LAUSR

TO THE EDITOR: In a large trans-ethnic meta-analysis GWAS of childhood acute lymphoblastic leukemia (ALL), we identified three putatively novel susceptibility loci (MYB/HBS1L, TET1, and NRBF2/JMJD1C) and two additional independent risk associations at previously reported loci (CEBPE and 17q12). The known and novel ALL risk alleles together explained about 24% of familial relative risk in both nonLatino white (NLW) and Latino (LAT) populations, respectively. Genome-wide association studies (GWAS) have identified a number of risk loci for childhood acute lymphoblastic leukemia (ALL) [1–6] and estimated the heritability to be 21% [7]. The known risk loci together account for a relatively small proportion of the total variance in genetic risk [7], suggesting that additional susceptibility alleles may be discovered. Furthermore, published GWASs predominantly investigated only European-ancestry individuals, despite the racial and ethnic disparities in ALL incidence and outcomes [8, 9]. Given this context, we performed a trans-ethnic GWAS of childhood ALL in a discovery panel of 76,317 individuals, including 3482 cases and 72,835 controls distributed across four ethnic cohorts (African Americans, AFR; East Asian Americans, EAS; Latino Americans, LAT; Non-Latino White Americans, NLW; Supplementary Information). After quality control filtering, our dataset consisted of 124, 318, 1878, 1162 cases and 2067, 5017, 8410, 57,341 controls in AFR, EAS, LAT, and NLW, respectively (Table S1; Fig. S1, Supplementary Information). Furthermore, we tested the association at 7,628,894 imputed SNPs, including low frequency (minor allele frequency, MAF, between 1–5%) variants that were not previously systematically tested. We aggregated summary statistics across the four ethnic groups in a fixed-effect metaanalysis (Fig. S1). The genomic control inflation factor was 1.022 after excluding 16 previously reported ALL-associated loci (Table S2), suggesting our meta-analysis was reasonably robust to any confounding due to population stratification (Fig. S2). Of 16 previously published risk loci, all were nominally associated with ALL (P < 0.05) or have a SNP nearby with strong association (Table S2). Most of these risk SNPs showed consistent direction of effects across ethnic groups and little evidence of heterogeneity (but note C5orf56 and TLE1 in Table S3). Given the larger sample size and trans-ethnic analysis, the best associated variants in our analysis may reflect the more likely causal/shared association across populations. More importantly, we discovered three putatively novel susceptibility loci: one at 6q23 and two at 10q21 (Table S4, Fig. S2). The strongest association signal in 6q23 is at rs9376090 (P= 8.23 × 10, OR= 1.27) in the intergenic region between MYB and HBS1L (Fig. 1A). A locus in 10q21 was identified with the lead SNP rs9415680 (P= 7.27 × 10, OR= 1.20), within a broad association peak and apparently long-range LD with SNPs overlapping NRBF2, JMJD1C, and REEP3 (Fig. 1B). A second 10q21 locus was identified 5Mb upstream with lead SNP rs10998283 (P= 3.92 × 10, OR= 1.15) in an intronic region in TET1 (Fig. 1C). Each of the three putatively novel loci harbors genes and/or variants with a role in hematopoiesis and leukemogenesis, and is within larger chromatin regions containing several genes in a B-lymphoblastoid cell line that mirrors the differentiation state of the majority of childhood ALL (Supplemental Information, Fig. S3–S5). We found little difference in association with different ALL subtypes (Table S5). We tested for association of the three novel variants and their LD proxies (with P < 5 × 10; n= 141 variants) in the independent COG/WTCCC and CCLS replication cohorts (Fig. S1, Supplementary Information). For MYB/HBS1L, where the association with ALL was driven by NLW in the discovery cohort, we replicated the signal in COG/WTCCC (rs9376090, PCOG= 4.87 × 10 , PCOG+discovery analysis = 1.23 × 10; Table S6), but not in CCLS likely owing to its small NLW cohort. For TET1, in which the association was driven by LAT in the discovery cohort, three of the four SNPs with P < 5 × 10 in the discovery cohort nominally replicated in CCLS (lead SNP