(M1430-02-10) Population Pharmacokinetic Modeling of Time-Varying Effects of Anti-Drug Antibody Formation on the Pharmacokinetics of a Human Monoclonal Antibody in a Rodent Model

Purpose: Administration of human monoclonal antibodies (mAb) to animals during preclinical drug development may be accompanied by an immune response. This may lead to the formation of anti-drug antibodies (ADA) against the mAb. ADA forms an immune complex with the mAb and may reduce the systemic exposure of the mAb by triggering an immune complex-mediated clearance. This complicates the accurate assessment of preclinical pharmacokinetics (PK) and may prevent toxicology studies from establishing safe exposure ranges of the mAb in study animals. The objective of this work is to model the time-course of ADA formation and its impact on the PK profile in rats after administration of a human mAb at two different dose levels and aggregation states, and under concomitant administration of different immunosuppressive regimens.

Methods: A population PK (PPK) model was developed using mAb (erenumab used as model drug) concentration and ADA data from 80 Sprague Dawley rats in two 12-week long preclinical studies (n=40 in each study). In study 1, rats were randomly divided into 5 groups that received subcutaneous administration of 10 mg/kg of the mAb in monomeric form. Animals in the first group only received mAb monomer, whereas animals in the other groups received one of four different immunosuppressive regimens of methotrexate or a combination of tacrolimus and sirolimus (TAC/SIR). In study 2, animals received 1 mg/kg mAb monomer or 1 mg/kg mAb in aggregated form with and without TAC/SIR immunosuppression. A nonlinear mixed effects model was fit to study 1 and 2 data to explain the impact of ADA on mAb exposure over time. The best model was chosen based on lowest objective function value (OFV), Akaike information criterion and goodness-of-fit plots. The model was validated with a non-parametric bootstrap analysis to determine precision of parameter estimates.

Results: The PK of the mAb was best described by a one-compartment model with linear elimination (CL/F 1.84 mL/day [3.8% RSE]) and uniform distribution (Vd/F 28.16 mL [6.6% RSE]). The absorption was characterized using a first-order process (Ka 1.14 day^-1 [8.3% RSE]) with lag time (0.022 day [18.8% RSE]). Two separate residual error models were used for the two studies. Significant covariate effects were identified on CL and Ka as indicated by a drop in OFV of 3.84 or greater and a drop in inter-individual variability of greater than 10%. Aggregation type was identified as a covariate for CL and Ka. mAb in its aggregated form led to ~2.54-fold (5.6% RSE) higher CL and ~0.09-fold (14.1% RSE) lower Ka compared to mAb monomer (Table 1). ADA was incorporated into the model as a continuous, time-varying covariate for CL, where the ADA effect was driven by the magnitude (quantified as Signal/Noise [S/N]) of response at each of the 35 observation time points (Equation 1). ADA magnitude was found to have a significant impact on the CL and mAb systemic exposure. The final PPK model parameter estimates were contained within the 90% confidence interval determined by a bootstrap analysis based on 1000 replicates (Table 1).

Conclusion: The PPK model adequately described the PK and impact of ADA magnitude on the mAb’s PK in rats. This analysis well-captured the time-varying impact of ADA formation on the PK of a mAb.

Table 1: Parameter estimates from PPK model of erenumab in rats


Equation 1