Construction of m6A-Related lncRNA Prognostic Signature Model and Immunomodulatory Effect in Glioblastoma Multiforme

Abstract

Background: Glioblastoma multiforme (GBM), the most prevalent and aggressive of primary malignant central nervous system tumors (grade IV), has a poor clinical prognosis. This study aimed to assess and predict the survival of GBM patients by establishing an m6A-related lncRNA signaling model and to validate its validity, accuracy and applicability.

Methods: RNA sequencing data and clinical data of GBM patients were obtained from TCGA data. First, m6A-associated lncRNAs were screened and lncRNAs associated with overall survival in GBM patients were obtained. Subsequently, the signal model was established using LASSO regression analysis, and its accuracy and validity are further verified. Finally, GO enrichment analysis was performed, and the influence of this signature on the immune regulation response and anticancer drug sensitivity of GBM patients was discussed.

Results: The signature constructed by four lncRNAs AC005229.3, SOX21-AS1, AL133523.1, and AC004847.1 is obtained. Furthermore, the signature proved to be effective and accurate in predicting and assessing the survival of GBM patients and could function independently of other clinical characteristics (Age, Gender and IDH1 mutation). Finally, Immunosuppression-related factors, including APC co-inhibition, T-cell co-inhibition, CCR and Check-point, were found to be significantly up-regulated in GBM patients in the high-risk group. Some chemotherapeutic drugs (Doxorubicin and Methotrexate) and targeted drugs (AZD8055, BI.2536, GW843682X and Vorinostat) were shown to have higher IC50 values in patients in the high-risk group.

Conclusion: We constructed an m6A-associated lncRNA risk model to predict the prognosis of GBM patients and provide new ideas for the treatment of GBM. Further biological experiments can be conducted on this basis to validate the clinical value of the model.

Keywords: Glioblastoma multiforme; Immunotherapy; LncRNAs; Prognosis; m6A.

Figure 1

Identification of m6A-related lncRNAs in GBM. (A) Flow chart of this study. (B) Sankey diagram showed the expression correlation between m6A-related genes and m6A-related lncRNAs in GBM (|r|>0.4, P < 0.001).

Figure 2

The prognostic risk model established by LASSO regression analysis. (A) Forest plot shows m6A-related lncRNAs that influence GBM patients’ OS screened by univariate Cox regression analysis (P < 0.05). (B) The tuning parameters of OS-related proteins to cross-verify the error curve. (C) Perpendicular imaginary lines to calculating the minimum criteria. (D) Heat map of expression correlation between 4 lncRNAs involved in model construction and m6A-related genes. (E) Kaplan-Meier curves showed differences in overall survival of GBM patients with high-risk and low-risk in the training group or the testing group (P < 0.001). *: P<0.05, **: P<0.01, ***: P<0.001.

Figure 3

Verification of the signal model in the training and the testing groups. (A, D) Distribution of the risk score of each patient in the training group (A) and the testing group (D) ranked by risk score from lowest to highest. (B, E) Distribution of the survival status of each patient in the training group (B) and the testing group (E) ranked by risk score from lowest to highest. (C, F) Expression of the four m6A-related lncRNAs in the high-risk and low-risk GBM patients in the training group (C) and the testing group (F) ranked by risk score from lowest to highest.

Figure 4

Independent Prognostic Analysis and accuracy verification of the signal model. (A, B) Univariate and multivariate independent prognostic analysis of risk score and clinical variables. (C) Time-dependent ROC curves to evaluate the accuracy of risk scores for predicting 1-year, 3-year, and 5-year survival. (D, E) Time-dependent ROC curves and C-index curves assess the accuracy of risk scores, age, and gender for predicting GBM patients’ survival.

Figure 5

Nomogram and clinical grouping verification of the signal model. (A) The risk score, age, and gender were combined to construct a Nomogram to predict the 1-year, 3-year, and 5-year survival probabilities of GBM patients. (B) The calibration curve was used to evaluate the accuracy of the Nomogram. (C) In different clinical groups (age: >65 or ≤65, gender: female or male), the consistency of the model to predict OS was verified. (D) Principal Component Analysis is used to evaluate and compare the discrimination of all genes, m6A-related genes, m6A-related lncRNAs, and model lncRNAs between high-risk and low-risk GBM patients.

Figure 6

GO enrichment analysis and immune regulation of the signal model. (A) GO enrichment analysis of differentially expressed genes in high-risk and low-risk groups. (B) The heat map showed immune function analysis of high and low-risk groups. (C) Violin Plot showed the difference in TIDE scores between the high-risk and low-risk groups. *: P<0.05, **: P<0.01, ***: P<0.001.

Figure 7

Sensitivity analysis of anti-tumor drugs based on the signal model. (A) Differences in the sensitivity (IC50 value) of broad-spectrum anticancer drugs between GBM patients in the high-risk and low-risk groups. (B) Differences in the sensitivity (IC50 value) of targeted anticancer drugs between GBM patients in the high-risk and low-risk groups.

Figure 8

Gene mutation frequency of the signal model. (A, B) Top-20 gene mutation frequency in high (A) and low (B) risk groups. (C) Violin Plot showed the difference in the tumor mutation burden (TMB) between the high-risk and low-risk groups. (D) Survival curves of the high-TMB group and low-TMB group. (E) Survival curves of the high and low-TMB groups and high- and low-risk groups.