Introduction

Gliomas are the most prevalent tumors which occurs in the glial or precursor cells of the brain and central nervous system (CNS)1,2, and constitute 81% of all malignant brain tumors. In 2007, the World Health Organization (WHO) classified gliomas into grades 1 to 4 based on their histological characteristics for CNS tumors3. However, in the updated 2021 WHO categorization, gliomas are categorized based on their molecular characteristics, such as isocitrate dehydrogenase (IDH) mutations and 1p/19q status4. Based on their histological and molecular characteristics, gliomas can be classified into astrocytomas, oligodendrogliomas, or glioblastomas (GBMs)1,4,5.

Genome-wide association studies (GWAS) have been conducted to verify regions correlated with the risk of glioma. Risk of glioma associated variants at 27 loci, including 8 loci correlated with all glioma (3q26.2, 5p15.33, 7p11.2, 8q24.21, 9p21.3, 11q23.3, 17p13.1, and 20q13.33), 7 loci correlated with GBM (1q31.3, 11q14.1, 12q23.3, 12q23.33, 16p.12.1, 16q.13.3, and 22q13.1), and 12 loci correlated with non-GBM glioma (1q32.1, 1q44, 2q33.3, 3p14.1, 10q24.33, 10q25.2, 11q21, 11q23.2, 12q21.2, 14q12, 15q24.2, and 16q13.3) have been reported by previous studies6,7,8,9.

Telomeres, which are located at the ends of eukaryotic chromosomes, consist of TTAGGG repeats. Telomeres play an important role in protecting chromosomes from degradation, end-to-end fusion, and abnormal recombination. Therefore, telomeres are essential for preserving chromosome unity and genomic stability10. Telomere length is maintained by telomerase, which adds the telomeric repeat sequence to the single-strand 3’ overhang. This process preserves the ends of telomeres, which gradually shorten with each cell division11. Telomerase expression is particularly low in most normal human somatic cells, however, it is higher in malignant glioma12. TERT is the most crucial factor as catalytic subunit of telomerase in determining telomerase expression13.

To follow the results of the GWAS for TERT on 5p15.33, we selected single nucleotide polymorphisms (SNPs) around association signals in TERT for glioma. In addition to studying gliomas, we conducted association analyses to identify associations with glioma subgroups, considering various clinical characteristics such as histological features, molecular properties, and grades.

TERT variants associated with glioma risk have been demonstrated by multiple studies. For instance, in the Han Chinese population, rs2853677 is associated with an increased risk of glioma14. Similarly, in European and Han Chinese populations, rs2736100 is associated with an increased risk of glioma15,16,17. In the US and UK, rs2853676 has been reported to be significantly associated with gliomas6. rs4975538 has been reported to be significantly associated with IDH-wild-type and TERT mutated gliomas in multiple ethnic groups18. In a meta-analysis, rs10069690 was found to be associated with an elevated risk of glioma regardless of sex19. Previous studies have confirmed the association between common polymorphisms of TERT and the risk of heritable glioma. However, the investigation of the association between TERT SNPs and the risk of glioma has not yet been done in Korean population.

TERT variants can cause chromosomal instability, resulting in the development of cancer. Therefore, these SNPs may be associated with telomere malfunction and cancer. For this study, prostate cancer-associated SNPs (rs13172201, rs2242652, and rs7713218)20,21,22, lung carcinoma-associated SNP (rs13167280)23, decreased telomere length in T cell-associated SNP (rs33961405)24, leukocyte telomere length-associated SNP (rs7705526)25, and epigenetic age acceleration-associated SNP (rs2736099)26 were selected.

TERT promoter hotspot mutations (C250T and C228T) are frequently observed in patients with GBM and oligodendroglioma27. These mutations create binding sites for ETS transcription factors. GABP, a member of the ETS family, enhances TERT transcription28. The increased transcription of TERT contributes to the immortality of cancer cells through the expression of telomerase29. The prevalence of TERT promoter hotspot mutations in patients with GBM varies by ethnicity30,31,32.

