Glioma prognosis risk prediction model based on disulfide death related gene and application of glioma prognosis risk prediction model

Abstract

The invention discloses a glioma prognosis risk prediction model based on a disulfide death related gene and application thereof, and belongs to the technical field of biomedical detection. According to the invention, three glioblastoma queues are analyzed based on TCGA and CGGA databases, and a risk scoring model containing nine core genes is finally obtained by using a prognosis model constructed by using disulfide death related genes based on an LASSO-cox algorithm. The model is excellent in prediction performance, AUC values of one year, three years and five years are 0.876, 0.944 and 0.905 respectively, and verification results show that the model shows high prediction ability in an internal test set and an independent verification set and is significantly related to clinical features of GBM patients. A clinical prediction column diagram further combined with clinical characteristics shows better performance, and AUC values in 1-5 years are all higher than 0.9. The prognosis model provided by the invention is helpful for improving personalized treatment and prognosis evaluation of patients with glioma GBM, so that the clinical curative effect is improved, and an important basis is provided for personalized prognosis evaluation and immunotherapy of GBM.

Description

Technical Field

[0001] The present invention relates to the field of biomedical detection technology, and in particular to a glioma prognosis risk prediction model based on disulfide death-related genes and its application. Background Art

[0002] Glioma (GBM) is a global malignant disease that is believed to develop from neural stem cells or progenitor cells carrying tumor-initiating gene changes. It is one of the most common primary brain tumors in the world, with extremely high morbidity and mortality. Glioma is clinically multiple and is an intracranial tumor originating from glial cells, accounting for approximately 80% of all brain malignancies. The high heterogeneity and invasiveness of gliomas lead to their high mortality rate, high malignancy, rapid proliferation, easy recurrence and poor prognosis. It is considered one of the most difficult tumors to treat. According to the World Health Organization (WHO) classification of central nervous system (CNS) tumors, gliomas can be divided into grades I-IV, namely low-grade gliomas (grades I-II) and high-grade gliomas (grades III-IV), and different types of gliomas have certain differences in biological behavior and clinical manifestations. Currently, glioma treatments have achieved some progress and applications in treatments such as surgery combined with chemotherapy or radiotherapy, electric field therapy, and immunotherapy, and the prognosis of GBM patients has improved. However, due to the immunosuppressive state and high heterogeneity of the tumor microenvironment (TME), the overall treatment effect is not ideal, and the patient prognosis remains poor, especially for glioblastoma patients, with a 5-year survival rate of only 6.8%. Therefore, exploring the relationship between GBM and related potential factors and identifying GBM molecular biomarkers with high sensitivity and specificity are of great significance for revealing its pathological mechanisms and formulating clinical interventions to improve patient survival and prognosis.

[0003] Recent evidence suggests that disulfide death (DSD) is associated with the development and progression of cancer. Unlike other cell death pathways such as pyroptosis, necroptosis, and copper death, DSD is a newly discovered metabolic-related cell death pathway. Liu et al. found that under glucose starvation, cancer cells with high expression of the cystine transporter solute carrier family 7, member 11 (SLC7A11) experience abnormal accumulation of intracellular cystine and other disulfides, inducing disulfide stress and ultimately leading to cell dysfunction and death. This cell death mechanism has been implicated in various diseases and is particularly prominent in metabolically abnormal cells, such as cancer cells, suggesting its potential role in tumor development and progression. DSD-related genes (DRGs) play an important role in various biological processes. This suggests that DSD-related genes could serve as new biomarkers to identify cancer patients and predict their survival status and duration, potentially providing an effective strategy for cancer prevention, diagnosis, and treatment. However, the current research on the biological functions and comprehensive analysis of DRGs involved in disulfide death in GBM is still limited, and the DRGs used to predict GBM prognosis and targeted therapy are still unclear. Therefore, establishing a prognostic model of disulfide death-related genes that can predict the prognosis of GBM patients has important clinical significance and role in the treatment and prognosis of GBM patients. Summary of the Invention

[0004] To solve the above problems, the present invention uses bioinformatics knowledge and technology to obtain relevant data and corresponding clinical information of glioma patients from the Cancer Genome Atlas (TCGA) database, constructs a prognostic model for GBM based on DRGs, and screens out key prognostic genes. The accuracy and stability of the model are verified based on the mRNAarray_301 dataset and mRNAseq_325 dataset in the CGGA database. In addition, the potential role of DRGs as a biomarker for prognosis and targeted therapy is further explored to provide a certain basis for the clinical treatment and prognosis of glioma.

