Stefan Van Den Berg1,2, Philine Adolfsen1,2, Thomas Dorlo3, Theo Rispens1,2
1Amsterdam UMC Location Vrije Universiteit Amsterdam, Molecular Cell Biology and Immunology, 2Sanquin Research Amsterdam, Department of Immunity and Inflammation, 3Uppsala University, Department of Pharmacy
Introduction: Antibodies and their derivatives are widely used in treating various diseases, including the fields of oncology, hematology, infectious and autoimmune diseases. By May 2023, 173 therapeutic monoclonal antibodies (mAbs) had been approved [1]. Population pharmacokinetic (popPK) models are typically developed for each individual mAb. Typically, two-compartment models are employed, with the primary route of mAb clearance attributed to non-specific proteolytic degradation following linear kinetics. Additional elimination pathways include target-mediated drug disposition (TMDD) and anti-drug antibody-mediated disposal. However, sparse datasets with limited PK information and high inter-individual variability (IIV) pose challenges in deriving accurate parameter estimates or even achieving parameter identifiability. Given that biologics share similarities in size, shape, and susceptibility to lysosomal degradation, model estimates have been shown to be consistent across different mAbs [2]. We propose that deriving a generic model from published mAb PK models can aid in parameter estimation and model development. Such a generic model for clotting factor VIII has proven effective in accurately forecasting individual PK [3]. Objectives -Do similarities in PK parameter estimates persist among different model structures (with or without nonlinearity, with or without absorption depot) or based on characteristics of mAbs (subtype of IgG, origin of antibody)? -Assess the IIV in parameters across models, as well as the variability in typical parameter estimates. Is the median IIV similar to the variability in typical estimates observed for a given PK parameter? Methods The IMGT database of mAbs was queried to obtain canonical therapeutic mAbs of the immunoglobulin G type, without conjugated or fused drugs, and market approval [4]. PubMed was queried using the retrieved mAbs to locate published popPK models. Selection criteria were; consisting of two compartments, linear clearance, non-pediatric populations, and not more than 2 PK parameters fixed. Models were included if either intravenous (i.v.) or i.v. and subcutaneous (s.c.) administration, without assuming 100% absorption. Each paper was reviewed thoroughly, and the final PK model along with other pertinent data was collected for analysis. Estimates of core PK parameters of the model (CL, VC, VP, Q, ka, F) were summarized using the median and interquartile range IQR). Variability in parameter estimates across models was expressed as the percent coefficient of variation (%CV) in the log-domain. Estimates of PK parameters were compared (Wilcoxon rank tests) across model structures using all models, and mAb characteristics using the median estimate for each individual mAb. Results A total of 220 models were included in our meta-analysis, which, after exclusions, covered 160 two-compartment popPK models of 69 different mAbs. We confirmed a general consistency in parameter estimates across models. Median estimates for the core parameters were linear CL = 0.22 L/day, VC = 3.42 L, VP = 2.68 L, Q = 0.54 L/day. Median estimates of 52 models with s.c. administration were ka = 0.25 L/day, and F = 69%. We confirmed that these and subsequent results remained unchanged after normalizing the models to a 70 kg reference using allometric scaling. Our analysis reveals that models including nonlinear kinetics estimate 26% lower linear CL than models with only linear CL (0.18 vs. 0.25 L/day). Chimeric mAbs, with variable mouse regions, had 50% higher linear CL compared to humanized and human mAbs (0.33 vs 0.21 vs 0.22 L/day), potentially caused by unaccounted immunogenicity. No differences in PK parameters among IgG subtypes (IgG1, IgG2, IgG4) were found, except for a higher VP in IgG1 compared to IgG4 (2.7 vs 2.1 L). We summarized the most commonly used covariates in mAb models, including weight and sex, and demonstrated the presence of confounding effects between them. The IIV of mAbs have consistently been large. Our findings indicate that the variability in typical parameter estimates across different studies is similar to the IIV, with the variability between studies versus IIV being 55% vs. 43% for CL, and 25% vs. 30% for VC. Therefore, a generic popPK model could prove to be highly valuable. We briefly present a generic mAb popPK model on two or three different cases. Conclusion Our analysis reveals evidence of parameter compensation, indicating that the absence of one parameter led to other parameters adjusting to account for its effect, which may distort the interpretation of their contributions. Furthermore, our findings demonstrate the general similarity in mAb PK, supporting the use of a generic modeling approach. It is our view that this can assist in pharmacometrics, among which during early drug development, model building, pharmacodynamic modeling, and model-informed dosing.
[1] Luo, S., & Zhang, B. (2024). Benchmark glycan profile of therapeutic monoclonal antibodies produced by mammalian cell expression systems. Pharmaceutical research, 41(1), 29-37. https://doi.org/10.1007/s11095-023-03628-4 [2] Dirks, N. L., & Meibohm, B. (2010). Population pharmacokinetics of therapeutic monoclonal antibodies. Clinical pharmacokinetics, 49, 633-659. https://doi.org/10.2165/11535960-000000000-00000 [3] McEneny-King, A., Chelle, P., Foster, G., Keepanasseril, A., Iorio, A., & Edginton, A. N. (2019). Development and evaluation of a generic population pharmacokinetic model for standard half-life factor VIII for use in dose individualization. Journal of Pharmacokinetics and Pharmacodynamics, 46, 411-426. https://doi.org/10.1007/s10928-019-09634-7 [4] IMGT/mAb-DB, the international database of therapeutic monoclonal antibodies, available at https://www.imgt.org/mAb-DB (Accessed 2024, November 28).
Reference: PAGE 33 (2025) Abstr 11721 [www.page-meeting.org/?abstract=11721]
Poster: Methodology - New Modelling Approaches