TERT promoter mutation associated with multifocal phenotype and poor prognosis in patients with IDH wild-type glioblastoma

Abstract Background Although mutations in the promoter region of the telomerase reverse transcriptase (TERTp) gene are the most common alterations in glioblastoma (GBM), their clinical significance remains unclear. Therefore, we investigated the impact of TERTp status on patient outcome and clinicopathological features in patients with GBM over a long period of follow-up. Methods We retrospectively analyzed 153 cases of GBM. Six patients with isocitrate dehydrogenase 1 (IDH1) or H3F3A gene mutations were excluded from this study. Among the 147 cases of IDH wild-type GBM, 92 (62.6%) had the TERTp mutation. Clinical, immunohistochemical, and genetic factors (BRAF, TP53 gene mutation, CD133, ATRX expression, O6-methylguanine-DNA methyltransferase [MGMT] promoter methylation) and copy number alterations (CNAs) were investigated. Results GBM patients with the TERTp mutation were older at first diagnosis versus those with TERTp wild type (66.0 vs. 60.0 years, respectively, P = .034), and had shorter progression-free survival (7 vs. 10 months, respectively, P = .015) and overall survival (16 vs. 24 months, respectively, P = .017). Notably, magnetic resonance imaging performed showed that TERTp-mutant GBM was strongly associated with multifocal/distant lesions (P = .004). According to the CNA analysis, TERTp mutations were positively correlated with EGFR amp/gain, CDKN2A deletion, and PTEN deletion; however, these mutations were negatively correlated with PDGFR amp/gain, CDK4 gain, and TP53 deletion. Conclusions TERTp mutations were strongly correlated with multifocal/distant lesions and poor prognosis in patients with IDH wild-type GBM. Less aggressive GBM with TERTp wild type may be a distinct clinical and molecular subtype of IDH wild-type GBM.

TERT promoter mutation associated with multifocal phenotype and poor prognosis in patients with IDH wild-type glioblastoma Glioblastoma (GBM) is the most common primary malignant tumor affecting the central nervous system in adults. 1 Despite of radical surgery combined with concomitant chemoradiation therapy based on temozolomide, the median survival of patients is approximately 18 months. 2 According to the World Health Organization revised neuropathological criteria, these tumors are divided into 2 categories, namely isocitrate dehydrogenase (IDH) wildtype and IDH-mutant GBMs. In addition, recent reports indicated that 70%-80% of GBM genomes harbor either C228T or C250T mutations in the promoter region of the telomerase reverse transcriptase (TERTp) gene. 3,4 These mutations are associated with enhanced telomere maintenance. [5][6][7] Although several studies reported the prognostic significance of TERTp mutation in patients with GBM, its clinical and pathological roles remain unclear. [3][4][5][6] Recently, GBM patients with unmethylated O 6methylguanine-DNA methyltransferase (MGMT) and TERTp mutation have a worse prognosis than those with TERTp wild type. 3,8 However, the mechanism of interaction of TERTp mutation and MGMT promoter methylation is not well established.
Regarding imaging analysis, necrosis detected through magnetic resonance imaging (MRI) has been reported to indicate the presence of TERTp mutation. 9 However, predicting the TERTp status by preoperative imaging study alone remains difficult.
A recent systematic review and meta-analyses stated that the incidence of solitary GBM is 83%. 10 Other previous studies showed that 20% of patients with GBM had multiple lesions and their prognosis was worse than that recorded in patients with a single lesion. 11 In this study, we analyzed GBMs to determine whether the TERTp status was associated with other clinical and molecular factors. Particularly, this study utilized MRI to determine the development of multifocal/distant lesions during the clinical course.