[0005] The first object of the present invention is to provide a glioma prognosis risk prediction model, wherein the prediction model evaluates the predictive performance of glioma prognosis based on the disulfide death-related gene (DRGs) risk score, and the DRGs risk score conforms to the following formula:

[0006] Wherein, Exp is the expression value of each DRGs in the model (normalized expression level value), coef is the regression coefficient of each DRGs in the model, i represents the index of DRGs in the model, and n represents the number of DRGs included in the model. The DRGs include IL17RC gene, KIF18A gene, TFPI gene, ISG20 gene, HOXA2 gene, FRMPD1 gene, P2RX6 gene, PELI2 gene, and TXN2 gene.

[0007] Furthermore, the Gene ID of the IL17RC gene is 84818, the Gene ID of the KIF18A gene is 81930, the Gene ID of the TFPI gene is 7035, the Gene ID of the ISG20 gene is 3669, the Gene ID of the HOXA2 gene is 3199, the Gene ID of the FRMPD1 gene is 22844, the Gene ID of the P2RX6 gene is 9127, the Gene ID of the PELI2 gene is 57161, and the Gene ID of the TXN2 gene is 25828.

[0008] Furthermore, the regression coefficients corresponding to the IL17RC gene, KIF18A gene, TFPI gene, ISG20 gene, HOXA2 gene, FRMPD1 gene, P2RX6 gene, PELI2 gene, and TXN2 gene were 0.305, 0.269, 0.156, 0.127, 0.068, -0.101, -0.107, -0.332, and -0.711, respectively.

[0009] A second object of the present invention is to provide a method for constructing the prediction model, comprising the following steps:

[0010] S1. Obtain prognostic samples from glioma patients and control prognostic samples, analyze significantly differentially expressed genes with high prognostic correlation, and obtain prognostic-related genes, wherein the prognostic-related genes are IL17RC gene, KIF18A gene, TFPI gene, ISG20 gene, HOXA2 gene, FRMPD1 gene, P2RX6 gene, PELI2 gene, and TXN2 gene;

[0011] S2. Multivariate Cox regression analysis was performed on prognosis-related genes, and the risk score of each patient was calculated based on the regression coefficient and expression level of each gene;

[0012] S3. Evaluate the predictive performance of the prognostic risk prediction model based on the risk score.

[0013] The third object of the present invention is to provide a glioma prognosis risk prediction system, wherein the prediction system comprises the glioma prognosis risk prediction model and a reagent for detecting DRGs expression level.

[0014] The DRGs include IL17RC gene, KIF18A gene, TFPI gene, ISG20 gene, HOXA2 gene, FRMPD1 gene, P2RX6 gene, PELI2 gene, and TXN2 gene.

[0015] The fourth object of the present invention is to provide the application of the glioma prognosis risk prediction model or glioma prognosis risk prediction system in the preparation of glioma prognosis detection products.

[0016] The fifth object of the present invention is to provide a biomarker for predicting the prognosis risk of glioma, wherein the biomarker is a combination of IL17RC gene, KIF18A gene, TFPI gene, ISG20 gene, HOXA2 gene, FRMPD1 gene, P2RX6 gene, PELI2 gene and TXN2 gene.

[0017] The sixth object of the present invention is to provide a reagent for detecting the expression level of the biomarker for use in preparing a glioma prognosis risk detection product.

[0018] Furthermore, the reagent for detecting the expression level of the biomarker comprises primers for amplifying IL17RC gene, KIF18A gene, TFPI gene, ISG20 gene, HOXA2 gene, FRMPD1 gene, P2RX6 gene, PELI2 gene and TXN2 gene.

[0019] The seventh object of the present invention is to provide a detection kit, which contains reagents for detecting the expression level of the biomarker.

[0020] Furthermore, the detection step includes:

[0021] (1) Extracting genomic DNA from the sample to be tested and quantifying the IL17RC gene, KIF18A gene, TFPI gene, ISG20 gene, HOXA2 gene, FRMPD1 gene, P2RX6 gene, PELI2 gene, and TXN2 gene;

[0022] (2) Determine the prognosis of glioma based on gene expression levels.

[0023] Furthermore, in step (2), the prognostic model is established based on the gene expression level to determine the prognosis of glioma.

