Background
Diffuse gliomas are one of the most common primary neoplasms of the central nervous system, accounting for approximately 81% of all malignant brain tumors, leading to a high rate of mortality and morbidity [
1,
2]. These aggressive and heterogeneous tumors are generally associated with poor outcomes, due to their complexity and resistance to therapeutic approaches [
3].
In the last years, improvements in molecular techniques have been important tools to update the knowledge about the genetic profile of gliomas. These progresses, led in 2016, the World Health Organization (WHO) classification of Central Nervous System Tumors to include Isocitrate dehydrogenase (
IDH) mutations and 1p/19q codeletion as central biomarkers for the diagnosis of diffuse gliomas [
4]. This new classification breaks the principle of diagnosis based exclusively on microscopy, allowing a more accurate determination of the patient’s prognosis [
4,
5]. Nevertheless, this new classification has limitations to characterize these heterogeneous tumors. New biomarkers for diagnostic, prognostic and response to therapy are a major concern for the management of patients with gliomas [
6]. In this context, different potential biomarkers for diffuse gliomas have been proposed, such as:
TERT (telomerase reverse transcriptase) promoter mutations, amplification/mutations in
EGFR (epidermal growth factor receptor) gene, mutations/deletions in
PTEN (phosphatase and tensin homologue) and
MGMT (O-6-methylguanine-DNA methyltransferase) promoter methylation.
TERT promoter mutations are present in a high percentage of gliomas (80–90%), which makes it an interesting target gene to be studied [
7]. This gene encodes the catalytic subunit of telomerase, an enzyme that maintains the length of telomeres during cell division [
8]. In addition,
TERT promoter mutations are associated with increased levels of telomerase activity allowing the indefinite proliferation of tumor cells [
8‐
10]. The amplification of
EGFR was identified in approximately 40–50% of all cases of glioblastoma (GBM), 2007 WHO grade IV, the most malignant of diffuse gliomas [
11,
12]. This molecular alteration determines the over-activation of an important signaling pathway, phosphatidylinositol-3-kinase - protein kinase B (PI3K-AKT), which regulates a wide range of cellular processes such as cell proliferation, migration, angiogenesis, differentiation and apoptosis [
13].
PTEN deletion is present in approximately 30–40% of GBM [
14,
15], however there is no unanimity regarding the prognostic value of this alteration in diffuse gliomas [
16,
17], as well as, regarding
TERT promoter mutations [
18‐
20] and
EGFR amplification [
21‐
23].
MGMT promoter methylation has been described as a predictive biomarker in GBM with benefit from chemotherapy based on temozolomide [
24‐
26]. Moreover, this benefit is higher in patients with
IDH-wildtype gliomas, particularly in old patients (aged ≥70 years) [
3,
27,
28].
MGMT promoter methylation is predominant in
IDH-mutant gliomas, representing a favorable prognostic factor, although this biomarker is not associated with the benefit from either temozolomide or radiotherapy in this molecular subgroup [
27].
Currently, these genes are not included in the 2016 WHO classification of diffuse gliomas, although these genetic alterations could be relevant in the diagnostic routine, patient management and on the choice of the treatments [
4,
29].
In the present study, we aimed to reclassify a 444 cohort of diffused gliomas based on the 2016 WHO classification of Central Nervous System Tumors. Subsequently, we used this reclassified cohort to evaluate the impact of TERT promoter mutations, PTEN deletion, EGFR amplification and MGMT promoter methylation in diagnosis, prognosis and response to therapy.
Discussion
In the present study we evaluated the impact of the new 2016 WHO classification of Central Nervous System Tumors in a 444 diffuse gliomas cohort, previously classified according to the 2007 WHO classification based on histological features.
Our results showed a decrease in the percentage of oligodendrogliomas, from 18.5% of the samples previously diagnosed using the histological classification, to 11% of the samples according to the new classification. On the other hand, there was an increase in the percentage of astrocytomas (from 11 to 22.1% of the samples). This main alteration in glioma subgroups was associated with the introduction of 1p/19q codeletion and
IDH status, which were decisive in the subdivision of astrocytoma and oligodendroglioma as well as the disintegration of the oligoastrocytoma group. These results are in accordance with the study of Iuchi et al., which reported astrocytoma and oligodendroglioma subgroups as the main targets of the 2016 WHO classification effect [
32]. However, according to Tabouret and co-authors, the reclassification of the French cohort showed a similar frequency of oligodendrogliomas before and after the reclassification of gliomas (31.6–34.5%, respectively), while the number of GBM (33.8–50.3%) and astrocytomas (7–16.2%) increased [
33]. The differences observed between our study and the French cohort, could be related with the reclassification of oligoastrocytomas, since in our study most oligoastrocytomas were reclassified as astrocytoma (
n = 35), while in the study of Tabouret et al. the vast majority of oligoastrocytomas were considered GBM [
33].
