Abstract
Background
Tumor hypoxic environment can regulate pyroptosis and immune escape by affecting multiple signaling pathways, and synergistic targeting this interaction may be a promising strategy for tumor treatment. However, their connection in glioblastoma (GBM) remains unclear. This study aimed to construct hypoxia- and pyroptosis-related signatures for exploring their values in predicting prognosis of GBM.
Methods
GBM expression profiles, hypoxia genes, and pyroptosis genes were collected from the following public databases: The Cancer Genome Atlas (TCGA), Gene Expression Omnibus (GEO), Genecards, and gene set enrichment analysis (GSEA). After obtaining the hypoxia- and pyroptosis-related differentially expressed genes (DEGs) within GBM vs. normal, least absolute shrinkage and selection operator (LASSO) regression analysis was performed to establish prognostic models. The models were utilized to differentiate patients into low-risk and high-risk groups, and then difference in survival time between the two groups was assessed. Biological significance of these genes was explored by immune infiltration and drug response analyses. Finally, the correlation between hypoxia- and pyroptosis-related biomarkers was examined.
Results
A total of 13 hypoxia- and 13 pyroptosis-associated genes were included in the risk models, respectively. Two models had high accuracy and stability in predicting GBM prognosis. Low-risk and high-risk groups identified by the model had different immune cell infiltration patterns. In the high-risk group, the abundance level of naïve B cells, plasma cells, and native CD4 T cells was inhibited, while levels of neutrophils were increased. In addition, the risk models could predict chemotherapy sensitivity. Of interest, pyroptosis-related genes showed a significant negative correlation with most hypoxia-related genes.
Conclusions
These constructed prognostic models can effectively identify high-risk patients, and drugs targeting the hypoxia-pyroptosis axis have a broad application prospect in GBM.
Keywords: Glioblastoma (GBM), hypoxia, pyroptosis, prognostic model
Highlight box.
Key findings
• This study constructed two prognostic models (hypoxia- and pyroptosis-related) for glioblastoma (GBM) using 13 genes each, demonstrating high accuracy in predicting patient survival.
• We identified distinct immune infiltration patterns, with high-risk patients showing suppressed naïve B cells, plasma cells, and native CD4 T cells, whereas elevated neutrophils.
• This study revealed a significant negative correlation between pyroptosis-related genes (e.g., BRD4, PINK1) and most hypoxia-related genes, suggesting a potential crosstalk mechanism.
What is known and what is new?
• Hypoxia and pyroptosis independently contribute to GBM progression and therapy resistance. Hypoxia regulates immune escape, while pyroptosis influences inflammatory responses.
• This study establishes the first combined hypoxia-pyroptosis gene signatures for GBM prognosis, links their interaction to immune modulation, and identifies chemotherapy-sensitive subgroups.
What is the implication, and what should change now?
• The models enable risk stratification, guiding personalized therapy by targeting hypoxia-pyroptosis genes or immune cells (e.g., neutrophils). Drugs like BI.2536 and BIBR.1532 may benefit specific subgroups.
• Clinical validation of these models and exploration of hypoxia-pyroptosis axis inhibitors (e.g., BRD4/MDM2 blockers) are warranted to improve GBM outcomes.
Introduction
Glioblastoma (GBM) is the most common type of malignant primary brain tumor in adults (1). This refractory and highly aggressive cancer has a poor prognosis, with a survival rate of only 15 months after diagnosis (2). Despite different therapeutic approaches, the 5-year relative survival rate for GBM remains one of the lowest of all cancer types (less than 5%), and almost all GBM recur at or around the site of disease (3,4). Besides, the progressive understanding of the GBM’s biology over the past decades has not led to significant improvements in patient treatment and prognosis. The main reason for GBM treatment failure comes from its unique molecular characteristics, especially the ability of glioma stem cells to self-renew, which leads to tumor cells easily developing resistance to radiotherapy or chemotherapy (5). Therefore, the in-depth exploration of the molecular pathogenesis of GBM has important clinical significance.
Hypoxia refers to the reduced availability of oxygen, which plays a crucial role in neuronal development (6,7). It is a determinant that drives the aggressive phenotype of tumors and is also a major feature that causes rapid expansion and proliferation of GBM (8). In GBM, hypoxia conditions promote cell proliferation, stimulate angiogenesis, and confer therapeutic resistance by inducing HIF-1 stabilization and pro-angiogenic gene activation (9,10). Pyroptosis is a type of programmed cell death characterized by the release of cytokines and activation of inflammatory-immune response cascades by cell swelling (11). Pyroptosis participates in a variety of inflammatory responses in GBM, and drug-induced pyroptosis has great potential in tumor therapy, indicating that the research of pyroptosis is valuable for the prognostic assessment and therapeutic plan formulation of GBM (12,13). Notably, recent evidence has highlighted that hypoxia can trigger pyroptosis of cells in the tumor microenvironment by modulating multiple signaling pathways (14). For example, hypoxia condition in cancer cells causes programmed death-ligand 1 (PD-L1) to transport into the nucleus, inducing an increase in GSDMC expression (a pyroptosis marker), thereby converting caspase-8-induced apoptosis into pyroptosis (15). The association between hypoxia and pyroptosis has attracted considerable attention due to its value in tumor therapy, but the mechanisms of this association are complex and many unanswered questions remain. Meanwhile, the role of their link in GBM is poorly studied.
