A Minimal Physiologically Based Pharmacokinetic Model with a Nested Endosome Compartment for Novel Engineered Antibodies

Abstract

We proposed here a minimal physiologically based pharmacokinetic (mPBPK) model for a group of novel engineered antibodies in mice and humans. These antibodies are designed with altered binding properties of their Fc domain with neonatal Fc receptor (FcRn) or the Fab domain with their cognate targets (recycling antibodies) in acidic endosomes. To enable simulations of such binding features in the change of antibody pharmacokinetics and its target suppression, we nested an endothelial endosome compartment in parallel with plasma compartment based on our previously established mPBPK model. The fluid-phase pinocytosis rate from plasma to endothelial endosomes was reflected by the clearance of antibodies in FcRn dysfunctional humans or FcRn-knockout mice. The endosomal recycling rate of FcRn-bound antibodies was calculated based on the reported endosomal transit time. The nonspecific catabolism in endosomes was fitted using pharmacokinetic data of a human wild-type IgG₁ adalimumab in humans and B21M in human FcRn (hFcRn) transgenic mice. The developed model adequately predicted the pharmacokinetics of infliximab, motavizumab, and an Fc variant of motavizumab in humans and the pharmacokinetics of bevacizumab, an Fc variant of bevacizumab, and a recycling antibody PH-IgG₁ and its non-pH dependent counterpart NPH-IgG₁ in hFcRn transgenic mice. Our proposed model provides a platform for evaluation of the pharmacokinetics and disposition behaviors of Fc-engineered antibodies and recycling antibodies.

Keywords: FcRn; endosome; mPBPK model; pharmacokinetics; recycling antibody.

Fig. 1

mPBPK model with a nested endosomal compartment for simulation of engineered antibodies

Fig. 2

Model development for human. (a) Pharmacokinetics of human IgG in rheumatoid arthritis (RA) patients with functional FcRn and in a familiar hypercatabolic hypoproteinemia patient with mutant FcRn. (b) Model predictions of adalimumab pharmacokinetics in RA patients. Symbols are observations and curves are model predictions

Fig. 3

Model predictions of infliximab pharmacokinetics in rheumatoid arthritis patients. Symbols are observations and curves are model predictions

Fig. 4

Model predictions of motavizumab and its Fc variant YTE pharmacokinetics in healthy humans at a dose of 0.3 mg/kg (a), 3 mg/kg (b), 15 mg/kg (c), and 30 mg/kg (d). Note that the FcRn binding affinity was assumed to be equal to adalimumab for motavizumab (both are IgG1 wild type) and 10-fold higher than adalimumab for YTE. Symbols are observations and curves are model predictions

Fig. 5

Model fitting of CL_e to B21M pharmacokinetics in hFcRn Tg32 homozygous transgenic mice

Fig. 6

Model predictions of bevacizumab and its Fc variant Xtend pharmacokinetics in hFcRn Tg276 heterozygous transgenic mice. Symbols are observations and curves are model predictions

Fig. 7

Model predictions of pharmacokinetics of anti-hsIL6-R antibody and hsIL6-R in hFcRn Tg32 homozygous transgenic mice. (a) A conventional antibody NPH-IgG1; (b) A recycling antibody PH-IgG1. Symbols are observations and curves are model predictions

Fig. 8

Simulated effect of enhanced FcRn binding and target dissociation in endosomes on plasma profiles of adalimumab and its cognate target TNF-α. (a, b) Ten-fold stepwise increase of adalimumab binding affinity to FcRn at pH 6 by either increasing association constant k1_on (a) or decreasing dissociation constant k1_off (b) up to 1,000-fold. (c, d) Ten-fold stepwise decrease of adalimumab-TNF-α binding affinity at pH 6 by either decreasing the association constant ke_on (c) or increasing the dissociation constant ke_off (d) up to 1,000-fold. Solid lines and dotted lines show the pharmacokinetics of adalimumab and TNF-α, respectively. Solid arrows and dash-dot arrows show the change of direction for adalimumab and TNF-α, respectively. Black: 1x; red: 10x; blue: 100x; brown: 1,000x