Predicting inhibitor development using a random peptide phage-display library approach in the SIPPET cohort.

Inhibitor development is the most severe complication of hemophilia A (HA) care and is associated with increased morbidity and mortality. This study aimed to use a novel immunoglobulin G epitope mapping method to explore the factor VIII (FVIII)-specific epitope profile in the SIPPET cohort population and to develop an epitope mapping-based inhibitor prediction model. The population consisted of 122 previously untreated patients with severe HA who were followed up for 50 days of exposure to FVIII or 3 years, whichever occurred first. Sampling was performed before FVIII treatment and at the end of the follow-up. The outcome was inhibitor development. The FVIII epitope repertoire was assessed by means of a novel random peptide phage-display assay. A least absolute shrinkage and selection operator (LASSO) regression model and a random forest model were fitted on posttreatment sample data and validated in pretreatment sample data. The predictive performance of these models was assessed by the C-statistic and a calibration plot. We identified 27 775 peptides putatively directed against FVIII, which were used as input for the statistical models. The C-statistic of the LASSO and random forest models were good at 0.78 (95% confidence interval [CI], 0.69-0.86) and 0.80 (95% CI, 0.72-0.89). Model calibration of both models was moderately good. Two statistical models, developed on data from a novel random peptide phage display assay, were used to predict inhibitor development before exposure to exogenous FVIII. These models can be used to set up diagnostic tests that predict the risk of inhibitor development before starting treatment with FVIII.