Currently, there is growing evidence that establishment of prognostic model based on molecular mechanisms can improve the clinical prognosis of GBM by recognizing high-risk patients and providing them with precise therapy (16,17). In this work, we established prognostic models related to hypoxia and pyroptosis through bioinformatics analysis, and evaluated the value of these models in predicting the GBM prognosis, understanding the immune microenvironment, and assessing therapeutic response. In addition, we analyzed the associations between the characterized genes in the two models. These studies provide a scientific basis for revealing the relationship among hypoxia, pyroptosis, and GBM prognosis, as well as helping clinicians tailor treatment strategies for GBM patients. We present this article in accordance with the TRIPOD reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1367/rc).
Methods
Public data collection
GBM-related datasets, including five normal and 167 disease samples, were acquired from The Cancer Genome Atlas (TCGA, https://www.cancer.gov/) database and served as the training set. The survival information of GBM patients was also collected. In addition, GSE83300 (50 GBM cases) and GSE74187 (60 GBM cases) were downloaded from the Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/geo) database as test sets for hypoxia- and pyroptosis-related models, respectively. Hypoxia-related genes were extracted from two databases (Genecards, https://www.genecards.org, gene set enrichment analysis (GSEA, https://www.gsea-msigdb.org/gsea/index.jsp), and integration resulted in 6,167 genes for subsequent analysis. The genes associated with pyroptosis were collected from Genecards, and a total of 534 genes were obtained in this study. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.
Identification of differentially expressed hypoxia- or pyroptosis-related genes in GBM
Based on the training set, the limma method in R software (version 4.3.2) was utilized to perform the differential expression of GBM and normal samples, with differentially expressed genes (DEGs) identified as P<0.05 and |log2fold change| >1. Venn analysis was used to integrate these DEGs with hypoxia genes, and the shared genes were used as hypoxia-related DEGs. For pyroptosis-related DEGs, the differences in expression levels of pyroptosis-related genes in GBM and normal cases were evaluated by t-test, with P<0.05 as the threshold.
Protein-protein interaction (PPI) analysis
Hypoxia- or pyroptosis-related DEGs were imported into the Search Tool for the Retrieval of Interacting Genes (STRING) database, and PPI pairs were predicted based on combined_score >0.4. These pairs were mapped using Cytoscape software to create a PPI network. Further, hub genes in the PPI network were screened by the molecular complex detection (MCODE) plug-in.
Establishment of prognostic model through the least absolute shrinkage and selection operator (LASSO) analysis
To screen gene sets with core prognostic value from hypoxia- or pyroptosis-related DEGs, LASSO regression analysis was conducted using glmnet to determine highly relevant genes for model construction. The formula for risk score was ∑(Xi*Coefi), where Coefi is the coefficient of LASSO analysis, and Xi is the normalized expression level of the gene. After calculating the risk score for each sample, the population was divided into low-risk and high-risk groups based on its median. Difference in survival time between the two groups was assessed by combining the Kaplan-Meier and log-rank test. In addition, the receiver operating characteristic (ROC) curves were plotted to reflect the ability of the prognostic model in predicting overall survival (OS) for 1-, 2-, and 3-year. We further confirmed the predictive capability of the model in the test set.
Immunoinfiltration analysis
CIBERSORT is a method for characterizing complex cell composition in tissues based on gene expression patterns (18). In this study, to infer the distribution of immune cells in each sample, we employed the CIBERSORT algorithm to estimate the infiltration levels of 22 types of immune cells. Further, t-test was applied to evaluate the differences in immune cell infiltration between the low-risk and high-risk groups. P<0.05 as the screening criterion for differential immune cells. Results were also verified in the test set.
Sensitivity analysis of chemotherapeutic agents
To apply the results to guide clinical treatment, we compared the sensitivity of low-risk and high-risk patients to chemotherapy drugs. First, oncoPredict was employed to predict patients’ response to drugs [half-maximal inhibitory concentration (IC50) value], with t-test to evaluate the difference between the two groups. Second, the correlation between the drugs and model genes was calculated by Pearson analysis, and P<0.05 was regarded to be significant.
Correlation analysis of hypoxia- and pyroptosis-related model genes
To reveal the potential link between hypoxia and pyroptosis, Pearson correlation analysis was conducted on genes in the two constructed models. P<0.05 was used as a correlation screening threshold.
Statistical analysis
All statistical analyses were conducted using R software (version 4.3.2). For exploration of DEGs and immune cells, comparisons between the two groups were performed using t-test. The OS rates of the high- and low-risk groups were compared through Kaplan-Meier survival analysis, with the log-rank test employed to assess differences between the two groups. In addition, the ROC curve and area under the ROC curve (AUC) value were applied to evaluate the predictive performance of the risk score. The correlation analysis between variables was conducted using Pearson’s method. Statistical significance was set at P<0.05.
