Chiara Fornari 1, Anna Largajolli, Chiara Nicolò, Orla Cunningham, Emmanuel Vial
1 Certara (, Italy)
INTRODUCTION: GTX-B001 is a humanized bispecific antibody designed to selectively target and inhibit mast cells by binding to c-Kit and CD203c. GTX-B001 is expected to be best-in-class for the treatment of Chronic Urticaria (CU). We present a novel modeling and simulation framework based on the integration of in vitro and in vivo data.
OBJECTIVES:
– To develop a quantitative framework that integrates in vitro binding dynamic data with in vivo pharmacokinetic (PK) data in Non-Human Primates (NHP) and enables to predict GTX-B001:c-Kit:CD203c formation in skin (target organ in CU).
– To aid the interpretation of GTX-B001 pharmacological activity by identifying optimal GTX-B001 concentration range to maximize skin trimer concentrations (driver of efficacy).
– To support FIH study design via translation to humans.
METHODS: GTX-B001 in vitro binding dynamic model was based on the work by Sengers [1] and relied on experimentally measurable parameters (N=10) and assumptions from literature [1] (18 parameters in total).
GTX-B001 in vivo minimal physiologically based pharmacokinetic (mPBPK) model was based on the work by Cao and Zhao [2, 3]. System-specific parameters (N=9) were informed using literature-retrieved values [2, 3]. Drug-specific parameters (N=6) were estimated from two NHP studies (N=10; GTX-B001 1, 10, and 50 mg/kg administered as single and repeated dose administration; SD, RD). Model fitting was performed in R with nlmixr2 package [4].
The GTX-B001 in vitro model was then integrated with GTX-B001 in vivo mPBPK model assuming that: (i) GTX-B001 binding to its targets occurs in the skin compartment of mPBPK model, (ii) only cis-binding is happening, (iii) the two targets are uniformly mixed and exposed to the antibodies, (iv) binding dynamics do not directly modify GTX-B001 PK. Assumption (iv) was supported by the absence of clear target-mediated drug disposition features in the observed concentration–time profiles.
Lastly, translation to humans was based on species-specific physiology [2, 3] and allometry [5, 6]. Clinical regimens (0.11, 0.33, 1, 3, 9, 27 mg/kg, administered as SD via IV infusion of 30 minutes) were simulated to inform FIH study.
RESULTS: GTX-B001 in vitro model characterized GTX-B001 binding dynamics and showed that the profile of trimer concentrations versus GTX-B001 concentrations follows a bell-shaped curve, typical for bispecific antibodies. At low GTX-B001 concentrations (10-2-10-1 nM), cell binding is dominated by cross-linked antibody, whereas monovalent binding increases at higher concentrations (104-105 nM), when cross-linking is complete. Ternary complexes are favored at intermediate antibody concentrations (100-103 nM), which are model predicted GTX-B001 concentration for efficacy.
GTX-B001 mPBPK model characterized GTX-B001 PK in NHP. The model, using GTX-B001 plasma concentrations as input, predicted concentrations in skin, from which direct observations were not available. Consistent with antibody distribution, skin concentrations resulted several folds lower than in plasma. GTX-B001 plasma clearance (CLp) was 5.9E-4 L/h, plasma volume of distribution (Vp) was 1.2E-1 L. The model also included a non-linear CL from plasma compartment, which helped to better reproduce GTX-B001 PK profiles associated with the 1 mg/kg dose administration and suggesting dose dependent exposure due to target-mediated clearance.
GTX-B001 in vitro – in vivo model bridged the in vitro system with the in vivo system and predicted trimer concentrations and fraction occupancy in NHP, which are key to interpret GTX-B001 pharmacology activity. In terms of trimer formation and fraction occupancy, 1 and 10 mg/kg resulted the best regimens, 50 mg/kg RD regimen resulted sub-optimal. Trimer concentrations tended to decrease after RD administration.
The human model enabled to simulate clinical dosing regimen of interest. 3 and 9 mg/kg (administered as SD via IV infusion of 30 minutes) resulted as best regimens in terms of trimer formation and likely efficacy. Doses ≥ of 3 mg/kg achieved skin concentrations which remained above GTX-B001 in vitro IC50 threshold (5.2 nM) for the simulated period of 4 weeks. 1 and 0.33 mg/kg resulted in some trimer formation, but exposure was equal or below the in vitro IC50 threshold (even at Cmax), so they were not considered therapeutic.
CONCLUSIONS: We developed a novel mathematical framework which integrates in vitro data of GTX-B001 cellular binding with GTX-B001 PK data in NHP and is key to interpret pharmacological activity. The model provides in vivo predictions of GTX-B001:c-Kit:CD203c formation dynamics in skin. This framework was used to inform FIH study via translation to humans.
References:
[1] Sengers et al., MABS, 2016, VOL. 8, NO. 5, 905–915
[2] Cao et al., 2013, J Pharmacokinet Pharmacodyn, 40, 597–607
[3] Zhao et al., 2015, Pharm Res., 32(10), 3269–3281
[4] