TERT promoter mutations: a novel independent prognostic factor in primary glioblastomas

Abstract

Background

Activating somatic mutations in the promoter region of the telomerase reverse transcriptase gene (TERT) have been detected in several cancers. In this study we investigated the TERT promoter mutations and their impact on patient survival in World Health Organization grade IV glioblastoma multiforme (GBM).

Methods

The TERT core promoter region containing the previously described mutations and a common functional polymorphism (rs2853669) was sequenced in tumors and blood samples from 192 GBM patients. O(6)-methylguanine-DNA methyltransferase (MGMT) promoter methylation status was assessed by pyrosequencing in 177 (92.2%) cases. Relevant clinical data were obtained from a prospectively maintained electronic database.

Results

We detected specific (−124 C>T and −146 C>T) TERT promoter mutations in 143/178 (80.3%) primary GBM and 4/14 (28.6%) secondary GBM (P < .001). The presence of TERT mutations was associated with poor overall survival, and the effect was confined to the patients who did not carry the variant G-allele for the rs2853669 polymorphism. An exploratory analysis suggested that TERT mutations might be prognostic only in patients who had incomplete resections and no temozolomide chemotherapy.

Conclusions

In this study, specific TERT promoter mutations were markers of primary GBM and predicted patient survival in conjunction with a common functional polymorphism. The prognostic impact of TERT mutations was absent in patients with complete resections and temozolomide chemotherapy. If confirmed in additional studies, these findings may have clinical implications, that is, TERT mutations appear to characterize tumors that require aggressive treatment.

Glioblastoma multiforme (GBM), World Health Organization (WHO) grade IV, is a highly malignant and regrettably frequent intracerebral tumor. Approximately 16% of primary brain and CNS tumors are GBM.¹ Clinically, about 5% of GBM cases develop from previously diagnosed low-grade or anaplastic gliomas. These tumors are termed secondary glioblastomas (secGBM).² Standard GBM treatment consists of a gross total resection whenever safely possible, conventional fractionated radiotherapy, and in most cases temozolomide (TMZ) chemotherapy.^3,4 The role of subtotal resections is more controversial.⁵ Older patients may benefit from hypofractionated radiotherapy regimens.⁶ No specific treatment guidelines have been developed for secGBM. Adjuvant therapy options may be limited because patients may have already undergone radiation or chemotherapy for the preexisting lower-grade glioma at the time of the GBM diagnosis.

Molecular genetic findings differ considerably between primary (pGBM) and secGBM. Isocitrate dehydrogenase 1 (IDH1) mutations are rare in pGBM but occur in >70% of clinically diagnosed secGBM. TP53 mutations are more frequent in secGBM than pGBM (60%–90% vs 20%–35%). Primary GBM is characterized by frequent EGFR gene amplification (∼40%), and deletions in CDKN2A (30%–50%) and PTEN (>20%). More recently, ATRX mutations have been observed in 60%–80% of secGBM but in only 5% of pGBM.² Telomerase activation is another molecular hallmark of GBM. Increased telomerase activity is seen in >60%–70% of GBM and has been associated with an adverse prognosis.^7–9 In one study, secGBM displayed higher levels of telomerase activity than pGBM.¹⁰ Telomerase activation is also seen in many other cancers, including nonglial CNS tumors such as malignant and atypical meningiomas.^9,11

Telomerase counteracts physiological telomere shortening. The enzyme is constituted mainly by a catalytic reverse transcriptase and an RNA component encoded by the TERT and telomerase RNA component (TERC) genes, respectively. It is believed that the unlimited mitotic capacity of malignant tumors requires telomere maintenance through telomerase activation or alternative lengthening of telomeres.^12,13 Support for the concept of mutation-driven telomerase activation during tumorigenesis came from the recent discovery of activating mutations in the promoter region of the TERT gene in melanomas and several other cancers, including gliomas. The −124 C>T (C228T) and −146 C>T (C250T) mutations account for a majority of the somatic TERT promoter alterations identified. Both result in increased gene expression through creation of de novo binding sites for Ets/ternary complex factor transcription factors.^14–20 TERT promoter mutations have been associated with increased mRNA expression or telomerase activity in various tumors.^16,18