Patients and Samples
This retrospective study was conducted with the approval of the Ethics Committees of the Tohoku University School of Medicine and Yamagata University School of Medicine. Written informed consent was provided by all patients prior to their participation in the study.
Between January 2009 and October 2019, a total of 153 patients (89 treated at Yamagata University Hospital [Yamagata cohort] and 64 treated at Tohoku University Hospital [Tohoku cohort]) were analyzed. All patients met the following inclusion criteria: (1) diagnosis of GBM, World Health Organization grade IV; (2) no history of lower-grade tumors; (3) availability of genomic DNA; and (4) availability of information regarding events, such as recurrence or death during the follow-up period, or absence of such events for ≥12 months of follow-up. Patients who had previously undergone biopsies were excluded from the study. Tumor specimens were obtained from a lesion that exhibited enhancement on gadolinium-enhanced MRI and immediately stored at −80°C until DNA extraction ( Figure 1).

Classification of GBM According to Preoperative MRI
MRI sequences were acquired on a 1.5-T or 3.0-T scanner and typically included axial T1-weighted, T2-weighted fast spin-echo, and fluid-attenuated inversion-recovery sequences as well as a postcontrast 3-dimensional spoiled gradient-recalled acquisition in the steady state T1-weighted sequence. Contrast-enhanced lesions (CELs) were assessed to clarify whether they were in contact with the subventricular zone, as previously described. 12

Importance of the Study
Mutations in the promoter region of the telomerase reverse transcriptase (TERTp) gene are the most common mutations in isocitrate dehydrogenase (IDH) wild-type glioblastoma (GBM). While TERTp mutations are correlated with poor prognosis, aggressive clinicopathological characteristics, and metastasis in other cancers, their clinical significance in GBM remains unclear. Here, we analyzed GBMs to determine whether the TERTp status is associated with other clinical and molecular factors. Particularly, this study focused on whether multifocal/distant lesions were observed during the clinical course. In this study, we demonstrated that TERTp-mutant GBMs are strongly associated with the prognosis and multifocal/distant lesions during a long follow-up period. In addition, TERTp mutation was positively correlated with EGFR amp/gain, CDKN2A deletion, and PTEN deletion; however, it negatively correlated with PDGFR amp/gain, CDK4 gain, and TP53 deletion. Less aggressive GBM with TERTp wild type could be distinct clinical and molecular subtype of IDH wild-type GBM.

Definition of Multifocal/Distant Lesions
One or more enhancing noncontiguous lesions >1 cm distant from the original tumor on preoperative MRI were defined as multifocal/distant lesions at diagnosis. 13 In addition, as previously reported, "multifocal/distant lesions at recurrence" were defined as distant or multifocal recurrence. Recurrence was characterized by the development of new CEL centered >3 cm distant from the primary resection cavity or at the margins of the primary residual tumor, or at more than 1 site, with each lesion having a well-defined border and the patient exhibiting normal brain signals. 14,15

Clinical Parameters
The clinical profiles of patients were obtained from their medical records. The majority of patients underwent radical surgery followed by chemotherapy (nimustine hydrochloride [ACNU] or temozolomide) and radiotherapy.
Total surgical resection was defined as the disappearance of CEL according to pre-and postoperative gadoliniumenhanced MRI studies. In cases in which the primary tumor recurred, patients underwent salvage surgery, second-line chemotherapy, radiotherapy, or palliative therapy. The Ki-67 labeling index was determined by immunohistochemical staining of resected specimens with the Ki-67 antigen (Dako, Agilent Technologies). We also analyzed the expression of CD133 (Miltenyi Biotec), p53 (Dako, Agilent Technologies), and ATRX (Abcam) by immunohistochemical staining. The expression of CD133 in 144 patients among the Yamagata and Tohoku cohorts was previously reported. 16  (C) Local recurrence during the entire follow-up period. (D) Multifocal/distant lesion at second recurrence. Seven months after surgery, local recurrence was observed adjacent to the resection cavity. Ten months later, an enhanced lesion was detected at a distant location (arrow).
as the time between the day of the first operation and the day of death or final follow-up.

