PAGE. Abstracts of the Annual Meeting of the Population Approach Group in Europe.
PAGE 28 (2019) Abstr 8831 [www.page-meeting.org/?abstract=8831]
Poster: Drug/Disease modelling - Absorption & PBPK
Stephan Schaller (1), Shayne Watson (2), Doug Marsteller (2), Micha Levi (2)
(1) esqLABS, Germany, (2) Teva Pharmaceuticals, USA
Creation of molecules that provide more optimal characteristics for consumers continuously evolves in the drug development industry. Monoclonal antibodies (mAb) provide better target specificity and typically a longer systemic exposure resulting in fewer negative effects from the intervention and more prolonged positive effects due to the persistent maintenance of concentration levels. When testing new molecules with these attributes, the inherent risk associated with unknown safety profiles in human also exists. Thus, developing a strategy to preferentially degrade this type of molecule using competition at the FcRn receptor could provide one tool to mitigate this risk. The modelling approach assessed this hypothesis by using exogenous IgG to reduce the time-concentration profile of an extended half-life monoclonal antibody.
A human physiologically-based pharmacokinetic (PBPK) model for mAbs was built based on extrapolation from a previously developed (1) Non-Human Primate (NHP) PBPK model using PK-Sim® as part of the Open Systems Pharmacology Suite (OSPS), version 7.4. (2). PK data was extracted from literature (3,4) using WebPlotDigitizer (5). The PK-Sim® mAb (i.e protein-PK) model extrapolates to different species by considering species-specific measures of in-vitro FcRn binding affinity. It leverages an in-built endosomal degradation model with FcRn-based antibody-recycling considering competition by endogenous IgG. With these underlying assumptions, a test for how exogenous IgG utilized saturation of and competition for FcRn binding to change mAb concentrations. The predictivity of the PBPK software for the IgG competition at FcRn was first validated using experimental data in mice where an unspecific mAb was co-administered with a single iv IgG dose to reduce mAb exposure (6). The human PBPK model on MEDI-524-YTE was then used to evaluate clinical feasibility and applicability of iv IgG competition at FcRn with both a repeated IV bolus and a continuous iv infusion of IgG and to provide insight into the limitations and risks associated with an IgG-based preferential degradation in humans.
The translated PBPK model for MEDI-524-YTE from NHP to humans slightly under-predicted terminal half-life for higher doses after sc administration. Yet, importantly the model illustrated the difference in half-life for NHP (29 days) to humans (86 days). For validation of the exogenous IgG competition at FcRn, the PBPK software captured the time-concentration profile of the unspecified mAb in mice (data pulled from (6)), both with and without iv IgG reasonably well with a slight over prediction of the initial distribution phase in both scenarios. Nevertheless, the clearance phase and importantly the decrease in concentration of the unspecific mAb from the IgG bolus was accurately predicted, demonstrating the ability to predict alterations in clearance rate and the resulting concentration profile through the competitive binding at FcRn. Additional simulations in mice revealed that both, repeated bolus doses and continuous IV infusions, can sustainably reduce the concentrations through a constant reduction in half-life. Following this proof of concept, the exogenous IgG infusion was extrapolated for MEDI-524-YTE in humans. Half-life decreased with an increase in dose of IgG (100-1000mg/kg/day) and showed dose-independency across different doses of MEDI-524-YTE (10-1000mg). The half-life reduction plateaued with an IgG infusion nearing 1000 mg/kg/day and took typical half-life from 86 days down to approximately 4-5 days.
The use of exogenous IgG to reduce concentrations of a mAB provides one opportunity for an interventive approach to decrease the risk of testing new molecules in humans. The risk associated with the intervention should be moderated in the context of the event itself. Importantly, this endeavour suggests this modelling platform provided evidence to mitigate risk when translating into humans.