There are some data to suggest that in addition to somatic alterations, TERT germline variants may play a role in tumor biology. Genome-wide association studies have mapped susceptibility to glioma and other cancers to the TERT gene region.^21–23 Interestingly, in gliomas the strength of this association appears to increase with the histological grade, that is, the TERT glioma susceptibility locus is a GBM rather than a general glioma locus.²⁴ A recent study has shown that TERT promoter mutations affect survival and disease recurrence in bladder cancer in conjunction with a common genetic polymorphism (rs2853669) located within a preexisting Ets2 binding site of the TERT promoter.²⁵ The variant G-allele disrupts the binding site, resulting in decreased telomerase activity.²⁶ Interestingly, TERT mutations conferred an adverse prognosis only in patients without the variant rs2853669 allele.²⁵

In conclusion, several lines of evidence suggest a prominent role for telomerase activation and the TERT gene in the biology and therefore possibly also the clinical course of GBM. In this study we have characterized a large cohort of diffuse gliomas (94.6% GBM) for TERT promoter mutations. TERT mutations were detected much more frequently in pGBM than in secGBM. TERT mutations in conjunction with the rs2853669 polymorphism were shown to predict patient survival. We also found that TERT mutations seemed to affect survival primarily in patients with residual tumor after surgery and no postoperative TMZ chemotherapy.

Materials and Methods

Patients and Clinical Data

For the present study, 192 patients >18 years carrying a histological diagnosis of GBM were identified for whom appropriate tumor tissue had been deposited in the tumor bank of the Department of Neurosurgery at the University of Bonn. The cohort includes 178 pGBM as defined by clinical data (ie, regardless of IDH1 mutational status; see below). All patients underwent surgery at the Department of Neurosurgery at the University of Bonn between 1995 and 2012. All histopathological diagnoses were made at the Institute for Neuropathology/German Brain Tumor Reference Center at the University of Bonn Medical Center. Eighty-four cases (43.8%) were simultaneously enrolled in the German Glioma Network.²⁷

Clinical data were taken from a prospectively maintained electronic database. The clinical and demographic variables studied were age, preoperative Karnofsky performance index (KPI), histopathology (primary vs secondary glioblastoma), tumor location, extent of resection (biopsy vs partial vs gross total resection), radiotherapy (54–60 Gy vs early termination ≤54 Gy vs other radiotherapy vs none), and chemotherapy (TMZ radiochemotherapy according to the EORTC 26981–22981/NCIC CE3 trial protocol, intent to treat; TMZ radiochemotherapy including >3 adjuvant TMZ cycles vs other TMZ chemotherapy vs other chemotherapy vs none). Extent of resection was assessed on the first postoperative MRI scans; 47.4% of cases had early (<72 h) postoperative MRI scans. Patients were routinely followed as outpatients of the Department of Neurosurgery in Bonn, with MRI scans every 3–6 months. Some follow-up information was retrieved by contacting patients, their families, or their physicians directly. The study was approved by the University of Bonn Medical Center institutional review board for human research.

Molecular Analysis

Genomic DNA was isolated from tumor tissue and corresponding blood samples using standard methods. Tumor tissues were routinely assessed for tumor cell content using hematoxylin/eosin-stained frozen sections by an experienced neuropathologist (M.G.). Only samples containing >90% neoplastic cells were used for DNA isolation.

The core TERT promoter region was sequenced from the positions −278 to +65 from the ATG start site that included the site of recurrent somatic mutations and the rs2853669 polymorphism using PCR and Sanger sequencing as described.²⁵ O(6)-methylguanine-DNA methyltransferase (MGMT) promoter methylation was assessed using pyrosequencing as previously described.²⁸ Mutational status of IDH1 was investigated by direct sequencing.²⁹

Statistical Analysis

All statistical analyses were performed using commercially available software (SPSS version 21, IBM). Standard procedures (Fisher exact test, chi-square test, trend tests, Student t-test) were used for univariate analyses as indicated. Rates of overall survival (OS) and progression-free survival (PFS) were studied using univariate and multivariate Cox proportional hazards modeling (inclusion procedure). Survival endpoints were also analyzed with Kaplan–Meier estimates using the log-rank test for comparisons.

