Long telomeres in need of a SNP: Germline contributions of telomere maintenance to glioma

Epidemiologic research has identified a number of endogenous factors associated with differences in adult glioma risk, including male sex, older age, European ancestral origin, positive family history, and common risk alleles at more than 20 genetic loci.¹ Despite the success of recent genome-wide association studies (GWAS) of glioma predisposition, the biologic mechanisms through which risk alleles in target genes exert tumorigenic effects remain poorly characterized. However, common glioma risk alleles have been identified near 4 telomere-related genes (TERC, TERT, OBFC1, RTEL1),² and rare loss-of-function variants in a fifth (POT1) have been linked to a monogenic glioma predisposition syndrome.³ Integrative genomic analyses show that these glioma risk alleles are associated with increased telomere length, highlighting telomere maintenance as a key pathway in gliomagenesis.^3,4 These epidemiologic findings complement tumor profiling studies that identified activating hotspot mutations in the TERT gene promoter in ~80% of glioblastoma and oligodendroglioma patients, where they reactivate telomerase expression and enable replicative immortality.⁵

Genomic data suggest that ~25% of glioma (narrow-sense) heritability is attributable to common genetic variants, but known risk loci account for only 6% of glioma heritability.² While ever-larger GWAS studies are likely to identify novel risk alleles of small effect, alternative analytic approaches have recently emerged that can integrate genomic, epigenomic, and transcriptomic data to potentially reveal the sources of this “missing heritability.” A recent report in the journal effectively implemented this approach to interrogate the relationship between genetic determinants of telomere length and glioma etiology.⁶

Saunders et al begin by performing 2-sample Mendelian randomization (MR) analyses using 16 genome-wide significant variants associated with leukocyte telomere length (LTL) from a published GWAS (N = 78 592). These instrumental variables were applied to a large case-control dataset of 12 488 glioma cases (N = 6183 glioblastoma; 5820 lower-grade glioma) and 18 169 controls. Genetic predisposition to longer telomere length was consistently associated with significantly elevated risk of glioma, glioblastoma, and lower-grade glioma across multiple MR models and sensitivity analyses.

Standard MR tests the overall hypothesis that LTL is associated with glioma susceptibility, but offers limited insight into the effects of individual variants. To investigate the latter, the authors leverage the summary Mendelian randomization (SMR) framework and perform heterogeneity in dependent instruments (HEIDI) tests at LTL-associated loci to evaluate whether regional associations harbor a shared causal signal. These analyses provide evidence of concordant effects on telomere length and glioma susceptibility at PARP1, TERC, CARMIL1, PRRC2A, POT1, and OBFC1, while HEIDI tests suggested the presence of multiple signals at TERT, RTEL1, and ATM. Five of these genes were previously linked to gliomagenesis (TERC, TERT, OBFC1, RTEL1, POT1) and five are novel (PARP1, DCAF4, CARMIL1, PRRC2A, ATM). With the exception of ATM, the allele associated with longer telomere length was also associated with increased risk of glioma at all these loci.

These intriguing findings merit follow-up studies to more fully assess the extent to which LTL-associated variants confer glioma risk. A major challenge lies in disentangling vertical pleiotropy—where a genetic variant influences telomere length, which in turn influences glioma risk—from horizontal pleiotropy, which occurs when a variant affects glioma directly or through multiple pathways not limited to telomere maintenance. Future work could benefit from colocalization methods that explicitly model linkage disequilibrium structure, test for multiple causal variants, and conduct multi-trait fine-mapping.⁷ Summary-level GWAS data have also been used to quantify the proportion of cancer risk for a given variant that is mediated by telomere length.⁸

Saunders et al also present SMR analyses incorporating transcriptome and methylome data to explore mechanisms that may be contributing to glioma susceptibility in LTL-associated regions. Several variants, including those near POT1 and PARP1, were observed to be methylation quantitative trait loci (meQTLs), associated with differences in the methylation status of DNA from normal brain tissues. An additional set of variants were associated with both LTL and gene expression levels in blood and brain tissues, including near PARP1. These exploratory results provide increased rationale for studying epigenetic determinants of glioma risk to discover genes that may be overlooked by GWAS and instead confer risk through methylation-induced changes in gene expression.

Overall, Saunders et al integrate large GWAS of glioma predisposition and of telomere length to explore the shared genomic architecture of these 2 traits. They confirm the etiologic relevance of genetic predisposition to longer telomere length and identify novel associations between glioma risk and LTL-associated variants near several relevant genes, including PARP1 and ATM. It is likely that several of these loci will emerge at genome-wide statistical significance in future, larger GWAS studies of adult glioma. Indeed, a prior MR-based analysis of glioma risk that leveraged 8 LTL-associated variants first identified OBFC1 as a novel risk locus,⁴ which was later validated by GWAS.²

Although the increased risk of adult glioma conferred by telomere maintenance is now well established, several outstanding questions remain. Saunders et al observe significant heterogeneity between GBM and non-GBM tumors, which suggests that further refining these associations by molecular subtypes may reveal effect modification by the tumor’s IDH1/2 status—and particularly by the presence of TERT promoter and/or ATRX mutations. Telomere length is also known to differ significantly by both sex and by race/ethnicity, but the analyses presented did not explore effect modification by sex and included exclusively European-ancestry subjects. A recent multi-ethnic MR study of osteosarcoma observed that genetic predisposition to longer telomere length conferred elevated risk and also revealed significant interaction with race/ethnicity, with the strongest effects observed among Hispanic subjects.⁹

Future studies must expand genetic predictors of telomere length to include over 100 recently discovered loci and to incorporate ancestry-specific determinants of LTL.¹⁰ Although longer telomeres are often seen as biomarkers of healthy aging, the strong and consistent association with increased risk of several cancers, particularly glioma, suggests that increased capacity for telomere maintenance at the cellular level may provide pre-malignant cells with enhanced replicative capacity, buying them the time needed to acquire the TERT or ATRX mutations that appear necessary for immortalization.