[0024] Beneficial effects of the present invention:

[0025] This study successfully constructed a prognostic model for glioma patients using bioinformatics analysis based on DRGs, revealing the important role of DRGs in the tumor immune microenvironment. These findings provide important clues for understanding the potential role of DRGs as biomarkers for prognosis and targeted therapy, and for exploring the clinical treatment and prognosis of glioma. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 808 disulfide death-related genes (DRGs) associated with glioma were identified by high-throughput CRISPR screening analysis of glucose-starved cells.

[0027] Figure 2 Construction and performance analysis of the DRGs prognostic model. (A) Univariate Cox regression analysis identified 808 DRGs associated with overall survival (OS) in GBM patients in the training set; (B-C) LASSO algorithm screened 18 DRGs associated with OS; (D) Forest plot of the 9 DRGs used in the prognostic model constructed using multivariate Cox regression analysis; (E) Time ROC curve showing the ROC curves and AUC values ​​for patients in the training set at 1, 3, and 5 years; (F) Survival curves for patients in the high- and low-risk groups in the training set; (G) Distribution of risk scores, survival outcomes, and time for samples in the high- and low-risk groups in the training set, with a heat map showing the differential expression of DRGs in the model between the high- and low-risk groups in the training set.

[0028] Figure 3 To verify the stability of the DRGs prognostic model in the internal test set and the external independent validation set. (A) The TimeROC curve shows the ROC curves and AUC values ​​of patients in the internal test set at 1, 3, and 5 years; (B) The survival curves of patients in the high-risk and low-risk groups in the internal test set; (C) The TimeROC curve shows the ROC curves and AUC values ​​of patients in mRNAarray_301 at 1, 3, and 5 years; (D) The survival curves of patients in the high-risk and low-risk groups in mRNAarray_301; (E) The TimeROC curve shows the ROC curves and AUC values ​​of patients in mRNAseq_325 at 1, 3, and 5 years; (F) The survival curves of patients in the high-risk and low-risk groups in mRNAseq_325; The distribution of risk scores and survival outcomes of samples in different risk groups in the internal test set (G) and mRNAarray_301 (H), where the heat map shows the expression differences between the model and the high-risk and low-risk groups in the dataset;

[0029] Figure 4The DRG risk score for disulfiramylase-associated death is an independent prognostic factor. (A) Forest plot showing the univariate Cox analysis results of the DRGs risk score and major clinicopathological characteristics for patient prognosis in TCGALGG_GBM; (B) Forest plot showing the multivariate Cox analysis results of the factors that were significant in the univariate Cox analysis for patient prognosis in TCGALGG_GBM; (C) Forest plot showing the univariate Cox analysis results of the DRGs risk score and major clinicopathological characteristics for patient prognosis in the mRNAarray_301 dataset; (D) Forest plot showing the multivariate Cox analysis results of the factors that were significant in the univariate Cox analysis for patient prognosis in mRNAarray_301.

[0030] Figure 5 To construct and evaluate a clinical prediction nomogram. (A) A clinical prediction nomogram was constructed based on independent prognostic factors identified by multivariate Cox analysis of the TCGA LGG_GBM dataset; (B) A calibration plot validates the accuracy of the nomogram constructed based on the TCGA LGG_GBM dataset; (C) Receiver operating characteristic (ROC) curve of the score calculated from the nomogram; (D) A clinical prediction nomogram was constructed based on independent prognostic factors identified by multivariate Cox analysis of the mRNAarray_301 dataset.

[0031] Figure 6 Correlation analysis of the risk score with immune status and tumor microenvironment. (A) Heat map showing the correlation between the risk score and the infiltration of various immune cells and stromal cells in multiple datasets, including LGG_GBM, mRNAarray_301, and mRNA-array_325. Scatter plots showing the correlation between the risk score and Th2 cells (B), MI macrophages (C), microenvironment score (D), and immune score (E) in the TCGA LGG_GBM cohort. (F) Heat map showing the correlation between the risk score and immune checkpoint genes in multiple datasets, including LGG_GBM, mRNAarray_301, and mRNA-array_325. Scatter plots showing the correlation between the risk score and the expression of SIGLEC7 (G), LILRB2 (H), and HAVCR2 (I) genes in the TCGA LGG_GBM cohort.

