Co-expression prognostic-related genes signature base on propofol and sevoflurane anesthesia predict prognosis and immunotherapy response in glioblastoma

Abstract

Objectives: Anesthetic drugs had been reported may impact the bio-behavior of the tumor. Propofol and sevoflurane are common anesthetics in the operation for glioblastoma (GBM). This study aims to establish a co-expression prognostic-related genes signature base on propofol and sevoflurane anesthesia to predict prognosis and immunotherapy response in GBM.

Method: GPM tissues with different anesthetics gene expression profiles (GSE179004) were obtained from the Gene Expression Omnibus (GEO) database. Core modules and central genes associated with propofol and sevoflurane anesthesia were identified by weighted gene coexpression network analysis (WGCNA) and establish a risk score prognostic model. Immune cell signature analysis in TCGA datasets was predicted via CIBERSORT. At last, serum methylation level of O6-methylguanine-DNA methyltransferase (MGMT) promoter was detected in GPM patient in different time during perioperative period.

Results: The burlywood1 group screened was significantly associated with sevoflurane-treated GBM tissue. 22 independent prognostic differential genes were construct a prognostic-related genes risk score in GBM, and showed good predictive ability. The risk score was strongly correlated with the age of the patients, but not with the sex of the patients. In addition, the differential responses to immunotherapy in high and low risk groups were analyzed, indicating that sevoflurane signature genes were consistent in the classification of gliomas. High-risk patients have high T-cell damage score and are less sensitive to immunotherapy. At last, serum methylation level of MGMT promoter was decreased in GBM patients during propofol and sevoflurane anesthesia.

Conclusions: Propofol and sevoflurane anesthesia associated impact on the gene expression of GBM, included the methylation level of MGMT promoter. Propofol and sevoflurane anesthesia-based risk score prognostic model, which has good prognostic power and is an independent prognostic factor in GBM patients. Therefore, this model can be used as a new biomarker for judging the prognosis of GBM patients.KEY MESSAGESPropofol and sevoflurane anesthesia-based risk score prognostic model has good prognostic power and is an independent prognostic factor in GBM patients.High Propofol and sevoflurane anesthesia-based risk score GBM patients have high T-cell damage scores and are less sensitive to immunotherapy.Serum methylation level of MGMT promoter decrease during propofol and sevoflurane anesthesia in GBM patients.

Keywords: Glioblastoma; MGMT; prognosis; propofol; sevoflurane.

Figure 1.

Co-expression networks were analyzed by WGCNA. (A) Sample dendrogram and trait heatmap. (B) Analysis of the scale-free index for various soft-threshold powers (β). (C) Heatmap of the correlations between the clinical traits and MEs of glioma. (D) Burlywood1 module correlation with INHA.

Figure 2.

Biological characteristics of different clusters in TCGA-GBM patients. (A) Consensus matrix of TCGA-GBM patients by non-negative matrix factorization. (B) Overall survival analyses for different clusters. (C) Progression free survival analyses for different clusters. (D) Gene set enrichment analysis for different clusters.

Figure 3.

PCA scores were associated with TCGA-GBM prognosis. (A) Principal component analysis for TCGA-GBM patients. (B) PCA scores in different clusters. (C) Overall survival analyses for different clusters. (D) Progression free survival analyses for different clusters.

Figure 4.

Generation of a gene expression signature to predict patient survival. (A) The 10-fold cross-validation for variable selection in the LASSO model. (B) The LASSO coefficient profile. (C) ROC curves to evaluate the predictive ability of the risk model in the train cohort. (D) ROC curves to examine the robustness of the risk model in the test cohort. (E) AUC change curves of the train cohort in 3 years. (F) AUC change curves of the test cohort in 3 years.

Figure 5.

Risk score is an independent risk factor for prognosis. (A) Univariate Cox regression analyses of traits. (B) Multivariate Cox regression analyses of traits. (C) Survival analysis of risk scores in train group. (D) Survival analysis of risk scores in test group.

Figure 6.

Prognostic value of the co-expression genes after propofol and sevoflurane anesthesia in GBM. (A) ACOT7; (B) GALE; (C) NUAK2; (D) ACTA1; (E) EEF1B22; (F) NMNAT3; (G) RPL39L; (H) PCDHB3; (I) GUCA1A; (J) MICALL2; (K) SLC35G5; (L) MGMT; (M) TSPAN4; (N) NOL3; (O) NLRP12; (P) RENBP.

Figure 7.

Correlation between risk score and clinical traits and subgroup survival analysis. (A) Correlation between risk score and age. (B) Correlation between risk score and Gender. (C) Survival analysis of risk scores in patients with age > 65. (D) Survival analysis of risk scores in patients with male. (E) Survival analysis of risk scores in patients with age< =65. (F) Survival analysis of risk scores in patients with female.

Figure 8.

Immune microenvironment and immunotherapy. (A) Alluvial diagram showing the changes of risk score, cluster, and PCA score. (B) Differences of immune cell infiltration in different risk groups by CIBERSORT. (C) The TIDE values in high and low risk score groups. (D) The likelihood of the clinical response to antiPD1 and anti-CTLA4 therapy for high and low risk score patients from the TCGA cohorts. True represents immunotherapy responders, while false represents immunotherapy nonresponders. (E) Correlation analysis of risk scores and immune-related markers.

Figure 9.

Serum methylation level of MGMT promoter before and after propofol and sevoflurane anesthesia. The serum methylation level of MGMT promoter in TIVA group and INHA group. (ns: no significant difference).