Ignacio González-GarcÃa1, John Hood1, Nicholas White1, Leeron Marshall2, Vincent F S Dubois1, Paolo Vicini1, Paul G Baverel1
1Clinical Pharmacology, Drug Metabolism and Pharmacokinetics, MedImmune, Cambridge, UK; 2 University of Cambridge
Objectives: A joint pharmacokinetic (PK) and pharmacodynamic (PD) model was built in NONMEM, version 7.3 from a first-in-human trial of a novel biologic, MEDI7836. MEDI7836 is a human immunoglobulin G1 lambda (IgG1λ-YTE) monoclonal antibody (mAb), with an Fc modification to reduce clearance. MEDI7836 specifically binds to and functionally neutralizes interleukin-13 (IL-13) by preventing interactions with both IL-13 receptors (IL-13Rα1 and IL-13Rα2).
Methods: Thirty-two healthy male adults were enrolled into this single dose-escalation clinical trial. Four different active doses were tested (30 mg, 105 mg, 300 mg, and 600 mg) with 6 patients each and 8 patients received placebo. Following single subcutaneous administration, individual time courses of MEDI7836 concentrations, and the resulting IL-13 modulation in vivo (PD) were quantified. PK and PD samples were taken from day 1 to day 281. The average number of samples taken per subject was 14 PK and 15 PD samples, respectively. Baseline body weight, baseline age, race and immunogenicity status were recorded and tested as covariates. The performance of the final PK-PD model was evaluated by conducting a visual predictive check (VPC) and bootstrap. All post-processing graphical and statistical analyses were completed with R version 3.5.1.
Results: A first-order absorption and 2-compartment disposition with linear elimination model adequately described the PK data. Residual variability was modelled using a combined (additive + proportional) error model. Mean (%CV) of CL, V1, V2 and Q were estimated at 0.437 L/d (52%), 3.96 L (67%), 7.46 L (46%), and 1.06 L/d. Covariate analysis revealed an impact of anti-drug antibody (ADA) on CL, with treatment emergent ADA positive subjects showing a 74% CL increase (0.76 L/d [ADA +] vs 0.437 L/d [ADA -])). A binding PK-PD indirect response model was built to characterize the exposure-driven modulation of the target (IL-13) over time. Validation data indicated the bioanalytical assay quantified the level of target that was not in-complex with drug (i.e. a free IL-13 assay). However, reported data suggested dose-dependent increase in IL-13 plasma-concentration over time, indicative of a total IL-13 assay. The target time course was thus modelled as a linear combination of free target and a percentage (to be estimated from data) of the drug-target complex. Complex formation/dissociation parameters were assumed constant based on in vitro experiments and sequential PKPD modelling was utilized. A relatively fast turnover and low baseline levels of IL-13 were estimated, and the fraction of complex detected by the assay was predicted to be around 10%. No relationship was found between any of the covariates and PD parameters. Expectation-maximization (SAEM with interaction and IMP) estimation methods were selected to achieve a successful minimization avoiding convergence problems. VPC (n = 1,000) plots showed that the final model could describe both MEDI7836 kinetics and IL-13 modulation with reasonable accuracy across the dose range tested. All final model estimates were close to the median values of the bootstrap estimates (1,000 resampled datasets) and were within the 95% confidence interval.
Conclusions:
A population PKPD binding model linking exposure of MEDI7836 with IL-13 modulation in the serum adequately characterized the exposure-response observed in a first-in-human trial and helped rationalize unexpected results from a bioanalytical free PD assay kit.
Reference: PAGE 28 (2019) Abstr 8976 [www.page-meeting.org/?abstract=8976]
Poster: Drug/Disease Modelling - Other Topics