Molecular Analysis
Genomic DNA was extracted with the QIAamp DNA mini kit (Qiagen), according to the instructions provided by the manufacturer. The isocitrate dehydrogenase1/2 (IDH1/2), H3F3A, HIST1H3B, TP53, BRAF, and TERTp genes were amplified via polymerase chain reaction (PCR), and sequencing was conducted as previously described. 18,19 In the MGMT promoter methylation analysis, we performed methylation-specific PCR or quantitative methylationspecific PCR following the bisulfite modification of tumor DNA. 19 To assess copy number alterations (CNAs), we performed Multiplex Ligation-dependent Probe Amplification (MLPA) using the SALSA MLPA KIT P105 (version D2), in accordance with the manufacturer's protocol (MRC Holland). 20 The P105 kit is designed to detect CNAs typically found in gliomas and includes probes against the PDGFRA, EGFR, CDKN2A, PTEN, TP53, CDK4, MDM2, and NFKBIA genes. Based on the previous publications, the CNA categories were classified according to the following thresholds: homozygous deletion (x ≤ 0.4), hemizygous deletion (0.4 < x ≤ 0.7), gain (1.3 ≤ x < 2.0), and amplification (x ≥ 2.0). 20,21 We used OncoPrinter, a tool provided by the cBioPortal for Cancer Genomics (cbioportal.org/oncoprinter), to visualize and analyze our data with some modifications. 22,23 Statistical Analysis Statistical analyses were performed using the SPSS (IBM Japan) software. The relationship between 2 variables was evaluated using the Mann-Whitney U test and Fisher's exact test. Estimates of PFS and OS were calculated with the Kaplan-Meier method, and the Log-rank (Mantel-Cox) test was used to evaluate differences between the groups. Cox regression was used for the multivariate analysis. The significance level was set at P < .05.

Population and Tumor Characteristics on MRI
A total of 153 patients, including 82 males and 71 females with a median age of 63 years (range: 27-86 years) and median preoperative Karnofsky Performance Status of 80 (range: 30-100), were included in the present study. Patients in the Yamagata cohort were older than those in the Tohoku cohort (P < .001) (Supplementary Table 1). Genomic DNA and paraffin-embedded samples were obtained from all patients. The median duration of the follow-up period was 17 months (range: 1-152 months), and 119 patients (77.8%) expired. Total surgical resection was achieved in 96 patients (62.7%). In this group, IDH1, H3F3A, and BRAF gene mutations were detected in 4 (2.6%), 2 (1.3%), and 1 patient (0.65%), respectively; however, neither IDH2 nor HIST1H3B gene mutations were detected. TERTp gene mutations were detected in 92 patients (60.1%), including 65 (42.5%) and 27 (17.6%) with C228T and C250T mutations, respectively. Although the frequency of TERTp gene mutations in the Yamagata cohort was higher than that in the Tohoku cohort (P = .019, Supplementary Table 1), there was no significant difference in the mutation frequency in older patients (age ≥60) between the 2 cohorts (P = .348) (data not shown). MGMT gene promoter methylation was found in 62 patients (40.5%). Postoperative treatments consisted of radiation alone for 6 patients, while the remaining 147 patients received combined radiation and chemotherapy with temozolomide (n = 123), ACNU (n = 14), or other agents (n = 10). Bevacizumab was administered as first-and second-line therapy in 1 and 50 patients, respectively. There were no significant differences observed in PFS and OS between patients treated with ACNU and temozolomide (data not shown). TP53 gene mutations and/or strong immunoreactivity of p53 were found in 63 patients (43.4%) (Figure 2). Eleven of 88 patients (12.5%) displayed the loss of ATRX expression. The major CNAs frequently observed in 139 GBMs included EGFR amp/ gain (66.2%), CDKN2A deletion (60.4%), and PTEN deletion (51.8%) ( Figure 2 and Table 1).

