Adaptive Dosing Trial Simulation Framework to Inform Dose Optimization: Case Study Using Longitudinal PKPD Models...

Population models to identify optimal dosages typically include the evaluation of a drug's pharmacokinetics and the relationships with efficacy and safety. Dose modifications due to intolerable toxicities are frequently required for oncology therapeutics taken chronically and may impact antitumor activity and subsequent occurrences of adverse events. Model-based simulations must account for the impact of dose modifications to provide meaningful predictions that capture the dynamic interplay between pharmacokinetics, safety, and efficacy. While static dose modification schedules may be incorporated into simulations, these pre-determined scenarios are empiric and may not adequately capture the time-dependent and stochastic nature of adverse events that lead to dose modifications. PF-06804103 is an anti-HER2 antibody-drug conjugate that demonstrated dose-dependent antitumor activity and safety events. PF-06804103 appeared to have a narrow therapeutic window, which made it difficult to identify the optimal starting dosing regimen and dose modification schedule. Thus, quantitative modeling approaches were explored to better understand exposure-response relationships. An adaptive clinical trial simulation framework consisting of pharmacokinetic, tumor dynamic, and repeated time-to-event peripheral neuropathy longitudinal models was developed to guide the dose optimization of PF-06804103. This framework provides the advantage of accounting for adverse event-driven dose modifications and their real-time impact of safety on PK and efficacy, and can be adapted to other oncology therapeutics with time-dependent and exposure-related safety and efficacy profiles. The framework could be particularly beneficial to inform dose optimization, by integrating multi-dimensional early clinical data to evaluate different dosage regimens and define the regimen with the best benefit/risk ratio for further investigation.