[0032] Figure 7The risk score is associated with glioma progression. Heat maps show the association of the risk score with cancer-related signaling pathways (A) and biological functions (B) in GSEA analysis results based on the LGG_GBM, mRNAarray_301, and mRNA-array_325 datasets. GSEA plots demonstrate the enrichment of the DRG risk score in several important signaling pathways (C) and biological functions (D). Heat maps show the intersection of the correlation analysis results between the risk score and oncogene expression (E) and antitumor drug sensitivity (F) in the three GBM datasets. DETAILED DESCRIPTION

[0033] The present invention will be further described below with reference to the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it. However, the embodiments are not intended to limit the present invention.

[0034] The scheme that the present invention relates to is as follows:

[0035] The prognostic model constructed based on 9 DRGs in the present invention has good accuracy and stability. When the risk score of the model is considered alone, the AUC of the 3-year survival prediction in the training set and the internal test set is greater than 0.93, and the AUC of the 3-year survival prediction in another independent validation set is 0.86. The clinical prediction nomogram after incorporating other independent prognosis-related clinical pathological features shows better performance, with the AUC of the 3-year survival prediction reaching 0.9. Some risk prediction models based on transcriptome data have been reported previously, such as a model constructed based on TGFβ signaling-related lncRNA, whose AUC values ​​for the 1-year, 3-year and 5-year ROC curves were 0.858, 0.899 and 0.825, respectively. Another study constructed a model based on neurotrophic factor-related genes with an AUC of 0.734 and 0.823 for 1-year and 3-year OS, respectively. Compared with some published literature, the model constructed by the present invention has better performance and is of great significance in clinical application.

[0036] Further mechanistic studies have shown that the risk score is also significantly correlated with the infiltration of multiple immune cells, such as Th2 and M1 macrophages, and the expression of multiple immune regulatory genes, such as SIGLEC7, suggesting that it has a regulatory effect on the immune microenvironment and immune response. Studies have confirmed that there is a significant correlation between tumor cells and the tumor microenvironment, which also plays an important role in the occurrence, development, and metastasis of tumors. Further studies have found that the DRGs risk score is significantly associated with multiple cancer-related signaling pathways, such as DNA replication, glycosylation, cell adhesion molecules, and mismatch repair, suggesting that high-risk gliomas may have stronger proliferative activity and genomic instability, which is consistent with their aggressive clinical behavior. Currently, tumor cells resist targeted drugs through different mechanisms, which is a major obstacle to genotype-based precision medicine. The DRGs risk score is also significantly correlated with the expression of key oncogenes such as PAX3. These genes play an important role in tumor malignant progression, invasion and metastasis, and treatment resistance, suggesting that this model can serve as a reliable prognostic assessment tool and provide a molecular basis for predicting patient sensitivity to targeted therapy and chemotherapy.

[0037] In summary, this study constructed a DRGs risk score model based on the TCGA database and validated it using different datasets. This model can serve as an independent risk factor for survival and prognosis in GBM patients. It also provides a new and effective biomarker and reference direction for the treatment of GBM patients, potentially offering a promising approach for the treatment of gliomas.

[0038] Example 1

[0039] 1. Data Collection

[0040] RNA sequencing data and clinical information from 659 glioma patients were retrieved and downloaded from The Cancer Genome Atlas (TCGA) database. Clinical data included patient age, sex, KPS score, tumor stage, risk score, radiotherapy status, and survival time. Data that met inclusion and exclusion criteria were randomly divided into a training set (n = 463) and an internal test set (n = 196) in a 7:3 ratio. All samples were complete data sets, and expression profiles, clinical information, and results were openly available. To validate the model, two independent validation sets (mRNAarray_301 dataset (n = 284) and mRNAseq_325 dataset (n = 313)) were obtained from the TCGA database. Low-quality sequencing data were removed and normalized to ensure data reliability and accuracy.