Correlation Analyses Between the TERTp Status and Other Prognostic Factors
Six patients with IDH1 or H3F3A mutations were excluded from this study. Therefore, we analyzed 147 GBM patients with IDH wild type to determine the factors correlated with the TERTp mutation. The median age was higher in GBM patients with TERTp mutation than those with TERTp wild type (P = .034) ( Table 1).
In terms of MRI characteristics, 21 of the 147 patients (14.3%) had multifocal/distant lesions at diagnosis (Table 1). During the follow-up, 129 patients (87.7%) experienced the first recurrence, which included local recurrence and multifocal/distant recurrence in 99 (67.3%) and 30 (20.4%) patients, respectively. Among the patients with a well-controlled first recurrent lesion, 15 patients (10.2%) had new multifocal/distant lesions at second recurrence. Neither local nor distal recurrence was observed at the time of the last observation in the remaining 18 patients (12.3%).
Although multifocal/distant lesions at diagnosis or recurrence were weakly correlated with TERTp mutations (P = .087 and P = .096, respectively), these lesions were significantly more common in patients with TERTp-mutant GBM than in patients with TERTp wild-type GBM during the entire follow-up period (P = .004, Table 1).

Neuro-Oncology Advances
To determine whether the TERTp mutation was negatively correlated with PFS and OS in the non-multifocal/ distant group, we analyzed the survival of the 81 patients in the non-multifocal/distant group. The median PFS and OS were 9 and 23 months, respectively, with no significant correlation of PFS and OS with the TERTp mutation (P = .129 and P = .148, respectively) (data not shown).
We also investigated the prognostic value of TERTp mutation in combination with MGMT promoter methylation. Among patients with TERTp mutation, unmethylated MGMT was significantly associated with poor PFS and OS (P < .0001 and P < .0001, respectively) ( Figure 3C and D). However, among patients with TERTp wild type, there was no significant difference of PFS and OS between patients with and without MGMT promotor methylation (P = .938 and P = .699, respectively) ( Figure 3C and D).

Factors Associated With Multifocal/ Distant Lesions
We investigated several factors to determine whether they correlated with multifocal/distant lesions. As shown in Supplementary Table 2, TERTp mutations, the expression of CD133, and PTEN deletion were significantly associated with multifocal lesions (P = .004, P = .004, and P = .004, respectively).

Discussion
TERTp mutation is the most common alteration in GBM; however, the clinical impact of TERTp mutations in GBM remains unclear. To understand the poor prognosis of GBM with TERTp mutations, we hypothesized that malignant clinical features exist in this group. Long-term follow-up revealed that the cumulative incidence of multiple/distant lesions was significantly higher in GBM with TERTp mutations than in patients with TERTp wild-type GBM. Conversely, the non-multifocal/distant group did not show any differences in PFS and OS based on TERTp status. Therefore, we, for the first time, demonstrated that GBM with TERTp mutations has a poor prognosis because of its clinically aggressive behavior. In accordance with this finding, several studies regarding other cancers demonstrated that these mutations were correlated with a poor prognosis, aggressive clinicopathological characteristics, and metastasis. [24][25][26][27][28] Xing et al. found that TERTp mutation strongly correlated with vascular invasion in patients with papillary thyroid cancer. 25 Yuan et al. reported that thyroid cancer patients with the TERTp mutation have a 4-fold higher risk of distant metastasis than those with TERTp wild type. 27 The frequency of TERTp mutations in our study was 62.6%, which is lower than that of previous reports from North America and European countries, which reported mutation

