Deciphering gliomagenesis from genome-wide association studies

The etiology of gliomas is poorly understood. Differences in the incidence of glioma between regions have suggested environmental risk factors: Many have been suspected—including pesticides, nonionizing radiations, diet, pollution, socioeconomic status, etc.—but the only validated associations are ionizing radiation (which increases risk in both adults and children) and history of allergies (which decreases risk in adults). Regarding genetics, around 3%–5% of glioma patients have a first-degree relative affected also by glioma. It is estimated that 1%–2% of adult and 4% of pediatric cases are due to known Mendelian disorders, mostly due to loss-of-function mutations in tumor suppressor genes, such as tumor protein p53 (TP53), mismatch repair (MMR) genes, and neurofibromin 1 (NF1) mutations.¹ Besides these known predisposition syndromes, agnostic unbiased approaches performed over the last 15 years by genome-wide association studies (GWAS) have identified now more than 26 unequivocal genetic loci associated with the risk of gliomas, the last study is based on 12 496 cases and 18 190 controls.² Overall heritability is estimated to represent 25% of the occurrence of gliomas, as estimated by genome-wide complex trait analysis.³ These loci are mapped by single nucleotide polymorphism (SNP) which are anonymous variants, mostly located in noncoding regions, and not associated with biological function, but rather supposed to be genetically linked to a variant involved in gliomagenesis. It is particularly striking to note that GWAS, which is an agnostic approach devoid of any biological presupposition, leads to unequivocal loci mapping in the close vicinity of genes involved in gliomagenesis such as epidermal growth factor receptor (EGFR), cyclin-dependent kinase inhibitor 2A/B (CDKN2A/B), telomerase reverse transcriptase (Tert), TP53, and isocitrate dehydrogenase 1 (IDH1).

Most of these SNP are associated with a very mild relative risk (RR) variation, with usually RR < 1.5. Interestingly several loci are associated specifically with IDHwt, others more numerous to IDHmut gliomas, suggesting distinct mechanistic pathways.⁴ This is of particular interest as IDH mutant enzyme modifies epigenetics, reshapes chromatin structure, and may unravel cis-interactions. Overall, the number and the genetic weight of the polymorphisms identified so far is higher for IDHmut gliomas, which affects younger people compared to IDHwt, an observation that can be put in parallel with the observation that the genetic part of glioma predisposition is higher in children compared to adults. Genotyping these common SNPs in an individual allows to calculate of a polygenic risk score, and this may be suitable to assess risk to develop IDHmut gliomas in young adult patients.⁵

One of these loci, located on 8q24, associated with IDHmut gliomas, and especially oligodendrogliomas, is of particular relevance, and its biological significance has been recently elucidated. Fine mapping of the 8q24 SNP identified rs55705857, whose risk allelic frequency is 0.06, as the SNP with the highest effect.⁶ It confers a 6-fold greater risk of IDHmut glioma, and a 9-fold risk for oligodendroglioma with 1p19q codeletion,⁴ being one of the highest reported inherited genetic associations with cancer, comparable in terms of RR with inherited BRCA1 gene mutations for breast and ovary cancer.

Unlike the majority of SNP identified by GWAS, rs55705857 has probably a causal effect in gliomagenesis of IDHmut gliomas. It resides in an intron of the long noncoding RNA CCDC26. The risk allele G disrupts OCT2/4 binding, increasing the interaction of CCDC26 with the Myc promoter and resulting in an increased Myc expression. It increases tumor penetrance in the context of mutant Idh1R132H, Trp53, and Atrx mutant astrocytoma mouse model.⁷ There is no currently available mouse model reproducing faithfully the oligodendrogliomagenesis and 1p19q codeletion, but it is likely, as suggested by the GWAS⁴ that the effect of the G allele of rs55705857 would be even stronger in triggering oligodendrogliomagenesis, than demonstrated on this astrocytoma model.

Several observations support the link between rs55705857 and IDH mutant gliomas. It is active in oligodendrocyte precursor cells. ATAC-seq shows chromatin accessibility at rs55705857 almost exclusively in IDHmut low-grade gliomas. IDH mutation reshapes tridimensional structure of chromatin and interestingly Hi-C (chromosome conformation capture) data revealed a significant interaction between the risk SNP rs55705857 and MYC in IDH mutant gliomas.⁴

In this issue of Neuro-Oncology, Alpen et al. analyzed genome-wide data from the Australian Genomics and Clinical Outcomes of Glioma (AGOG) consortium for 560 glioma cases and 2237 controls of European ancestry.⁸ The study, limited by the lack of genetic profiling of the tumors (cases were simply classified as glioblastoma, nonglioblastoma, astrocytoma, or oligodendroglioma, but not according to the WHO 2021) stratified the risk by sex. Replicating the results of former larger GWAS studies,⁹ confirmed a greater RR of glioma for women versus male for s55705857 in the region on 8q24.21. They concluded that a greater risk for women associated with variants at 8q24 might provide biological insight into gliomagenesis with implications for risk prediction and personalized treatments. The highest RR associated with women could be related to epigenetic differences between male and female in the context of IDH mutant gliomas.¹⁰

Certainly elucidating the biological basis of this difference is beyond the scope of this confirmatory study. This new GWAS has been performed on a population of European Ancestry, and not surprisingly confirmed most of the previous data performed on a similar population. However, the estimated risks may not be applicable to other populations. GWAS are needed in more diverse populations. As the next challenges, interaction studies are expected to elucidate how epigenetics and environmental factors may interact with genetic susceptibilities. Finally, the biological mechanism that underlies these genetic associations has yet to be elucidated as it has been for s55705857 on 8q24.

Funding

There was no funding related to the preparation of this manuscript.

Conflict of interest statement

L.B. and M.S. have no conflict of interest.

Author Contribution

L.B. and M.S. conceived and wrote the manuscript.