Population PK–PD model for Fc-osteoprotegerin in healthy postmenopausal women

Osteoporosis is a metabolic bone disease resulting from increased bone resorption and characterized by low bone mass that leads to increased bone fragility and risk of fracture, particularly of the hip, spine and wrist. Bone resorption is dependent on receptor activator of NF-kappa B ligand (RANKL), which binds to RANK receptor on preosteoclasts to initiate osteoclastogenesis and maintains osteoclast function and survival. To neutralize the effects of RANKL, the body naturally produces the protein osteoprotegerin (OPG), which acts as a decoy receptor for RANKL and contributes to bone homeostasis. We describe the piecewise development of a three-compartment pharmacokinetic model with both linear and Michaelis–Menten eliminations, and an indirect pharmacodynamic response model to describe the pharmacokinetics and pharmacodynamics, respectively, of the fusion protein, Fc-osteoprotegerin (Fc-OPG), in healthy postmenopausal women. Subsequently, model verification was performed and used to address study design questions via simulation. The model was developed using data from eight cohorts (n = 13 subjects/cohort; Fc-OPG:placebo = 10:3) classified by dose level (0.1, 0.3, 1.0, or 3.0 mg/kg) and route of administration (intravenous [IV] or subcutaneous [SC]). Fc-OPG serum concentrations and urinary N-telopeptide/creatinine ratios (NTX) following both IV and SC administration were available. The model provided an adequate fit to the observed data and physiologically plausible parameter estimates. Model robustness was tested via a posterior predictive check with the model performing well in most cases. Subsequent clinical trial simulations demonstrated that a single 3.0-mg/kg SC dose of Fc-OPG would be expected to produce, at 14 days post-dose, a median NTX percentage change from baseline of  −45% (with a 95% prediction interval ranging from −34% to −60%). Lastly, model ruggedness was evaluated using local and global sensitivity analysis methods. In conclusion, the model selection and simulation strategies we applied were rigorous, useful, and easily generalizable.