We initially selected SNPs of TERT to investigate this association. Additionally, we conducted an association analysis to identify the relationship between susceptibility alleles and glioma subgroups, considering various clinical characteristics such as grades, histological features, and molecular properties.

Results

Patient characteristics

Of the 317 glioma patients, the mean age was 53.43 ± 12.02 years and 50.7% were male. Patients were categorized into different subgroups according to histological characteristics: astrocytoma WHO grade 2, astrocytoma WHO grade 3, astrocytoma WHO grade 4, oligodendroglioma WHO grade 2, oligodendroglioma WHO grade 3 and GBM. Based on molecular properties of the 2021 WHO classification for CNS tumors, 84 glioma patients had IDH mutations and 67 had 1p/19q co-deletion (Table 1). The PC group comprised 480 individuals aged over 40, with a mean age of 54.8 ± 9.5 years and 49.4% male. The detailed categorizations of the patients are presented in Table 1.

Table 1 Clinical characteristics of study subjects.
Full size table

Genotyping TERT genetic variants

Genotyped TERT SNPs are located on chromosome 5p15.33, as shown on physical map in Supplementary Fig. 1A. There were three LD blocks, each consisting of six, five, and five haplotypes, with a frequency of > 5% (Supplementary Fig. 1B). Detailed information, including coordinates, SNP alleles, and positions, is provided in Table 2.

Table 2 Logistic association regression analysis of TERT SNPs with glioma.
Full size table

Associations between TERT SNPs and glioma risk

Logistic regression analysis was conducted to confirm the causal TERT variants associated with glioma using an additive model, with adjustments for age and sex as covariates, as shown in Table 2. Eight SNPs (rs2853676, rs2736100, rs10866498, rs7713218, rs4975538, rs2242652, rs10069690, and rs56345976) were significantly associated with the risk of glioma (P < 0.009; Table 2). In addition, 16 haplotypes which frequency is over 5 were used for logistic regression analysis.

Block2-ht1 and block3-ht3 are associated with increased risk of glioma (P = 0.02 and 0.04, respectively). And block2-ht2, block2-ht3, and block3-ht1 are associated with decreased risk of glioma (P = 0.01, 0.02 and 0.01 respectively) (Supplementary Table 1).

Within the Korean population of our study, we observed the genetic effects of previously reported glioma associated TERT variants (rs2853677, rs2736100, rs10069690, rs2853676, and rs4975538). The results are summarized in Table 3.

Table 3 Replication of previously reported glioma-associated TERT variants in this study.
Full size table

Analysis of independency and causality of TERT variants on glioma risk

Stepwise and pairwise conditional analyses were conducted to identify the independent association among the eight significant TERT SNPs and risk of glioma. The rs56345976 SNP remained after stepwise analysis (P = 0.007) in the model with a parametric discriminant P-value (0.01). Pairwise conditional analysis also indicated that rs56345976 could be a causal variant among the eight significant variants (Table 4) when considering the remaining significances in four out of seven conditional analyses.

Table 4 Stepwise and conditional analysis of glioma-associated TERT variants.
Full size table

The genetic effects of rs56345976 (G > A) were subsequently investigated using additive, dominant, recessive, and referent models in the glioma subgroups (Table 5). Although significant associations were detected in all four analysis models, the A allele showed dose-dependent generic risk for glioma, e.g., the AA genotype (OR = 2.59, P = 0.0004) had a higher OR and stronger association compared to those in the GA genotype (OR = 1.16, P = 0.34) (referent model). Additionally, we investigated the genetic association among subgroups of glioma in relation to clinical characteristics, including WHO grade and histological features and molecular properties. The results indicate an association between this variant and an increased risk of glioma in the WHO grade 4, GBM, IDH wild-type, and 1p/19q non-codeletion subgroups (Table 5; Fig. 1). Stratification analyses were performed in all, male and female groups. Although the direction the trends were same, the genetic association were more apparent in female patients (Supplementary Table 2).