Results
Screening of hypoxia-related DEGs and development of PPI network
A total of 927 DEGs, containing 289 down-regulated and 638 up-regulated genes, were identified within GBM vs. normal (Figure 1A). The top 30 DEGs (ranked by |log2fold change|) were displayed in a heatmap, indicating that these genes can clearly separate disease from control samples (Figure 1B). After integration with hypoxia-related genes, 266 hypoxia-related DEGs were obtained (Figure 1C). Further, PPI analysis was utilized to reveal the connections between the proteins encoded by the 266 genes. Figure 1D illustrates the core network selected by MCODE, including 26 genes, of which RPS2, RPL5, and RPL4 had high degree values.
Figure 1.
Screening for core hypoxia-related genes in the GBM and normal samples. (A) Volcano plot displaying DEGs between GBM and normal samples. Red and blue nodes represent up- and down-regulated genes, respectively. (B) Heatmap showing top 30 DEGs ranked by |log2fold change| value. (C) Venn diagram integrating DEGs and hypoxia-related genes. (D) Core PPI network of hypoxia-related DEGs screened by the MCODE. DEGs, differentially expressed genes; GBM, glioblastoma; MCODE, Molecular Complex Detection; PPI, protein-protein interaction.
Development of a 13-hypoxia signature for predicting GBM prognosis
LASSO regression analysis indicated that 13 hypoxia-related DEGs were identified to be strongly associated with the prognosis of GBM, and could be used as characteristic genes to construct a risk model (Figure 2A). Risk score =0.257 × LYAR − 0.120 × RPL15 − 0.188 × EIF3E − 0.067 × RPS11 + 0.102 × RPL27A + 0.107 × RPL23A + 0.004 × RPS2 − 0.04 × RPL29 − 0.099 × EEF1G − 0.115 × RPL4 − 0.241 × RPL5 + 0.206 × RPL34 + 0.118 × TPT1. Notably, LYAR, RPL27A, RPL23A, RPS2, RPL34, TPT1 could serve as prognostic risk factors, whereas RPL15, EIF3E, RPS11, RPL29, EEF1G, RPL4, and RPL5 were considered protective factors. The samples were separated into low-risk and high-risk groups based on median risk score, and Kaplan-Meier analysis revealed markedly poor survival times in patients from the high-risk group (Figure 2B). Moreover, the ROC curve showed that this model could effectively evaluate the prognosis of GBM patients, demonstrating an area under the ROC curve (AUC) of 0.628, 0.734, and 0.691 for 1-, 2-, and 3-year OS (Figure 2B). Further, the same results were observed in the test dataset (GSE83300), that is, patients in the high-risk group had poor outcomes, and the AUC values for the model predicting OS at 1, 2, and 3 years were 0.675, 0.721, and 0.941, respectively (Figure 2C). These results suggest that the model can predict GBM prognosis stably and accurately.
Figure 2.
Establishment and validation of hypoxia-associated prognostic model. (A) Lambda and cross-validation plots from the LASSO regression analysis. (B,C) Kaplan-Meier curve of different risk groups and ROC curve based on the train (TCGA) and test (GSE83300) datasets. AUC, area under the ROC curve; LASSO, least absolute shrinkage and selection operator; ROC, receiver operating characteristic; TCGA, The Cancer Genome Atlas.
Immune cell differences between two risk groups in a hypoxia-related model
We demonstrated the infiltration levels of 22 immune cells in high-risk and low-risk samples in both training and test datasets, and found a relatively high proportion of activated mast cells and M2 macrophages in samples (Figure 3A). Then, differences in cellular infiltration between two risk groups were compared and revealed that only the naïve B cells showed differences in both datasets, which were significantly suppressed in the high-risk group (Figure 3B).
Figure 3.
Differences in immune cell between the two risk groups from the hypoxia-related model. (A) Landscape of immune infiltration in the high- and low-risk samples from the train (TCGA) and test (GSE83300) datasets. (B) Differences in immune cell abundance between the high- and low-risk groups. *, P<0.05; **, P<0.01. NK, natural killer; ns, no significance; TCGA, The Cancer Genome Atlas.
Screening of pyroptosis-related DEGs and development of PPI network
Of the 534 pyroptosis-related genes, 302 exhibited significant differences between GBM and normal samples, including 73 down-regulated and 229 up-regulated genes (Figure 4A). In addition, the PPI network screened by MCODE analysis contained 39 core genes, especially NFKB1, STAT3, and TLR2 with high connectivity (Figure 4B).
Figure 4.
Screening of key pyroptosis-related DEGs between GBM and normal samples. (A) Heatmap revealing pyroptosis-related DEGs within GBM vs. normal. (B) Core PPI network of pyroptosis-related DEGs screened by MCODE. DEGs, differentially expressed genes; GBM, glioblastoma; MCODE, Molecular Complex Detection; PPI, protein-protein interaction.
