EPID-19. SHARED GENOMIC ARCHITECTURE OF GLIOMA AND NEURO-COGNITIVE AND NEURO-PSYCHIATRIC TRAITS REVEALED BY LD-SCORE REGRESSION

Abstract

Listen

Translate article

Abstract BACKGROUND Genome-wide analyses estimate glioma heritability at 25%, yet known risk loci account for just one-third of this risk and suggest that sporadic glioma is a highly polygenic disease with hitherto unaccounted for genomic architecture. LD-score regression leverages results from genome-wide association studies (GWAS) and known patterns of linkage disequilibrium (LD) to estimate correlation between the genetic architecture of multiple phenotypes. We applied LD-score regression to identify associations with neuro-cognitive and neuro-psychiatric traits not amenable to study in prior glioma case-control analyses. METHODS GWAS summary statistics were obtained from the Glioma International Case-Control Consortium (GICC) meta-analysis (Melin, et al. 2017) and for 64 neuro-cognitive and neuro-psychiatric traits primarily from the UK Biobank. Included SNPs had quality scores ≥0.70 and minor allele frequency ≥0.01. Pairwise genetic correlations between traits were estimated using the LDSC package. P-values < 7.8x10-4 (i.e. 0.05/64) were considered significant. RESULTS Significant negative correlations were identified between the genetic architectures of glioma and bipolar disorder (Rg=-0.41, P=1.4x10-9) and schizophrenia (Rg=-0.29, P=7.1x10-9), consistent in both GBM and LGG. Significant positive correlations were identified with measures of educational attainment, including age at educational completion (Rg=0.11, P=2.0x10-4), obtaining a college degree (Rg=0.086, P=4.9x10-4), college attendance (Rg=0.086, P=5.9x10-4), and years of education (Rg=0.081, P=7.7x10-4). These associations were notably stronger with LGG. A number of additional nominal associations were observed, including with anorexia (anti-correlated) and performance on a pair-matching cognitive test (positively correlated). Importantly, LD-score regression did not identify an association between glioma risk in GICC data and Townsend deprivation index in UK Biobank data. CONCLUSIONS These results implicate a shared genetic basis for glioma and several psychiatric and cognitive traits. Further research is needed to understand these associations and to explore underlying mechanisms, including the mediating effects of neuro-inflammation, developmental differences in neural‒glial cell circuitry, and inter-individual variation in myelination.