Application of Two-Target Quasi-Steady-State (QSS) Model in Population Pharmacokinetic and Pharmacodynamic (PK-PD) Modeling of MNRP1685A in Cynomolgus Monkeys
Yan Xin (1), Hong Xiang (1), Denise Jin (1), Manish Gupta (1), Amita Joshi (1), Shuang Bai (1)
(1) Genentech, Inc. South San Francisco, CA, 94080, USA
Objectives: MNRP1685A (anti-NRP1) is a fully human IgG1 monoclonal antibody against neuropilin-1 (NRP1), a protein necessary for blood vessel maturation. Anti-NRP1 showed strong nonlinear PK across a wide dose range in preclinical species. It is currently evaluated in Phase I studies as a single agent and in combination with bevacizumab with or without paclitaxel. MNRP1685A binds to both membrane-bound (mNRP1) and circulating (cNRP1) targets. The purpose of the study was to develop a mechanism-based model to simultaneously describe the kinetics of MNRP1685A and total cNRP1 in monkeys; thus to understand the nonlinear PK of MNRP1685A and to support pharmacological dose selection.
Methods: MNRP1685A was dosed in monkeys at 0.5, 3, 15, or 50 mg/kg (single dose) or 10, 30, or 100 mg/kg (9 weekly doses). Serum samples for MNRP1685A and total cNRP1 were collected at predose and various time points from 15 minutes up to 56 days post dose. The analysis was performed in NONMEM using FOCE with interactions. The two-target QSS model was applied to describe cNRP1 and mNRP1. Model performance was tested with diagnostic methods including visual predictive check. Free cNRP1 and mNRP1 levels were simulated based on model fitting results.
Results: The concentration-time profiles of both MNRP1685A and total cNRP1 after single-dose or multiple-dose administrations in monkeys were well described by the two-target QSS model. Parameters for MNRP1685A, mNRP1, and cNRP1 were estimated with reasonable precision. Nonspecific CL and V1 were estimated to be 3.26 mL/day/kg and 38.2 mL/kg, respectively. Maximum elimination rate through mNRP1 was 98.8 nM/day. For cNRP1, the model-derived rates of degradation, synthesis, and internalization were 1.53 day-1, 3.8 nM/day, and 0.260 day-1, respectively. QSS constants for mNRP1 and cNRP1 were 6.94 and 2.8 nM, respectively. Simulation results suggest that both mNRP1 and cNRP1 can be blocked by MNRP1685A in a dose and dosing frequency-dependent manner.
Conclusions: The two-target QSS model provides mechanistic understanding of the nonlinear PK of MNRP1685A. Modeling results suggest minimal impact of cNRP1 but major impact of mNRP1 on MNRP1685A PK; cNRP1 has higher binding affinity to MNRP1685A than mNRP1. The model was able to provide precise estimates of both soluble and membrane target parameters, which is potentially useful to select clinical dose/regimen that can saturate targets.
References:
[1] Yanmei Lu, Hong Xiang, Peter Liu, et al., mAbs 1:4, 364-369; July/August 2009.[2] Leonid Gibiansky, Ekaterina Gibiansky, J Pharmacokinet Pharmacodyn, July 2010.