Modification of Predicted Inhibitor Risk in Non-Severe Hemophilia-a By in silico Analysis of Human Proteome Homology with Wild-Type, FVIII-Derived Peptides

Neutralising antibodies (inhibitors) are increasingly recognized to be a life-time risk in non-severe hemophilia A patients exposed to factor VIII concentrates. It is currently not possible to reliably identify the variable inhibitor risk between individuals. Risk appears to differ between F8 genotypes but also for individuals living with the same F8 genotype. As a T cell dependent process, the wild-type, factor concentrate-derived peptide sequences spanning the F8 mutation position are presented by class II MHC and responsible for driving the T helper response and subsequent B cell response. We hypothesize that the primary sequences of the 20,469 proteins in the human proteome will, coincidentally, contain short primary sequences homologous to key immunogenic peptides derived from the therapeutic factor VIII. We present in silico data and correlation with published registry inhibitor data (Fisher's exact test) to demonstrate the potential impact of such "proteome protection" and future potential to more reliably stratify individuals between low/negligible risk and more significant risk of inhibitor formation.

We utilize a well-validated, computational tool, NetMHC-II, to enable large scale, computational comparison of predicted antigen presentation between endogenous, mutated FVIII derived peptides and factor-concentrate derived, wild-type FVIII peptides spanning all 520 F8 missense mutations listed on www.hadb.org. NetMHC-II analyses peptide presentation by 14 class II MHC HLA-DR alleles, resulting in analysis of 7,280 (520 x 14) permutations of F8 -MHC-II. We identify 56% (n=4,077) of these permutations to be at low/negligible risk of inhibitor formation, at a binding threshold of 500nM, defined as absence of a novel peptide-MHC surface capable of driving a helper T cell response (p=0.005). When cross referenced with potential homologous sequences buried anywhere in the human proteome (http://www.ebi.ac.uk/reference_proteomes), a further 1,237 F8 -MHC-II combinations are afforded "proteome protection" due to direct sequence homology between FVIII-derived peptide and peptide(s) derived from other proteins. This increases the total number of F8 -MHC-II combinations predicted to be unable to drive a T cell response to 73% (n=5,314). The residual 1,966 (27%) F8 -MHC-II combinations are predicted to retain the ability to present novel-interface FVIII-derived peptides to T cells with an IC50<500nM (higher affinity).

Previous work exploring in silico prediction of FVIII-derived peptide presentation may have overestimated the number of F8 -MHC II combinations deemed to be at risk of contributing to inhibitor formation. Our data suggests an additional mechanism "protecting" a larger proportion of those living with non-severe HA from inhibitor formation. The contribution of "proteome protection" further reduces the "at risk" F8 -MHC II permutations to be more representative of clinically observed inhibitor rates. We identify a potential novel tolerance mechanism and provide key data for future in vitro validation strategies.