Xia Li, Carlos Pérez Ruixo, Honghui Zhou, Juan Jose Perez Ruixo, Anne-gaëlle Dosne
Janssen Research & Development, Beerse, Belgium
Objectives: Daratumumab, a human immunoglobulin Gκ monoclonal antibody targeting CD38, has been developed and approved by FDA and EMA for treatment of multiple myeloma. Previously developed population pharmacokinetic (popPK) model was an empirical 2-compartment structure with parallel linear and Michaelis-Menten (MM) nonlinear elimination pathway using time-dependent Vmax to describe the decrease of the total target receptor (CD38) over time. Currently, we aimed to integrate the mechanistic target-mediated drug disposition (TMDD) process into the popPK model and describe dynamic alteration of receptor in patients with multiple myeloma after intravenously (i.v.) administered daratumumab.
Method: The popPK model development started with the datasets of i.v. daratumumab monotherapy studies (GEN501 and MMY2002) in multiple myeloma subjects. Study GEN501 was a Phase 1/2 study with a dose-escalation phase (from 0.005 up to 24 mg/kg i.v.) followed by a single arm phase (additional subjects treated at a dose of 8 or 16 mg/kg i.v. administered once weekly (QW), followed by every 2 weeks (Q2W) for 16 weeks, and then every 4 weeks (Q4W) for up to 72 weeks). Study MMY2002 was a Phase 2, open-label, multicenter study with subjects randomized to 8 mg/kg i.v. once every 4 weeks (Q4W) or to 16 mg/kg QW for 8 weeks, followed by Q2W for 16 weeks, and thereafter Q4W. In both studies patients received daratumumab monotherapy (i.e. no other background therapies).
The model development was using nonlinear mixed-effects modeling (NONMEM, Version 7.3, ICON plc, Dublin, Ireland) with the first-order conditional estimation method with interaction estimation method. The TMDD with quasi-steady-state (QSS) approximation, quasi-equilibrium (QE)1 and previously developed MM models2 were tested.
The relationship between physiological or pathological covariates and parameter estimates were explored. Model prediction was qualified using a visual predictive check (VPC). Model evaluation was based on OFV if models were nested (AIC otherwise), acceptable parameter precision, and goodness-of-fit.
Results: The popPK analysis was based on 2,572 PK samples from 223 subjects. The two-compartment TMDD structure with parallel linear and nonlinear elimination pathway (assuming QSS approximation) was found to better (AIC=-2857; residual unexplained variability (RUV)=22.7%) describe the PK of daratumumab compared to the previously developed MM models (AIC=-2072; RUV=27.4%). The model was parameterized in terms of non-specific linear clearance for IgG (CL, 0.0051 L/h), volume of distribution in the central compartment (V1, 3.76 L), inter-compartmental clearance (Q, 0.0329 L/h), volume of distribution in the peripheral compartment (V2, 3.31 L/h), as well as the parameters of QSS approximation, i.e., the synthesis rate constant (Ksyn, 0.178 mg/L·h−1) , the degradation rate constant (Kdeg, 0.0082 h−1), the complex internalization rate constant (Kint, 0.092 h−1) and the steady-state constant (Kss, 1.79 h−1). Kint was far higher than Kdeg, indicating that the total number of receptors (CD38) in subjects would decrease after taking daratumumab, and well capturing the empirical time-dependency of Vmax in MM model.
The estimated linear clearance was very close to the reported clearance of nonspecific endogenous IgG in the literature3, and the volume of distribution of central compartment (3.76 L) was close to plasma volume; both parameters were related to body weight, as expected for monoclonal antibodies.
The following covariates on linear clearance were identified as statistically significant: body weight, albumin concentration, and type of myeloma [IgG versus non-IgG] (ie, the 95% CI of the estimated covariate effect did not cover zero [ie, no effect]). The following covariates on volume of distribution in the central compartment were identified as statistically significant: body weight and sex.
The relative standard errors of the estimates for each parameter were small, confirming the quality of the model. The good predictive performance of the TMDD model was assessed based on graphical and quality criteria, together with VPC and comparison of the predictions to those from a previously developed empirical popPK model.
Conclusions: The developed semi-mechanistic TMDD model with QSS approximation describes the interaction of daratumumab with its target CD38, and consequently well describes the behavior of both daratumumab and CD38 kinetics. Further work is ongoing to evaluate how this model behaves in patients receiving SC dosing and different background therapies, potentially reflective of different disease status.
References:
[1] Mager DE, Krzyzanski W (2005). Quasi-equilibrium pharmacokinetic model for drugs exhibiting target-mediated drug disposition. Pharm Res 22(10): 1598-1596.
[2] Luo, Man Melody et al. Exposure-Response and Population Pharmacokinetic Analyses of a Novel Subcutaneous Formulation of Daratumumab Administered to Multiple Myeloma Patients. Journal of clinical pharmacology vol. 61,5 (2021): 614-627. doi:10.1002/jcph.1771
[3] Ryman JT, Meibohm B. Pharmacokinetics of monoclonal antibodies. CPT Pharmacometrics Syst Pharmacol. 2017;6(9):576-588.
Reference: PAGE 29 (2021) Abstr 9701 [www.page-meeting.org/?abstract=9701]
Poster: Drug/Disease Modelling - Oncology