B. Wang,1 C. Wu,1 L. Roskos2; 1AstraZenica/MedImmune, Mountain View, CA, 2AstraZenica/MedImmune, Gaithersburg, MD
BACKGROUND: The development of a new drug is a lengthy and costly process with low probability of success. Continuum of “learn/confirm/predict” by translational and clinical modeling and simulations (M&S) was implemented at every decision point for mavrilimumab, a human monoclonal antibody targeting GM-CSFRα. Mavrilimumab is currently being developed for the treatment of rheumatoid arthritis.
METHODS: At discovery stage, a translational model integrating biology, receptor internalization kinetics and endogenous IgG kinetics was utilized to set an antibody affinity goal. The binding affinity of the lead molecule and nonhuman primate data were incorporated in the model to predict the pharmacokinetics (PK) and receptor occupancy (RO) in humans for first-time-in-human (FIH) starting dose recommendation. Mechanistic PK modeling facilitated the transition from single weight-based intravenous dosing for FIH to multiple fixed subcutaneous (SC) dosing of mavrilimumab for a Proof-of-Principle (PoP) study. Efficacy modeling and stochastic simulations identified 150 mg as the top dose for a Proof-of-Concept (PoC) study.
RESULTS: The lead molecule, mavrilimumab, met the affinity goal and the PK in RA patients was well predicted. PK and efficacy outcome from the PoP study confirmed appropriate selection of SC doses and every-two-week dosing interval. Although 150 mg was not evaluated in the PoP study, improved efficacy was observed at this dose level in PoC as predicted by a priori clinical simulations. Informative dropouts were further included in the PoC efficacy model for selection of optimal Phase III dose.
CONCLUSION: By rational recommendations of an antibody affinity goal, safety margin and clinical dosing regimens, M&S greatly facilitated the discovery and development of mavrilimumab.