[0041] 2 Research Methods

[0042] 2.1 Construction and validation of the prognostic model

[0043] The present invention uses univariate and multivariate Cox regression models and least absolute shrinkage and selection operator (LASSO) regression analysis to establish a predictive risk model. The univariate Cox model was used to study the relationship between the continuous expression level of DRGs and overall survival (OS), and the hazard ratio (HR) and P value of the univariate Cox regression analysis were used to determine the candidate DRGs associated with survival. After screening out the variables with potential prognostic value, they were included in the multivariate Cox regression model. By the stepwise regression method, the variables that had a significant impact on prognosis were further screened out to construct its predictive model. The LASSO regression model analysis was performed using the "glmnet" software package. Finally, the risk score of each GBM patient was calculated based on the expression values ​​of DRGs (Expi) and the Cox coefficient (coefi). The specific formula is:

[0044] Based on the median risk score, each patient was divided into high- and low-risk groups, and survival was analyzed and compared between the two groups. Kaplan-Meier survival curves were then generated to assess the survival rates of patients in the high- and low-risk groups within each dataset, visually demonstrating the survival differences between the two groups. Statistical testing using the log-rank test validated the model's ability to stratify prognostic outcomes in the training, internal test, and validation sets.

[0045] In addition, the present invention draws the receiver operating characteristic (ROC) curve and calculates the area under the ROC curve (AUC) at different time points to evaluate the model's predictive accuracy for patient prognosis; calculates the consistency index (C-index), which comprehensively considers the model's ability to rank the prognostic risks of all patients, further verifying the model's predictive ability.

[0046] 2.2 Gene set enrichment analysis (GESA)

[0047] Based on correlation analysis between risk scores and all mRNAs, samples were divided into high- and low-risk groups. GSEA enrichment analysis was further performed using the "ClusterProfiler" R package (Version 4.1.1) to explore the potential mechanisms of DRGs. Differentially expressed genes (DEGs) between the high- and low-risk groups were identified using the "limma" R package, using a log(fold change) > 1 and P < 0.05 as thresholds. These DEGs were then input into the DAVID online tool (https: / / david.ncifcrf.gov / ) for pathway and biological process enrichment analysis.

[0048] 2.3 Construction and evaluation of nomograms

[0049] An alignment diagram, also known as a nomogram, is a tool that graphically displays the predictions of multivariate models and is primarily used to provide personalized survival predictions for patients. Based on the results of multivariate Cox regression analysis, this study used the "rms" R package to construct a nomogram, constructing a comprehensive model that visually demonstrates the impact of risk scores and clinical characteristics on patient survival. The nomogram combines DRGs risk scores with independent prognostic factors derived from multivariate Cox regression analysis to generate survival probability predictions for GBM patients.

[0050] In addition, the predictive performance of the nomogram was evaluated by calculating the C-index (consistency index). A calibration curve was used to assess the applicability of the model in actual clinical practice. Decision curve analysis (DCA) was also used to evaluate the clinical decision-making value of the nomogram.

[0051] 2.4 Correlation Analysis

[0052] In order to further explore the biological role and clinical significance of the DRGs prognostic model, the risk score was correlated with the immune status and tumor microenvironment.

[0053] Oncogenes were obtained from the ONGene database (http: / / www.ongene.bioinfo-minzhao.org), while immune checkpoint genes (ICGs) and immunomodulatory genes (IMGs) were obtained from previous studies. Finally, correlation analysis was performed using the Spearman method based on the "psych" package.

[0054] 2.5 Data Statistical Analysis

[0055] All analyses were performed using R software (Version 4.1.1) and its corresponding base packages. Lasso regression analysis was performed using the "glmnet" R package, and univariate and multivariate Cox analyses and Kaplan-Meier survival curves were performed using the "survival" and "survminer" R packages. Receiver operating characteristic (ROC) curves were drawn using the "TimeROC" R package, and the area under the curve (AUC) was calculated using the "survivalROC" R package. Nomograms were created using the "rms" R package.

[0056] GraphPad V8.3.0 software (GraphPad Software, LLC) was used for graphing and statistical analysis. Data are presented as mean ± standard deviation. Two-group comparisons were performed using the Student's t-test, multiple-group comparisons were performed using analysis of variance (ANOVA), and categorical variables were analyzed using chi-square analysis to determine whether there were statistically significant differences between groups. All statistical tests were two-sided, and P < 0.05 was considered statistically significant.

[0057] Here are the results:

[0058] (1) Data collection

[0059] The present invention utilized three GBM cohorts from the TCGA and CGGA databases and their corresponding clinical data. Tables 1 and 2 summarize the demographic and clinical characteristics of the training set, internal test set, and independent validation set. After excluding samples with missing clinical information in the TCGA-LGGGBM dataset, a total of 659 GBM patients were included, of whom 417 were alive at the end of the follow-up period and 242 had died (median follow-up time: 2.29 years). The dataset was randomly divided into a training set (n = 463) and an internal test set (n = 196) in a 7:3 ratio. As expected, there were no significant differences in the main clinicopathological characteristics between the training set, test set, and the entire TCGA-LGGGBM cohort (see Table 1). In addition, the present invention also included the CGGA datasets mRNAarray_301 and mRNAseq_325, which included 284 and 313 GBM patients, respectively, with a median follow-up time of 4.53 and 3.98 years at the end of follow-up, respectively.

