Complex Relationship of Blood Group H, ABO Glycans, and Von Willebrand Factor Levels

Background: ABO(H) is a carbohydrate blood group system expressed in multiple tissues including red blood cells, blood vessels, and mucosal surfaces. ABO is the largest known genetic modifier of plasma VWF level (VWF:Ag). It has been hypothesized that the effect of ABO on VWF is mediated by H glycan density. FUT2, the gene underlying Secretor phenotype, encodes a glycosyltransferase synthesizing H antigen in mucosal tissues, and variation in the gene has also previously been associated with VWF:Ag, but past studies have been conflicting. To clarify these relationships, we studied the relationship between VWF:Ag, ABH glycans, and FUT2 genotype.

Methods:

The primary study group was a representative cohort of US blood donors from the Retrovirus Epidemiology Donor Study (REDS, N=499). A validation cohort of unrelated individuals was created from the Genes and Blood Clotting Study (GABC), healthy siblings between ages 14 and 35 years from University of Michigan, Ann Arbor, by choosing a random individual from each family (N=488).

VWF:Ag was determined by ELISA in platelet poor plasma. Forensic techniques were adapted to detect ABH glycans in whole blood (REDS) or RBC-rich, frozen buffy coat (GABC). A and B glycans were detected using anti-A or anti-B (Immucor). A biotinylated Ulex europaeus agglutinin (UEA lectin, Vector Labs) was used to detect H. Relative A, B, and H antigen density was quantified on dot blots with ImageQuant (GE). In REDS, functional FUT2 alleles (Secretor) were determined by Sanger sequencing of FUT2 exon 2. FUT2 copy-number was assayed with real-time quantitative PCR. In GABC, FUT2 genotypes were determined through SNP genotyping (Illumina).

In REDS, genotype data was phased (BEAGLE) to identify functional haplotypes. In GABC, functional alleles were inferred from the genotypes of two SNPs (rs601338, rs1047781) that determined secretor status in nearly all samples in REDS (see below). Multivariate regression was applied to VWF:Ag as a function H antigen density and O vs. non-O blood group, both separately and within the same model.  All models were adjusted for age, gender, and self-reported ethnicity.

Results:

ABO blood group frequencies in both cohorts were similar to that of the US population. VWF:Ag differed significantly between ABO blood groups with lower values in blood group O versus non-O (REDS: ratio = 0.75, 95% CI [0.71, 0.80], p<2.2e-16, log-linear regression). H glycan density also differed between ABO blood group (p<2.2e-16, Kruskal-Wallis; O vs. non-O, ratio = 2.03, 95% CI [1.88, 2.19], p<2.2e-16). A significant fraction of H glycan density variation was attributable to ABO (REDS r²=0.45; GABC r²=0.34), but a wide range of H glycan density was observed in all blood groups, including within blood group O. In multivariate regression, both H glycan density and O versus non-O blood group, considered separately or in the same model, were significantly associated with VWF:Ag (REDS combined model: p = 2.9e-5 and 2.7e-4 for H glycan and blood group, respectively). In GABC but not in REDS, regression was also significant for an interaction between H glycan density and blood group (p = 0.031). Both cohorts estimated a stronger association of VWF:Ag with H in non-O blood groups.

FUT2 sequencing in the REDS cohort identified 22 distinct FUT2 haplotypes. 99.5% (497/499) of individuals could be accurately assigned FUT2 genotypes based on the common W154* (rs601338) polymorphism alone. In GABC, this decreased to 94% (476/488) due to higher frequency of a hypomorphic allele (rs1047781) more commonly found in Asian individuals. In REDS, quantitative PCR did not identify copy-number variation at FUT2. There was no significant association between FUT2 genotype-predicted function and VWF:Ag or H glycan density in either cohort.

Conclusion:

H glycan density correlated with VWF:Ag in both cohorts.  H glycan density could mediate the association between ABO blood group and VWF:Ag.  If so, this suggests a non-linear, complex relationship between H and VWF and a separate ABO mechanism cannot be excluded. In both cohorts, there was no association of FUT2 (Secretor) genotype with VWF. Our data suggest prior focused approaches to FUT2 genotyping are at risk to undercall non-functional alleles, particularly in cohorts with non-European ancestries. Taken together, these indicate blood group H may be a significant modifier of VWF:Ag, and that this effect is not due to the influence of Secretor phenotype.