Results

TERT Mutations and Rs2853669 Genotypes in GBM

We investigated 192 GBM cases for mutations in the TERT promoter. Overall, somatic mutations were detected in 147 (76.6%) tumors with −124 C>T in 108 and −146 C>T in 39. The 2 mutations were mutually exclusive. The TERT promoter mutations were far more common in pGBM (80.3%) than in secGBM (28.6%) (chi-square test: P < .001; Table 1). Primary GBM patients with TERT mutations in tumors were significantly older than those harboring nonmutated tumors (mean age, 62.2 ± 11.4 y vs 55.5 ± 17.2 y; Student t-test: P < .006; see also Table 2).

Table 1.

TERT mutations and rs2853669 genotypes in pGBM and secGBM

	pGBM				secGBM
	rs2853669				rs2853669
	AA	AG	GG	All	AA	AG	GG	All
TERT mutation	66	60	17	143 (80.3%)	3	1	0	4 (28.6%)
TERT wild type	18	11	6	35 (19.7%)	5	5	0	10 (71.4%)
	84/178 (47.2%)	71/178 (39.9%)	23/178 (12.9%)	178 (100%)	8/14 (57.1%)	6/14 (42.9%)	0/14 (0%)	14 (100%)

Table 2.

Demographic, clinical, and molecular findings in n = 176 primary glioblastomas vs TERT mutational status

Parameter		All	TERT Mutation	No TERT Mutation	P
Sex	Male	104/176 (59.1%)	85/141 (60.3%)	19/35 (54.3%)	NS
Age, y	Mean ± SD	60.9 ± 13.0	62.2 ± 11.4	55.5 ± 17.2	.006
	Median, range	64, 24–89	65, 29–84	59, 24–89
Preoperative KPI	Mean ± SD	79.3 ± 11.9	79.4 ± 11.9	79.0 ± 11.9	NS
	Median, range	80, 40–100	80, 40–100	80, 50–100
Tumor location	Right	91/176 (51.7%)	75/141 (53.2%)	16/35 (45.7%)	NS
	Left	76/176 (43.2%)	59/141 (41.8%)	17/35 (48.6%)
	Bilateral	9/176 (5.1%)	7/141 (5.0%)	2/35 (5.7%)
	Frontal	65/176 (36.9%)	51/141 (36.2%)	14/35 (40.0%)	NS
	Temporal	59/176 (33.5%)	47/141 (33.3%)	12/35 (34.3%)
	Parieto-occipital	26/176 (14.8%)	22/141 (15.6%)	4/35 (11.4%)
	Insula	13/176 (7.4%)	10/141 (7.1%)	3/35 (8.6%)
	Multilocular	8/176 (4.5%)	8/141 (5.7%)	0
	Other	5/176 (2.8%)	3/141 (2.1%)	2/35 (5.7%)
	SVZ§	93/164 (56.7%)	74/131 (56.5%)	19/33 (57.6%)	NS
Extent of resection	Gross total	77/176 (43.8%)	60/141 (42.6%)	17/35 (48.6%)	NS
	Partial	96/176 (54.5%)	79/141 (56.0%)	17/35 (48.6%)
	Biopsy	3/176 (1.7%)	2/141 (1.4%)	1/35 (2.9%)
Radiation therapy	54–60 Gy	145/174 (83.3%)	115/139 (82.7%)	31/35 (88.6%)	NS
	<54 Gy	8/174 (4.6%)	7/139 (5.0%)	0
	No radiotherapy	20/174 (11.4%)	17/139 (12.2%)	3/35 (8.6%)
	Radiosurgery	1/174 (0.6%)	0	1/35 (2.9%)
Chemotherapy	Radiochemotherapy (TMZ), intent to treat*	125/173 (72.3%)	101/138 (73.2%)	24/35 (68.6%)	NS
	Radiochemotherapy (TMZ) >3 adjuvant cycles*	77/173 (44.5%)	57/138 (41.3%)	20/35 (57.1%)	NS
	Any TMZ chemotherapy	129/173 (74.65%)	104/138 (75.4%)	25/35 (71.4%)	NS
	No chemotherapy	42/173 (24.3%)	33/138 (23.9%)	9/35 (25.7%)	NS
	Other chemotherapy	2/173 (1.2%)	0	2/35 (5.7%)
MGMT	Promoter hypermethylation	80/162 (49.4%)	66/132 (50.0%)	14/30 (46.7%)	NS

Abbreviation: NS, nonsignificant.

