Population pharmacokinetics of the humanised monoclonal antibody, HuHMFG1 (AS1402), derived from a phase I study on breast cancer

Abstract

Background: HuHMFG1 (AS1402) is a humanised monoclonal antibody that has undergone a phase I trial in metastatic breast cancer. The aim of this study was to characterise the pharmacokinetics (PKs) of HuHMFG1 using a population PK model.

Method: Data were derived from a phase I study of 26 patients receiving HuHMFG1 at doses ranging from 1 to 16 mg kg(-1). Data were analysed using NONMEM software and covariates were included. A limited sampling strategy (LSS) was developed using training and a validation data set.

Results: A linear two-compartment model was shown to be adequate to describe data. Covariate analysis indicated that weight was not related to clearance. An LSS was successfully developed on the basis of the model, in which one sample is collected immediately before the start of an infusion and the second is taken at the end of infusion.

Conclusion: A two-compartment population PK model successfully describes HuHMFG1 behaviour. The model suggests using a fixed dose of HuHMFG1, which would simplify dosing. The model could be used to optimise dose level and dosing schedule if more data on the correlation between exposure and efficacy become available from future studies. The derived LSS could optimise further PK assessment of this antibody.

Figure 1

Semi-logarithmic representation of concentration–time profiles obtained from 26 patients during first administration of HuHMFG1. Administered doses were 1 mg kg⁻¹ (white triangle, solid line), 3 mg kg⁻¹ (black square, solid line), 9 mg kg⁻¹ (cross, dashed line) and 16 mg kg⁻¹ (open circle, solid line).

Figure 2

Goodness-of-fit obtained with the model objectified through observed concentrations vs (A) predicted (PRED) observations and through (B) weighted residuals (WRES) vs predicted (PRED) observations.

Figure 3

Accuracy of the final model evaluated by posterior visual predictive check assessment obtained after 1000 simulations. (A–D, respectively) Data correspond to the first four administrations, which are representative of further administrations. Solid lines correspond to 5th and 95th percentiles; dashed line corresponds to the median.

Figure 4

(A) Q–Q plot of the NPDE obtained after 1000 Monte Carlo simulations of the model. The solid line represents the identity line. (B) Shows the frequency distribution of the NPDE (histograms) compared with the theoretical normal distribution (solid line).