(T-082) Optimizing Clinical Dosing Strategies to Mitigate Corneal Toxicity: Ocular PBPK Model-Based Evaluation of the Extent and Rate of Therapeutic Protein Distribution in the Human Cornea

Leticia Arrington, PhD – Senior Principal Scientist, Amgen; Di Zhou, PhD – Senior Principal Scientist, Amgen; Vijay Upreti, PhD, FCP – Executive Director, Amgen

Objective: Cornea-related adverse events pose a significant risk for oncology patients treated with systemic therapeutic proteins (TPs), including monoclonal antibodies (mAbs). Data on TP distribution in the human cornea is scarce. Nonclinical studies often lump cornea and limbus with other eye tissues, obscuring detailed biodistribution insights in these key tissues [1]. Recent ocular-PBPK models, show less than 4% of serum levels of exogenous mAbs reach the cornea via aqueous humor, while early research on endogenous immunoglobulins indicate that up to 70% of serum levels can reach cornea through the limbal vasculature [2,3]. The objective of this work is to evaluate the impact of these vastly different assumptions on clinical dosing strategies to mitigate corneal toxicity in oncology patients.

Methods: A Physiologically Based Pharmacokinetic model with detailed description of ocular sub-structures (ocular-PBPK) was used to describe distribution of full-length mAbs in the rabbit cornea. The model was calibrated in rabbits using data in cornea, aqueous humor, and serum following single and multiple intravenous administration of exogenous full-length mAbs and translated to humans. The impact of the varying extent and rate of distribution of TP in the human cornea and their implications for clinical dosing strategies were assessed by simulating and comparing various dosing regimens (QW, Q2W, Q3W, & Q4W).

Results: The rabbit ocular-PBPK model, without accounting for limbal vasculature contribution, accurately predicted serum and aqueous humor concentrations of full-length mAbs but underestimated cornea concentrations. The calibrated model best described the observed cornea levels and was successfully translated to humans. The human ocular-PBPK model indicate that mAbs accumulate slowly in the cornea after multiple intravenous dosing, achieving steady state in 2-4 months or 4-6 months with and without limbal vasculature contributions, respectively. These findings align with the reported clinical observations of delayed ocular toxicity. Model predictions indicate that the distribution of mAbs in the human cornea at steady state is 1.5% without contributions from limbal vasculature, and 52.4% with it marking the best- and worst-case cornea exposure scenarios. Furthermore, model simulations showed minimal relative differences in peak cornea concentrations (Cmax) at steady state between Q2W and Q3W dosing regimens across the exposure scenarios tested.

Conclusions: Using ocular-PBPK modeling, we evaluated how the extent and rate of distribution of full-length mAbs in the human cornea affects clinical dose optimization in oncology patients to mitigate corneal adverse events. The analysis also highlighted key knowledge gaps, suggesting targeted experiments in the future to better understand distribution of mAbs, as a class of TPs, in the cornea.

Citations: [1] Shivva, Vittal, et al. "Antibody format and serum disposition govern ocular pharmacokinetics of intravenously administered protein therapeutics." Frontiers in Pharmacology 12 (2021): 601569.
[2] Bussing, David, and Dhaval K Shah. "Development of a physiologically-based pharmacokinetic model for ocular disposition of monoclonal antibodies in rabbits." Journal of pharmacokinetics and pharmacodynamics 47 (2020): 597-612.
[3] Naware, Sanika, David Bussing, and Dhaval K. Shah. "Translational physiologically-based pharmacokinetic model for ocular disposition of monoclonal antibodies." Journal of Pharmacokinetics and Pharmacodynamics (2023): 1-16.