one-carbon (1C) pool by folate; trehalose degradation; and glycerol phosphate shuttle) were excluded because they were not significant in the metabolic pathway analysis (red color). 3. The common items in both cases (R-b C and O-d C) and databases (KEGG and SMPDB) were considered vital and further discussed. Two items were obtained from the KEGG database (nitrogen metabolism, alanine, aspartate, and glutamate metabolism) and three items were obtained from the SMPDB database (alanine metabolism, aspartate metabolism, and malate-aspartate shuttle) (Tables 7 and S19-S22). Table 7. The final results from the metabolic pathway analysis. Further discussion of two items from the KEGG database (nitrogen metabolism, alanine, aspartate, and glutamate metabolism) and three from the SMPDB database (alanine metabolism, aspartate metabolism, and malate-aspartate shuttle) due to vitality. "+": accept; "--": reject.

Joint Pathway Analysis
We used joint pathway analysis to simultaneously analyze eleven identified essential genes and 182 distinctive metabolites within metabolic pathways. Three types of results were obtained based on the topology measure used ( Table 8). The results were analyzed and summarized based on four criteria: 1. The selection of the first 10 cases in each of the centralities based on the highest score obtained in the impact parameter. 2. In the next step, the results were classified based on FDR, and five cases (1C pool by folate, p53 signaling pathway, phosphatidylinositol signaling system, longevity regulating pathway, and mitophagy-animal) were excluded because they were not significant in the joint pathway analysis. 3. Considering the results obtained from all three of the centralities, the two items observed in all three (citrate cycle and arginine biosynthesis) were selected and further discussed (Tables 9 and S23-S26). The results were combined in one table to obtain a comprehensive overview of the pathway enrichment analysis at different levels (eleven essential proteins, five miRNAs, and 182 metabolites), pathways, and joint pathway analyses (Table 10). Closeness centrality Table 9. The final results from joint pathway analysis. The two items observed in all three centralities (citrate cycle and arginine biosynthesis) were selected and further discussed. "+": presence; "--": absence.

Gene-Metabolite Interaction Network
Drawing the interaction network between the genes and metabolites showed that the APP and TP53 genes are related to each other through five factors (adenosine triphosphate, ADP, glycerol, L-glutamic acid, and L-lysine). Moreover, the connection between RELN, CTNNB1, and CSNK2A1 and APP was shown through Gamma-aminobutyric acid (GABA), palmitic acid, and glycerol, respectively ( Figure 12).

Gene-Metabolite Interaction Network
Drawing the interaction network between the genes and metabolites showed that the APP and TP53 genes are related to each other through five factors (adenosine triphosphate, ADP, glycerol, L-glutamic acid, and L-lysine). Moreover, the connection between RELN, CTNNB1, and CSNK2A1 and APP was shown through Gamma-aminobutyric acid (GABA), palmitic acid, and glycerol, respectively ( Figure 12).

Identification of SNPs-Related Metabolites and Genes
Using eleven critical genes as input data in the mGWAS-Explorer database, six metabolites and 23 SNPs were identified (Table 11 and Supplementary Table S27). Additionally, we determined the top 25 SNPs via 182 metabolites as input data in the MetaboAnalyst 5.0 database ( Figure 13).

Identification of SNPs-Related Metabolites and Genes
Using eleven critical genes as input data in the mGWAS-Explorer database, six metabolites and 23 SNPs were identified (Tables 11 and S27). Additionally, we determined the top 25 SNPs via 182 metabolites as input data in the MetaboAnalyst 5.0 database ( Figure 13).

Discussion
Gene sequencing studies have made it possible to learn more about GBM genetics and epigenetics in recent years [46]. A practical approach for diagnosing and treating GBM is to use molecular biomarkers. The current study used genes, miRNAs, and metabolites to develop a panel of predictive biomarkers for GBM.
Based on the biological pathways, the identified genes in the GBM biomarker panel were classified into different signaling pathways such as the AKT (MYC, BMI1, EGFR, PIK3CA, PTK2, and UBC), the inflammation (IRAK1 and APP), the P53 (HDAC1, P53, and TRIM28), the WNT (CTNNB1), and the mitochondrial signaling pathways (DAB1, PINK1, and RELN). AKT regulates angiogenesis and metabolism [47]. Numerous human tumors have elevated levels of MYC, including GBM [48]. There is a correlation between Myc expression and glioma grade [49,50], and it was shown that inhibiting Myc in gliomas reduces proliferation and increases apoptosis [51]. The BMI-1 is another important gene related to the AKT signaling pathway. The BMI1 gene belongs to the polycomb group (PcG) gene family and is a transcriptional repressor of several genes that govern cell proliferation and differentiation throughout life [52][53][54][55][56]. In GBM, the simultaneous targeting of EZH2 and BMI1 was more effective than either agent alone due to the presence of both proneural and mesenchymal GSCs [57].
There is evidence that EGFR overexpression is associated with more aggressive GBM phenotypes in most primary GBMs and some secondary GBMs [58]. An analysis of the TCGA GBM database uncovered a subgroup with EGFR amplification and TP53 mutations. Both of these functions are almost mutually exclusive, suggesting EGFR regulates the function of wild-type p53 (wt-p53). EGFR signaling inhibits the function of wt-p53 in GBM by facilitating the interaction between p53 and the DNA-dependent protein kinase catalytic subunit (DNA-PKcs) [59].
It has been reported that 6-15% of glioblastomas contain activating mutations in the PIK3CA. There is evidence that PIK3CA-activating mutations are associated with an earlier recurrence of GBM in adults and a shorter survival time [60]. Additionally, immunohistochemical analysis of most anaplastic astrocytomas and glioblastomas demonstrated a strong expression of the PTK2 protein [61]. Elevated PTK2 (focal adhesion kinase 1 (Fak1)) protein levels were detected in astrocytic gliomas [62].

