Abstract
Lower grade gliomas are invasive brain tumors that are difficult to completely resect neurosurgically. They often recur following resection and progress, resulting in death. Although previous studies have shown that specific germline variants increase the risk of tumor formation, no previous study has screened many germline variants to identify variants predictive of survival in patients with glioma. In this study, we present an approach to identify the small fraction of prognostic germline variants from the pool of over four million variants that we variant called in The Cancer Genome Atlas whole-exome sequencing and RNA sequencing datasets. We identified two germline variants that are predictive of poor patient outcomes by Cox regression, controlling for eleven covariates. rs61757955 is a germline variant found in the 3' UTR of GRB2 associated with increased KRAS signaling, CIC mutations, and 1p/19q codeletion. rs34988193 is a germline variant found in the tumor suppressor gene ANKDD1a that causes an amino acid change from lysine to glutamate. This variant was found to be predictive of poor prognosis in two independent lower grade glioma datasets and is predicted to be within the top 0.06% of deleterious mutations across the human genome. The wild-type residue is conserved in all 22 other species with a homologous protein. IMPLICATIONS: This is the first study presenting an approach to screening many germline variants to identify variants predictive of survival and our application of this methodology revealed the germline variants rs61757955 and rs34988193 as being predictive of survival in patients with lower grade glioma.