Neuro-Oncology Advances
frequencies of 73%-75% in IDH wild-type GBMs. [3][4][5] Other reports from Japan also showed relatively low frequencies of TERTp mutations among IDH wild-type GBM, ranging from 50% to 70%. 6,9,29 Thus, racial differences in the frequency of TERTp mutations may exist. One possible explanation for the low frequency of TERTp mutations in the Japanese cohort is that other mechanisms involved with replicative immortality in TERTp wild-type GBM. One such mechanism is TERTp hypermethylation, and the other is ATRX or SMARCAL1 gene mutation. TERTp hypermethylation can aberrantly activate telomerase in cancer, 30 and the ATRX or SMARCAL1 gene mutations are strongly associated with the maintenance of telomere length, referred to as alternative lengthening of telomeres. 31 Indeed, our results indicated the frequent loss of ATRX expression in TERTp wild-type GBM. The other explanation is potential inclusion of other IDH wild-type high grade gliomas such as anaplastic astrocytoma with piloid features. 32 Although our cases were histologically confirmed as GBM, further molecular testing may be required to classify into novel entities.
The prognostic significance of the TERTp mutation remains controversial in patients with GBM. 3,[33][34][35] In the present study, univariate and multivariate analyses showed that the TERTp mutation was significantly associated with both PFS and OS. In accordance with previous reports, we also found that unmethylated GBM with TERTp mutations presented a poor prognosis. 3,8 However, among patients with TERTp wild type, there was no significant difference of PFS and OS between patients with and without MGMT promotor methylation. The reason may be that GBM tumors with the TERTp mutation form multifocal/distant lesions by invading various directions. Nevertheless,     Neuro-Oncology Advances those with methylated MGMT were sensitive to treatment with alkylating agents, such as temozolomide. Therefore, TERTp mutated GBM patients with methylated MGMT may survive longer than those with unmethylated MGMT.
Recently, GBMs were divided into 2 groups according to the IDH mutation status. Although IDH mutation is frequently found in lower-grade diffuse glioma, only 5%-10% of patients with GBM had this mutation. 36,37 In addition, GBM patients with the IDH mutation are usually young and diagnosed with progression from a lower grade of diffuse astrocytoma. Thus, TERTp mutation, frequently found in GBM is more useful for predicting survival and clinical behavior, such as the pattern of invasion.
Our data showed that TERTp mutations were significantly associated with EGFR amp/gain, CDKN2A deletion, and PTEN deletion and were typically found in IDH wildtype GBM; conversely, the TERTp wild type was associated with PDGFR amp/gain, CDK4 gain, and TP53 deletion. Recently, Williams et al. reported TERTp wild-type GBMs showed frequent PI3K pathway and BAF complex gene family (ATRX, SMARCA4, SMARCB1, and ARID1A) mutations. 38 Our data also suggest that TERTp wild-type GBMs are genetically distinct from TERTp-mutant GBMs.
The present study had some limitations. First, since this was a retrospective study, patients were not treated in the same manner. Although we performed a multivariate analysis, differences in treatment may have affected the pattern of recurrence. Second, we demonstrated the malignant features of GBM with the TERTp mutation based on clinicopathological characteristics, but patients with oligodendroglioma (the most benign diffuse glioma) also had the TERTp mutation. 6 Third, it has been reported that PTEN, PI3K3A mutation and the expression of CD133 are associated with distant recurrence in patients with GBM. 16,17,39,40 In the present study, there was no significant association between CD133 expression and the TERTp mutation, but PTEN deletion was significantly correlated with TERTp mutations and multifocal/distant lesions. The mechanism of invasiveness based on the TERTp mutation warrants further investigation.

Conclusion
We retrospectively investigated whether the TERTp mutation was associated with multifocal/distant lesions in GBM. The results suggested that the TERTp mutations strongly correlated with the multifocal phenotype and poor prognosis in patients with IDH wild-type GBM. We further demonstrated that TERTp mutations were significantly associated with EGFR amp/gain, CDKN2A deletion, and PTEN deletion, whereas the TERTp wild type was correlated with PDGFR amp/gain, CDK4 gain, and TP53 deletion. The IDH wild-type GBM with and without TERTp mutations may be a distinct clinical and molecular subtype.

Supplementary Data
Supplementary data are available at Neuro-Oncology Advances online.