Even with the introduction of molecular biomarkers, the distribution of patients previously diagnosed with oligoastrocytomas remains a difficult task, which is demonstrated by the variability between studies [
32,
33]. In our study 10 samples of oligoastrocytomas were included into the NOS subgroup. The analysis of alpha thalassemia/mental retardation syndrome X-linked (
ATRX) loss and tumor protein 53 (
TP53) mutations was not performed, which constitutes a limitation of this study. The analysis of these both biomarkers is suggested in the 2016 WHO classification only in doubtful cases [
4], and they are currently done in our Institute by immunochemistry. Actually, the mutational status of these genes is not determined in diagnosis of gliomas for two main reasons: i) by itself are unable to identify the subtype of glioma sample and because; ii) it is expensive, they constitute long genes, becoming difficult their analysis using the conventional molecular techniques.
In total, 41 samples of our cohort were inserted into the NOS glioma subgroup, highlighting the need for new biomarkers, in order to be possible to classify gliomas with 1p or 19q deletion and gliomas IDH-wildtype with 1p/19q codeletion. Although, it is important to note the most of them are included in this subgroup due to technical issues.
Here, as expected, GBM constituted the most prevalent type of glioma (57.7%), like previously reported by Iuchi et al. (66%), Tabouret et al. (50%) and Ostrom et al. (45%) [
1,
32,
33]. However, GBM
IDH-mutant accounted for only 2.5% of all GBM, slightly less than the 10, 17.2 and 7.8% previously reported [
32‐
34]. In addition, astrocytomas
IDH-wildtype and
IDH-mutant showed similar frequencies (9.7 and 12.4% respectively), slightly different from the results reported by Iuchi and co-authors (13.7% of astrocytomas
IDH-wildtype and 6.7% of astrocytomas
IDH-mutant) and Tabouret et al., (11% astrocytomas
IDH-mutant and 5.3% of astrocytoma
IDH-wildtype) [
32,
33]. The differences detected in astrocytomas diagnosis could be related to their reclassification in GBM, which presently depends only on the histological features of the tumor, introducing some variability between studies.
Notably, 1p/19q codeleted gliomas were the molecular subgroup associated with the longest overall survival (OS:198 months), regardless of whether they were classified as oligodendrogliomas grade II (OS:172 months) or grade III (OS:97 months). These results suggested that 1p/19q codeletion is a strong biomarker of prognosis and even better than histological classification, since could embrace less aggressive tumors. In addition, these statements are in accordance with previous studies indicating 1p/19q codeletion was associated with a better prognosis when compared to non-codeleted tumors [
35,
36].
Furthermore,
IDH mutational analysis divided the astrocytoma group into two subgroups with distinct prognoses (
p < 0.001), as previously reported [
4,
37]. Here,
IDH mutations did not had prognostic impact in GBM (
p = 0.092) (Table
5), which could be explained by the reduced number of GBM
IDH-mutant samples (
n = 11).
We also evaluated the impact of
EGFR amplification,
PTEN deletion,
TERT promoter mutations and
MGMT promoter methylation in the diagnosis, prognosis and response to therapy of patients with diffuse gliomas. As reported by other studies,
EGFR amplification was more common in
IDH-wildtype gliomas [
38‐
40]. To date, most studies evaluated the prognostic value of
EGFR amplification using only the histological diagnosis, instead of the molecular subgroups of gliomas [
21‐
23]. In this work, we reported for the first-time that
EGFR amplification had no significant prognostic value in molecular subgroups of gliomas -
IDH-wildtype GBM and astrocytomas. Nevertheless, we only evaluated the presence of
EGFR amplification and not
EGFR activating mutations. Previously, it was described that tumors with both
EGFRvIII overexpression and
EGFR amplification constitute an indicator of poor prognosis in GBM patients [
23]. The prognostic value of
EGFR was also associated with patient’s age, seeming to be correlated with worse outcomes in younger patients [
41]. However, none of these studies considered the mutational status of
IDH, which means that the effect on prognosis by
EGFR amplification may be dependent of
IDH mutations.
