Identification of a Prognostic Microenvironment-Related Gene Signature in Glioblastoma Patients Treated with Carmustine Wafers

Abstract

Despite the state-of-the-art treatment, patients diagnosed with glioblastoma (GBM) have a median overall survival (OS) of 14 months. The insertion of carmustine wafers (CWs) into the resection cavity as adjuvant treatment represents a promising option, although its use has been limited due to contrasting clinical results. Our retrospective evaluation of CW efficacy showed a significant improvement in terms of OS in a subgroup of patients. Given the crucial role of the tumor microenvironment (TME) in GBM progression and response to therapy, we hypothesized that the TME of patients who benefited from CW could have different properties compared to that of patients who did not show any advantage. Using an in vitro model of the glioma microenvironment, represented by glioma-associated-stem cells (GASC), we performed a transcriptomic analysis of GASC isolated from tumors of patients responsive and not responsive to CW to identify differentially expressed genes. We found different transcriptomic profiles, and we identified four genes, specifically down-regulated in GASC isolated from long-term survivors, correlated with clinical data deposited in the TCGA-GBM dataset. Our results highlight that studying the in vitro properties of patient-specific glioma microenvironments can help to identify molecular determinants potentially prognostic for patients treated with CW.

Keywords: carmustine wafers; glioblastoma; patient’ derived in vitro model; transcriptomics; tumor microenvironment.

Figure 1

Comparison of the gene expression profile of GASC-LS (n = 6) versus GASC-SS (n = 5), using a customized RT2 PCR Array. Volcano plot shows the gene expression changes of 78 protein-coding transcripts analyzed using the RT2 Profiler™ PCR Array. The volcano plot highlights significant gene expression changes by plotting the log2 of the fold changes in gene expression on the x-axis versus their statistical significance on the y-axis. The center vertical line indicates unchanged gene expression, while the two outer vertical lines indicate the selected fold regulation threshold (abs (≥2)). The horizontal line indicates the selected p-value threshold (p ≤ 0,05). Genes with data points in the far upper left (down-regulated, green dots) and far upper right (up-regulated, red dots) sections meet the selected fold regulation and p-value thresholds.

Figure 2

Gene expression profiling of the four genes down-regulated in GASC-LS in the TCGA-GBM RNA-seq dataset. Box plots represent the gene expression levels in the TCGA-GBM dataset (n = 163) and in the TCGA-GTEx matched normal samples (n = 207) of the genes with a significant down-regulation in GASC-LS, identified with the RT2 Profiler array. Data were obtained and plotted from the GEPIA web server (|Log2FC| Cutoff: 1; * p-value < 0.01).

Figure 3

Clinical relevance of the 4 genes down-regulated in GASC-LS. The prognostic value of ALPL, GPR68, VGF, and NETO1 was assessed independently in the TCGA-GBM RNA-seq dataset evaluating the Overall Survival (OS) and was represented by Kaplan–Meier plots. Afterward, the analysis was repeated considering the four-genes signature (median expression). Patients were stratified based on the optimal cut-point.