Glioma immune microenvironment change during tumor recurrence

Document Type

Conference Proceeding

Publication Date


Publication Title

Cancer Res


Background: Gliomas are the most common malignant brain tumor, have a very aggressive behavior, and invariably relapse and progress. Despite the recent advances, only a few drugs are approved and they present limited success. Currently, there are numerous clinical trials evaluating the efficacy of immunotherapy for gliomas, which are not completed yet. Deciphering the composition of the tumor microenvironment (TME) can have an important and immediate impact on therapeutic interventions and on the development of prognostic and predictive biomarkers for gliomas immunotherapy. To investigate the molecular dynamics over time and in response to therapeutic pressures, the Glioma Longitudinal AnalySiS (GLASS) Consortium, a multinational collaboration, is investigating epigenome-wide molecular data from primary and recurrent matched pairs. Objective: Our aim is to evaluate glioma TME using the deconvolution method methylCIBERSORT applied to DNA methylation data from GLASS. Methods: We generated and validated a customized reference signature defining 10 cell types to predict the relative proportions of immune cell type in the TME of 370 glioma specimens, including 132 longitudinal pairs (initial and recurrent tumors) in association with clinical features (recurrence, survival etc). Results: We found that the TME differs across gliomas of different subtypes. In general, IDHmut subtypes (Codel, GCIMP-high, and GCIMP-low) presented less immune infiltration than IDHwt (Classic-like, Mesenchymal-like, and PA-like). The most abundant estimated infiltrated cell types in IDHmut and IDHwt gliomas were TCD4 cells and macrophages, respectively. Post-treatment (chemo+radiotherapy), we found a decrease of TCD4 and an increase of TCD8 cells in recurrent Codel and G-CIMP-high subtypes; and an increase of macrophages in classic recurrent tumors. High frequency of macrophages and TCD8 cells were associated with poorer overall survival in the IDHwt patients (log-rank p=0.040, hazard ratio (HR) = 1.38; log-rank p=0.046, HR = 2.37, respectively). Conclusions: Using a DNA methylation-based deconvolution approach, we have described the TME of longitudinal gliomas. We found a TME diversity across glioma molecular subtypes and an association with IDH mutation and overall survival. Our findings indicate that the epigenomic deconvolution of TME has a potential therapeutic and prognostic implication to guide the management of patients with gliomas.

PubMed ID

Not assigned.