Background: Genome-wide association studies (GWAS) have identified several genes associated with childhood acute lymphoblastic leukemia (ALL) that are also implicated in variation in hematological traits. We performed a comprehensive study of the shared heritability of blood cell traits and ALL, and investigated whether genetic variation in blood cell traits may underlie ALL risk.
Methods: We leveraged genome-wide single nucleotide polymorphism (SNP) data from: 1) 335,332 European ancestry, cancer-free subjects without immune deficiencies or hematopoietic disorders in the UK Biobank (UKB); and 2) a childhood ALL GWAS meta-analysis including 2,121 cases and 59,965 controls of European ancestry, to investigate the shared genetic basis of blood cell trait variation and ALL susceptibility.
Co-heritability between blood cell profiles and ALL susceptibility was evaluated by LD score regression. To formally test variation in blood cell traits as potential causal pathways in ALL development, we conducted Mendelian randomization (MR) analyses, which use genetic predictors of blood cell phenotypes to overcome potential confounding and reverse causation in directly measured blood counts. Genetic instruments for MR were developed using a two-stage GWAS of 6 cell types (lymphocytes, monocytes, neutrophils, eosinophils, basophils, platelets), and their ratios (lymphocyte/monocyte - LMR; neutrophil/monocyte - NLR; platelet/lymphocyte - PLR) in the UKB. Next we applied multi-trait analysis of GWAS (MTAG) to improve power for identifying trait-specific loci by exploiting the correlated nature of blood cell traits. Causal odds ratios (OR) for ALL were estimated per 1 standard deviation increase in normalized cell counts (109cells/L) or 1-unit increase in cell type ratios.
Results: Using genome-wide SNP data, we found that ALL has a heritability of hg=0.235 (95% confidence intervals, CI:0.203-0.268) in European ancestry individuals. In LD score regression, ALL susceptibility was correlated at the genetic level with overall blood cell counts (rg=0.070, P=0.003), lymphocyte counts (rg=0.088, P=0.004), LMR (rg=0.065, P=0.012), and PLR (rg= -0.072, P=0.001).
Genetic instruments for MR were selected from independent (LD r2<0.05) variants with P<5×10-8 in the discovery blood cell trait GWAS/MTAG (N=234,778) and P<0.05 in the replication analysis (N=100,554). The resulting genetic predictors explained between 4.0% (basophils, NSNP=144) and 23.9% (platelets, NSNP=661) of trait variation.
Our findings lend support to a modest but significant causal effect of variation in lymphocyte counts (OR=1.16, 95% CI:1.01-1.33), LMR (OR=1.23, 95% CI:1.07-1.41), NLR (OR=0.67, 95% CI:0.54-0.83), and PLR (OR=0.80, 95% CI:0.70-0.92) on ALL risk. There was no evidence of directional pleiotropy based on the MR Egger intercept test (P>0.05). However, significant heterogeneity among SNP-specific causal effects (Cochran's Q P<0.05) indicated potential for confounding due to balanced pleiotropy. In sensitivity analyses using MR-PRESSO framework, which corrects for distortion in causal effects due to pleiotropy, the associations with ALL risk became attenuated for lymphocytes (OR=1.14, 95% CI:0.98-1.32), but persisted for LMR (OR=1.18, 95% CI:1.01-1.38), NLR (OR=0.76, 95% CI:0.60-0.97), and PLR (OR=0.82, 95% CI:0.70-0.95).
Next, we searched for individual variants that may underlie the effects on ALL risk using heterogeneity-based clustering of genetic instruments. This analysis confirmed that regulation of blood cell traits partially mediates the effects of known ALL risk variants: rs4948492, rs4245597 (ARID5B), rs2239630 (CEBPE), and rs78697948 (IKZF1). We also identified putative novel ALL risk loci at chromosome 2q22 (OR=1.28, P=2.5×10-6), chr6q23 (OR=1.21, P=9.3×10-6), and chr13q12 (OR=1.79, P=3.2×10-5), though further analyses are required to confirm these associations and identify likely causal variants and genes at these loci.
Conclusions: Our findings suggest that dysregulation of blood cell profiles characterized by a genetic predisposition to increased lymphocyte counts, particularly in relation to neutrophils, monocytes, and platelets, confers a modest but significant increase in ALL risk. Additional research is required to determine whether a genetic propensity to higher lymphocyte levels is associated with ALL subtypes or clinical outcomes in ALL patients.
Ma:Celgene/BMS: Research Funding; BMS: Consultancy.
Asterisk with author names denotes non-ASH members.