The use of physiologically-based pharmacokinetic modeling in the design of pH-dependent target binding antibodies
C. Niederalt (1), H.U. Siegmund (1), J. Lippert (1), C. Gao (2), B. Agoram (3)
(1) Bayer Technology Services GmbH, Leverkusen, Germany; (2) MedImmune, Gaithersburg, MD, USA; (3) MedImmune, Cambridge UK
Objectives: Reducing the binding affinity of therapeutic proteins to their target within acidic endosomal space is a strategy to increase the half-life of these drugs [1,2]. The aim of the study was to investigate if physiologically-based pharmacokinetic (PBPK) modeling can be used to design high potency pH-dependent recycling antibodies.
Methods: The PBPK model for proteins implemented in PK-Sim® software was used with additional modifications as necessary. The pharmacokinetic data [2] of pH-dependent binding variants of tocilizumab (TCZ), an antibody against the IL-6 receptor (IL-6R), was predicted in this case study. For experiments with soluble target (hsIL-6R) in wild-type mice the standard PK-Sim® [3] model for proteins was extended by hsIL-6R binding in interstitial space (pH 7.4) and endosomal space of vascular endothelium (assumed pH 6). For experiments with effective membrane bound IL-6R (transgenic mice & cynomolgus) the standard PK-Sim® mouse & monkey models for proteins were extended by membrane-bound and endosomal IL-6R in tissue cells. The processes of biosynthesis, endocytosis & recycling were introduced. The affinity requirements of pH-dependent binding antibodies directed towards soluble and cell-surface targets was explored through simulations using the final model to help guide design of these compounds.
Results: The simulated plasma concentration time profiles of TCZ, its variant PH2 and of hsIL-6R were in good agreement with the experimental data. The plasma concentration time profiles of hsIL-6R with co-administration of TCZ and PH2 were adequately described by assuming that the in vivo affinities were ~4-10-fold different from reported in vitro values. This variance between estimated and reported affinities was true for transgenic mice and cynomolgus data also. Simulations using the final model showed that a complex relationship exists between affinity and potency for pH-dependent binding systems and higher affinity ratios at pH 6 to pH 7.4 were not always better at increasing in vivo potencies.
Conclusions: PBPK modeling can be used to guide design of pH-dependent binding variants for various targets. However, accurate translation of in vitro affinities to in vivo parameters and description of endogenous IL-6 interaction are potential limitations in the process.
References:
[1] C.S. Sarkar et al., Nat. Biotechnol. 20, 908-913 (2002)
[2] T. Igawa et al., Nat. Biotechnol. 28, 1203-1207 (2010)
[3] S. Willmann et al., Biosilico. 1, 121-124 (2003)
