ARID5B Genetic Polymorphisms Contribute to Racial Disparities In Childhood Acute Lymphoblastic Leukemia: A Children's Oncology Group Study

Abstract 8

Acute lymphoblastic leukemia (ALL) is the most common type of cancer in children. Substantial racial differences exist not only in the risk of developing childhood ALL, but also in the outcome of ALL therapy, e.g. children with Hispanic ethnic background exhibit the highest incidence of ALL and the lowest survival rate among major racial/ethnic groups in the US. Although both genetic and non-genetic factors are thought to be important, the exact cause(s) of such racial disparities remains largely unknown. Taking a genome-wide approach, we previously identified single nucleotide polymorphisms (SNPs) in the ARID5B gene that are strongly associated with ALL risk in whites (e.g. rs10821936, P=1.4×10^-15, Nat. Genet. 2009, 41:1001), which was independently replicated in several studies (Nat. Genet. 2009, 41:1006; Blood 2010 115:1765). Interestingly, the ALL risk allele (allele C) at the ARID5B SNP is also associated with ALL risk in black children, but is much less prevalent in blacks than whites (allele C frequency, 18% in blacks and 34% in whites), providing a genetic basis for the lower risk of ALL in black children (Leukemia 2010 24:894). In this study, we evaluated the contribution of ARID5B genetic variations to the disparities in ALL incidence and outcome between Hispanic and white children. To identify genetic variations associated with ALL in each population, we first compared genotype frequencies at 51 ARID5B SNPs between ALL cases and ancestry-matched controls, i.e., between 330 Hispanic children with ALL and 134 Hispanic controls; and between 978 white children with ALL and 1,046 white controls. After adjusting genetic ancestry to control for population stratification, we observed that 10 of 51 ARID5B SNPs were significantly associated with ALL in both populations, and 6 and 10 SNPs were significant only in whites and Hispanics, respectively. In both race groups, the strongest association was observed at rs10821936 (P= 8.3×10^-20 in whites and P=3.7×10^-8 in Hispanics). Interestingly, the frequency of the ALL risk allele (allele C) at rs10821936 was higher in Hispanics (43%) than in whites (33%), consistent with the higher incidence of ALL in Hispanics. In fact, for all 10 SNPs significant in both populations, the risk alleles were consistently more common in Hispanics than whites and conferred greater risk of ALL (higher odds ratios in Hispanics than whites). Additionally, we performed forward selection-based multivariate analyses in each population in which ARID5B SNPs compete against each other on the basis of independent associations with ALL. In whites, rs10821936 was the first and only SNP that entered the multivariate model, suggesting a single causal variant at the ARID5B locus tagged by this SNP. In contrast, in Hispanics, 3 ARID5B SNPs were independently associated with ALL in the multivariate model even after adjusting for genotype at rs10821936. Moreover, these 3 ARID5B SNPs were located in linkage disequilibrium blocks distinct from that tagged by rs10821936, arguing for multiple possible causal variants in the ARID5B gene in Hispanics. Finally, we tested whether ARID5B polymorphisms are related to treatment outcome in 1,605 children enrolled on the COG P9904 and P9905 studies. Of 26 ARID5B SNPs associated with ALL risk, 5 were also associated ALL relapse (P<0.05), with the alleles related to disease risk always linked to poorer treatment outcomes. For instance, the T allele at rs6479778 was more frequent in ALL cases than in healthy controls (P=0.0029 in whites and P=0.0031 in Hispanics), and patients carrying the T allele also exhibited an increased risk of ALL relapse (P=1.3×10^-4). Because the T allele is more common in Hispanics (24%) than whites (14%), this polymorphism is likely to contribute to the racial differences in both disease risk and treatment outcome of ALL. In conclusion, our results indicate that genetic polymorphisms in the ARID5B gene are important determinants of racial disparities in childhood ALL.