Establishment of a 13-pyroptosis signature for predicting GBM prognosis
In the LASSO regression analysis, after controlling the risk of overfitting and conducting cross-validation, 13 key factors optimally predicting the prognosis of GBM were selected (Figure 5A). Among them, HSP90B1, PINK1, BRCA1, BCL3, CALM3, AGER, SPP1, IFI16, EZH2, MDM2, and BRD4 as potential risk factors, while IRF9 and BCL6 as protective factors. The risk score for each sample was calculated by combining gene expression and LASSO coefficients, with its median value subsequently used to categorize patients into low-risk and high-risk groups. The specific formulate was: risk score = 0.049 × IFI16 + 0.127 × CALM3 + 0.015 × BRD4 + 0.435 × HSP90B1 + 0.196 × BRCA1 − 0.141 × BCL6 + 0.148 × BCL3 − 0.046 × IRF9 + 0.018 × MDM2 + 0.119 × AGER + 0.097 × SPP1 + 0.335 × PINK1 + 0.0325 × EZH2. As expected, patients in the high-risk group had significantly lower survival rates than those in the low-risk group. The AUC values for predicting GBM survival at 1, 2, and 3 years were 0.757, 0.765, and 0.828, respectively (Figure 5B). The predictive performance of the model was also validated in the test dataset (GSE74187). Patients in the high-risk group demonstrated remarkably shorter survival. Besides, the corresponding AUC values for survival time at 1, 2, and 3 years were 0.617, 0.715, and 0.86, respectively (Figure 5C).
Figure 5.
Development and validation of a pyroptosis-associated prognostic model. (A) Lambda and cross-validation plots from the LASSO regression analysis. (B,C) Kaplan-Meier curve of different risk groups and ROC curve for predicting 1, 2, and 3-year survival based on the train (TCGA) and test (GSE74187) datasets. AUC, area under the ROC curve; LASSO, least absolute shrinkage and selection operator; ROC, receiver operating characteristic; TCGA, The Cancer Genome Atlas.
Immune cell differences between two risk groups in a pyroptosis-related model
Immune landscape analysis showed that the infiltration degree of neutrophils and M2 macrophages was abundant in the GBM samples (Figure 6A). In the training and test sets, three types of cells, including plasma cells, native CD4 T cells, and neutrophils, displayed significant differences in the high-risk and low-risk groups (Figure 6B). Specifically, in the high-risk group, infiltration levels of plasma cells and native CD4 T cells were dramatically inhibited, while the proportion of neutrophils was elevated.
Figure 6.
Differences in immune cell between the two risk groups from the pyroptosis-related model. (A) Landscape of immune infiltration in the two risk groups from the train (TCGA) and test (GSE74187) datasets. (B) Differences in immune cell abundance between the high and low risk groups. *, P<0.05; **, P<0.01; ***, P<0.001. NK, natural killer; ns, no significance; TCGA, The Cancer Genome Atlas.
Differences in drug sensitivity between the two risk groups
Differences in drug response between the two risk groups were further explored. IC50 values were estimated for 138 chemotherapeutic agents in the low-risk and high-risk groups. A total of 86 and 52 differential drugs were found in the two risk groups from hypoxic- and pyroptosis-related models, respectively. The top four representative drugs with the most significant differences are shown in Figure 7. For the hypoxic model, BIBR.1532, OSI.027, tozasertib, and obatoclax mesylate may not be appropriate for patients in the high-risk group (Figure 7A). For the pyroptosis model, BI.2536, BIBR.1532, sabutoclax, and BMS.345541 were potential drugs for the treatment of patients from the low-risk group (Figure 7B).
Figure 7.
Differences in drug sensitivity between the low- and high-risk groups. (A) Two risk groups in hypoxia-associated prognostic model. (B) Two risk groups in pyroptosis-associated prognostic model (only showing representative top 4 drugs ranked by P value). IC50, half-maximal inhibitory concentration.
Correlation between hypoxic- and pyroptosis-related signature
To reveal the link between hypoxia and pyroptosis, correlation analysis of model genes was carried out. The results showed that there was a significant relationship between genes in the two models (Figure 8). For example, the pyroptosis-related genes, especially BRD4, PINK1, BCL6, and BRCA1, presented negative connection with most hypoxia-related genes.
Figure 8.
Correlation analysis of hypoxia- and pyroptosis-related prognostic genes. Red and blue indicate positive and negative relationship, respectively. *, P<0.05; **, P<0.01; ***, P<0.001.
Discussion
It is generally believed that tumor hypoxia regulates immune checkpoint expression and pyroptosis through activating HIF signaling, thus targeting the hypoxia-pyroptosis axis in combination with immunotherapy may be a promising therapeutic strategy (14). In this study, two risk models were constructed to predict the prognosis of GBM patients based on hypoxia- and pyroptosis-related genes. In addition, these prognostic models were closely associated with tumor immunity and can predict the sensitivity to treatment in different risk groups.
In the hypoxia-related prognostic model, Kaplan-Meier and ROC curve analyses indicated that the model was accurate and robust, predicting patient survival. The model contained the following 13 key genes: YAR, RPL15, EIF3E, RPS11, RPL27A, RPL23A, RPS2, RPL29, EEF1G, RPL4, RPL5, RPL34, TPT1. We observed that most of these genes encode ribosomal proteins (RPs). RPs are molecules required for the construction of ribosomal subunits and mature ribosomes, as well as are involved in a variety of biological functions, such as tumorigenesis, development, and immune signaling (19). In a variety of malignant tumors, upregulation of RPs transcripts is correlated with clinical stage, predicting increased resistance to standard therapy and poor prognosis for patients (20). Meanwhile, hypoxia can affect the composition of RPs and their incorporation into the ribosomes (21). The role of RPs dysregulation in GBM has been revealed. Under cellular stress conditions, RPs promote stem cell-like phenotype and invasiveness in GBM cells (22). Several key RPs-related genes, including RPS11, RPL27A, RPL23A and RPL34, are abnormally up-regulated in GBM and are related to adverse prognosis and recurrence in patients (23-26). RPL5 has distinct deletions and mutations in GBM and is found to be a candidate tumor suppressor in GBM (27). In addition, LYAR, EIF3E, EEF1G, and TPT1 are discovered to be prognostic biomarkers for GBM (28-31). Given that these genes play a crucial role in cancer growth and metastasis, further functional validation is necessary.
