Parth J. Upadhyay1, Adrian Devitt Lee2, Rocio Lledo Garcia2, Ivan Matthews2, Victoria Flores2, Akash Khandelwal3, Pinky Dua2
1UCB Pharma B.V. , 2UCB Pharma Ltd., 3UCB Pharma GmbH
Introduction: Minimal physiology-based pharmacokinetic (mPBPK) models are a valuable tool in the pharmacometrician arsenal at the earliest stages of therapeutic antibody (Ab) development, when preliminary in vitro investigations and literature are available to project dose ranges required for in vivo and clinical studies. An estimate of clinical dose range in the preclinical stage is crucial for formulation development and design of toxicology studies [1]. Furthermore, an indication of uncertainty around the dose range projection is also vital to prevent foreseeable delays during development due to suboptimal formulation development. Antibodies offer predictable PK characteristics of distribution and clearance, which can be extrapolated from pre-clinical stages, when a good understanding of target levels and binding kinetics across species is available. mPBPK models provide a framework that can characterize the disposition of Abs through plasma, lymph and tissues, including tumor, with high and low extravasation (leaky and tight tissues, respectively) and to characterize target mediated drug disposition (TMDD) in relevant compartments [2]. We developed an r Shiny application that leverages the common properties of Abs combining drug and binding specific properties available from in vitro data to project dose ranges in preclinical species and in human based on receptor occupancy. Methods: A mPBPK model framework, as developed by Cao and Jusko [2], has been implemented in rShiny using mrgsolve. The knobs function for sensitivity analysis was used to input typical ranges for Ab clearance and extravascular coefficients into tight and leaky tissues, lymphatic tissues and tumor [3]. Partition into tissues was also governed by a partition coefficient of 0.8 as is typical for IgG1 [2]. A tumor compartment was incorporated such that distribution of Ab into tumor interstitial fluid governed target interaction [4]. In the application, the user can input values for drug (dose), target (R0, kout) and binding (KD, koff, kint) specific properties obtained from in vitro investigations and literature. TMDD can be selectively characterized in compartments of interest, which requires input on initial target concentration for each compartment. The model assumes a well-stirred distribution of target and does not consider avidity, which may be important for membrane bound multispecifc Abs. Results: The output of the r Shiny application is divided into two parts. The first part projects doses required to achieve a peak % target occupancy (TO) of 20, 50, 90 and 99% after the first dose. For multispecific Abs, if complexation to multiple targets is required for efficacy, an option is available to use complex formation and not TO as a metric for dose projections. In this scenario, doses are calculated based on the relevant complex (eg. ternary) for 20, 50,90 and post peak 90% of peak complex formation. To assist toxicology studies, preclinical doses required to sustain 10x, 50x and 100x fold margins from human Cmax reaching 90% TO, are also output. The second part provides an in-depth local sensitivity analysis for key parameters in the mPBPK model and the resulting impact of the parameter variability on exposure and TO%. Finally, the output can be downloaded in the form of a report, along with the simulations as a spreadsheet for onward processing. As a test case, select parameters from a mPBPK model of pembrolizumab (KD = 35 nM, koff = 2.6 d-1, kint = 0.43 d-1, R0 = 0.31 nM, kdeg = 0.48d-1) were inputted in the rShiny application and a dose range of 0.03 – 13 mg/kg (peak TO 20 – 99% in tumor) was projected, with >99% TO expected in serum at doses above 0.8 mg/kg. This corresponded reasonably well with the 0.005 – 10 mg/kg dose range investigated for the first in human (FIH) clinical study KN001 and target saturation achieved in serum at 1 mg/kg [5]. Conclusions: In conclusion, an application was developed to support initial dose range projection for preclinical studies and provide an early indication on dose ranges required for FIH studies at pre-candidate stages based on preliminary information available on therapeutic antibodies in the pipeline. The application allows quick turnaround of results which will serve to inform early development of Abs.
[1] Maurer TS, Smith D, Beaumont K, et al. Dose Predictions for Drug Design. J Med Chem. 2020 Jun 25;63(12):6423-6435. doi: 10.1021/acs.jmedchem.9b01365. Epub 2020 Jan 22. PMID: 31913040. [2] Cao Y, Jusko WJ. Incorporating target-mediated drug disposition in a minimal physiologically-based pharmacokinetic model for monoclonal antibodies. J Pharmacokinet Pharmacodyn. 2014 Aug;41(4):375-87. doi: 10.1007/s10928-014-9372-2. Epub 2014 Jul 31. PMID: 25077917; PMCID: PMC4167346. [3] Cao Y, Jusko WJ. Survey of monoclonal antibody disposition in man utilizing a minimal physiologically-based pharmacokinetic model. J Pharmacokinet Pharmacodyn. 2014 Dec;41(6):571-80. doi: 10.1007/s10928-014-9374-0. Epub 2014 Aug 22. PMID: 25146360; PMCID: PMC4226811. [4] Baxter LT, Zhu H, Mackensen DG, et al. Physiologically based pharmacokinetic model for specific and nonspecific monoclonal antibodies and fragments in normal tissues and human tumor xenografts in nude mice. Cancer Res. 1994 Mar 15;54(6):1517-28. PMID: 8137258. [5] Li TR, Chatterjee M, Lala M, et al. Pivotal Dose of Pembrolizumab: A Dose-Finding Strategy for Immuno-Oncology. Clin Pharmacol Ther. 2021 Jul;110(1):200-209. doi: 10.1002/cpt.2170. Epub 2021 Mar 8. PMID: 33462831.
Reference: PAGE 33 (2025) Abstr 11714 [www.page-meeting.org/?abstract=11714]
Poster: Methodology - Other topics