Genome-wide association study identifies multiple new loci associated with Ewing sarcoma susceptibility

Abstract

Ewing sarcoma (EWS) is a pediatric cancer characterized by the EWSR1-FLI1 fusion. We performed a genome-wide association study of 733 EWS cases and 1346 unaffected individuals of European ancestry. Our study replicates previously reported susceptibility loci at 1p36.22, 10q21.3 and 15q15.1, and identifies new loci at 6p25.1, 20p11.22 and 20p11.23. Effect estimates exhibit odds ratios in excess of 1.7, which is high for cancer GWAS, and striking in light of the rarity of EWS cases in familial cancer syndromes. Expression quantitative trait locus (eQTL) analyses identify candidate genes at 6p25.1 (RREB1) and 20p11.23 (KIZ). The 20p11.22 locus is near NKX2-2, a highly overexpressed gene in EWS. Interestingly, most loci reside near GGAA repeat sequences and may disrupt binding of the EWSR1-FLI1 fusion protein. The high locus to case discovery ratio from 733 EWS cases suggests a genetic architecture in which moderate risk SNPs constitute a significant fraction of risk.

Fig. 1

Manhattan plot of meta-analysis –log₁₀ P-values for the association of each SNP with EWS risk. Association -log10 p-values for each tested genetic variant are plotted. Chromosomes are plotted sequentially across the x-axis with the scale proportional to chromosomal size. Colors are used to visualize differences in chromosome. The dotted line indicates genome-wide significance (P<5×10^-8)

Fig. 2

Conditional analysis at the 20p11.22-23 region. Overall meta-analysis –log₁₀ P-values are plotted in gray in the background. In the foreground, meta-analysis –log₁₀ p-values when the top tagging SNP is the region (rs6047482) is conditioned on is plotted in blue. A second independent signal, tagged by rs6106336, remains