*Radiochemotherapy with TMZ (EORTC-NCIC protocol).³ Intent to treat analysis.

^†Radiochemotherapy with TMZ (EORTC-NCIC protocol)³ with >3 adjuvant TMZ cycles.

^§Tumors involving the SVZ.

The sequenced region of the TERT promoter also included the rs2853669 polymorphism. The overall frequencies for the AA, AG, and GG genotypes were 47.9%, 40.1%, and 12.0%, respectively (Table 1). Observed genotype frequencies did not differ significantly from published data (see also http://www.ncbi.nlm.nih.gov/projects/SNP/snp_ref.cgi?rs=2853669).²⁵ No associations were observed between TERT mutations (individually and any mutation vs no mutation) and the rs2853669 polymorphism (genotypes and AA genotype vs AG/GG = G-allele carrier status).

Correlations of tert Mutations with mgmt Promoter Methylation and idh1 Mutations

MGMT promoter hypermethylation is a frequent finding in malignant gliomas and is an important prognostic and predictive biomarker. We therefore also investigated possible relations between TERT mutations, the rs2853669 polymorphism, and MGMT hypermethylation in 177 patients (92.2%) from this series. No significant correlations between MGMT status and TERT mutations or the rs2853669 polymorphism were observed in the unselected cohort and in the pGBM subset, respectively.

The high rate of TERT mutations in pGBM versus secGBM predicts an inverse correlation with IDH1 mutations. IDH1 mutation status was assessed in a subset of 97 cases from the present series. IDH1 codon 132 mutations were found in 9/13 (69.2%) secGBM cases but in only 4/84 (4.8%) pGBM (chi-square test: P < .001). IDH1 mutations were much more frequent in wild-type TERT cases (11/27 = 40.7%) than in TERT-mutated tumors (2/70 = 2.9%; Fisher exact test: P < .001). No correlation was seen between IDH1 mutational status and the rs2853669 polymorphism.

Impact of TERT Mutations on Patient Survival

In order to identify a potential impact of TERT mutations on patient survival, clinical data from 176 patients with pGBM were analyzed after exclusion of 2 cases for which the TERT mutational status was assessed only in recurrent rather than the primary tumor. Demographic data for this cohort and key clinical findings versus TERT mutational status are detailed in Table 2. There was no association between tumor location and the TERT mutational status (Table 2). Specifically, there was no correlation between TERT mutations and a presumed tumor origin in the subventricular zone (SVZ). This latter location is considered a stem cell niche, and GBM contacting the SVZ may carry a particularly adverse prognosis.³⁰

One hundred fifty-five patients (88.1%) were followed until tumor recurrence; 153 (86.9%) cases until death. Mean follow-up was 16.5 ± 15.3 months (median: 12.0; range: 1–97). Overall survival at 1 year was 52.9% (median: 13.0 ± 0.5; 95% CI: 11.0–15.0). Progression-free survival at 1 year was 30.6%, and median PFS was 7.0 ± 0.5 months (95% CI: 5.9–8.1). TERT mutations were seen in 141/176 (80.1%) cases. The secGBM subset of our cohort was deemed too small (n = 14) and too heterogeneous to allow for a meaningful survival analysis.

