A Focal Adhesion-Related Gene Signature Predicts Prognosis in Glioma and Correlates With Radiation Response and Immune Microenvironment

Abstract

Background: Glioma is the most frequent brain malignancy presenting very poor prognosis and high recurrence rate. Focal adhesion complexes play pivotal roles in cell migration and act as hubs of several signaling pathways.

Methods: We used bioinformatic databases (CGGA, TCGA, and GEO) and identified a focal adhesion-related differential gene expression (FADG) signature by uniCox and LASSO regression analysis. We calculated the risk score of every patient using the regression coefficient value and expression of each gene. Survival analysis, receiver operating characteristic curve (ROC), principal component analysis (PCA), and stratified analysis were used to validate the FADG signature. Then, we conducted GSEA to identify the signaling pathways related to the FADG signature. Correlation analysis of risk scores between the immune checkpoint was performed. In addition, the correlation of risk scores and genes related with DNA repair was performed. CIBERSORT and ssGSEA were used to explore the tumor microenvironment (TME). A nomogram that involved our FADG signature was also constructed.

Results: In total, 1,726 (528 patients diagnosed with WHO II, 591 WHO III, and 603 WHO IV) cases and 23 normal samples were included in our study. We identified 29 prognosis-related genes in the LASSO analysis and constructed an eight FADG signature. The results from the survival analysis, stratified analysis, ROC curve, and univariate and multivariate regression analysis revealed that the prognosis of the high-risk group was significantly worse than the low-risk group. Correlation analysis between risk score and genes that related with DNA repair showed that the risk score was positively related with BRCA1, BRCA2, RAD51, TGFB1, and TP53. Besides, we found that the signature could predict the prognosis of patients who received radiation therapy. SsGSEA indicated that the high-risk score was positively correlated with the ESTIMATE, immune, and stromal scores but negatively correlated with tumor purity. Notably, patients in the high-risk group had a high infiltration of immunocytes. The correlation analysis revealed that the risk score was positively correlated with B7-H3, CTLA4, LAG3, PD-L1, and TIM3 but inversely correlated with PD-1.

Conclusion: The FADG signature we constructed could provide a sensitive prognostic model for patients with glioma and contribute to improve immunotherapy management guidelines.

Keywords: focal adhesion; glioma; immune checkpoints; prognosis; radiation response; tumor microenvironment.

Figure 1

Flow chart of our study.

Figure 2

Identification of candidate genes and LASSO-COX analysis (A) Volcano plot of differentially expressed genes in glioma from the GEO database. (B) Heatmap of differentially expressed genes in glioma from the GEO database. (C) Univariate regression analysis of FADGs. (D) Venn plot of differentially expressed genes and focal adhesion related genes. (E) Protein-protein interacting network. (F, G) Construction and validation of focal adhesion related signature. (H) Multivariate cox regression analysis of eight candidate genes.

Figure 3

Analysis of the CGGA training cohort. (A) Cutoff of low-and high-risk patients (B) Heatmap of expression of candidate genes. (C) Survival status of low-and high-risk patients. (D) Survival analysis of low- and high-risk patients. (E) PCA of low-and high-risk patients. (F) ROC of 1-, 3-, and 5-year OS.

Figure 4

Analysis of the testing cohort and validation cohort. (A, B) Survival status of low-and high-risk patients in the CGGA testing and TCGA cohorts. (C, D) Survival analysis of low- and high-risk patients in the CGGA testing and TCGA cohorts. (E, F) PCA of low-and high-risk patients in the CGGA testing and TCGA cohorts. (G–H) ROC of 1-, 3-, and 5-year in the CGGA testing and TCGA cohorts.

Figure 5

(A–O) Stratified survival analysis of low- and high-risk patients in the CGGA database, by age, gender, grade, 1p19q codeletion, IDH mutant, PRS, and MGMT status.

Figure 6

Related pathway were analyzed by GSEA. (A) Hallmark analysis of our signature (B) KEGG analysis of our signature. (C) GO analysis of our gene signature.

Figure 7

Correlation of risk score and radiation response genes. (A) Correlation of risk score and BRCA1. (B) Correlation of risk score and BRCA2. (C) Correlation of risk score and RAD51. (D) Correlation of risk score and TGFB1. (E) Correlation of risk score and TP53. (F) Risk score of the CR and PD groups.

Figure 8

Relationship between immune cell infiltration and risk score. (A) Heatmap of ssGSEA and correlation between risk and ESTIMATE, immune, and stromal scores, and tumor purity. (B) Correlation of risk and ESTIMATE scores. (C) Correlation of risk and stromal scores. (D) Correlation of risk and immune scores. (E) Correlation of risk and tumor purity scores.

Figure 9

Correlation analysis of risk score and immune checkpoints. (A) Correlation analysis of risk score and B7-H3 levels. (B) Correlation analysis of risk score and CTLA4 levels. (C) Correlation analysis of risk score and LAG3 levels. (D) Correlation analysis of risk score and gene PD-1 levels. (E) Correlation analysis of risk score and PD-L1 levels. (F) Correlation analysis of risk score and gene TIM3 levels.

Figure 10

Nomogram was constructed with the risk score and clinical characteristics (A) Nomogram of the clinical features and risk score. (B) Calibration curve of the actual 3-year OS. (C) DCA analysis of the FADGs signature.