Introduction
Glioblastoma multiforme (GBM) is a devastating brain tumor with extremely poor prognosis because of its diffusive and infiltrative nature, which is marked cytological heterogeneity. It is one of the most aggressive and common malignant brain tumors accounting for more than 50% of all gliomas [
1]. GBM tumors are histologically and molecularly diverse, exhibiting heterogeneity both between patients and within individual tumors [
2]. Several distinct GBM subtypes have been identified based on the gene expression-based molecular classification. Three glioblastoma subtypes were defined based on patient prognosis and gene expression clustering; proneural, proliferative, and mesenchymal subtypes [
3]. Subsequently, gene expression studies from The Cancer Genome Atlas (TCGA) dataset defined four distinct glioblastoma subclasses, including proneural, neural, classical, and mesenchymal [
4]. Kim et al. classified GBMs into three prognostic groups, which were different from the previously identified subtypes, and identified a 42 probe set of gene signatures, which associated with tumor aggressiveness and related with the epithelial-mesenchymal transition (EMT) process [
5].
Currently, the standard treatment for GBM includes surgical resection followed by radiation therapy, chemotherapy, or a combination of these therapies. Despite the use of these therapies, the median survival time in patients remains dismal, at approximately 1 year. GBM remains an incurable disease with few therapeutic advances over the past several decades. Novel approaches are under development targeting glioma cell surface antigens and receptors, tumor invasion, angiogenesis, proliferation, immune escape, and tumor recurrence [
6]. A relatively new form of therapy involves tumor targeting, which relies on the identification of unique or over-expressed cell surface receptors or antigens on tumor cells. One of the extensively studied cell-surface targets is interleukin-13 receptor alpha-2 chain (IL-13Rα2). This receptor is one of the two chains of the IL-13R complex [
7]. IL-13Rα2 binds IL-13 with high affinity and has received significant attention in brain tumor therapy because it is overexpressed by high-grade glioblastomas, but not expressed at significant levels by normal brain tissue [
6‐
8]. In addition, it has been shown that IL-13Rα2 promotes tumor invasion and metastasis in mouse models of human pancreatic and ovarian cancers [
9,
10]. It is also shown that IL-13Rα2 can protect tumor cells from apoptosis thereby contributing to tumor growth [
11]. The IL-13Rα2 chain has become a new and attractive target in immunotherapy using monoclonal antibodies [
12], IL-13Rα2-peptide pulsed dendritic cells [
13], and IL-13R-targeted chimeric antigen receptor modified T cells [
14‐
16]. To target IL-13R, we have developed a recombinant fusion protein composed IL-13 and a mutated form of
Pseudomonas exotoxin (IL-13-PE) [
17]. The IL-13-PE was found to be highly selective and potent in killing human GBM cells in vitro and in animal models of glioma tumors [
18‐
22]. Based on these preclinical studies, several Phase I/II clinical trials targeting IL-13Rα2 in GBM by IL-13-PE were undertaken (
https://clinicaltrials.gov). A randomized controlled Phase 3 clinical trial was completed [
23] and additional clinical trials are planned.
Despite advances in the understanding of IL-13Rα2 biology in glioma tumors and clinical trials targeting this receptor for therapy, the functional significance of
IL-13Rα2 expression in malignant glioblastoma is not well understood. We and others have shown that IL-13Rα2 may be associated with the increase in glioma malignancy grade and associated with poor patient prognosis [
8,
24,
25]. In order to demonstrate a possible correlation between
IL-13Rα1 and
α2 expression in GBM with the clinical outcomes, we analyzed datasets publicly available at NCI’s TCGA database, which was established by NCI/NIH to generate the comprehensive catalog of genomic abnormalities (
https://tcga-data.nci.nih.gov/tcga/). The TCGA data provided detailed clinical information of a large number of GBM patients. These datasets were downloaded and an association between expression of
IL-13Rα2 and clinical outcomes in GBM patients was studied. We also examined a possible association between expression of
IL-13Rα1 and clinical outcomes in GBM patients. Our analysis found that the level of
IL-13Rα1 and
α2 expression is associated with poor patient survival, particularly long-term survival and GBM recurrence. Furthermore, some immune regulatory genes seem to be associated with
IL-13Rα2 expression. Our findings have important implications in the understanding of the role of IL-13R in pathogenesis and evaluating possible therapeutic interventions for patients suffering from GBM.
