Genome-Wide Association Study (GWAS) Identifies a Significant Acute Myeloid Leukemia (AML) Susceptibility Locus Near BICRA

Introduction: AML is the most common form of acute leukemia in adults, but hereditary predisposition to AML has been insufficiently studied. To date, investigations of AML heritability have focused on families with a clear history of AML and identified rare highly penetrant germline mutations that segregate with disease. However, the role of common polymorphisms in susceptibility to AML is not clear. We therefore conducted the first large-scale GWAS to identify risk alleles for adult de novo AML.

Methods: Whole-genome genotyping was performed on two independent American case-control sets, using Illumina Infinium Omni-1 Quad-bead arrays. GWAS quality control measures consisted of excluding samples with <94% genotyping yield or <90% European ancestry, and variants with <94% yield, differences among batches or significant (P<10^-6) deviation from Hardy-Weinberg equilibrium. Phasing and imputation yielded a total of 16,058,266 autosomal variants. Association tests were performed using logistic regression including age, sex, and ten principal components as co-variates. Selected variants were genotyped in a German validation case-control set using an Agena MassARRAY System. Results were combined using METAL with fixed effects. P<5x10^-8 was considered the threshold for genome-wide significance.

Results: Two separate GWASs were performed in a total of 1,183 American AML patients from the Alliance for Clinical Trials in Oncology and 2,369 population- and age-matched controls from independent case-control sets (USA1: 894 cases and 1,714 controls; USA2: 289 cases and 655 controls). After combining the results from these two data sets in a meta-analysis, five independent loci with a total of 11 polymorphisms were suggestively associated with AML (P<10^-6), located at: 1q41 (P=7.2x10^-7), 13q22.1 (P=3.1x10^-7), 18q22.3 (P=7.4x10^-7), 19p13.11 (P=4.0x10^-7) and 19q13.33 (P=3.0x10^-7) (Figure 1). Six variants with the lowest P-values from three loci (13q22.1, 19p13.11 and 19q13.33) were selected and genotyped in an additional 350 German patient samples and compared with 1,600 population- and age-matched controls from the German LIFE-Adult study. In the German sample set alone, both the 19p13.11 and the 19q13.33 loci showed support for association with AML (P<0.05). Results from the three independent sample sets were combined in an overall meta-analysis to calculate the final odds ratios (ORs) and significance (Table 1). The overall P-value for the 19q13.33 variant, rs75797233, was 4.15x10^-8 (OR=2.28), which meets the threshold for genome-wide significance (P<5x10^-8), making it the first reported common germline variant significantly associated with AML risk, with a risk allele frequency (RAF) in the cases of 3.1% compared to a RAF of 1.8% in the controls.

rs75797233 is located 13kb 5′ of the BICRA gene, which encodes a key component in a newly described SWI/SNF chromatin remodeling complex that can impact on cancer cell proliferation. GTEx expression data revealed that individuals who have the rs75797233 AML risk allele (thymine [T]) have significantly lower expression of BICRA in whole blood samples compared with individuals who are homozygous for the non-risk rs75797233 allele (adenine [A]; P=0.0002). As rs75797233 is located in a region of open chromatin marked by histone H3 lysine 4 mono-methylation, we examined ENCODE chromatin immunoprecipitation sequencing (ChIP-seq) data for transcription factor binding. Indeed, the GATA2 transcription factor showed a clear ChIP-seq binding peak that encompassed rs75797233 and the surrounding region. Analysis of transcription factor motifs using MotifbreakR indicated that when the non-risk allele of rs75797233 (A) is present it is in a critical position of a GATA2 binding site; however, GATA2 is not able to bind to this site when the rs75797233 risk allele (T) is present (Figure 2).

Conclusions: Our study sheds light on the lower-penetrance heritability of AML and identifies rs75797233 as the first common AML risk allele. rs75797233 might associate with AML risk through mechanisms involving regulation of BICRA by GATA2, and ongoing characterization of the risk locus could yield novel insights into AML disease susceptibility.

View largeDownload slide

View largeDownload slide

Close modal