Nathalie Perdaems, Sylvain Fouliard, Marylore Chenel
Clinical PK and Pharmacometrics division, Servier, France
Objectives: In preclinical development, monkey is the reference species to predict the pharmacokinetics (PK) of monoclonal antibodies (mAb) in human. Allometric scaling can beused to transpose the PK model from monkeys to support the first in human (FIH) study in healthy volunteers.
Methods: A population PK model was built in monkeys, using data for a mAb, from a pharmacological study after single administration of 3 doses after iv (0.1, 1 and 10 mg/kg) and 1 dose after sc (1 mg/kg). Ten cynomolgus monkeys with a total of 78 plasma concentrations were included in the model. NONMEM version 7.3 was used for the population analysis.
Allometric scaling was used to transpose the population PK model in human and predict plasma concentrations for the FIH study. Allometric scaling factors were 0.85 for clearance, 1 for volume and -0.25 for the absorption constant after sc administration [1] [2] [3].
An ex vivo PK/PD relationship between the concentration and the receptor occupancy (RO) in human cells was used to predict the RO in the FIH study (3 experiments with 2 or 3 replicates were performed with 16 levels of concentrations from 0 to 10000 ng/mL). Phoenix WinNonlin version 6.4 was used to describe this relationship.
This approach allowed to justify the doses planned in the FIH study and the predicted concentrations and the predicted RO were compared to the observed one.
Results: The population PK model developed in monkeys allowed to well describe plasma concentrations of the mAb after the iv and sc administration. The model structure for the PK modelling in monkeys was a 2-compartment model with 3 elimination pathways (2 linear clearances (CL and CLADA describing ADA-related drug elimination) and a non-linear clearance) with no covariate. All the parameters were well estimated (RSE < ~40 %), except the intercompartmental clearance and the concentration at half the maximum rate of elimination (Vmax) (Km) with RSE < 85 %. Interindividual varibility (IIV) was estimated on the volume of the central compartment (VC) and a proportional error model (31.6 %) best described the residual error.
Allometric scaling were used for CL, Q, volumes (VC and volume of the peripheral compartment (VP)) Vmax and Ka. The monkey parameters were used for the bioavailability (F) after sc administration and Km for the human predictions. ADA-related clearance was not considered in the transposition.
The ex vivo PK/PD relationship was described using an inhibitory effect sigmoïd model (WNL5 classical PD model 107) relying the log-transformed concentrations and the percentage of normalized free receptors.
Observed concentrations are available for the first doses of the FIH study and were consistent with predictions.
Also the ex vivo PK/PD relationship allowed to predict the in vivo RO in human (less than 20 % for the first dose and less than 100 % for the second dose).
Conclusions: A population PK model in monkeys was built and the use of allometric scaling for clearances (including Vmax), volumes and Ka allowed to well predict plasma concentrations in human for a mAb and associated receptor occupancy.
References:
[1] Deng R. Projecting human pharmacokinetics of therapeutic antibodies from nonclinical data – what have we learned ? mAbs (2011) 3:1, 61-66.
[2] Dong J. Quantitative Prediction of Human Pharmacokinetics for Monoclonal Antibodies Retrospective Analysis of Monkey as a Single Species for First-in-Human Prediction. Clin Pharmacokinet; (2011) 50 (2): 131-142.
[3] Singh P. Quantitative Prediction of Human Pharmacokinetics for mAbs Exhibiting Target-Mediated Disposition. AAPS J. (2015) Vol. 17, No. 2 43:252-259
Reference: PAGE 28 (2019) Abstr 8834 [www.page-meeting.org/?abstract=8834]
Poster: Methodology - Other topics