[0060] Based on the results of high-throughput CRISPR screening analysis of glucose-starved cells, a total of 808 disulfide death-related genes (DRGs) were identified when the normZ score was set to ±2, including 399 disulfide death-cooperating genes and 409 suppressor genes ( Figure 1 ).

[0061] Table 1 Demographic and clinical data of the training and internal test sets

[0062]

[0063] Table 2 Demographic and clinical data of the two external validation sets

[0064]

[0065]

[0066] (2) Construction of a DRGs-based prognostic model in GBM patients

[0067] Based on the training data, we first screened out 520 DRGs significantly associated with prognosis through univariate Cox regression analysis ( Figure 2 A). Subsequently, the LASSO algorithm was used for variable selection, and 18 key prognostic-related DRGs were finally identified ( Figure 2 BC). Through stepwise multivariate Cox regression analysis, a risk scoring model including 9 core genes (IL17RC, KIF18A, TFPI, ISG20, HOXA2, FRMPD1, P2RX6, PELI2, TXN2) was finally constructed ( Figure 2 B) The risk score was calculated as follows: Risk score = IL17RC Exp × (0.305) + KIF18A Exp × (0.269) + TFPI Exp × (0.156) + ISG20 Exp × (0.127) + HOXA2 Exp × (0.068) + FRMPD1 Exp × (-0.101) + P2RX6 Exp × (-0.107) + PELI2 Exp × (-0.332) + TXN2 Exp × (-0.711). The model showed excellent predictive performance in the training set. ROC curve analysis showed that the AUC values ​​for 1-, 3-, and 5-year survival rates reached 0.876, 0.944, and 0.905, respectively. Figure 2 E). Survival analysis further confirmed that the prognosis of patients in the high-risk group was significantly worse than that in the low-risk group ( Figure 2 F), and their overall survival time is shorter and the mortality rate is higher ( Figure 2 G). In addition, heat map analysis revealed the differential expression patterns of these 9 key genes in high-risk and low-risk groups ( Figure 2 G), further verifying the biological plausibility of the model.

[0068] (3) Validation of the DRGs-based prognostic model in GBM patients

[0069] To verify the robustness of the model, an independent test set was used for further analysis. The ROC curve results showed that the risk score had excellent predictive efficacy for the patient's 1-, 3-, and 5-year survival rates, with AUC values ​​reaching 0.902, 0.931, and 0.858, respectively. Figure 3 A). Survival analysis showed that the prognosis of patients in the low-risk group was significantly better than that in the high-risk group (P<0.001, Figure 3 B). To comprehensively evaluate the robustness and universality of the model, further validation was performed in two independent validation sets, mRNAarray_301 and mRNAseq_325. The results showed that the risk score had a stable predictive efficacy for the 1-, 3-, and 5-year survival rates of patients, especially for the mid- and long-term prognosis. The 3- and 5-year AUC values ​​in the two datasets were 0.860 and 0.834, respectively (mRNAarray_301, Figure 3 C) and 0.732 and 0.763 (mRNAseq_325, Figure 3 E). Group analysis based on the median risk score showed that the overall survival of patients in the low-risk group was significantly better than that in the high-risk group (P<0.001, Figure 3 D, F). Further analysis found that the mortality rate of patients in the high-risk group in the mRNAarray_301 and mRNAseq_325 datasets was significantly higher than that in the low-risk group, indicating that the model has good prognostic stratification capabilities. Heat map analysis intuitively shows the differential expression patterns of key genes between high- and low-risk groups ( Figure 3 GH). These results fully demonstrate that the GBM risk score constructed based on DRGs is a stable and reliable prognostic indicator.