Table 5 Logistic analysis of association of rs56345976 in TERT with the risk of glioma and subtypes.
Full size table
Fig. 1
Fig. 1The alternative text for this image may have been generated using AI.
Full size image

The association result of rs56345976 between glioma subgroups and PCs. Logistic regression between glioma subgroups and PCs (n = 480) under additive model, adjusted by age and sex as covariates, was used for calculating ORs (95% CI) and P-values at rs56345976. Each plot indicates the point estimate of ORs on the X-axis shown with 95% CI on the error bars. Significant associations are bolded. PC population control, WHO world health organization, AST astrocytomas, GBM glioblastomas, IDH-mutant IDH1 or IDH2-mutated gliomas, IDH-wildtype IDH-wildtype gliomas, 1p/19q (+) 1p/19q non-codeletion, OR odds ratio, PCs population control, CI confidence interval.

We extended our investigation by examining whether SNP rs56345976 was linked to gene expression in brain tissue using eQTL calculators within the GTEx database (https://gtexportal.org/home/testyourown)33. As results, the risk allele showed 1.12 times higher gene expression than the reference allele (G) (P = 0.042) (Table 6), which possibly corresponds to higher telomerase activity and mRNA levels in malignant glioma12,34. Human Splicing Finder (https://www.genomnis.com/access-hsf), indicated that “A” allele of rs56345976 developed 4 new exonic splicing enhancer motifs and 1 deleted enhancer motif site (Table 7). And another software, FORGE2-TF (https://forge2-tf.altiusinstitute.org/) indicates that regulatory factor X 1 (RFX1) binds to rs56345976 (Table 8).

Table 6 Result of single-tissue expression quantitative trait loci (eQTL) of rs56345976 in TERT.
Full size table
Table 7 In Silico analysis of the rs56345976 by human splicing finder.
Full size table
Table 8 In Silico analysis of the rs56345976 by Forge 2 TF motif information.
Full size table

Discussion

TERT is crucial for reconstituting telomerase activity and enabling its function. In most human cancers, telomerase reactivation occurs during carcinogenesis owing to TERT expression35. TERT somatic mutations are the most common non-coding mutations in human cancer cells. Overexpression of TERT is linked to the development of cancers and the creation of tumors36,37, and TERT regulation is important for cancer cell maintenance.

Our results indicated that a specific locus in TERT (rs56345976) is associated with an increased risk of glioma. Furthermore, rs56345976 might have causal, independent, and dose-dependent genetic effects on the risk of glioma. In the subgroup analysis, the genetic effects in the WHO grade 4, astrocytoma, and GBM subgroups were obvious. In this study consisted of Korean population, we could replicate the genetic effects of 5 SNPs, which were previously reported in Chinese Han, European, US, UK, and multiethnic populations. In silico analysis, eQTL calculators, Human Splicing Finder, and FORGE2-TF revealed different gene expressions, splicing enhancer motifs, and transcript factor that binds to rs56345976 which possibly corresponds to higher telomerase activity and mRNA levels in malignant glioma12,34. Two hotspot TERT promoter mutations (C228T and C250T), which is prevalent in glioma38, have been reported that create binding sites of Ets/TCF transcription factors, and it is associated with 2 to 4 times increased transcription28. Previous study examined interaction between TERT promoter hotspot mutation and TERT SNP by comparing mRNA expression39. Similarly, the transcription factor that binds to TERT can interact with rs56345976, thereby increasing TERT transcription efficiency. ‘A’ allele of rs56345976 increases mRNA level in basal ganglia tissue (Table 6). RFX1 is a tumor suppressive transcription factor that directly downregulates CD44 expression in glioblastoma40. ‘A’ allele of rs56345976 may affect RFX1 binding and transcript efficiency. These results could be a complement to the association between rs56345976 and risk of glioma.

