Genetic Susceptibility Loci for Childhood Acute Lymphoblastic Leukemia Among Japanese

Scrutiny of the human genome through evaluation of common genetic variants has revealed hundreds of disease susceptibility loci. In childhood acute lymphoblastic leukemia (ALL), six regions that have replicated in several populations are now considered known susceptibility loci (ARID5B, IKZF1, CEBPE, CDKN2A, PIP4K2A, and GATA3), but their effects have yet to be fully confirmed in populations of non-European ancestry. Targeted validation attempts based on the same SNPs originally identified in European ancestral populations have been performed in East Asians, but findings have been inconsistent. This may be due to differences in linkage disequilibrium patterns, allele frequency, and/or magnitude of effect between Europeans and East Asians; thus a comprehensive characterization of genetic variation across the targeted genetic loci is required for an appropriate validation attempt in different populations.

Using a large network of hospitals within the Tokyo Children's Cancer Study Group, saliva samples from previously diagnosed childhood ALL patients (aged 0-19 years) were collected between December 2012 and May 2015. Genome-wide single nucleotide polymorphism (SNP) genotyping was performed and resulted in the inclusion of a total of 570 ALL patients, with genetic data available for up to about 500,000 SNPs after quality control exclusions. Control genome-wide data were available for 2,712 previously genotyped samples from the Nagahama Study Group and Aichi Cancer Center Study, Japan. SNP imputation was performed on the combined case-control dataset using ShapeIT and Minimac3, and the 1000 Genomes Project Phase I Version 3 as the reference population. Tests of association between childhood ALL and all available SNP genotypes across the six genes (mentioned above) implicated in previous genome-wide association studies was performed using logistic regression and assuming a log-additive model of inheritance.

Of the six genomic regions examined, SNPs within the IKZF1, ARID5B, and PIP4K2A genes showed a statistically significant association with childhood ALL risk after Bonferroni correction. SNPs with the strongest evidence of association for these three genes included rs7090445 (ARID5B, OR=1.75, P =3.7x10-¹⁷), rs12533431 (IKZF1, OR=1.43, P =4.3x10^-5), and rs11013045 (PIP4K2A, OR=0.76, P =9.5x10^-5). Further examination of these regions indicated a second independently associated locus within ARID5B. Furthermore, we observed that the same previously reported primary ALL susceptibility SNPs for IKZF1 (e.g. rs4132601, rs11978267) and PIP4K2A (e.g. rs10828317, rs7088318) were not associated in Japanese. This highlights the importance of considering regional genetic variation comprehensively when testing the role of previously implicated candidate regions in a different racial/ethnic population. Characterization of the role of CEBPE, CDKN2A, and GATA3 genetic variation in Japanese may benefit from greater statistical power and potentially additional coverage of SNPs within these regions.