Estimation of Intra-vitreal Half-Lifes in the Rabbit Eye with Semi-mechanistic Equations

Quantitative intravitreal pharmacokinetics in mouse as a step towards inter-species translation

Although mouse models are widely used in retinal drug development, pharmacokinetics in mouse eye is poorly understood. In this study, we applied non-invasive in vivo fluorophotometry to study pharmacokinetics of intravitreal fluorescein sodium (molecular weight 0.38 kDa) and fluorescein isothiocyanate-dextran (FD-150; molecular weight 150 kDa) in mice. Intravitreal half-lives of fluorescein and FD-150 in mouse eyes were 0.53 ± 0.06 h and 2.61 ± 0.86 h, respectively. These values are 8-230 times shorter than the elimination half-lives of similar compounds in the human vitreous. The apparent volumes of distribution in the mouse vitreous were close to the anatomical volume of the mouse vitreous (FD-150, 5.1 μl; fluorescein, 9.6 μl). Dose scaling factors were calculated from mouse to rat, rabbit, monkey and human translation. Based on pharmacokinetic modelling and compound concentrations in the vitreous and anterior chamber, fluorescein is mainly eliminated posteriorly across blood-retina barrier, but FD-150 is cleared via aqueous humour outflow. The results of this study improve the knowledge of intravitreal pharmacokinetics in mouse and facilitate inter-species scaling in ocular drug development.

Quantitative systems pharmacology of the eye: Tools and data for ocular QSP

Good eyesight belongs to the most-valued attributes of health, and diseases of the eye are a significant healthcare burden. Case numbers are expected to further increase in the next decades due to an aging society. The development of drugs in ophthalmology, however, is difficult due to limited accessibility of the eye, in terms of drug administration and in terms of sampling of tissues for drug pharmacokinetics (PKs) and pharmacodynamics (PDs). Ocular quantitative systems pharmacology models provide the opportunity to describe the distribution of drugs in the eye as well as the resulting drug-response in specific segments of the eye. In particular, ocular physiologically-based PK (PBPK) models are necessary to describe drug concentration levels in different regions of the eye. Further, ocular effect models using molecular data from specific cellular systems are needed to develop dose-response correlations. We here describe the current status of PK/PBPK as well as PD models for the eyes and discuss cellular systems, data repositories, as well as animal models in ophthalmology. The application of the various concepts is highlighted for the development of new treatments for postoperative fibrosis after glaucoma surgery.

Evaluation of Molecular Properties versus In Vivo Performance of Aflibercept, Brolucizumab, and Ranibizumab in a Retinal Vascular Hyperpermeability Model

Purpose

To evaluate the molecular, pharmacokinetic, and pharmacological properties of three anti-vascular endothelial growth factor (VEGF) agents—aflibercept, brolucizumab, and ranibizumab—and to provide a prediction of the optimal design of an intravitreal VEGF challenge in rabbits to assess the preclinical in vivo activity of the different anti-VEGF agents.

Methods

Biochemical analyses and cellular and animal models of retinopathy were used to characterize anti-VEGF efficacy. Anti-VEGF biochemical binding affinity was determined through a kinetic exclusion assay. The in vitro potency was investigated by a calcium mobilization assay. Pharmacokinetic parameters were estimated for each drug to predict intraocular exposure relationships among the agents. The in silico modeling efforts informed the design of an in vivo rabbit model of VEGF-induced retinal hyperpermeability to determine the extent of VEGF neutralization in vivo. Consequently, data generated from the in vivo study enabled pharmacokinetic analysis and the generation of a logistical model describing the impact of the anti-VEGF agents on the VEGF-induced vascular leakage in rabbits.

Results

The three anti-VEGF agents ranked from most efficacious to least efficacious as aflibercept, brolucizumab, and ranibizumab, with results consistent and significant within each individual characterization experiment.

Conclusions

This composite study demonstrated how the molecular properties of aflibercept, brolucizumab, and ranibizumab translate into differences of in vivo efficacy, with results in line with the reported literature.

Translational Relevance

In silico, in vitro, and in vivo integrated studies provide information that enables the enhanced characterization of translational properties of anti-VEGF agents currently used for the treatment of retinal diseases.

Mechanistic Model for the Prediction of Small-Molecule Vitreal Clearance Combining Diffusion-Limited and Permeability-Limited Clearance

The discovery of new small-molecule drugs for intravitreal administration would benefit from simple models to predict vitreal clearance (CL). The current models available have limitations in their applicability to small-molecule drugs and translatability to humans. We developed a mechanistic model combining the diffusion rate of the molecule in the vitreous and permeability across posterior segment tissues and applied it to 30 small molecules with observed CL available mostly from literature. We used Caco-2 permeability as a surrogate for ocular tissue permeability. The model predicted rabbit vitreal CL well, with 80% of the predictions being within a 2-fold range of the observed CL. For an accurate prediction, it was crucial to consider the anterior diffusion CL from the vitreous to the aqueous and a limiting diffusion CL for the whole eye. We observed no major differences in model accuracy when using literature permeability values from retinal pigment epithelial cell models. Importantly, by adopting the specific dimensions of the human eye, the model was able to accurately predict vitreal CL of four compounds for which human vitreal CL data are available. In summary, this mechanistic model enables a simple, accurate, and translatable estimation of small-molecule vitreal CL.

Influence of Charge, Hydrophobicity, and Size on Vitreous Pharmacokinetics of Large Molecules

Purpose

Development of therapeutics for retinal disease with improved durability is hampered by inadequate understanding of pharmacokinetic (PK) drivers following intravitreal injection. Previous work shows that hydrodynamic radius is correlated with vitreal half-life over the range of 3 to 7 nm, and that charge and hydrophobicity influence systemic clearance. Better understanding the molecular attributes affecting vitreal elimination half-life enables improved design of therapeutics and enhances clinical translatability.

Methods

Impacts of charge and hydrophobicity on vitreal PK in the rabbit were systematically assessed using antibody and antibody fragment (Fab) variant series, including ranibizumab, altered through amino acid changes in hypervariable regions of the light chain. The impact of molecule size on vitreal PK was assessed in the rabbit, nonhuman primate, and human for a range of molecules (1–45 nm, net charge −1324 to +22.9 in rabbit), including published and internal data.

Results

No correlation was observed between vitreal PK and charge or hydrophobicity. Equivalent rabbit vitreal PK was observed for ranibizumab and its variants with isoelectric points (pI) in the range of 6.8 to 10.2, and hydrophobicities of the variable domain unit (FvHI) between 1009 and 1296; additional variant series had vitreal PK similarly unaffected by pI (5.4–10.2) and FvHI (1004–1358). Strong correlations were observed between vitreal half-life and hydrodynamic radius for preclinical species (R² = 0.8794–0.9366).

Conclusions

Diffusive properties of soluble large molecules, as quantified by hydrodynamic radius, make a key contribution to vitreal elimination, whereas differences in charge or hydrophobicity make minor or negligible contributions.

Translational Relevance

These results support estimation of vitreal elimination rates based on molecular size in relevant preclinical species and humans.