[0070] (4) Disulfide death-related genes (DRGs) risk score is an independent prognostic factor

[0071] Univariate Cox regression analysis of the TCGA LGG_GBM dataset revealed that the DRGs risk score was significantly correlated with clinicopathological characteristics such as age (HR=1.05), KPS score (HR=1.12), and tumor grade (HR=2.87) (all P<0.05, Figure 4 A). Multivariate analysis further confirmed that after adjusting for other confounding factors, the DRGs risk score still maintained its independent prognostic value (HR=7.29, P<0.001, Figure 4 B). In the mRNAarray_301 dataset, univariate Cox regression analysis showed that DRGs risk score (HR=3.12), tumor grade (HR=2.35), gender (HR=1.68), and IDH mutation status (HR=0.62) were all significantly associated with patient prognosis (all P<0.001, Figure 4C). After multivariate adjustment, the DRGs risk score still maintained its independent prognostic significance (HR=2.96, P<0.001).

[0072] (5) Construct and evaluate clinical prediction nomograms

[0073] To ensure the practicality of the DEGs prognostic model, a prognostic nomogram for predicting the overall survival of glioma patients was established based on the TCGALGGGBM dataset ( Figure 5 A). The nomogram included independent prognostic factors based on multivariate COX regression, including age, KPS score, tumor grade, and risk score. The performance of the nomogram was evaluated by calibration curve and Bootstrap C-index. The results showed that the C-index was >0.86. ( Figure 5 B). The ROC curve confirmed that the score calculated based on the nomogram had a high predictive ability for the overall survival of patients, with the AUC values ​​of 0.903, 0.955, and 0.908 at 1, 3, and 5 years in the TCGA LGGGBM dataset, respectively ( Figure 5 C). Furthermore, a prognostic nomogram for predicting overall survival of glioma patients was established based on the mRNAarray_301 dataset, and the Grade and risk scores were incorporated into the nomogram ( Figure 5 D). Calibration curves and receiver operating characteristic (ROC) curves confirmed that the score calculated based on the nomogram was highly predictive of overall survival, especially for long-term survival, with 3- and 5-year AUC values ​​of 0.876 and 0.847, respectively.

[0074] (6) Correlation analysis between risk score, immune status and tumor microenvironment

[0075] The body's immune status plays an important role in the occurrence and development of tumors. To evaluate the correlation between DRGs risk score and the immune status of GBM patients, we found a significant correlation between DRGs risk score and the infiltration of various immune cells and stromal cells in three independent cohorts based on the XCELL algorithm ( Figure 6 A), which includes T helper 2 cells (Th2) (r=0.692 in TCGA cohort, Figure 6 B), M1 macrophages (r=0.599 in TCGA cohort, Figure 6 C) etc. Similarly, this risk score is also correlated with the Microenvironment score (r=0.525 in the TCGA cohort, Figure 6 D) and immune score (r=0.448 in TCGA cohort, Figure 6E) was positively correlated. Similarly, we examined the correlation between DRGs risk score and immune checkpoint gene expression in these three glioma datasets. The results showed that the risk score was significantly correlated with the expression of multiple immune checkpoint genes ( Figure 6 F), including SIGLEC7 (r=0.583 in TCGA cohort, Figure 6 G), LILRB2 (r=0.532 in TCGA cohort, Figure 6 H) and HAVCR2 (r = 0.519 in the TCGA cohort, Figure 6 I).

[0076] (7) Risk score is associated with glioma progression

[0077] Furthermore, we evaluated the correlation between DRGs risk score and glioma biological functions. Based on the GSEA algorithm, we analyzed three GBM datasets to analyze the biological functions and signaling pathways associated with DRGs risk score. In the signaling pathway analysis, the results showed that the risk score was associated with multiple cancer-related signaling pathways such as DNA replication, cell cycle, glycosylation, cell adhesion molecules and mismatch repair ( Figure 7 AB). For biological functions, the risk score is associated with immune response activation, DNA replication, cell adhesion regulation, and immune response activation ( Figure 7 CD). Based on correlation analysis, we also found significant correlations between DRGs risk scores and cancer gene expression in the three datasets ( Figure 7 E), including PAX3, FND3B, ECT2, and CDKN3. In addition, to explore the relationship between risk scores and anti-tumor drug sensitivity and better establish treatment strategies for GBM patients, we calculated the sensitivity of each sample in the three cohorts to multiple anti-tumor drugs based on the Oncopredict package. Correlation analysis showed that the risk score was significantly correlated with the sensitivity of these drugs ( Figure 7 F), including dashatinib, 5-fluorouracil, linsitinib, etc.

