Population PK of Monoclonal Antibodies
Compared to small molecule drugs, monoclonal antibodies typically exhibit less inter- and intra-subject variability of the standard PK parameters geographic location can differ from a selected group of “normal” subjects. These covariates can have substantial influence on PK parameters. Therefore, good therapeutic practice should always be based on an understanding of both the influence of covariates on PK parameters as well as the PK variability in a given patient population. With this knowledge, dosage adjustments can be made to accommodate differences in PK due to genetic, environmental, physiological or pathological factors, for instance in case of compounds with a relatively small therapeutic index. The framework of application of population PK during drug development is summarized in the FDA guidance document entitled “Guidance for Industry—Population Pharmacokinetics” (www. fda.gov).
For population PK data analysis, there are generally two reliable and practical approaches. One approach is the standard two-stage method, which estimates parameters from the drug concentration data for an individual subject during the first stage. The estimates from all subjects are then combined to obtain a population mean and variability estimates for the parameters of interest. The method works well when sufficient drug concentration-time data are available for each individual patient. A second approach, the non-linear mixed effect modeling (NONMEM) attempts to fit the data and partition the unpredictable differences between theoretical and observed values into random error terms. The influence of fixed effect (i.e., age, sex, body weight,etc) can be identified through a regression model building process.
The original scope of using NONMEM was that it is applicable even when the amount of time-concentration data obtained from each individual is sparse and conventional compartmental PK analyses are not feasible. This is usually the case during the routine visits in Phase III or IV clinical studies. Currently, this approach is applied far beyond its original scope due to its flexibility and robustness. It has been used to describe data-rich Phase I and Phase IIa studies or even preclinical data to guide and expedite drug development from early preclinical to clinical studies (Aarons et al., 2001; Chien et al., 2005).
There is increasing interest in the use of population PK and PD analyses for different anti-body products (i.e., antibodies, antibody fragments, or antibody fusion proteins) over the past 10 years (Lee et al., 2003; Nestorov et al., 2004; Zhou et al., 2004; Yim et al., 2005; Hayashi et al., 2006). One example involving analysis of population plasma concentration data involved a dimeric fusion protein, etanercept (Enbrel ). A one-compartment first-order absorption and elimination population PK model with interindividual and interoccasion variability on clearance, volume of distribution, and absorption rate constant, with covariates of sex and race on apparent clearance and body weight on clearance and volume of distribution, was developed for etanercept in rheumatoid arthritis adult patients (Lee et al., 2003). The population PK model foretanercept was further applied to pediatric patients with juvenile rheumatoid arthritis and established the basis of the 0.8 mg/kg once-weekly regimen in pediatric patients with juvenile rheumatoid arthritis (Yim et al., 2005). Unaltered etanercept PK with concurrent methotrexate in patients with rheumatoid arthritis has been demonstrated in a Phase IIIb study using population PK modeling approach (Zhou et al., 2004). Thus, no etanercept dose adjustment is needed for patients taking concurrent methotrexate. A simulation exercise of using the final population PK model of subcutaneously administered etanercept in patients with psoriasis indicated that the two different dosing regimens (50 mg every week vs. 25 mg every other week) provide a similar steady-state exposure (Nestorov et al., 2004). Therefore, their respective efficacy and safety profiles are likely to be similar as well.
An added feature is the development of a population model involving both PK and PD. Population PK/PD modeling has been used to characterize drug PK and PD with models ranging from simple empirical PK/PD models to advanced mechanistic models by using drug-receptor binding principles or other physiologically based principles. A mechanism-based population PK and PD binding model was developed for a recombinant DNA-derived humanized IgG1 monoclonal antibody, oma-lizumab (Xolair ) (Hayashi et al., 2006). Clearance and volume of distribution for omalizumab varied with body weight, whereas clearance and rate of produc-tion of IgE were predicted accurately by baseline IgE and overall, these covariates explained much of the inter-individual variability. Furthermore, this me-chanism-based population PK/PD model enabled the estimation of not only omalizumab disposition, but also the binding with its target, IgE, and the rate of production, distribution and elimination of IgE.
Population PK/PD analysis can capture uncer-tainty and the expected variability in PK/PD data generated in preclinical studies or early phases of clinical development. Understanding the associated PK or PD variability and performing clinical trial simulation by incorporating the uncertainty from the existing PK/PD data allows projecting a plau-sible range of doses for future clinical studies and final practical uses.