(S-033) The Dose and Dosing Schedule Selection for an Investigational Immuno-Oncology Drug Based on a Pharmacokinetics and Pharmacodynamic Model of Tolerance

Wenbo Chen – Quantatitive Pharmacology and Pharmacometrics – Merck & Co., Inc.; Runchen Zhao – 1Quantatitive Pharmacology and Pharmacometrics – Merck & Co., Inc.; Konstantin Dobrenkov – Clinical Research – Merck & Co., Inc.; Amy Beebe – Biology-Discovery – Merck & Co., Inc.; Fang Liu – Biostatistics – Merck & Co., Inc.; Yiwei Zhang – Biostatistics – Merck & Co., Inc.; Mark Ayers – Biomarkers – Merck & Co., Inc.; Song Zhai – Biostatistics – Merck & Co., Inc.; Tomoko Freshwater – Quantatitive Pharmacology and Pharmacometrics – Merck & Co., Inc.

Objectives: An investigational drug is a humanized agonist monoclonal antibody to provide a costimulatory signal that enhances T cell-mediated response and administered as an intravenous (IV) infusion tested in a first-in-human (FIH) study. Current clinical data indicate that full target engagement may not be needed for the agonist to trigger efficacy. T cell activation biomarkers data also suggest there is more pronounced T cell activation at lower doses compared to higher doses. We present an integrated population (pop) pharmacokinetic - pharmacodynamic (PK-PD) model that describes the development of acute tolerance to the effect of the drug, which supports the dose and dosing schedule selection for further development.

Methods: Data were obtained from the FIH study which contains dose escalation arms and expansion arms in particular tumor types. Individual serum concentrations over time data and biomarkers concentrations over time data were explored graphically and prepared for developing the pop PK - pop PD model. Based upon the observed decrease in PD responses with higher exposure and frequent dosing schedule during multiple dosing, particularly the pronounced reduction in peak response following the second dose and thereafter, several model types were evaluated for their ability to describe this tolerance mechanism. Models that were explored included both direct effect and indirect response model structures. Model evaluation was assessed by using visual individual fits, likelihood-based criteria, the precision and correlations of parameter estimates, and various diagnostic plots including prediction corrected visual predictive checks. Trial simulations were performed using the final model to assess and to compare biomarker changes from baseline following continuous treatment of various dose regimens.

Results: Because there was not a strong and consistent delay observed between exposures and associated responses, a direct effect model was selected as the most plausible and parsimonious PK-PD model to describe both the main drug effects and the development of tolerance. To describe the development of tolerance, the stimulatory effect was modified by a hypothetical noncompetitive antagonist that counteracted and reduced the magnitude of drug effects. The turnover of the antagonist was described by a first-order production and elimination process. The final parameters were well estimated with RSE < 30 %. The integrated PK-PD model simulations indicated intermediate dose levels with longer dosing interval may reduce the development of full tolerance and thus promoting more sustainable PD responses for target engagement and bio-signal elicitation, compared to high dose levels and/or frequent dosing schedules.

Conclusions: The integrated pop PK - pop PD model well characterized the PK and PD concentration time profile data following repeated dosing of various dose levels and schedules. The predictions from the final model were used to support dose regimen selections in the late-stage immuno-oncology drug development, along with early phase planning in similar programs which drugs exhibit tolerance mechanisms.

Citations: NA

Keywords: PK-PD model, Tolerance mechanism, Oncology.