Rami Ayoun Alsoud (1), Natacha Le Moan (2), Lars Holten-Andersen (2), Tom Knudsen (2), Ulrika Simonsson (1)
(1) Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden, (2) Catalyst Bioscience, South San Francisco, California, USA
Introduction:
Initial dose selection in first-in-man studies relies on extrapolation of animal drug exposure and drug safety. It is well-established that many physiological processes are correlated to body size regardless of species. This relationship serves as the basis of inter-species allometric scaling that translates the pre-clinical pharmacokinetic (PK) parameters to human[1]. A power function is used in allometric scaling to describe the relationship between the PK parameter and, for example, body weight[2]. The exponents of the power function of 0.75 for clearance (CL) and 1 for volume of distribution (V) are commonly accepted and used for small molecules if data from one species is available. However, larger molecules, such as therapeutic proteins, exhibit additional clearance mechanisms wherefore the exponents might be different and some species may not even be of translatable value. Different approaches for obtaining the allometric exponents have been suggested such as linear regression of noncompartmental analysis (NCA) PK parameters from different species versus body weight on the log-log scale but it is difficult to evaluate how far away from the concept of allometry a species can be. With pharmacometric modelling, a population PK (popPK) model can be evaluated using multi-species data in order to estimate the PK parameters and the allometric exponents in order to maximize the use of information. Deviations from the concept of allometry can be evaluated using formal hypothesis testing of species-specific PK parameters in order to inform which species are most informative for the prediction of human PK.
Objective:
The aim was to compare a three-species simple allometric regression scaling approach to a full popPK approach for allometric scaling where PK species differences were evaluated in order to inform a safe starting dose towards a first human clinical trial.
Methods:
Pharmacokinetic data for CB 4332, a large therapeutic protein, in three species; mice, rats, and nonhuman primates (NHPs), were included from different dose ranging studies of intravenous (IV) and subcutaneous (SC) administration. The allometric regression scaling approach was done by first obtaining the NCA-based mean PK parameters for each species and then fitting a linear regression line of CL and V versus bodyweight on log-log scale. The obtained exponents were used in a popPK model developed using the CB 4332 PK data after IV and SC administration in NHP only.
The full popPK approach for allometric scaling included simultaneous fitting of all PK data from all three species and estimation of the allometric exponents. Species differences in the PK parameters, in addition to allometric scaling, were explored for all PK parameters. Initially, only IV data were included in the model development in order to better describe the elimination PK parameters. The SC data was subsequently added to the model in order to describe the SC absorption.
The NCA was conducted in Phoenix WinNonlin™, 8.2[3]. The population analyses and simulations were performed using NONMEM, 7.5.0[4] using the FOCEI estimation method, supplemented with the PsN, 5.0.0[5]. R, 4.1.2[6] was used for general scripting, data management, and goodness of fit and graphical analyses.
Results:
The three-species simple allometric scaling approach yielded exponents for CL and V of 0.81 and 1.07, respectively. The 2-compartment popPK model using these exponents provided a good fit to the NHP data. For the full popPK approach for allometric scaling and using data from all three species, allometric scaling with body weight did not describe well the PK data from all three species. Evaluation of species-dependent PK parameters in addition to allometric scaling, revealed that the allometrically scaled CL from rat was statistically significantly higher than the other two species. Thus, a rat-specific CL parameter was included in the final model. The allometric scaling exponents for the CL and V terms were estimated to be 0.78 and 1.15, respectively. It was not possible to capture the discrepancy seen in the rat CL using the three-species simple allometry.
Conclusion:
The model-based analysis of all species revealed statistically significant deviations in the rat CL from the concept of allometry which was not identified using the three-species simple allometry approach, thereby providing a framework for definition of safe starting dose and moving towards a first human clinical trial of CB 4332.
References: [1] Mahmood, I. Pharmacokinetic Allometric Scaling of Oligonucleotides. Nucleic Acid Ther. 2011: 21(5):315-321. [2] Mordenti, J., et al. Interspecies Scaling of Clearance and Volume of Distribution Data for Five Therapeutic Proteins. Pharm Res. 1991; 8(11): 1351–1359. [3] Phoenix WinNonlin version 8.2 (Certara USA, Inc., Princeton, NJ). [4] Beal, S. L.; Sheiner, L. B.; Boeckmann, A. J.; Bauer, R. J. eds. NONMEM 7.5 Users Guides (1989-2020); ICON Development Solutions: Gaithersburg, MD, USA, 2020. [5] Lindbom, L.; Pihlgren, P.; Jonsson, E. N. PsN-Toolkit–a collection of computer intensive statistical methods for non-linear mixed effect modeling using NONMEM. Comput Methods Programs Biomed 2005, 79, 241–257. [6] R Core Team. R: a language and environment for statistical computing; R Foundation for Statistical Computing: Vienna, Austria, 2019.
Reference: PAGE 30 (2022) Abstr 9997 [www.page-meeting.org/?abstract=9997]
Poster: Methodology - Covariate/Variability Models