Discussion
We demonstrate that GBM tumors can be classified into three different distinct groups based on the analysis of gene expression data from 428 glioma subjects at the TCGA database for IL-13R (α1 and α2) gene expression. Group I tumors do not express IL-13Rα1 and α2 mRNA; group II tumors expressed mild to moderate levels of IL-13Rα1 and α2 mRNA, while groups III tumors expressed high level of IL-13Rα1 and α2 mRNA. A large % of GBM samples (76%) expressed mild to moderate levels (Log2 > 0 to < 2) of IL-13Rα1, while 28% samples expressed mild to moderate levels of IL-13Rα2 mRNA. More than 42% of GBM samples were highly positive for IL-13Rα2 mRNA (Log2 ≥ 2) while only 16% samples were highly positive for IL-13Rα1 mRNA. Patients with group III tumors had the shortest overall survival irrespective of treatment compared to group I and group II patients. Thus overexpression of IL-13Rα1 and α2 gene expression in tumors is associated with decreased patient survival and poor patient prognosis. In addition, patients with highest expression of both IL-13Rα1 and α2 mRNA in tumors showed poorest survival in GBM patients. IL-13Rα2 mRNA expression was confirmed by RNA-seq technology. These results indicate that IL-13R expression in glioma tumors is associated with poor patient prognosis and it is possible that IL-13Rs are prognostic indicator for GBM. Furthermore, when IL-13Rα2 expression was compared between lower-grade glioma and high grade GBM, both IL-13Rα1 and IL-13Rα2 gene expression levels were significantly lower in lower-grade glioma compared with GBM. These results suggest that IL-13Rα2 gene expression may also be associated with GBM malignancy grade.
Temozolomide-based therapy is the standard of care for patients with GBM. However, the efficacy of standard temozolomide chemotherapy and radiation therapy for patients with GBM is compromised by resistance to these therapies. In vitro studies demonstrated that
CD133 positive GBM cells show strong tumor’s resistance to chemotherapeutic agents, including TMZ [
36]. It was found that expression levels of
CD74 in high grade gliomas were inversely associated with TMZ resistance in GBM xenograft lines, suggesting a role in TMZ resistance [
37]. In addition, the resistance to TMZ has been shown to be modulated by the DNA repair protein
O6-methylguanine-DNA methyltransferase (
MGMT). It has been shown that elevated
MGMT protein levels or lack of
MGMT promoter methylation is associated with TMZ resistance in some, but not all GBM tumors [
38]. Although DNA methylation of the MGMT gene promotor was shown to be a prognostic marker for treatment response of temozolomide in GBM [
39], Brennan et al. reported that
MGMT status distinguishes responders from non-responders to TMZ only among samples classified as classical subtype of GBM (n = 96), but not among other samples classified as proneural, mesenchymal, and neural subtypes of GBM (a total of n = 225) [
40]. Their data indicate that
MGMT DNA methylation can only be used as a prognostic marker for the classical subtype of GBM, but not for any other subtypes of GBM [
40]. In our study, we observed for the first time that the patients with over expressed
IL-13Rα2 treated with TMZ chemotherapies had shorter overall survival time compared with the patients with
IL-13Rα2 negative expression treated with TMZ, implicating that
IL-13Rα2 mRNA expression is associated with GBM resistance to TMZ chemotherapy. In addition, we did not find any correlation of
IL-13Rα2 mRNA expression with the
MGMT expression (Supplementary Table S2), indicating that the influence of
IL-13Rα2 expression on TMZ response was independent of the expression of
MGMT. These data suggest that
IL-13Rα2 may be a new modulator of TMZ response, representing a distinct mechanism of TMZ resistance from
MGMT.
Targeting
IL-13Rα2 has motivated the development of highly effective therapies and novel administration strategies. So far, a total of six clinical trials using IL-13-PE in patients with various malignant gliomas have been completed in the United States (
https://clinicaltrials.gov). Early clinical trials (Phase I) targeting
IL-13Rα2, by IL-13-PE38QQR via CED in combination with the current standard of care (surgery, radiotherapy, and temozolomide) showed promising safety and efficacy profiles (
http://clinicaltrials.gov). A completed phase III randomized clinical trial of convection enhanced delivery (CED) of IL-13-PE38QQR vs. an FDA approved drug carmustine-releasing Gliadel wafers (GW) for recurrent glioblastoma, showed that IL-13-PE was well tolerated, but it did not show superiority over GW in overall survival. Retroactive data analysis of time-to-progression was significantly higher with IL-13-PE compared to GW [
23]. However, tumor specimens from the original surgery were not evaluated for the presence of IL-13 receptors in the enrolled patients in this trial. The outcome could potentially be improved if enrolled patients could be limited only to ones with over expressed IL-13Rα2 in tumor tissue. Furthermore, IL-13-PE treatment data in the TCGA dataset showed that IL-13-PE treated patients had a longer median survival (657 days) compared with median survival of the reference patient group without IL-13-PE treatment (384 days) (Supplementary Fig S8). Although the sample size of IL-13-PE treated patients in TCGA was extremely small (n = 6), these preliminary results have significant implications and indicate that targeting of
IL-13Rα2 in GBM treatment holds a promise, specifically for the
IL-13Rα2 positive group of patients.
The mechanism of poor survival of patients with GBM tumors expressing high levels of
IL-13Rα1 and
IL-13Rα2 is not clear.
