Genetic determinants of glucose-6-phosphate dehydrogenase activity in Kenya

Abstract

Background: The relationship between glucose-6-phosphate dehydrogenase (G6PD) deficiency and clinical phenomena such as primaquine-sensitivity and protection from severe malaria remains poorly defined, with past association studies yielding inconsistent and conflicting results. One possibility is that examination of a single genetic variant might underestimate the presence of true effects in the presence of unrecognized functional allelic diversity.

Methods: We systematically examined this possibility in Kenya, conducting a fine-mapping association study of erythrocyte G6PD activity in 1828 Kenyan children across 30 polymorphisms at or around the G6PD locus.

Results: We demonstrate a strong functional role for c.202G>A (rs1050828), which accounts for the majority of variance in enzyme activity observed (P=1.5×10⁻²⁰⁰, additive model). Additionally, we identify other common variants that exert smaller, intercorrelated effects independent of c.202G>A, and haplotype analyses suggest that each variant tags one of two haplotype motifs that are opposite in sequence identity and effect direction. We posit that these effects are of biological and possible clinical significance, specifically noting that c.376A>G (rs1050829) augments 202AG heterozygote risk for deficiency trait by two-fold (OR = 2.11 [1.12 - 3.84], P=0.014).

Conclusions: Our results suggest that c.202G>A is responsible for the majority of the observed prevalence of G6PD deficiency trait in Kenya, but also identify a novel role for c.376A>G as a genetic modifier which marks a common haplotype that augments the risk conferred to 202AG heterozygotes, suggesting that variation at both loci merits consideration in genetic association studies probing G6PD deficiency-associated clinical phenotypes.

Figure 1

Fine mapping association study of G6PD activity.a, P-values and b, estimated allelic effect sizes from unbiased association tests (covariate: gender), with different colors indicating strength of linkage disequilibrium with c.202G>A (purple square). Gene graphic shows exons (blue) and introns (yellow) for genes annotated at UCSC Genome Browser (hg19); IKBKG=inhibitor of nuclear factor kappa-B kinase subunit gamma.

Figure 2

Association testing for effects independent of c.202G>A.a, -Log10 P-values and b, estimated allelic effect sizes from ‘controlled’ association tests (covariates: gender, c.202G>A genotype). The twelve polymorphisms exhibiting significant effects at P<0.001 are colored according to whether they exhibit positive ES (green) or negative ES (red). Gene graphic shows exons (blue) and introns (yellow) for genes annotated at UCSC Genome Browser (hg19); IKBKG=inhibitor of nuclear factor kappa-B kinase subunit gamma.

Figure 3

Haplotype-based association testing for effects independent of c.202G>A. Results of association testing of a, full-length and b, partial/twelve-site haplotypes are shown for all haplotype tests suggestive of association (P<0.025, covariates: gender, c.202G>A genotype), with haplotypes defined by patterns of ancestral (blue) and derived (yellow) alleles. SNP labels for the twelve sites that comprise the partial haplotypes are colored according to whether they exhibited positive ES [green] or negative ES (red) in single SNP controlled tests (Figure 2). c, G6PD activity distribution in c.202G>A wild-type homozygotes (202G or 202GG) as stratified by carriage of H+, H-, or neither haplotype. Pairwise t-test results are shown via asterisk (*: P<0.05 and ***: P=9.1×10⁻⁵). In box-and-whisker plot, center of box represents median, edges of box represent upper and lower quartile bounds, and whiskers represent either minimum and maximum values or 1.5 times the interquartile range, whichever is less extreme.

Figure 4

Genetic determinants of G6PD deficiency status.a, G6PD activity distribution stratified by genotype at c.202G>A. b, G6PD activity in 202AG heterozygotes stratified by genotype at c.376A>G (NB: the 202A derived allele always exists on a 376G derived allele background), with contingency table (inset) for qualitative test of association at c.376A>G in 202AG heterozygotes. Also shown in both plots is the reference line (blue) dividing normal vs. deficient for qualitative tests of G6PD deficiency. In box-and-whisker plots, center of box represents median, edges of box represent upper and lower quartile bounds, and whiskers represent either minimum and maximum values or 1.5 times the interquartile range, whichever is less extreme.