Prediction of Monoclonal Antibodies Pharmacokinetics in Human: Identification of a Reference Neonatal Fc Receptor...

Prediction of monoclonal antibody (mAb) pharmacokinetics (PK) in drug development remains challenging due to the lack of a standardized method for predicting elimination based on mechanistic pathways. Among the processes implemented in the physiologically based pharmacokinetic (PBPK) models for large molecules, FcRn-mediated recycling constitutes the predominant mechanism influencing the elimination of mAbs. In the present study, we assessed the predictivity of a generic value for the dissociation constant (Kd) for FcRn (Kd_FcRn) in humans, identified based on clinical data, to provide means for mechanism-based PK projections for mAbs in first-in-human (FIH) trials. We compiled a database of digitalized linear PK profiles for 50 mAbs administered intravenously in humans. Subsequently, the database was randomly divided into a training and a test data set, using a 7:3 ratio. For each drug in the training data set, a generic PBPK model was set up in PK-Sim, and a drug-specific Kd_FcRn parameter was estimated through data fitting. The median of estimated drug-specific Kd_FcRn was 1.05 μM and was used for naïve predictions of the PK of the drugs in the test data set. Plasma exposure (AUC) and terminal half-life were accurately predicted for 80% and 60% of the drugs in the test data set, respectively, with a prediction error within the 0.80-1.25-fold range. Additionally, 100% of the test data set showed prediction errors within the 0.50-2.00-fold range for both plasma exposure and half-life. The median of the estimated drug-specific Kd_FcRn determined using the whole database with 50 mAbs was 1.07 μM and was retained after evaluation as a more accurate default Kd_FcRn value. The reported results provide a large database of mAbs PBPK models with estimated Kd_FcRn values using PK-Sim, and a validated default Kd_FcRn value of 1.07 μM to perform naïve predictions of mAbs linear PK in the context of FIH trials.