Further, the pyroptosis-related model also displayed accurate predictive ability for prognosis in GBM patients. The model contained 13 genes, IFI16, CALM3, BRD4, HSP90B1, BRCA1, BCL6, BCL3, IRF9, MDM2, AGER, SPP1, PINK1, and EZH2. IFI16 is an intracellular innate immune receptor involved in the formation of the inflammasome and exerts a key role in the pyroptosis process (32). It can activate the NF-kappa B pathway, thereby promoting the mesenchymal phenotypic transformation of GBM stem cells and causing radiotherapy resistance (32). BRD4 facilitates PPARγ degradation and pyroptosis through transcriptional activation MDM2 expression, and BRD4 or MDM2 inhibitors may help broaden the clinical treatment options for GBM (33-35). Besides, BRD4 and MDM2 are involved in the immune evasion mechanism of GBM cells, and their targeted inhibitors contribute to improved clinical outcomes (36,37). AGER induces cellular pyroptosis and its increased level is toxic to GBM cells (38,39). Similarly, PINK1-mediated mitophagy regulates pyroptosis of trophoblast cells by reducing ROS production and activating NLRP3 inflammasome (39). PINK1 is a negative regulator of cancer cell growth in GBM (40). CALM3, BRCA1, BCL3, and BCL6 are oncogenes that can drive cell survival and are associated with low OS in GBM patients (41-44). Moreover, HSP90B1, IRF9, and SPP1 are recognized as prognostic markers of GBM, and targeted inhibition can induce cancer cell death (45-47). Together, the genes identified in this study offer promising options for the precise treatment of GBM.
The relationship between hypoxia- and pyroptosis-related signatures was analyzed, and revealed the close link between them. It was further proved that hypoxia stimulation could contribute to the transcription of pyroptosis genes (48). Notably, pyroptosis-related genes displayed negative correlation with most hypoxia-related genes. In animal model of GBM, inhibition of hypoxia-induced HIFIα expression in macrophages is found to impair tumor progression (49). Benzimidazoles may exert satisfactory therapeutic performance by triggering pyroptosis of GBM cells (50). Combined with our findings, it is speculated that the development of hypoxia inhibition combined with pyroptosis induction drug may serve as an alternative medicine for GBM treatment.
We also revealed differences in the immune infiltration landscape in different risk populations classified by the prognostic models. Four types of immune cells caught our attention, among which naïve B cells, plasma cells, and native CD4 T cells were suppressed in the high-risk GBM patients, while neutrophils exhibited the opposite result. GBM is immunologically considered a “cold” tumor (51). B cells play a crucial role in the regression of GBM induced by immunotherapy, potentially acting as antigen-presenting cells to generate effective antitumor immune responses (52). These findings underscore that B cells serve as a positive prognostic indicator in GBM (53). Recent studies have revealed that plasma cells primarily secrete immunoglobulin G subtype antibodies, which can promote GBM stem cell proliferation at lower doses (54). However, the specific role of plasma cells in GBM remains largely unknown. CD4 T cells-mediated antitumor effects have attracted increasing attention (55). Research has demonstrated that the combination of chlorpromazine and vascular endothelial growth factor inhibitors exhibits significant therapeutic efficacy in GBM mouse models, primarily driven by the infiltration and activation of CD4 T and CD8 T cells (56). Therefore, maintaining a persistent presence of CD4 T cells in the tumor microenvironment is essential for achieving effective antitumor activity (57), which partly explains the reduced CD4 T cell levels observed in high-risk populations. Neutrophils are abundant immune cells in the circulatory system that frequently infiltrate tumor tissues in large numbers (58). In GBM, increased infiltration of neutrophils contributes to tumor growth, invasion, and treatment resistance (59). Taken together, high-risk patients display characteristics of an immune-cold tumor microenvironment, which partially accounts for their poor survival outcomes.
Our study is the first utilizing high-throughput data to establish the risk models related to hypoxia and pyroptosis for predicting the prognosis of GBM. Nevertheless, our research has several limitations. First, the data utilized in this study primarily originated from the TCGA and GEO databases. Although these databases provide abundant clinical and genomic data, their sample size remains limited and the transparency of sample collection and processing procedures is lacking, which may introduce bias. Second, although the constructed models were validated in the training and test sets, it still lacks large-scale external validation, which limits their general applicability. Third, our conclusions are based on bioinformatics analysis and lack functional experiments for verification. Meanwhile, this study focused on observing the expression patterns of hypoxia and pyroptosis-related genes in GBM, without exploring the molecular mechanisms underlying these correlations. Further research should include more independent datasets and experimental validation to enhance the reliability and clinical applicability of the models.