IL-13 and
IL-13Rα2 has been shown to be involved in immune evasion and tolerance mechanisms and thus it is possible that high
IL-13Rα1 and
IL-13Rα2 expression participates in systemic profound immunosuppression seen in GBM patients [
41]. In that regard, we found that several immunosuppressive genes were highly expressed in
IL-13Rα2 over expressed tumors, but not in
IL-13Rα2 negative tumors. These genes included
CCL2, which is known to attract MDSCs in cancer microenvironment. MDSCs represent one of the most important players mediating immunosuppression. These cells may not only inhibit an anti-tumor immunity but also directly stimulate tumorigenesis as well as tumor growth and expansion [
35]. MDSCs reduce antigen specific CD8 + T cell proliferation, increase T-cell death by apoptosis, and change the profile of cytokines secreted by activated T lymphocytes [
42]. It is possible that targeting MDSCs will have a favorable outcome in patients with GBM. A better understanding of the contribution of the tumor on systemic immune suppression is necessary for improved therapies, to monitor negative effects of novel treatments, and to improve patient outcomes.
Recently, various gene expression studies have identified gene signatures that are associated with various types of GBM as well as signatures that correlate with survival. Kim et al. identified 42 probe sets that show an association with tumor aggressiveness and patient survival [
5]. Verhaak et al. identified a gene signature associated with four subtypes (proneural, neural, classical, and mesenchymal) of GBM. An 840 gene signature (210 genes per class) was established. Each of the signatures was highly distinctive [
4]. These defined subtypes differ by the type of genetic abnormalities they carried and by the patient’s clinical characteristics. A high level of
EGFR expression and
EGHR amplification were mainly observed in the classical subtypes. The
IDH1 and
TP53 mutations were significantly frequent events in the proneural subtype.
IDH1 somatic mutation has been linked to a glioma-CpG island methylator phenotype (G-CIMP) [
43]. Turcan et al., have demonstrated that
IDH1 mutation is the cause of CIMP and leads to CIMP phenotype, and is sufficient to establish the glioma hypermethylator phenotype [
44]. G-CIMP tumors belong to the proneural subgroup in GBM and are more prevalent among lower-grade gliomas [
43,
45]. In addition,
PDGFRA was another gene, which was mutated and highly expressed only in the proneural subtype [
4]. Brown et al. reported that high
IL-13Rα2 gene expression is positively correlated with the mesenchymal signature gene expression and negatively correlated with the proneural signature gene expression [
24]. They further showed that
IL-13Rα2 expression is correlated, but not limited to the expression of mesenchymal signature genes. In our study, we did not find any correlation between
IL-13Rα2 mRNA expression and previously reported biomarkers of GBM subtypes such as IDH1, EGFR, MGMT, and PDGFRA (Supplementary Table S2). Furthermore, we also found that high expression of
IL-13Rα1 and
α2 genes are associated with poor prognosis in
IDH1-Wt/non-G-CIMP GBM (Supplementary Fig S9a and b). Our Data shows that approximately 70% GBM tumors express moderate to high level of
IL-13Rα2 mRNA, which is expressed in most of the subtypes identified by Verhaak et al. [
4]. Our recent studies in animals confirm our observations that
IL-13Rα2 is involved in tumor invasion, metastasis and poor survival of animals implanted with human pancreatic and ovarian cancers [
9,
10]. These results indicate that a single gene (
IL-13Rα2) may provide a stronger correlation with survival than a group of genes previously identified, thus making
IL-13Rα2 an important target for glioma therapy.
It is of interest to note that mRNA for three chains of IL-13 receptor (
IL-13Rα2, IL-13Rα1 and
IL-4Rα) are expressed at different levels in GBM samples.
IL-13Rα2 was highly expressed in more than 42% of total GBM samples, compare with that of
IL-13Rα1 only highly expressed in less than 16% of the total GBM. We did not find any correlation between the expression of
IL-13Rα2 and
IL-13Rα1, nor between
IL-13Rα2 and
IL-4Rα mRNA. These analyses indicate that over expression of
IL-13Rα2 mRNA in GBM is independent from expression of
IL-13Rα1 or
IL-4Rα suggesting that IL-13Rα2 does not seem to form a complex with either IL-13Rα1 or IL-4Rα chain. This is consistent with our previous work that summarized by Suzuki et al. and Joshi et al. [
29,
46]. In contrast, the expression of IL-13Rα1 mRNA positively correlated with IL-4Rα indicates that these two receptor chains form a type II IL-13R complex in GBM. Indeed, this receptor has been shown to mediate IL-13 signaling in cancer cells [
7,
10,
29].
In conclusion, we have found that high IL-13Rα1 and IL-13Rα2 mRNA expression is associated with poor patient prognosis, demonstrating an inverse relationship between IL-13R expression and overall survival. The similar inverse relationship seems to be also associated with days to GBM recurrence and long-term patient survival. Furthermore, we show for the first time that IL-13Rα2 expression is associated with GBM resistance to TMZ chemotherapy. These findings have important implications in understanding a possible role of IL-13R in GBM pathogenesis, development of targeted therapies, and define a patient population for immunotherapy or alternative therapies in clinical trials. Additional studies are ongoing to further confirm our observations.