Abstract 5221: Identifying acute lymphoblastic leukemia risk loci in latino children via admixture mapping

Acute lymphoblastic leukemia (ALL) is the most common childhood cancer and accounts for 25% of malignancies before the age of 20. Despite decades of inquiry into ALL causes, Latino children have up to 1.4 times the rate of ALL compared to their non-Latino White counterparts. The etiology of ALL is complex and this disparity in risk has not been fully explained by social factors alone, suggesting the role of genetic variants and ancestry. Previous literature has implicated variants in IKZF1, ARID5B, GATA3, PIP42KA, and ERG with greater risk allele frequencies or population-specific risk effects in Latinos. Indigenous American ancestry has also been previously associated with ALL risk, relapse, and poor prognosis. To elucidate the genetic and ancestral etiology of ALL, we conducted the first admixture mapping analysis of 1930 childhood ALL cases and 8520 controls of self-identified Latino ethnicity. We uncovered putative admixture associations on chromosomes 2, 7, 10, and 15 (P = 3.21 x 10−5 to 7.32 x 10−8 among top associations across the four loci). Two of the four loci, on chromosomes 2 and 15, were previously not known to be associated with ALL. Following imputation with the TOPMed reference panel, we fine-mapped each putative locus by testing the ALL risk association of each variant with a minor allele frequency greater than 1%. The top associated variants at each locus showed substantial frequency differences between ancestries from the gnomAD or 1000Genomes project. For example, the top associated variant on chromosome 2 has a risk allele frequency of 43% in Native Americans from the Human Genome Diversity Panel, but a frequency of 0.10% in Non-Finnish Europeans in gnomAD. Including these variants in the admixture mapping model greatly attenuated the admixture signals at each locus, suggesting that they are likely the causal SNPs or close proxies of the causal SNPs. The top associated variants from three out of the four loci passed regional multiple testing burden and mapped to genes MGAT5, IKFZ1 and ARID5B on chromosomes 2, 7 and 10, respectively. The top two SNPs on chromosome 15, which narrowly missed the regional threshold, mapped to genes RAB11A and MEGF11. We are pursuing replication of putative novel loci on chromosomes 2 and 15 in additional Latino GWAS datasets. Our results suggest that ALL risk variants with higher frequencies in individuals with Native American ancestry may contribute to the observed increased risk of ALL in Latino children