Whole Genome Sequencing to Elucidate Genetic Modifiers of FVIII Clearance Heterogeneity in Patients with Hemophilia a

Introduction: Significant inter-individual variation in FVIII clearance has been described in patients with haemophilia A (HA). The mechanisms underlying this heterogeneity remain poorly defined. Importantly, previous GWAS studies in healthy populations have described associations between SNPs and plasma FVIII and VWF levels.Furthermore, specific lectin and scavenger clearance receptors have been shown to regulate FVIII clearance in vivo.

Methods: In vivo clearance of rFVIII (Advate) was assessed in 49 adult patients with HA using MyPKFit. Age, ABO blood group, plasma VWF:Ag and VWF propeptide were determined for each patient. Whole genome sequencing (WGS) was performed and selected candidate SNPs of interest were assessed. Associations with FVIII clearance (dL/h/kg), half-life (h), and time to 1% (h) were evaluated, addressing both additive and non-additive effects. Linkage disequilibrium was evaluated to group SNPs.

Results and discussion: Overall, rFVIII half-life varied from 7.7 to 10 hours, and SNP variant allele frequency ranged from 0.07 to 0.74. First, known FVIII and/or VWF clearance receptors were investigated. Rs12229292 (STAB2) was significantly associated with FVIII clearance (β=0.004, p=0.033), and demonstrated a trend towards significance for FVIII half-life (β=-0.97 p=0.057) and time to 1% (β=-6.7, p=0.091) with an expected opposing directionality of effects. Comparable results were observed in a non-additive and multivariable model, including VWF:Ag, ABO blood group and age. Explained variance ranged from 5-9.3% across all models. Likewise, rs2289645 (ASGR2) demonstrated significant associations in an additive multivariable model for FVIII half-life (β=-1.0, p=0.013) and time to 1% (β=-7.8, p=0.016), and a trend towards statistical significance for FVIII clearance (β=0.003, p=0.069). Rs868875 (CLEC4M) was also significantly associated with FVIII clearance in an additive uni- and multivariable model with a β-coefficient of 0.004 (p=0.025) and 0.004 (p=0.012), respectively. Last, rs688 (LDLR) achieved statistical significance for time to 1% in a multivariable model (β=10.4, p=0.027), with a similar trend in the univariable model (β=10.4, p=0.059).

In addition, associations between these SNPs and VWF:Ag were evaluated, to assess the potential biological mechanism of our findings. Only rs12229292 (STAB2) demonstrated an additional significant association with VWF:Ag in both a uni- (β=-17.5, p=0.035) and multivariable model (β=-15.3, p=0.019). Crucially, correcting for VWF:Ag in the previous rs12229292 (STAB2) and rs868875 (CLEC4M) multivariable analyses did not fully eliminate effects, suggesting that part of the observed association between these SNPs and FVIII clearance is independent of VWF:Ag.

Interestingly, significant associations between rs10133762 (TC2N) and FVIII clearance, half-life and time to 1% were observed in a multivariable non-additive model with β-coefficients of -0.009 (p=0.026), 2.1 (p=0.023) and 18.8 (p=0.009), respectively. Explained variance ranged from 7.3 to 9.7%. TC2N has previously been associated with FVIII and VWF levels in GWAS, however, the in vivo function remains undetermined. Crucially, our results suggest that TC2N may impact FVIII and/or VWF clearance mechanisms.

Combined, these STAB2, ASGR2, CLEC4M, TC2N, and LDLR SNPs accounted for 26% of FVIII clearance heterogeneity. Crucially, this further increased to 56% when VWF:Ag, blood group and age were included in the model. The observed effect in this model was mainly driven by rs12229292 (STAB2; β=0.004, p=0.026) and rs868875 (CLEC4M; β=0.004, p=0.008).

Conclusion: In this comprehensive study of deeply characterized HA patients receiving the same rFVIII product with available FVIII PK and WGS, we describe for the first time that a combination SNP genotyping and clinical characteristics explain more than half of the observed FVIII clearance heterogeneity. Crucially, this explained variance is expected to increase by our current efforts to include additional SNPs of interest derived from GWAS studies. Moreover, as illustrated for TC2N, these genomic strategies have high potential to identify novel pathways associated with FVIII and/or VWF clearance, thus being of direct translational relevance.