Discussion
Gene sequencing studies have made it possible to learn more about GBM genetics and epigenetics in recent years [46]. A practical approach for diagnosing and treating GBM is to use molecular biomarkers. The current study used genes, miRNAs, and metabolites to develop a panel of predictive biomarkers for GBM.
Based on the biological pathways, the identified genes in the GBM biomarker panel were classified into different signaling pathways such as the AKT (MYC, BMI1, EGFR, PIK3CA, PTK2, and UBC), the inflammation (IRAK1 and APP), the P53 (HDAC1, P53, and TRIM28), the WNT (CTNNB1), and the mitochondrial signaling pathways (DAB1, PINK1, and RELN). AKT regulates angiogenesis and metabolism [47]. Numerous human tumors have elevated levels of MYC, including GBM [48]. There is a correlation between Myc expression and glioma grade [49,50], and it was shown that inhibiting Myc in gliomas reduces proliferation and increases apoptosis [51]. The BMI-1 is another important gene related to the AKT signaling pathway. The BMI1 gene belongs to the polycomb group (PcG) gene family and is a transcriptional repressor of several genes that govern cell proliferation and differentiation throughout life [52][53][54][55][56]. In GBM, the simultaneous targeting of EZH2 and BMI1 was more effective than either agent alone due to the presence of both proneural and mesenchymal GSCs [57].
There is evidence that EGFR overexpression is associated with more aggressive GBM phenotypes in most primary GBMs and some secondary GBMs [58]. An analysis of the TCGA GBM database uncovered a subgroup with EGFR amplification and TP53 mutations. Both of these functions are almost mutually exclusive, suggesting EGFR regulates the function of wild-type p53 (wt-p53). EGFR signaling inhibits the function of wt-p53 in GBM by facilitating the interaction between p53 and the DNA-dependent protein kinase catalytic subunit (DNA-PKcs) [59].
It has been reported that 6-15% of glioblastomas contain activating mutations in the PIK3CA. There is evidence that PIK3CA-activating mutations are associated with an earlier recurrence of GBM in adults and a shorter survival time [60]. Additionally, immunohistochemical analysis of most anaplastic astrocytomas and glioblastomas demonstrated a strong expression of the PTK2 protein [61]. Elevated PTK2 (focal adhesion kinase 1 (Fak1)) protein levels were detected in astrocytic gliomas [62].
In the case of GBM, ubiquitin-dependent mechanisms may be exploited as a therapeutic strategy. It was indicated that the ubiquitin system is involved in core signaling pathways, including EGFR, TGF-β, p53, and stemness-related pathways in GBM [63]. In addition, it was shown that the inhibition of ubiquitin signaling could reverse metabolic reprogramming and suppresses GBM growth. The regulation of protein stability by the ubiquitin-proteasome system (UPS) represents an important control mechanism of cell growth in various human cancers, including GBM [64].
Evidence from both in vitro and in vivo studies shows that the highest expressed form of IRAK1 in low-grade gliomas (LGG) inhibits cell apoptosis and increases malignancy [65]. GBM is positively associated with mortality in Alzheimer's disease (AD) [66]. There is an up-regulation of the HDAC class I isoforms HDAC1 and HDAC2 in GBM cell lines compared with non-neoplastic brain tissues [67,68]. Proliferating, migrating, and invading are inhibited in GBM cells when HDAC1 and HDAC2 expressions are silenced. Similarly, HDAC3 is overexpressed in aggressive glioma cell lines and is associated with poor prognosis and OS of GBM patients [69]. A selective histone deacetylase inhibitor induces autophagy and cell death in GBM cells by downregulating SCNN1A [70]. A significant correlation was found between most members of the HDAC family and glioma grade, IDH1 mutation, and 1p/19q co-deletion. Among the HDAC1-related signatures for precise prognosis prediction in glioma, HDAC1 indicates prognosis and immune infiltration [71].
GBM is commonly associated with TP53 mutations. Approximately 84% of GBM patients exhibit dysregulation of the p53-ARF-MDM2 pathway, a finding that is confirmed in 94% of GBM cell lines adopted for in vitro assays [72,73].