Additionally
, EGFR amplification has been appointed as one of the causes for the development of radio-resistance in gliomas [
42]. Most interestingly, we found that patients with GBM
IDH-wildtype and
EGFR amplification had a significantly better overall survival than those without
EGFR amplification, only when treated with radiotherapy alone, and not when treated with chemo-radiotherapy. At this point, the clinical significance of this finding, and the reasons why it did not occur with chemo-radiotherapy, are not fully understood. However, this result should be validated in other cohorts with a higher number of samples. Nevertheless, our observation is consistent with the previously described in non-small cell lung cancer, where the presence of
EGFR activating mutations and also
EGFR amplification were associated with a radiosensitive phenotype, inducing increased levels of pro-apoptotic proteins and reduced capability to repair DNA [
43‐
45].
The relative frequency and prognostic value of
PTEN deletion in diffuse gliomas were analyzed using histological diagnosis, which explains the variability between the reported studies [
16,
17]. Interestingly, the observed role of
PTEN deletion in prognosis of GBM and astrocytomas
IDH-wildtype has never been documented. In our study,
PTEN deletion was considered a factor of good prognosis in GBM
IDH-wildtype (
p = 0.042), although using a reduced number of samples without
PTEN deletion (
n = 38 vs
n = 184 with
PTEN deleted). Despite this, it was previously noticed that
PTEN loss could be associated with a more favorable prognosis, since it leads to a better response to chemotherapy by compromising homologous recombination of DNA, through the transcriptional regulation of Rad51 [
16,
46]. Another hypothesis to the observed result in our study may be the absence of an inverse correlation between
PTEN expression and AKT activity, as demonstrated in melanoma and breast cancer [
47,
48]. Moreover, this dual effect of
PTEN deletion in prognosis could be related with the specific tyrosine which is the target of
PTEN phosphorylation [
49]. This hypothesis would explain why
PTEN deletion predicts a good outcome in GBM
IDH-wildtype. In contrast, in astrocytomas
IDH-wildtype the deletion of
PTEN is a factor of poor prognosis, as expected, since this is a tumor suppressor gene. Further work should be undertaken to evaluate the mechanisms through which this molecular alteration differentially affects the prognosis of these both groups of gliomas.
In this dataset,
TERT promoter mutations did not have prognostic value in GBM
IDH-wildtype, which is consistent with Nguyen et al. and Eckel-Passow et al. previous studies [
17,
18]. Eckel-Passow et al. reported that
TERT promoter mutations are associated with a poor prognosis in the absence of
IDH mutations in grade II and III gliomas [
18]. Therefore, this study verified that
TERT promoter mutations would be an important biomarker in grade II and III gliomas. However, these results are not according to the 2016 WHO classification. The authors were not considering a subgroup of 1p/19q codeletion and
IDH mutation, as an independent group [
18]. Interestingly in our results 1p/19q codeleted gliomas showed a higher percentage of
TERT promoter mutations – 94%. Therefore, this suggests that the effect observed on overall survival by Eckel-Passow et al., could be associated with the difference between 1p/19q codeleted/
IDH mutant gliomas and astrocytomas independently of
TERT promoter mutations.
MGMT promoter methylation is a biomarker extensively studied in GBM. As previously mentioned, this biomarker is a prognostic factor of prolonged overall survival [
24,
26]. Here,
MGMT methylation was found mainly in
IDH-mutant gliomas, which is in accordance with the literature [
3,
27].
IDH mutations are responsible for increased levels of 2-hydroxiglutarate, which in turn determines the inhibition of several enzymes, such as Jumonji-C domain-containing histone lysine demethylases [
50]. It is already known that
MGMT methylation is associated with better outcomes in both
IDH mutant and
IDH wildtype GBM [
27,
28], although it only constitutes a predictive biomarker for the benefit to temozolomide chemotherapy and not to radiotherapy in patients with GBM-
IDH wildtype [
25]. Interestingly, Rivera et al., reported the predictive value of
MGMT methylation to radiotherapy response in GBM patients, independently of the
IDH mutational status [
51].