Etched in code: Revisiting glioma risk factors through the exome

What caused my brain tumor? This question is encountered daily in new patient visits in neuro-oncology clinic. Despite decades of research, the underlying causes of brain cancer are still largely unknown. Rare patients harbor a Mendelian disorder (eg, neurofibromatosis, hereditary nonpolyposis cancer syndrome) or have a known antecedent environmental exposure (eg, prior cranial radiation). However, for the majority of patients, a risk factor is not identified and they are left wondering, why me?

Over the years, numerous cancer-causing factors have been explored in neuro-oncology. Some of the earlier epidemiological studies focused on a potential association between glioma and N-nitroso compounds found in processed meat. Subsequent studies have failed to support these initial observations and have not demonstrated a link between glioma incidence and ingestion of cured meats.¹ Other dietary and nutritional exposures have also not been linked to glioma including intake of alcohol, coffee and tea, vitamin D, vitamin E, antioxidants (eg, vitamin A), smoking, or obesity.² Pooled data from 3 large prospective cohorts showed that no major food group, nutrient, or common healthy dietary pattern was associated with increased glioma risk.³ The search for a cause has also ventured into occupational risk. Initial studies of chemical exposure and pesticides suggested an increased risk in farmers or industrial workers due to chemical and pesticide exposure.⁴ Recent large studies have failed to confirm this association.⁵ In terms of protective factors, allergy and atopic conditions have been associated with lower incidence of glioma. However, the size of this effect may be small.⁶ There has been considerable public concern about the association between cell phones and brain tumors. To date, a link has not been confirmed between glioma and the effect of radiofrequency electromagnetic fields from mobile phones.⁷

Fortunately, technological innovations are changing the landscape of risk factor investigation. Large annotated datasets are more widely and readily available to answer epidemiologic questions with sufficient sample sizes. New computational approaches are also being utilized including “omic” and other genetic and epigenetic data that can be leveraged. Artificial intelligence and deep-learning methods are increasingly available to aid in extracting and exploring “big data” from electronic medical records.

In this issue, Claus et al take an interesting step in exploring risk factors for glioma.⁸ The authors examined whole-exome sequencing data from 1105 adult glioma samples including nearly 40% IDH-mutant gliomas and 60% IDH-wildtype. As expected they confirmed the importance of well-known genes frequently altered in glioma including IDH1, TP53, CIC, ATRX, PIK3CA, EGFR, and others. Interestingly, BRAF V600E mutations while of low prevalence demonstrated the highest sex-specific differences in cancer effect size. For the first time, these authors also report a potential association between haloalkane exposure and glioma risk.

There are several important conclusions to take away from this study. First, the authors found that cancer-causing mutations primarily originate as a consequence of endogenous rather than exogenous factors. The authors implicate aging as a key endogenous contributor to glioma risk. Second, they show that cancer-causing mutations in glioma may vary by sex. Sex-specific differences in glioma incidence and outcomes have been reported for decades. The finding in this study may help elucidate the mechanisms underlying these epidemiological observations. Finally, the authors report a rare association between glioma in men and the molecular signature for environmental exposure to haloalkanes.

This study confirms well-known mutational signatures in glioma. While not particularly novel, the authors demonstrate the reliability of a relatively new method for quantifying the importance of driver and other mutations. Here, the authors established a cancer effect size which was used to quantify the relative contribution of mutational variants on cellular survival, division, and tumorigenesis. This method could be impactful in other datasets. The study is also strengthened by the use of an existing, large whole-exome dataset from well-characterized case repositories (ie, National Cancer Institute’s Genomic Data Commons Data Portal and the GLASS Consortium).

The sex-specific differences reported in this study deserve attention. The incidence of glioma is 50% higher in males. Previous studies have posited an association between glioma and circulating hormones. However, investigations into the relationship between sex hormones and clinical outcomes have been inconclusive. For the first time, these data show that in 2 genes, PIK3CA and PIK3R1, the sites of mutation differ between sexes. Dysregulation of the phosphoinositide 3-kinase (PI3K) pathway is well known to contribute to gliomagenesis. Here, mutations in the PIK3CA gene occurred in the helical domain in females and kinase domain in males. In the PIK3R1 gene, 2 variants were observed to be associated with higher cancer effect size in males (eg, N564D and G376R) but not in females.

It is interesting that the sex-specific differences in PIK3CA and PIK3R1 were only observed in IDH-wildtype tumors and not IDH-mutant tumors. These findings raise interesting questions about whether pregnancy may impact tumor growth in some but not all gliomas, thus, explaining why some patients during pregnancy may experience tumor progression while others do not. This also raises questions about whether the ratio of androgens to estrogens may have differential effects on glioma biology based on the tumor’s underlying molecular profile.

Perhaps the most interesting observation is the haloalkane signature. Environmental causes of glioma have long been suspected but never confirmed due to difficulties in defining and measuring exposure and risk as well as methodologic challenges inherent in studies of rare diseases. Here, Claus et al suggest that rare environmental exposure to haloalkanes may account for the occurrence of glioma, particularly in males. Haloalkanes are used commercially in flame retardants and fire extinguishing equipment. Prior reports have suggested an increased risk of glioma in firefighters.⁹ Haloalkanes have been associated with the development of cholangiocarcinoma in a group of Japanese printing workers but not with development of glioma.¹⁰

While this study supports exploring the association between haloalkane environmental exposure and IDH-wildtype glioma, caution must be exercised in interpreting these data. The study was not designed to test the hypothesis that clinical haloalkane exposure contributes to glioma risk nor to determine a causal link. Epidemiologic studies are needed to follow up these genetically driven findings.

As the neuro-oncology community increasingly operates in an era of molecularly defined classification, we have an opportunity and an obligation to reassess age-old questions through a new lens. This study is a reminder of the need and benefit of revisiting investigations into risk factors for glioma and exploring how they may differ by molecular subtype.

Conflict of interest statement. Dr. R.E.S. serves as a consultant for Monteris Medical Inc and Novocure; he receives an editorial stipend as Section Editor of the Resident and Fellow Section of Neurology and has received research/grant support from the American Academy of Neurology, American Society for Clinical Oncology, American Board of Psychiatry and Neurology, and Jazz Pharmaceuticals.

Authorship statement. This text is the sole product of the author and no third party had input or gave support to its writing.