Diverse cellular functions are mediated by PI3K/Akt-WNT signaling interactions in GBM, including cell proliferation, EMT, metabolism, and angiogenesis [47]. A study has shown that inhibiting WNT-CTNNB1 signaling enhances the SQSTM1 expression and sensitizes GBM cells to autophagy blockers [74]. This pathway also regulates autophagy and mitophagy [74][75][76]. Mitophagy, a selective autophagy of mitochondria, is crucial for quality control since it can efficiently degrade, remove and recycle malfunctioning or damaged mitochondria [77]. It has been demonstrated that platelet-derived growth factor (PDGF) signaling induces N6-methyladenosine (m 6 A) accumulation in GSCs to regulate mitophagy. A PDGF ligand stimulates the transcription of early growth response 1 (EGR1), which promotes the proliferation and self-renewal of GSCs by inducing methyltransferase-like 3 (METTL3). By regulating the m 6 A modification of optineurin (OPTN), the PDGF-METTL3 axis inhibits mitophagy. In GBM patients, the forced expression of OPTN mimics the inhibition of PDGF, and higher OPTN levels predict a longer survival time [78].
In GBM, miR-221/222, which targets the p53 upregulated modulator of apoptosis (PUMA), was reported to induce cell survival [79]. There is evidence that the chronic miR-221/222-mediated downregulation of MGMT may result in cells being unable to repair genetic damage. The presence of miR-221/222 oncogenic potential may improve the prognosis of GBM [80]. Furthermore, decreased EGFR and increased miR-221 were associated with increased resistance to temozolomide (TMZ) and radiotherapy in GBM [81], compared to normal brain tissues (NBTs). MiR-30a-5p is overexpressed in glioma cell lines and glioma samples, with its expression level positively correlated with tumor grade [82]. According to researchers, the Wnt/β-catenin-miR-30a-5p-NCAM regulatory axis is essential in controlling glioma cell invasion and tumorigenesis. It was shown that the Wnt/β-catenin pathway activates miR-30a-5p through the direct binding of β-catenin/TCF4 to two sites in the promoter region of miR-30a-5p. In addition, miR-30a-5p can inhibit the expression of neural cell adhesion molecule (NCAM) by directly targeting two sites in the 3 -untranslated regions (3 -UTR) of NCAM mRNA [83].
The proliferation and invasion of GBM cells are mediated by several critical molecules, such as cell adhesion molecule 1 (CADM1). CADM1 expression is decreased in GBM patients and GBM cell lines, and CADM1 overexpression inhibits the proliferation of GBM cells. According to these findings, CADM1 effectively suppresses the proliferation of GBM. MiR-15a-5p was shown to promote the proliferation and invasion of T98G GBM cells by targeting CADM1 [84,85].
In the GBM biomarker panel, five categories of metabolites were identified: lipid metabolism (cholesterol), glutamate metabolism (glutamate and GABA), tricarboxylic acid (TCA) cycle (alanine), urea cycle (arginine), and the Leloir cycle (galactose). There is a link between metabolic syndrome and several types of cancer, including GBM.
An analysis of a New Zealand cohort of GBM patients showed that metabolic syndrome is associated with reduced OS. In light of this finding, there is a greater likelihood that GBM results from metabolic pathogenesis [86]. According to studies, lipid metabolism plays a critical role in the pathogenesis of GBM. Among the members of the apolipoprotein family, apolipoprotein C1 (ApoC1) is crucial for the metabolism of both very-low-density lipoprotein (VLDL) and high-density lipoprotein (HDL) cholesterols. The pre-surgery level of serum LDL cholesterol was a prognostic factor for the outcome of patients with GBM [87].
Cholesterol is another important molecule with the potential role of repurposed drugs. Due to the discovery that many cancers, including GBM, reprogrammed the cholesterol metabolism, the cholesterol metabolism has become a promising potential target for therapy [88]. Different strategies for inhibiting cholesterol metabolism have been proposed since GBM cells require external cholesterol for survival and lipid droplets for rapid growth. The activation of liver X receptors (LXRs) inhibits cholesterol uptake, promotes cholesterol efflux, disrupts cholesterol trafficking within cells, interferes with SREBP signaling, and impedes cholesterol esterification. They may potentially counteract glial tumor growth [89,90]. It has also been demonstrated that lipid accumulation and oxidation play a role in GBM. Monounsaturated fatty acids have been found to promote GBM proliferation by modulating triglyceride metabolism [91]. A knockdown of carnitine palmitoyltransferase 1A (CPT1A), a critical enzyme in fatty acid oxidation (FAO), also reduced tumor growth and increased survival, according to in vivo studies [92].