[0078] Obviously, the above embodiments are merely examples for clarity of explanation and are not intended to limit the implementation methods. Those skilled in the art will appreciate that other variations or modifications can be made based on the above description. It is not necessary and impossible to enumerate all implementation methods here. Obvious variations or modifications arising therefrom remain within the scope of protection of the present invention.

Claims

1. A glioma prognosis risk prediction model, characterized in that: The glioma prognosis risk prediction model evaluates the predictive performance of glioma prognosis based on the disulfide death-related gene DRGs risk score. The DRGs risk score conforms to the following formula: ， Where Exp is the expression value of each DRGs in the model, coef is the regression coefficient of each DRGs in the model, i represents the index of DRGs in the model, and n represents the number of DRGs included in the model. DRGs include IL17RC gene, KIF18A gene, TFPI gene, ISG20 gene, HOXA2 gene, FRMPD1 gene, P2RX6 gene, PELI2 gene and TXN2 gene.

2. The glioma prognosis risk prediction model according to claim 1, characterized in that: Contain at least one of the following characteristics: (1) The Gene ID of the IL17RC gene is 84818, the Gene ID of the KIF18A gene is 81930, the Gene ID of the TFPI gene is 7035, the Gene ID of the ISG20 gene is 3669, the Gene ID of the HOXA2 gene is 3199, the Gene ID of the FRMPD1 gene is 22844, the Gene ID of the P2RX6 gene is 9127, the Gene ID of the PELI2 gene is 57161, and the Gene ID of the TXN2 gene is 25828; (2) The regression coefficients corresponding to the IL17RC gene, KIF18A gene, TFPI gene, ISG20 gene, HOXA2 gene, FRMPD1 gene, P2RX6 gene, PELI2 gene, and TXN2 gene were 0.305, 0.269, 0.156, 0.127, 0.068, -0.101, -0.107, -0.332, and -0.711, respectively.

3. The method for constructing a glioma prognosis risk prediction model according to claim 1 or 2, characterized in that: The following steps are involved: S1. Obtain prognostic samples from glioma patients and control prognostic samples, analyze significantly differentially expressed genes with high prognostic correlation, and obtain DRGs, wherein the DRGs are IL17RC gene, KIF18A gene, TFPI gene, ISG20 gene, HOXA2 gene, FRMPD1 gene, P2RX6 gene, PELI2 gene, and TXN2 gene; S2. Multivariate Cox regression analysis was performed on DRGs to calculate the patient's DRGs risk score based on the regression coefficient and expression level of each gene; S3. Evaluate the predictive performance of the prognostic risk prediction model based on the DRGs risk score.

4. A glioma prognosis risk prediction system, characterized in that: The prediction system comprises the glioma prognosis risk prediction model according to claim 1 or 2, and a reagent for detecting the DRGs expression level. The DRGs include IL17RC gene, KIF18A gene, TFPI gene, ISG20 gene, HOXA2 gene, FRMPD1 gene, P2RX6 gene, PELI2 gene and TXN2 gene.

5. Use of the glioma prognosis risk prediction model according to claim 1 or 2 or the glioma prognosis risk prediction system according to claim 4 in the preparation of a glioma prognosis detection product.

6. A biomarker for predicting the prognosis risk of glioma, characterized in that: The biomarker is a combination of IL17RC gene, KIF18A gene, TFPI gene, ISG20 gene, HOXA2 gene, FRMPD1 gene, P2RX6 gene, PELI2 gene and TXN2 gene.

7. Use of a reagent for detecting the expression level of the biomarker according to claim 6 in the preparation of a glioma prognosis risk detection product.

8. A detection kit, characterized in that: The detection kit contains reagents for detecting the expression level of the biomarker according to claim 6.

9. The detection kit according to claim 8, characterized in that The reagents for detecting the expression levels of biomarkers include primers for amplifying IL17RC gene, KIF18A gene, TFPI gene, ISG20 gene, HOXA2 gene, FRMPD1 gene, P2RX6 gene, PELI2 gene and TXN2 gene.

10. The detection kit according to claim 8, characterized in that The testing steps include: (1) Extract genomic DNA from the sample to be tested and quantify the IL17RC gene, KIF18A gene, TFPI gene, ISG20 gene, HOXA2 gene, FRMPD1 gene, P2RX6 gene, PELI2 gene, and TXN2 gene; (2) Establishing the glioma prognosis risk prediction model according to claim 1 or 2 based on the gene expression level to determine the prognosis of glioma.

Images