Recently, progress in gene expression analysis, such as molecular profiling, has offered greater predictive information than the WHO classification of glioma41. Mutations in IDH1 and IDH2 are commonly detected in patients with astrocytoma and oligodendroglioma42. In addition, wild-type IDH is typically detected in GBM43. A previous study found that wild-type IDH was more frequent in primary GBM patients44. In this study, all GBM patients showed wild-type IDH. This indicated that the risk of GBM disease was associated with wild-type IDH. In this study, rs56345976 was found to be associated with wild-type IDH (P = 0.003). IDH wildtype is a molecular feature of glioblastoma and pediatric glioma4. Increased levels of TERT have been demonstrated to be associated with improved DNA repair mechanisms45. Recently, SNPs in genes which are related to base excision repair (BER) pathway, nucleotide excision repair (NER) pathway, developmental processes and differentiation of stem cells were reported as risk factors of pediatric glioma46,47,48. So, rs56345976 may be associated with pediatric glioma and BER or NER pathway.

A previous study demonstrated that introns can enhance transcription efficiency49. Another study reported that the transcription levels of genes with introns were higher compared to those without introns in yeast50. In mammals, gene expression involves complex interactions between promoters and enhancers, which are often located at considerable distances within the genome, recent research indicates that certain promoters may interact with regulatory sequences located within introns51. After splicing, most introns are degraded, however, some are only partially degraded and can give rise to functional non-coding RNA byproducts52. The factors that determine whether a splice site is recognized as genuine or as a pseudo splice site remain unclear, particularly given the significant diversity observed in actual splice site sequences53,54.

TERT promoter hotspot mutations, C250T and C228T, are commonly observed in patients with GBM and oligodendroglioma27. The prevalence of TERT promoter hotspot mutations in patients with oligodendroglioma is high55; however, in patients with GBM, the prevalence is specific to race (Supplementary Table 3)30,31,32. A previous study reported that 66.3% of patients diagnosed with GBM and 94.1% of those diagnosed with oligodendroglioma at Seoul National University Hospital have TERT promoter mutations32. Another study reported that TERT promoter hotspot mutations are associated with 1p/19q co-deletion55. However, many patients with GBM do not exhibit 1p/19q co-deletion, whereas oligodendroglioma patients typically do. The variant rs56345976 is significantly associated with GBM but not with oligodendroglioma or 1p/19q co-deletion. Therefore, rs56345976 may be a biomarker for GBMs with TERT promoter hotspot mutations, independent of the TERT promoter hotspot mutations associated with 1p/19q co-deletion status. However, the association between rs56345976 and TERT promoter hotspot mutations requires further study.

Owing to the limitations in statistical power, including small sample sizes in the subgroup analyses of gliomas, caution is required when interpreting the results of this study. While we used age- and sex-matched PCs, the study lacked comprehensive clinical information regarding susceptibility to glioma, which resulted in limited inclusion and exclusion criteria. In spite of using these PCs, and given the challenges in acquiring a large number of control samples, this study could be regarded as substitutive approach to ascertain the genetic influences on gliomas56. So, further clinical research, including mRNA and protein analyses, is required to determine the genetic effects of rs56345976 on gliomas in the Korean population. Additional support from functional studies would be required to establish if the identified SNP is causal.

This study aimed to explore the genetic link between TERT SNPs and glioma risk in a Korean population. This study provides initial evidence suggesting that the intronic polymorphism rs56345976 could play a role in glioma susceptibility in the Korean population. Moreover, it emphasizes that it is crucial for researchers to conduct association studies of TERT variants using glioma samples from local populations.

Materials and methods

Study subjects

This study analyzed 797 participants, comprising 317 patients and 480 controls. Patients with glioma (n = 317) were diagnosed between 2006 and 2016 and recruited from Yonsei University Severance Hospital and cooperating hospitals. The patients were categorized into subgroups of glioma based on histological and molecular properties, as classified by the WHO grading system for CNS tumors between 2007 and 20213,4. Patients with other cancers were excluded after reviewing their clinical records. The population control (PC) samples (n = 480), which excluded individuals with other types of cancer, were obtained from the Korean Genome and Epidemiology Study (KoGES) Consortium and National Biobank of Korea57. The controls, which consist of quality-controlled biospecimen collections, were obtained from population-based cohorts. This included a total of 10,038 blood donors whose ages ranged from 40 to 60 years from the Ansung-Ansan Community-based Cohort in 2001. The Wizard Genomic DNA Purification Kit (Promega, Madison, WI, USA) was used to extract genomic DNA from blood samples.

