Tumor antigens and immune subtypes of glioblastoma: the fundamentals of mRNA vaccine and individualized immunotherapy development

Abstract

Purpose: Glioblastoma (GBM) is the most common primary brain tumor in adults and is notorious for its lethality. Given its limited therapeutic measures and high heterogeneity, the development of new individualized therapies is important. mRNA vaccines have exhibited promising performance in a variety of solid tumors, those designed for glioblastoma (GBM) need further development. The aim of this study is to explore tumor antigens for the development of mRNA vaccines against GBM and to identify potential immune subtypes of GBM to identify the patients suitable for different immunotherapies.

Methods: RNA-seq data and the clinical information of 143 GBM patients was extracted from the TCGA database; microarray data and the clinical information of 181 GBM patients was obtained from the REMBRANDT cohort. A GBM immunotherapy cohort of 17 patients was obtained from a previous literature. GEPIA2, cBioPortal, and TIMER2 were used to identify the potential tumor antigens. Immune subtypes and gene modules were identified using consensus clustering; immune landscape was constructed using graph-learning-based dimensionality reduction analysis.

Results: Nine potential tumor antigens associated with poor prognosis and infiltration of antigen-presenting cells were identified in GBM: ADAMTSL4, COL6A1, CTSL, CYTH4, EGFLAM, LILRB2, MPZL2, SAA2, and LSP1. Four robust immune subtypes and seven functional gene modules were identified and validated in an independent cohort. Immune subtypes had different cellular and molecular characteristics, with IS1, an immune cold phenotype; IS2, an immune hot and immunosuppressive phenotype; IS3, a relatively immune cold phenotype, second only to IS1; IS4, having a moderate tumor immune microenvironment. Immune landscape revealed the immune distribution of the GBM patients. Additionally, the potential value of immune subtypes for individualized immunotherapy was demonstrated in a GBM immunotherapy cohort.

Conclusions: ADAMTSL4, COL6A1, CTSL, CYTH4, EGFLAM, LILRB2, MPZL2, SAA2, and LSP1 are the candidate tumor antigens for mRNA vaccine development in GBM, and IS1 GBM patients are best suited for mRNA vaccination, IS2 patients are best suited for immune checkpoint inhibitor. This study provides a theoretical framework for GBM mRNA vaccine development and individualized immunotherapy strategies.

Supplementary information: The online version contains supplementary material available at 10.1186/s40537-022-00643-x.

Keywords: Glioblastoma; Immune subtypes; Individualized immunotherapy; Tumor microenvironment; mRNA vaccines.

Fig. 1

Identification of potential tumor antigens in GBM. a The chromosomal distribution of up- and down-regulated genes in GBM. b Samples overlapping in the altered genome fraction group. c Samples overlapping in the mutation count group. d Genes with highest frequency in the fraction genome altered group. e Genes with highest frequency in the mutation count group

Fig. 2

Identification of tumor antigens associated with GBM prognosis. a Potential tumor antigens (total 1332) with both overexpression and mutation, and tumor antigens significantly associated with OS and RFS (total 9 candidates). Kaplan–Meier curves comparing OS for groups with different expression of ADAMTSL4 (b), COL6A1 (c), CTSL (d), CYTH4 (e), EGFLAM (f), LILRB2 (g), MPZL2 (h), SAA2 (i), and LSP1 (j) in GBM. Red lines represented high gene expression, blue represented low gene expression

Fig. 3

Identification of potential immune subtypes in GBM. Cumulative distribution function curve (a), delta area curve (b) and consensus heatmap (c) based on immune-related gene expression profile in the TCGA cohort. d The distribution of immune subtypes in GBM patients with different IDH1 status in TCGA cohort. e The distribution of immune subtypes in GBM patients with different MGMT status in the TCGA cohort. f The distribution of immune subtypes in GBM patients with different G-CIMP status in the TCGA cohort and the REMBRANDT cohort. g The distribution of immune subtypes in GBM patients with different molecular subtype in the TCGA cohort and the REMBRANDT cohort

Fig. 4

Association between immune subtypes and immunomodulators in GBM. Differences in expression levels of ICP-related genes among GBM immune subtypes in the TCGA cohort (a) and the REMBRANDT cohort (b). Differences in expression levels of ICD-related genes among GBM immune subtypes in the TCGA cohort (c) and the REMBRANDT cohort (d). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Fig. 5

Cellular and molecular characteristic of immune subtypes in GBM. Heatmaps of 28 previously reported immune cell signatures scores among GBM immune subtypes in the TCGA cohort (a) and the REMBRANDT cohort (b). Differences of 28 immune cell signatures scores among GBM immune subtypes in the TCGA cohort (c) and the REMBRANDT cohort (d). e The distribution of GBM four immune subtypes in the pan-cancer immune subtypes. f 21 immune-related molecular signatures with significant differences among GBM immune subtypes. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Fig. 6

The immune landscape of GBM. a The immune landscape of GBM. Each dot represents a patient, and different colors represent different immune subtypes. The horizontal axis represents the principal component 1, and the vertical axis represents the principal component 2. b Correlation between principal component 1/2 and 28 immune cell enrichment scores. c Immune landscape of the subgroups of GBM immune subtypes. Differences of 28 immune cells enrichment scores in the subgroups of IS3 (d) and IS4 (e). Immune landscape of samples from three extreme locations (f) and their prognostic status (g). ⁻p ≥ 0.1, ^·p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001

Fig. 7

Functional immune genes modules in GBM. Heatmaps of four immune subtypes and seven gene modules in the TCGA cohort (a) and the REMBRANDT cohort (c). Genes are ordered based on the gene modules, and patients are arranged based on their immune subtypes. Box plots of the expression patterns of seven gene modules of four immune subtypes in the TCGA cohort (b) and the REMBRANDT cohort (d). ****p < 0.0001

Fig. 8

Gene module characteristics in GBM. a The relationship between gene modules and OS in GBM patients. b Kaplan–Meier curve showing OS analysis of GM2 in the TCGA cohort. GO biological process enrichment analysis of GM2 (c) and GM1 (e). Correlation between GM2 score (d) and GM1 score (f) and principal component 1 in the immune landscape

Fig. 9

The role of immune subtypes in anti-PD-1 treatment cohort. Heatmaps of 28 previously reported immune cells infiltration among GBM immune subtypes in the pre-anti-PD-1 treatment group (a) and post-anti-PD-1 treatment group (b). c Distribution of immune subtypes among responders and non-responders in pre-anti-PD-1treatment samples. d Changes of immune subtypes in paired pre- and post-anti-PD-1 treatment samples. e Kaplan–Meier curve showing OS analysis of immune subtypes (IS1-IS3) in pre-anti-PD-1treatment samples. IS4 had only one sample and therefore was not included in this analysis. f Kaplan–Meier curve showing progression-free survival analysis of immune subtypes in pre-anti-PD-1treatment samples. IS4 had only one sample and therefore was not included in this analysis. Volcano plots showing DEGs between responders and non-responders in pre- (g) and post-anti-PD-1 treatment (i) samples. Red represented DEGs that were upregulated by responders relative to non-responders, while green represented DEGs that were downregulated. GO biological process enrichment analysis of DEGs between responders and non-responders in pre- (h) and post-anti-PD-1 treatment (j) samples