Germline genomic variants associated with childhood acute lymphoblastic leukemia

Abstract

Using the Affymetrix 500K Mapping array and publicly available genotypes, we identified 18 SNPs whose allele frequency differed significantly(P < 1 x 10(-5)) between pediatric acute lymphoblastic leukemia (ALL) cases (n = 317) and non-ALL controls (n = 17,958). Two SNPs in ARID5B not only differed between ALL and non-ALL groups (rs10821936, P = 1.4 x 10(-15), odds ratio (OR) = 1.91; rs10994982, P = 5.7 x 10(-9), OR = 1.62) but also distinguished B-hyperdiploid ALL from other subtypes (rs10821936, P = 1.62 x 10(-5), OR = 2.17; rs10994982, P = 0.003, OR 1.72). These ARID5B SNPs also distinguished B-hyperdiploid ALL from other subtypes in an independent validation cohort (n = 124 children with ALL; P = 0.003 and P = 0.0008, OR 2.45 and 2.86, respectively) and were associated with methotrexate accumulation and gene expression pattern in leukemic lymphoblasts. We conclude that germline variants affect susceptibility to, and characteristics of, specific ALL subtypes.

Figure 1. Genome-wide P values comparing allele frequency of SNPs between the ALL and combined non-ALL groups according to chromosome

The x axis shows the P values (−log₁₀) for 307,944 germline SNPs. The allele frequency of the SNPs was compared between patients with ALL in the discovery cohort (n=317) and the combined non-ALL control cohort (n=17,958). The SNP whose allele frequency differed most significantly between the two cohorts (P = 1.4 × 10⁻¹⁵) was localized to chromosome 10. The dashed line shows the threshold P value indicating genome-wide significance.