In view of the recent finding that TERT mutations have a powerful impact on patient outcome in bladder cancer cases without the variant G-allele of the rs2853669 polymorphism, we primarily investigated a potential role of TERT mutations as a prognostic parameter against the background of the patients' rs2853669 status. Kaplan–Meier estimates showed that the presence of a TERT mutation was a significant negative predictor of OS in the overall series (median OS with vs without TERT mutation = 11.0 ± 1.3 vs 16.0 ± 4.3 mo; log-rank test: P = .038; Fig. 1A). Interestingly, the impact of TERT mutational status on OS was indeed profoundly influenced by the rs2853669 genotype. The prognostic influence of the TERT promoter mutations was largely restricted to patients without the variant G-allele of the polymorphism (rs2853669 AA, median OS: TERT mutation vs no ∼ = 9.0 ± 0.7 vs 18.0 ± 6.1 mo; log-rank test: P = .04; rs2853669 AG/GG, TERT mutation vs no ∼ = 14.0 ± 0.7 vs 16.0 ± 1.3 mo; log-rank test: P = NS; Fig. 1B and C). We observed particularly poor survival in patients with TERT promoter mutations who did not carry the variant G-allele for the rs2853669 polymorphism (Kaplan–Meier estimates, median OS: no TERT mutation/rs2853669 AG/GG = 16.0 ± 1.2 mo; no TERT mutation/rs2853669 AA = 18.0 ± 6.1 mo; TERT mutation/rs2853669 AG/GG = 14.0 ± 6.7 mo; TERT mutation/rs2853669 AA = 9.0 ± 1.0 mo; overall log-rank test: P = .015; pairwise comparisons revealed significant survival differences only between the latter and all other patient subgroups).

Fig. 1.

Kaplan–Meier estimates of overall survival (OS) stratified for TERT mutational status for (A) all patients with primary glioblastomas (pGBM; log rank test: P = .038), (B) pGBM patients with the rs2853669 AA genotype (AA; log rank test: P = .04), and (C) pGBM patients with the rs2853669 AG/GG (variant) genotypes (log rank test: P = NS).

Additional exploratory univariate regression analyses identified age, preoperative KPI, extent of resection, postoperative radiation and chemotherapy, rs2853669 G-allele carrier status, and MGMT methylation status as additional factors associated with prognosis (Table 3). Confirming the presence of TERT mutations as an independent negative prognostic parameter were multivariate Cox regression models of OS using TERT mutational status, rs2853669 carrier status, age (<64 vs ≥64 y, median: 64), preoperative KPI (median: ≤80 vs >80), extent of resection (gross total vs residual tumor), radiotherapy (54–60 Gy vs all others), chemotherapy (any TMZ vs all others), and MGMT status as covariates (Table 4). The presence of TERT mutations was associated with a 2-fold increase of the risk of death.

Table 3.

Univariate predictors of survival in patients with primary glioblastomas

Parameter		P	HR	95% CI
Age ≥64 y	88/176 (50%)	.002	1.66	1.20–2.29
Sex (male)	104/176 (59.1%)	.979	1.00	0.73–1.39
KPI, 90–100	69/176 (39.2%)	.009	0.64	0.46–0.89
GTR	77/176 (43.8%)	<.001	0.44	0.32–0.62
Tumor location: SVZ	93/164 (56.7%)	.119	1.31	0.93–1.82
Radiotherapy 54–60 Gy	146/174 (83.9%)	<.001	0.27	0.17–0.42
Radiochemotherapy (TMZ), intent to treata	125/173 (72.3%)	<.001	0.49	0.34–0.70
Radiochemotherapy (TMZ), >3 cyclesb	77/173 (44.5%)	<.001	0.39	0.28–0.54
Any TMZ chemotherapy	129/173 (73.3%)	<.001	0.38	0.26–0.55
MGMT promoter hypermethylation	80/162 (49.4%)	<.001	0.45	0.32–0.64
IDH1 codon 132 mutation	4/84 (2.3%)	.690	0.81	0.30–2.24
TERT mutation	141/176 (80.1%)	.046	1.51	1.01–2.27
rs2853669 AA	84/176 (47.7%)	.047	1.38	1.01–1.90
TERT mutation + rs2853669 AA	66/176 (37.5%)	.003	1.63	1.18–2.26

Abbreviation: GTR, gross total resection.

^aRadiochemotherapy with TMZ (EORTC-NCIC protocol),³ intent to treat analysis.

^bRadiochemotherapy with TMZ (EORTC-NCIC protocol)³ with >3 adjuvant TMZ cycles.

Table 4.