Conclusions
In summary, we constructed the hypoxia as well as pyroptosis-related risk models with robust and accurate predictive performance on GBM prognosis. These models could recognize populations at different risks, which were strongly correlated with immune cell infiltration status and can assess patient response to treatment. Our study also provides new insights into the crosstalk mechanism between hypoxia and pyroptosis in GBM, and speculates that influencing the link between them may alter the poor prognosis of GBM.
Supplementary
The article’s supplementary files as
Acknowledgments
None.
Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.
Footnotes
Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1367/rc
Funding: This study was supported by National Natural Science Foundation of China (No. 82360596).
Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1367/coif). The authors have no conflicts of interest to declare.
References
- 1.Singh S, Dey D, Barik D, et al. Glioblastoma at the crossroads: current understanding and future therapeutic horizons. Signal Transduct Target Ther 2025;10:213. 10.1038/s41392-025-02299-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Khiabani NA, Doustvandi MA, Story D, et al. Glioblastoma therapy: State of the field and future prospects. Life Sci 2024;359:123227. 10.1016/j.lfs.2024.123227 [DOI] [PubMed] [Google Scholar]
- 3.Tan AC, Ashley DM, López GY, et al. Management of glioblastoma: State of the art and future directions. CA Cancer J Clin 2020;70:299-312. 10.3322/caac.21613 [DOI] [PubMed] [Google Scholar]
- 4.Minniti G, Niyazi M, Alongi F, et al. Current status and recent advances in reirradiation of glioblastoma. Radiat Oncol 2021;16:36. 10.1186/s13014-021-01767-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Rong L, Li N, Zhang Z. Emerging therapies for glioblastoma: current state and future directions. J Exp Clin Cancer Res 2022;41:142. 10.1186/s13046-022-02349-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Niklasson CU, Fredlund E, Monni E, et al. Hypoxia inducible factor-2α importance for migration, proliferation, and self-renewal of trunk neural crest cells. Dev Dyn 2021;250:191-236. 10.1002/dvdy.253 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Zhang K, Ma D, Wu Y, et al. Impact of Chronic Intermittent Hypoxia on Cognitive Function and Hippocampal Neurons in Mice: A Study of Inflammatory and Oxidative Stress Pathways. Nat Sci Sleep 2024;16:2029-43. 10.2147/NSS.S489232 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Domènech M, Hernández A, Plaja A, et al. Hypoxia: The Cornerstone of Glioblastoma. Int J Mol Sci 2021;22:12608. 10.3390/ijms222212608 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Park JH, Lee HK. Current Understanding of Hypoxia in Glioblastoma Multiforme and Its Response to Immunotherapy. Cancers (Basel) 2022;14:1176. 10.3390/cancers14051176 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Michelucci A, Sforna L, Franciolini F, et al. Hypoxia, Ion Channels and Glioblastoma Malignancy. Biomolecules 2023;13:1742. 10.3390/biom13121742 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Rao Z, Zhu Y, Yang P, et al. Pyroptosis in inflammatory diseases and cancer. Theranostics 2022;12:4310-29. 10.7150/thno.71086 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Zhao W, Zhang L, Zhang Y, et al. The CDK inhibitor AT7519 inhibits human glioblastoma cell growth by inducing apoptosis, pyroptosis and cell cycle arrest. Cell Death Dis 2023;14:11. 10.1038/s41419-022-05528-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Lin J, Lai X, Liu X, et al. Pyroptosis in glioblastoma: A crucial regulator of the tumour immune microenvironment and a predictor of prognosis. J Cell Mol Med 2022;26:1579-93. 10.1111/jcmm.17200 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Meybodi SM, Ejlalidiz M, Manshadi MR, et al. Crosstalk between hypoxia-induced pyroptosis and immune escape in cancer: From mechanisms to therapy. Crit Rev Oncol Hematol 2024;197:104340. 10.1016/j.critrevonc.2024.104340 [DOI] [PubMed] [Google Scholar]
- 15.Hou J, Zhao R, Xia W, et al. PD-L1-mediated gasdermin C expression switches apoptosis to pyroptosis in cancer cells and facilitates tumour necrosis. Nat Cell Biol 2020;22:1264-75. 10.1038/s41556-020-0575-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Chen W, Lei C, Wang Y, et al. Prognostic Prediction Model for Glioblastoma: A Ferroptosis-Related Gene Prediction Model and Independent External Validation. J Clin Med 2023;12:1341. 10.3390/jcm12041341 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Zhang B, Xie L, Liu J, et al. Construction and validation of a cuproptosis-related prognostic model for glioblastoma. Front Immunol 2023;14:1082974. 10.3389/fimmu.2023.1082974 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Newman AM, Liu CL, Green MR, et al. Robust enumeration of cell subsets from tissue expression profiles. Nat Methods 2015;12:453-7. 