The excitatory neurotransmitter glutamate plays a significant role in brain tumor cells' proliferation and growth. Glutaminase produces a large amount of glutamate in glioma cells, which converts glutamate from glutamine and increases intracellular Ca 2+ through P2 × 7Rs [93]. Moreover, high levels of glutamate have been found to cause brain edema and seizures in glioma patients. Glutamate and glutamine are linked to the proline pathway. L-proline is a multifunctional amino acid essential in the primary metabolism and physiological functions. Proline is oxidized to glutamate in the mitochondria, and the FAD-containing enzyme proline oxidase (PO) catalyzes the first step in the L-proline degradation pathway. It was shown that PO might play a regulatory role in glutamatergic neurotransmission by affecting the cellular concentration of glutamate [94]. Study results indicate that serine and glycine levels are higher in the low-nutrient regions of GBM tumors than in the other areas. A study of the metabolic and functional properties of GBM cells revealed that serine availability and 1C metabolism support glioma cells' survival following glutamine deprivation. Serine synthesis was mediated by autophagy rather than glycolysis [95].
ATP by glycolysis and the TCA cycle are associated with oxidative phosphorylation (OXPHOS) through the breakdown of pyruvate or fatty acids to meet the growing energy demand of cancer cells. Recent studies demonstrate that SMI EPIC-0412 can effectively perturb the TCA cycle. It was shown that in combination with the cytosolic phospholipase A2 (cPLA2) inhibitor AACOCF3 and the hexokinase II (HK2) inhibitor 2-DG, SMI EPIC-0412 disrupted GBM energy metabolism for targeted metabolic therapy. ATP production significantly declined in glioma cells when treated with a monotherapy (EPIC-0412 or AACOCF3), dual therapy (EPIC-0412 + AACOCF3), or triple therapy (EPIC-0412 + AACOCF3 + 2-DG) regimen [96]. Furthermore, it has been shown that patients with GBM with high levels of glycolysis-related genes, including HK2 and PKM2, and low levels of mitochondrial metabolism-related genes, such as SDHB and COX5A, related to TCA cycle and oxidative phosphorylation (OXPHOS), have poor survival rates. In contrast to LGG, the expression levels of genes involved in the mitochondrial oxidative metabolism in GBM are markedly increased; however, they are lower than those in normal brains [97].
It has been shown that dysregulated alanine could serve as a potential predictive marker for glioma [98]. Alanine, a glucogenic amino acid, enters the metabolic stream through enzymatic conversion to pyruvate to provide energy and replenish the nutrient reservoir for rapidly proliferating tumor cells [99]. Arginine is another amino acid substrate actively metabolized by tumor cells to promote tumor growth and immunosuppression [100,101]. Arginine transporters appear abundant in GBM, as evidenced by the accumulation of byproducts of arginine metabolism [102,103]. The results indicate that the arginine metabolism is functional and may be sensitive to targeted depletion. Recent research demonstrated that pegylated human recombinant ARG1 depleted arginine in glioma cells and induced cytotoxicity [104].
GBM tumor cell proliferation depends on the availability of extracellular nutrients. As a result of inadequate tumor perfusion, glucose and glutamine are in short supply. Due to this metabolic remodeling, GBMs scavenge alternative nutrients from the tumor microenvironment to sustain their growth and proliferation. Glut3 and Glut14 are sugar transporters expressed in GBM. GBM cells can scavenge galactose (Gal) from the circulation and extracellular space as a suitable substrate for Glut3/Glut14. The Leloir pathway provides GBM cells with an alternative energy source by transporting and metabolizing Gal at physiological Glc concentrations [105].

Conclusions and Future Direction
Research on GBM has made significant progress in recent years, partly due to the availability of large-scale multi-omics databases and other data sources that allow researchers to better understand the disease's genetic and molecular characteristics. It is important to identify genes, miRNAs, and metabolites that contribute to the progression of GBM. In our current investigations, we identified eleven important genes and five miRNAs that have a critical role in the pathogenesis of GBM. In addition, we found essential metabolites that drive GBM development. These findings highlight the importance of genes, miRNAs, and metabolites in GBM progression.
It is critical for both basic and clinical scientists to find out which genes or proteins lead to or are associated with metabolic changes. In future, transcriptome and proteomic analyses should identify key genes and proteins contributing to inflammation in GBM progression. Understanding how inflammation influences GBM progression is critical for developing new therapeutic strategies. Our team will focus on tailoring formulations and repurposing drugs to develop diagnostic and therapeutic biomarkers in the near future and open up new avenues for the personalized treatment of patients with GBM.