(M1330-05-29) Predicting Aflibercept Local Exposure in Humans Using an Ocular PBPK Model

Purpose: The intricate anatomy of the eye, its strong defense against external substances, and the difficulty of directly measuring drug levels at the intended site of action, make it challenging to evaluate the relationship between ocular physiology and drug product formulations. As a result, developing both innovative and generic ophthalmic drug products remains a significant hurdle for the pharmaceutical industry. Ocular physiologically based pharmacokinetic (O-PBPK) offer valuable insights into how drugs distribute across different eye tissues, especially those that are difficult or impossible to access in human studies. These models provide an alternative approach for studying the pharmacokinetics and pharmacodynamics of ophthalmic therapies. The utility of O-PBPK models has already been shown in predicting first-in-human doses for small molecule drugs delivered via ophthalmic topical administration. This study aims to demonstrate how an O-PBPK model, validated using preclinical data, can be used to estimate clinical local exposure levels of aflibercept (AFL), a biologic therapy used in treating Neovascular Age-Related Macular Degeneration.

Methods: AFL (MW = 115,000 g/mol) is a recombinant fusion protein that functions as a vascular endothelial growth factor (VEGF) inhibitor. To parameterize AFL's systemic distribution and elimination, its reported volume of distribution (6 liters) and elimination half-life (5.5 days following intravenous administration) were used. The Ocular Compartment Absorption and Transit (OCAT™) model within GastroPlus® v9.9 was then employed to build an Ocular PBPK (O-PBPK) model for AFL. OCAT model parameters (including tissue permeabilities and systemic absorption rates) were adjusted to accurately capture concentrations in the aqueous humor, vitreous humor, and retina in rabbits following intravitreal administration. After validation in rabbits, the O-PBPK model was applied to predict AFL pharmacokinetics in humans, with physiological parameters modified to reflect human anatomy. All drug-specific PK parameters were kept consistent across species. The dose, dose volume, and administration schedule were aligned with those used in human clinical trials. Simulated human ocular PK profiles were then compared with observed concentration data to assess the model’s ability to predict human ocular exposure based on preclinical data.

Results: The OCAT model successfully described the observed AFL aqueous humor, vitreous humor, and retina concentrations in rabbits following intravitreal administration (Figure 1). Adjusting the physiological parameters of the OCAT model to describe the human eye allowed for a reasonable prediction of AFL clinical systemic and ocular exposures following its intravitreal administration (Figure 2).

Conclusion: This case study demonstrates the potential of a mechanistic, model-based preclinical-to-clinical translation approach to predict both local and systemic exposure of a biologic drug in humans following its intravitreal administration. This methodology could significantly influence the development of both novel and generic biologic ophthalmic drug products.

Figure 1: Observed (dots) and simulated (lines) aflibercept aqueous humor, vitreous humor, and retina concentrations in rabbits following intravitreal administration.

Figure 2: Observed (dots) and simulated (lines) aflibercept plasma and aqueous humor concentrations in humans following intravitreal administration.