PB1190 - Factor VIII inhibitors exhibit either type I or type II kinetics in acquired haemophilia A

Background: Previous studies have reported differences in antibody (inhibitor) kinetics between acquired haemophilia A (AHA) and severe haemophilia A (SHA); with the latter demonstrating linear/first order (type I) kinetics and the former showing non-linear/second order (type II) kinetics with inability to completely inactive FVIII. Due to the rarity of AHA there is no large-scale systematic assessment of FVIII inhibitor kinetics.

Aims: To characterise kinetics of FVIII autoantibodies and evaluate proportions that display type I, II or indeterminate kinetics.

Methods: Adults diagnosed with AHA at the Royal Free Hospital between 2020-2023 were included. Inhibitor quantification was performed using the Nijmegen-Bethesda assay. Kinetics were evaluated using plasma dilution studies. Type I kinetics was defined based on linear FVIII inhibition with ≥97% inactivation of FVIII. Analyses were performed using non-linear regression and a sigmoidal dose-response model with a fitted variable slope.

Results: 34 patients were included (mean age 71.8 years, range 26-97 years). 14 patients (41.2%) exhibited type I kinetics, 17 patients (50%) type II kinetics, and 3 (8.8%) were indeterminate (Figure 1). Plateau mean residual FVIII:C was higher for antibodies displaying type II compared to type I kinetics (19.5 IU/dL vs 3.0 IU/dL, p=0.0001). Inhibitor kinetics were then evaluated using a non-linear dose-response curve without categorisation for kinetics types, yielding a model with poor fit, explaining only 39% of the variability. Refitting these regressions for type I or II inhibitors yielded models that explained 84 and 85% of the variability in the data respectively, indicating that both types of inhibitors are represented in this cohort.

Conclusion(s): FVIII auto-antibodies in this large cohort of AHA patients demonstrated either type I or type II kinetics in roughly equal proportions, supported by statistical analyses based on fitting of FVIII residuals using non-linear regression. Statistical modelling can support visual classification of FVIII inhibitor kinetics and for future studies of FVIII inhibitors.