Cox proportional hazards modeling (inclusion procedure) of OS in n = 158 patients with primary glioblastoma

Parameter	P	HR	95% CI
Age ≥64 y	.016	1.56	1.09–2.23
KPI, 90–100	.309	0.82	0.56–1.20
GTR	<.001	0.51	0.35–0.74
Radiotherapy 54–60 Gy	.001	0.66	0.51–0.85
TMZ chemotherapy	.001	0.48	0.31–0.74
MGMT promoter hypermethylation	<.001	0.37	0.26–0.54
TERT mutation	.002	2.05	1.30–3.23
rs2853669 AA	.456	1.14	0.80–1.63

Abbreviation: GTR, gross total resection.

Finally we also explored whether the impact of the TERT mutations on patient survival might vary with therapeutic variables. TERT mutations correlated with patient survival only in the presence of residual tumor (Kaplan–Meier estimates, median OS: complete resection, TERT mutation vs no ∼ = 18.0 ± 2.9 vs 21.0 ± 7.5 mo, log-rank test: P = NS; residual tumor, TERT mutation vs no ∼ = 8.0 ± 1.2 vs 14.0 ± 1.9 mo, log-rank test: P = .004; Fig. 2A and B). Comparing patients receiving any TMZ chemotherapy vs no or other chemotherapy, TERT mutational status was significantly predictive of OS only in the latter group (Kaplan–Meier estimates, median OS: any TMZ chemotherapy, TERT mutation vs no ∼ = 14.0 ± 1.0 vs 18.0 ± 4.8 mo, log-rank test: P = NS; no TMZ chemotherapy, TERT mutation vs no ∼ = 5.0 ± 1.1 vs 9.0 ± 8.7 mo, log-rank test: P = .016; Fig. 2C and D). In summary, the prognostic impact of TERT mutations was lost whenever patients had complete resections and postoperative TMZ chemotherapy.

Fig. 2.

Kaplan–Meier estimates of overall survival (OS) of patients with primary glioblastomas (pGBM) stratified for TERT mutational status and (A) gross total resection (P = NS, log rank test) vs (B) residual tumor (log rank test: P = .004), (C) temozolomide (TMZ) chemotherapy (log rank test: P = NS) vs (D) no or other chemotherapy (log rank test: P = .016).

Discussion

TERT promoter mutations, described initially in melanoma, have now been shown to occur frequently in a variety of cancers. The histological spectrum encompasses various carcinomas (in particular bladder and thyroid cancers), sarcomas, tumors of the CNS, and other neoplasms.^16–20 In the present study we found that the TERT promoter mutations are highly frequent (80.3%) in pGBM compared with secGBM and gliomas of WHO grades II and III. We also observed an inverse association between the TERT promoter mutations and IDH1 mutations as reported earlier.^16–18,20 These data provide additional support for the concept of separating pGBM from other diffuse gliomas. The principal novel finding of our study with potential clinical relevance is an association of the TERT promoter mutations with poor overall patient survival and modification of the effect by a common polymorphism within the core TERT promoter. We also show that the mutations are associated with increased patient age.

Based on the prevalence in different cancer types, it has been hypothesized that the TERT promoter mutations occur mainly in tumors that are derived from tissues with low rates of self-renewal.¹⁷ Even though exceptions to that hypothesis have been observed, the high rate of TERT mutations in pGBM fits well within the concept. Self-renewal is not a prominent feature of CNS tissues and glial cells. The relatively low frequency of TERT mutations in secGBM and other gliomas has been explained by the activity of an alternative mechanism of telomere lengthening in those tumors, possibly due to mutations of the ATRX gene.^17,31

The most important finding of this study is the association of TERT mutations with an adverse prognosis of pGBM patients in the absence of the variant G-allele of the rs2853669 polymorphism. In this study, in conformity with a previous report on bladder cancer,²⁵ we found that patients with somatic TERT promoter mutations who were noncarriers of the variant G-allele for the polymorphism carried the worst overall prognosis. On the other hand, the effect of the mutations was not statistically significant in patients who carried the variant allele of the polymorphism either in homozygous or heterozygous form. The rs2853669 polymorphism located at −245 bp upstream of the ATG site results in abrogation of a preexisting Ets2 site in the proximal region of the TERT promoter. Mutations at the Ets2 binding site have been shown to inhibit c-Myc binding to the E-box.³² The variant allele of the polymorphism has been previously shown to be associated with low telomerase activity in non–small cell lung carcinoma.²⁶

