Comparison of monoclonal antibody disposition predictions between different PBPK platforms
Pieter-Jan De Sutter, Elke Gasthuys, An Vermeulen
Department of Bioanalysis, Ghent University, Belgium
Objectives: Most monoclonal antibodies (mAbs) target soluble- or membrane bound antigens located outside the vasculature [1]. While these concentrations at the effect-site might be better predictors for efficacy and safety, they are generally inaccessible for clinical PK/PD sampling. One way to predict tissue concentrations is through physiologically based pharmacokinetic modelling (PBPK). Several PBPK modelling platforms have a specific module for the prediction of mAb disposition, but whether the models are of comparable accuracy has not been independently assessed. The primary aims of this work were to compare predicted plasma and tissue concentrations of mAbs between three PBPK modelling platforms and to evaluate the accuracy of predictions against observed data.
Methods: PBPK simulations were done in Simcyp® (v21), PK-Sim® (v11) and GastroPlus (v9.8.2) for three different mAbs: intravenous IgG (IgGIV), adalimumab and infliximab. For IgGIV, compound parameters (i.e. molecular weight and FcRn affinity) of endogenous IgG were used as specified in the simulators, while for adalimumab and infliximab, independent in-vitro data were used [2]. All simulations were done for a standard virtual male healthy volunteer (height: 176.6 cm, body weight: 80.7 kg). Concentration-time profiles in plasma and comparable tissues defined in all platforms (adipose, brain, heart, kidney, liver, lung, muscle, skin and spleen) were used as primary outputs. Area under the curve (AUC) and half-life (t½) were calculated with the PKNCA R package [3]. Parameters were quantitively compared between platforms using the percent coefficient of variation (%CV). Agreement between model outcomes and observed literature data was expressed as fold errors (FE) for individual observations and absolute average fold errors (AAFEs) for multiple observations. The proposed antibody distribution coefficients (ABC, i.e. the tissue/plasma concentration ratios) by Shah and Betts (total concentrations) [4] and Rafidi et al. (interstitial fluid (ISF) concentrations) [5] were used for the accuracy assessment of tissue distribution. To investigate the accuracy of predictions of mAb plasma kinetics, we compared model predictions with published observed data originating from five different clinical studies investigating the PK of adalimumab [6,7] and infliximab [8–10] at different intravenous doses.
Results: Differences in plasma kinetics of IgGIV between platforms were minimal for a dose of 1 mg/kg (9%CV for AUC0-672h, 5%CV for t½). The associated total tissue profiles differed significantly between the platforms: of the 9 comparable tissues, only 4 had a %CV on AUC0-672h below 50%: adipose (27%CV), brain (11%CV), heart (15%CV) and muscle (15%CV). Predicted ABCs for total tissue concentrations were within twofold of observed values except for brain (FEs: 8.12, 6.40 and 9.49 in Simcyp, PK-Sim and GastroPlus, respectively), lung (FE: 2.85 in PK-Sim), skin (FE: 0.36 in GastroPlus) and spleen tissue (FE: 2.44 in PK-Sim). For AUC0-672h values calculated from ISF tissue concentrations, only heart and lung values varied less than 50%CV between platforms (12 and 42 %CV, respectively). ABCs based on ISF concentration were within twofold except for adipose (FE: 3.65, Simcyp), brain (FE: <0.01 in Simcyp, FE: 2.94 in PK-Sim), heart (FE: 0.34 in Simcyp and GastroPlus, FE: 0.23 in PK-Sim), muscle (FE: 0.29 in GastroPlus) and skin (FE: 0.25 in GastroPlus).
The overall accuracy of the plasma profiles of adalimumab was similar across platforms; (AAFE of 1.36, 1.40 and 1.55 in Simcyp, PK-Sim and GastroPlus, respectively). For infliximab, the AAFEs were higher but also similar across platforms; 1.95 for Simcyp, 1.96 for PK-Sim and 2.28 for GastroPlus. Predicted AUCinf and t½ of both mAbs were higher than observed values. Incorporation of target-mediated-drug disposition did not resolve the overprediction of t½.
Conclusions: These results show that plasma predictions of mAbs are similar across the investigated PBPK platforms but tissue predictions vary considerably. When using PBPK to assess effect-site concentrations in tissues, the model verification procedure should not be confined to plasma only but also include an accuracy assessment on the tissue(s) of interest.
References:
[1] Tang Y, Li X, Cao Y. Which factors matter the most? Revisiting and dissecting antibody therapeutic doses. Drug Discovery Today. 2021;26:1980–90.
[2] Suzuki T, Ishii-Watabe A, Tada M, Kobayashi T, Kanayasu-Toyoda T, Kawanishi T, et al. Importance of Neonatal FcR in Regulating the Serum Half-Life of Therapeutic Proteins Containing the Fc Domain of Human IgG1: A Comparative Study of the Affinity of Monoclonal Antibodies and Fc-Fusion Proteins to Human Neonatal FcR. The Journal of Immunology. 2010;184:1968–76.
[3] Denney WS, Duvvuri S, Buckeridge C. Simple, Automatic Noncompartmental Analysis: The PKNCA R Package. Journal of Pharmacokinetics and Pharmacodynamics. 2015;42:11-107,S65.
[4] Shah DK, Betts AM. Antibody biodistribution coefficients. MAbs. 2013;5:297–305.
[5] Rafidi H, Rajan S, Urban K, Shatz-Binder W, Hui K, Ferl GZ, et al. Effect of molecular size on interstitial pharmacokinetics and tissue catabolism of antibodies. mAbs. Taylor & Francis; 2022;14:2085535.
[6] Broeder A den, Putte L van de, Rau R, Schattenkirchner M, Riel PV, Sander O, et al. A single dose, placebo controlled study of the fully human anti-tumor necrosis factor-alpha antibody adalimumab (D2E7) in patients with rheumatoid arthritis. The Journal of Rheumatology. The Journal of Rheumatology; 2002;29:2288–98.
[7] Weisman MH, Moreland LW, Furst DE, Weinblatt ME, Keystone EC, Paulus HE, et al. Efficacy, pharmacokinetic, and safety assessment of adalimumab, a fully human anti-tumor necrosis factor-alpha monoclonal antibody, in adults with rheumatoid arthritis receiving concomitant methotrexate: A pilot study. Clinical Therapeutics. 2003;25:1700–21
[8] Palaparthy R, Udata C, Hua SY, Yin D, Cai C-H, Salts S, et al. A randomized study comparing the pharmacokinetics of the potential biosimilar PF-06438179/GP1111 with Remicade® (infliximab) in healthy subjects (REFLECTIONS B537-01). Expert Review of Clinical Immunology. Taylor & Francis; 2018;14:329–36.
[9] Park W, Lee SJ, Yun J, Yoo DH. Comparison of the pharmacokinetics and safety of three formulations of infliximab (CT-P13, EU-approved reference infliximab and the US-licensed reference infliximab) in healthy subjects: a randomized, double-blind, three-arm, parallel-group, single-dose, Phase I study. Expert Review of Clinical Immunology. Taylor & Francis; 2015;11:25–31.
[10] Kavanaugh A, St Clair EW, McCune WJ, Braakman T, Lipsky P. Chimeric anti-tumor necrosis factor-alpha monoclonal antibody treatment of patients with rheumatoid arthritis receiving methotrexate therapy. J Rheumatol. 2000;27:841–50.