10.1038/nmeth.3337 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Zhou X, Liao WJ, Liao JM, et al. Ribosomal proteins: functions beyond the ribosome. J Mol Cell Biol 2015;7:92-104. 10.1093/jmcb/mjv014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Elhamamsy AR, Metge BJ, Alsheikh HA, et al. Ribosome Biogenesis: A Central Player in Cancer Metastasis and Therapeutic Resistance. Cancer Res 2022;82:2344-53. 10.1158/0008-5472.CAN-21-4087 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Brumwell A, Fell L, Obress L, et al. Hypoxia influences polysome distribution of human ribosomal protein S12 and alternative splicing of ribosomal protein mRNAs. RNA 2020;26:361-71. 10.1261/rna.070318.119 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Hide T, Shibahara I, Inukai M, et al. Ribosomes and Ribosomal Proteins Promote Plasticity and Stemness Induction in Glioma Cells via Reprogramming. Cells 2022;11:2142. 10.3390/cells11142142 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Yong WH, Shabihkhani M, Telesca D, et al. Ribosomal Proteins RPS11 and RPS20, Two Stress-Response Markers of Glioblastoma Stem Cells, Are Novel Predictors of Poor Prognosis in Glioblastoma Patients. PLoS One 2015;10:e0141334. 10.1371/journal.pone.0141334 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Verploegh ISC, Conidi A, Brouwer RWW, et al. Comparative single-cell RNA-sequencing profiling of BMP4-treated primary glioma cultures reveals therapeutic markers. Neuro Oncol 2022;24:2133-45. 10.1093/neuonc/noac143 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Shi Y, Sun Y, Cheng H, et al. EFNB1 Acts as a Novel Prognosis Marker in Glioblastoma through Bioinformatics Methods and Experimental Validation. J Oncol 2021;2021:4701680. 10.1155/2021/4701680 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Du C, Wang T, Jia J, et al. Suppression of RPL34 Inhibits Tumor Cell Proliferation and Promotes Apoptosis in Glioblastoma. Appl Biochem Biotechnol 2022;194:3494-506. 10.1007/s12010-022-03857-0 [DOI] [PubMed] [Google Scholar]
- 27.Fancello L, Kampen KR, Hofman IJ, et al. The ribosomal protein gene RPL5 is a haploinsufficient tumor suppressor in multiple cancer types. Oncotarget 2017;8:14462-78. 10.18632/oncotarget.14895 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Bhushan A, Kumari R, Srivastava T. Scouting for common genes in the heterogenous hypoxic tumor microenvironment and their validation in glioblastoma. 3 Biotech 2021;11:451. [DOI] [PMC free article] [PubMed]
- 29.Tian Y, Chen L, Jiang Y. LASSO-based screening for potential prognostic biomarkers associated with glioblastoma. Front Oncol 2022;12:1057383. 10.3389/fonc.2022.1057383 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Vastrad C, Vastrad B. Bioinformatics analysis of gene expression profiles to diagnose crucial and novel genes in glioblastoma multiform. Pathol Res Pract 2018;214:1395-461. 10.1016/j.prp.2018.07.015 [DOI] [PubMed] [Google Scholar]
- 31.Staflin K, Zuchner T, Honeth G, et al. Identification of proteins involved in neural progenitor cell targeting of gliomas. BMC Cancer 2009;9:206. 10.1186/1471-2407-9-206 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Gao Z, Xu J, Fan Y, et al. ARPC1B promotes mesenchymal phenotype maintenance and radiotherapy resistance by blocking TRIM21-mediated degradation of IFI16 and HuR in glioma stem cells. J Exp Clin Cancer Res 2022;41:323. 10.1186/s13046-022-02526-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Xu X, Qiu H. BRD4 promotes gouty arthritis through MDM2-mediated PPARγ degradation and pyroptosis. Mol Med 2024;30:67. 10.1186/s10020-024-00831-w [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Duan W, Yu M, Chen J. BRD4: New hope in the battle against glioblastoma. Pharmacol Res 2023;191:106767. 10.1016/j.phrs.2023.106767 [DOI] [PubMed] [Google Scholar]
- 35.Pellot Ortiz KI, Rechberger JS, Nonnenbroich LF, et al. MDM2 Inhibition in the Treatment of Glioblastoma: From Concept to Clinical Investigation. Biomedicines 2023;11:1879. 10.3390/biomedicines11071879 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Liu Y, Cai L, Wang H, et al. BRD4 promotes immune escape of glioma cells by upregulating PD-L1 expression. J Neurooncol 2025;171:669-79. 10.1007/s11060-024-04889-8 [DOI] [PubMed] [Google Scholar]
- 37.Arnoff TE, El-Deiry WS. MDM2/MDM4 amplification and CDKN2A deletion in metastatic melanoma and glioblastoma multiforme may have implications for targeted therapeutics and immunotherapy. Am J Cancer Res 2022;12:2102-17. [PMC free article] [PubMed] [Google Scholar]
- 38.Jandial R, Neman J, Lim PP, et al. Inhibition of GLO1 in Glioblastoma Multiforme Increases DNA-AGEs, Stimulates RAGE Expression, and Inhibits Brain Tumor Growth in Orthotopic Mouse Models. Int J Mol Sci 2018;19:406. 10.3390/ijms19020406 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Sun Y, Lv D, Xie Y, et al. PINK1-mediated mitophagy induction protects against preeclampsia by decreasing ROS and trophoblast pyroptosis. Placenta 2023;143:1-11. 10.1016/j.placenta.2023.09.010 [DOI] [PubMed] [Google Scholar]
