LAUSR

Background Resilience to brain damage describes the imperfect correlation between clinical symptoms and the extent of brain injury, whether traumatic, degenerative, or vascular in origin. DNA samples linked to electronic medical records (biobanks) offer unique opportunities to identify genes that underlie multiple brain resilience phenotypes. We use Traumatic Brain Injury (TBI) as our discovery phenotype, since over 8 million Americans sustain a TBI annually, and the pathology underlying TBI is hypothesized to overlap that of other neurological diseases, ranging from persistent headache to dementia. Methods We report the first GWAS of TBI risk, conducted in the Vanderbilt University Biobank, BioVU, which includes >270,000 patients. Results from single SNP analyses can be difficult to interpret and pay a heavy multiple testing burden. We therefore applied a gene-based tool, PrediXcan, to predict gene expression levels in 9155 European-ancestry BioVU patients from genome-wide genotype data in the target sample (BioVU) and a reference transcriptome dataset (GTEx). We tested 11005 genes in whole blood and 9164 genes in brain cortex for association with TBI defined by CDC criteria as one or more ICD codes for ‘Cerebral laceration and contusion’, ‘Concussion’, ‘Intracranial hemorrhage following injury’, ‘Skull fracture’, and ‘Other intracranial injury’. Analyses were adjusted for sex, age (at first TBI or at last ICD code in the medical record), and principal components 1–3. Genes significantly associated with TBI were tested for association with 1309 phenotype codes in a Phenome-Wide Association Analysis (PheWAS). Results TBI was diagnosed in 503 of 9155 patients. The median age at first TBI was 51.7 years, 6.8% of patients were Discussion Our analyses suggest that TMEM163 expression in cortex is associated with resilience to multiple types of brain damage. Further research is warranted to disentangle the complex temporal and potentially confounding relationships between phenotypes.