Rs2853669 is located 8.8 kb and 6.8 kb apart from rs2736100 and rs2853676, respectively. These latter single nucleotide polymorphisms (SNPs) have been shown to predispose to malignant glioma in a large genome-wide association study.²¹ We found no associations between the rs2736100 genotypes and patient outcome in a subset of patients who could be analyzed for clinical correlates.²⁴ The rs2853669 polymorphism has also been reported to be associated, though inconsistently, with an increased susceptibility to breast cancer.³³ Of note, the SNP rs2736098, which has also been associated with cancer susceptibility, is located only 1.2 kb away from rs2853669. The linkage disequilibrium between rs2736098 and both rs2736100 and rs2853676 is poor. In summary, it remains unclear at this point whether the (functional) polymorphism rs2853669 is behind any of the 5p15 associations with glioma and/or cancer susceptibility.²¹

An exploratory analysis investigating the prognostic impact of TERT mutations versus all generally accepted prognostic parameters, both clinical (age, KPI, extent of resection, adjuvant therapy, tumor location) and molecular (IDH1, MGMT promoter methylation), revealed TERT mutational status as an independent predictor of the patients' prognosis. This is the first study showing that TERT mutations impact on GBM patient survival in a large patient cohort for which detailed clinical data were made available. A correlation between patient survival and TERT mutations has also been described by Killela et al,¹⁷ albeit in a much smaller patient cohort. Most recently, Nonoguchi and coworkers¹⁹ investigated a relatively large GBM cohort for TERT mutations. In this study, TERT mutations could be linked to patient survival only in the univariate analysis.¹⁹ However, this study lacked the necessary detailed clinical data for determining any robust association.

The identification of a prognostic biomarker may have profound implications. While the overall prognosis of many patients with malignant gliomas remains dismal, the concept of individualized (or rather biomarker-based) therapy has improved survival in a selected patient subset. The efficacy of TMZ chemotherapy seems closely related if not restricted to tumors characterized by methylation of the MGMT promoter.³⁴ Among elderly patients, TMZ chemotherapy only is an option in patients with MGMT methylation, which spares some from adverse side effects of radiotherapy.^6,35 Having this background in mind, we explored whether the prognostic role of TERT mutations might vary with therapeutic variables. Our data appear to suggest a prognostic impact of TERT mutations in particular in patients with residual tumor who do not receive TMZ chemotherapy. However, these findings clearly require validation in an independent cohort because of the exploratory character of the analysis and the borderline statistical significance if one takes into account the issue of multiple testing. One also has to bear in mind that the patient subsets characterized by different treatment regimes (eg, partial vs gross total resection, TMZ chemotherapy yes vs no) were likely not balanced. More aggressive tumors might have been less often amenable to complete resections and chemotherapy—for instance, because of the respective patients' poor clinical condition. Some patients were treated in the pre-TMZ era. It is nevertheless tempting to speculate that surgery and TMZ chemotherapy in contrast to radiotherapy are effective against tumor cells responsible for the adverse prognosis associated with the TERT promoter mutations. Conversely the tumors with the TERT promoter mutations can be hypothesized to require a more aggressive surgical and chemotherapeutic strategy.

In conclusion, in this study we describe frequent TERT −124 C>T and −146 C>T mutations in pGBM but not in secGBM or gliomas of WHO grades II/III. These data provide further support for the concept of pGBM as a separate tumor entity fundamentally different from other diffuse gliomas. Furthermore, we show an association of the TERT promoter mutations with adverse outcome that is modified by the variant allele of a common polymorphism. TERT mutational status proved to be an independent predictor of patient survival. In addition, TERT mutations may have potential as a biomarker for treatment stratification.

Funding

This work was supported in part by Deutsche Forschungsgemeinschaft (DFG Az Si 552/2 to M.S. and J.S.), the University of Bonn (BONFOR O-126.0030 to M.S.), and Deutsche Krebshilfe (70/2385/WI2, 70/3163/WI3 to J.S.).