- 40.Agnihotri S, Golbourn B, Huang X, et al. PINK1 Is a Negative Regulator of Growth and the Warburg Effect in Glioblastoma. Cancer Res 2016;76:4708-19. 10.1158/0008-5472.CAN-15-3079 [DOI] [PubMed] [Google Scholar]
- 41.Wang A, Zhang G. Differential gene expression analysis in glioblastoma cells and normal human brain cells based on GEO database. Oncol Lett 2017;14:6040-4. 10.3892/ol.2017.6922 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Vassilakopoulou M, Won M, Curran WJ, et al. BRCA1 Protein Expression Predicts Survival in Glioblastoma Patients from an NRG Oncology RTOG Cohort. Oncology 2021;99:580-8. 10.1159/000516168 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Fabre MS, Stanton NM, Slatter TL, et al. The oncogene BCL6 is up-regulated in glioblastoma in response to DNA damage, and drives survival after therapy. PLoS One 2020;15:e0231470. 10.1371/journal.pone.0231470 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Wu J, Li L, Jiang G, et al. B-cell CLL/lymphoma 3 promotes glioma cell proliferation and inhibits apoptosis through the oncogenic STAT3 pathway. Int J Oncol 2016;49:2471-9. 10.3892/ijo.2016.3729 [DOI] [PubMed] [Google Scholar]
- 45.Liu Z, Yang L, Wu W, et al. Prognosis and therapeutic significance of IGF-1R-related signaling pathway gene signature in glioma. Front Cell Dev Biol 2024;12:1375030. 10.3389/fcell.2024.1375030 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Qian X, Tan H, Zhang J, et al. Identification of biomarkers for pseudo and true progression of GBM based on radiogenomics study. Oncotarget 2016;7:55377-94. 10.18632/oncotarget.10553 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Wang K, Xiao Y, Zheng R, et al. Immune cell infiltration and drug response in glioblastoma multiforme: insights from oxidative stress-related genes. Cancer Cell Int 2024;24:123. 10.1186/s12935-024-03316-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Yu LM, Zhang WH, Han XX, et al. Hypoxia-Induced ROS Contribute to Myoblast Pyroptosis during Obstructive Sleep Apnea via the NF-κB/HIF-1α Signaling Pathway. Oxid Med Cell Longev 2019;2019:4596368. 10.1155/2019/4596368 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Pang L, Guo S, Khan F, et al. Hypoxia-driven protease legumain promotes immunosuppression in glioblastoma. Cell Rep Med 2023;4:101238. 10.1016/j.xcrm.2023.101238 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Ren LW, Li W, Zheng XJ, et al. Benzimidazoles induce concurrent apoptosis and pyroptosis of human glioblastoma cells via arresting cell cycle. Acta Pharmacol Sin 2022;43:194-208. 10.1038/s41401-021-00752-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Singh K, Hotchkiss KM, Patel KK, et al. Enhancing T Cell Chemotaxis and Infiltration in Glioblastoma. Cancers (Basel) 2021;13:5367. 10.3390/cancers13215367 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Candolfi M, Curtin JF, Yagiz K, et al. B cells are critical to T-cell-mediated antitumor immunity induced by a combined immune-stimulatory/conditionally cytotoxic therapy for glioblastoma. Neoplasia 2011;13:947-60. 10.1593/neo.11024 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.Che Y, Lee J, Abou-Taleb F, et al. Induced B cell receptor diversity predicts PD-1 blockade immunotherapy response. Proc Natl Acad Sci U S A 2025;122:e2501269122. 10.1073/pnas.2501269122 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 54.Gao J, Gu D, Yang K, et al. Infiltrating plasma cells maintain glioblastoma stem cells through IgG-Tumor binding. Cancer Cell 2025;43:122-143.e8. 10.1016/j.ccell.2024.12.006 [DOI] [PubMed] [Google Scholar]
- 55.Alspach E, Lussier DM, Miceli AP, et al. MHC-II neoantigens shape tumour immunity and response to immunotherapy. Nature 2019;574:696-701. 10.1038/s41586-019-1671-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 56.Chryplewicz A, Scotton J, Tichet M, et al. Cancer cell autophagy, reprogrammed macrophages, and remodeled vasculature in glioblastoma triggers tumor immunity. Cancer Cell 2022;40:1111-1127.e9. 10.1016/j.ccell.2022.08.014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 57.Wang D, Aguilar B, Starr R, et al. Glioblastoma-targeted CD4+ CAR T cells mediate superior antitumor activity. JCI Insight 2018;3:e99048. 10.1172/jci.insight.99048 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 58.Maas RR, Soukup K, Fournier N, et al. The local microenvironment drives activation of neutrophils in human brain tumors. Cell 2023;186:4546-4566.e27. 10.1016/j.cell.2023.08.043 [DOI] [PubMed] [Google Scholar]
- 59.Zhong R, He H, Wang X. Novel neutrophil targeting platforms in treating Glioblastoma: Latest evidence and therapeutic approaches. Int Immunopharmacol 2025;150:114173. 10.1016/j.intimp.2025.114173 [DOI] [PubMed] [Google Scholar]