The IDH mutation and 1p/19q codeletion were analyzed by following established methods at Yonsei University Severance Hospital58. The molecular profiles of the patients were analyzed, including the analysis of O-6-methylguanine-DNA methyltransferase (MGMT) promoter methylation, IDH mutation status, and 1p/19q co-deletion. Immunostaining was performed to detect the IDH1-R132H mutation using a Ventana BenchMark XT Autostainer (Ventana Medical Systems Inc., Tucson, AZ, USA) and an anti-human IDH1 R132H mouse monoclonal antibody (Clone H09L, diluted 1:80; Dianova, Hamburg, Germany). When a mutation in IDH1-R132H was not detected by immunohistochemistry, the sequence of IDH1 codon 132 and IDH2 codon 172 was analyzed. To evaluate the 1p/19q status, fluorescence in situ hybridization (FISH) analysis was performed using LSI 1p36/1q25 and 19q13/19p13 Dual-Color Probe Kits (Abbott Molecular Inc., Abbott Park, IL, USA). Following the Euro-CNS protocols59, an experienced neuropathologist analyzed the acquired images. If the number of “deleted” nuclei was more than 50%, the tumor was classified as “deletion” for the targeted chromosome.

All procedures involving participants in this study were carried out in compliance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all the participants. The protocol for collecting blood samples from donors and isolating genomic DNA from the blood was reviewed and approved by the Public Institutional Review Board designated by the Ministry of Health and Welfare, Korea (P01-201507-31-004).

SNP selection and genotyping

Candidate TERT SNPs with a minor allele frequency (MAF) of over 5% in the Japanese and Han Chinese population from 1000 the Genomes database were selected for genotyping. Based on their positions and linkage disequilibrium (LD) with other SNPs, 27 SNPs in the TERT were selected. Additionally, five SNPs (rs2853677, rs2736100, rs10069690, rs2853676, and rs4975538) that were previously reported as glioma risk-associated SNPs were selected. The primer tool was designed using for Fluidigm SNP Type™ (San Francisco, CA, USA) to detect candidate SNPs. Genotyping was conducted on 797 subjects (317 patients and 480 controls) using the Fluidigm EP1 system (Fluidigm 96.96 SNPtype™). Priming was conducted using 96.96 Dynamic Array™ IFC Controller HX. Polymerase Chain Reaction (PCR) was performed using FC1™ Cycler. Signal intensities were measured using EP1™ Reader Data Collection Software after genotyping reaction. Genotype data were analyzed using the BioMark SNP Genotyping analysis software (version 4.3.2).

Statistical analysis

LD analysis of each genotyped SNPs was performed using Haploview v4.2 software developed by the Broad Institute (http://www.broadinstitute.org/mpg/haploview). To estimate each haplotypes, the PHASE 2.1 software was used60. Logistic regression analysis with an additive model was used to analyze the association with TERT variants by computing P-values, odds ratios (ORs), and 95% confidence intervals. Age and sex were adjusted as covariates using Golden Helix SVS8 software (Bozeman, MT, USA). The genotype distribution, Hardy-Weinberg equilibrium (HWE), and MAF of each SNP were compared between patients with glioma and PCs. Stepwise and conditional logistic analyses were performed using Statistical Analysis System (SAS) software (version 9.4; SAS Inc., Cary, NC, USA) to verify independent associations between significant TERT variants. Subsequently, to identify the detailed genetic effects, a reference model analysis based on the allele distribution of the SNP rs56345976 was conducted using Golden Helix SVS8 software. In silico analysis were performed using the GTEx eQTL calculator (https://gtexportal.org/home/testyourown), Human Splicing Finder (https://www.genomnis.com/access-hsf), and FORGE2-TF (https://forge2-tf.altiusinstitute.org/).