To the editor:
In this issue of Blood, Migliorini et al identify rs3824662 in GATA3 as a novel susceptibility locus for childhood acute lymphoblastic leukemia (ALL) via genome-wide association analyses of European children.1 Their results suggest that rs3824662 is strongly associated with nonhyperdiploid B-cell ALL (B-ALL), especially those cases lacking the TEL-AML1 fusion and having later age at diagnosis. Importantly, they and others have also linked the GATA3 region to poorer ALL outcomes.1,2 Hispanic children have a greater incidence of ALL,3 a later average age at diagnosis,4 and increased relapse compared with white children.5 We therefore investigated whether single nucleotide polymorphisms (SNPs) in GATA3 increase risk of ALL in Hispanics and whether they are associated with genetic ancestry by using genome-wide SNP data from 297 Hispanic children with B-ALL (97 hyperdiploid, 39 TEL-AML1 fusion–positive, 118 nonhyperdiploid/non-TEL-AML1 fusion–positive, 43 unknown molecular subtype) and 454 Hispanic controls from the California Childhood Leukemia Study.6
We found the most strongly associated variant from Migliorini et al (rs3824662; P = 8.62 × 10−12) has a significant effect in the same direction in our data (P = .046; odds ratio [OR], 1.23). The magnitude of this effect was not stronger in our nonhyperdiploid, non-TEL-AML1 fusion–positive cases (OR, 1.25). In total, 11 of 68 genotyped SNPs across the GATA3 region were associated with B-ALL in our sample (P < .05) (Figure 1), with the strongest association at rs1271899 (OR, 0.62; 95% confidence interval [CI], 0.46 to 0.84; P = 1.7 × 10−3). rs1271899 was also the most strongly associated variant when analysis was limited to nonhyperdiploid, non-TEL-AML1 fusion–positive cases, in whom it showed a slightly stronger effect than among all B-ALL (OR, 0.59; 95% CI, 0.40 to 0.88; P = 9.5 × 10−3).
Association of SNPs in the GATA3 locus with B-ALL risk among Hispanic children, by subtype. Association of 68 directly genotyped SNPs (black) and 477 imputed SNPs (gray) with B-ALL risk, adjusted for the first 5 principal components. Circles denote associations for B-ALL cases compared with controls. Triangles denote associations for nonhyperdiploid B-ALL cases without TEL-AML1 fusion compared with controls. Haplotype structure in the Hispanic sample appears below gene names, with darker shading indicating higher R2 values and greater correlation between SNP genotypes. The strongest association in the Hispanic samples is at rs1271899 located in haplotype block 5. The strongest association from Migliorini et al,1 rs3824662, is located between blocks 5 and 6 and is weakly but significantly linked to rs1271899 (R2, 0.10; P < .0001).
Association of SNPs in the GATA3 locus with B-ALL risk among Hispanic children, by subtype. Association of 68 directly genotyped SNPs (black) and 477 imputed SNPs (gray) with B-ALL risk, adjusted for the first 5 principal components. Circles denote associations for B-ALL cases compared with controls. Triangles denote associations for nonhyperdiploid B-ALL cases without TEL-AML1 fusion compared with controls. Haplotype structure in the Hispanic sample appears below gene names, with darker shading indicating higher R2 values and greater correlation between SNP genotypes. The strongest association in the Hispanic samples is at rs1271899 located in haplotype block 5. The strongest association from Migliorini et al,1 rs3824662, is located between blocks 5 and 6 and is weakly but significantly linked to rs1271899 (R2, 0.10; P < .0001).
Imputation to 1000 Genomes revealed 45 additional SNPs near GATA3 that were associated with B-ALL in our data (P < .05), although these variants had effect sizes similar to our top genotyped SNP, rs1271899. We found a modest association between the rs3824662 risk allele and later age at diagnosis (P = .085). This attenuated effect may be due to our smaller sample size or to a dilution of any age effect by the later average age at diagnosis experienced by Hispanic children with B-ALL.4
We have previously shown that B-ALL risk alleles in ARID5B, CDKN2A, CEBPE, and PIP4K2A are associated with Amerindian ancestry in Hispanics.7 Here, we observed that the GATA3 rs3824662 risk allele was also significantly associated with increased genome-wide Amerindian ancestry, in both Hispanic cases (r, 0.29; P < 1.0 × 10−4) and Hispanic controls (r, 0.23; P < 1.0 × 10−4). By using previously described methods,7 we quantified the contribution of this SNP to the increased ALL incidence observed in Hispanics relative to populations of exclusively European ancestry. Rs3824662 accounted for a 1.11-fold increased rate of B-ALL in Hispanics versus Europeans (95% CI, 1.02 to 1.21), equivalent to the ethnic incidence rate ratio previously observed for rs7089424 in ARID5B.7
Our results provide independent confirmation that variation near GATA3 confers risk of childhood ALL. Furthermore, GATA3 risk alleles contribute to the increased ALL incidence rate observed in Hispanics relative to Europeans and may underlie their poorer outcomes. Fine-mapping via SNP imputation localized the association peak in Hispanics to an ∼15-kb stretch in the shared promoter region of GATA3 and the GATA3 antisense RNA (GATA3-AS1).8 Although the GATA3 risk variants identified by Migliorini et al do not affect protein coding, a significant imputed SNP in our data (rs1244185) is predicted to affect transcription factor binding in GATA3-AS1 and may influence GATA3 regulation.9
Authorship
Acknowledgments: We wish to acknowledge Professor Mel Greaves for his collaborative spirit and helpful discussion about the GATA3 locus. The authors gratefully acknowledge the clinical investigators at the following collaborating hospitals for help in recruiting patients: University of California, Davis Medical Center (Dr Jonathan Ducore), University of California, San Francisco (Drs Mignon Loh and Katherine Matthay), Children’s Hospital of Central California (Dr Vonda Crouse), Lucile Packard Children’s Hospital (Dr Gary Dahl), Children’s Hospital Oakland (Dr James Feusner), Kaiser Permanente Roseville (formerly Sacramento) (Drs Kent Jolly and Vincent Kiley), Kaiser Permanente Santa Clara (Drs Carolyn Russo, Alan Wong, and Denah Taggart), Kaiser Permanente San Francisco (Dr Kenneth Leung), and Kaiser Permanente Oakland (Drs Daniel Kronish and Stacy Month). We also recognize the assistance of Yang Wang and Ling-i Hsu at the University of California, Berkeley. We mourn the unexpected passing of Pat Buffler, our consummate leader, whose life brought joy to those she knew and whose work brought health to countless more.
Contribution: K.M.W. and A.J.d.S. conceived and designed the study; L.B., P.B., A.P.C., C.M., W.R., K.M.W., and J.L.W. assisted in assembling the data; K.M.W. analyzed and interpreted the data; A.J.d.S. and K.M.W. wrote the manuscript; and all authors critically reviewed and edited the manuscript for intellectual content and gave final approval of the manuscript.
Conflict-of-interest disclosure: The authors declare no competing financial interests.
Correspondence: Kyle M. Walsh, University of California, San Francisco Helen Diller Family Comprehensive Cancer Center, Box 0520, 1450 3rd St, HD276, San Francisco, CA 94143-0520; e-mail: kyle.walsh@ucsf.edu.
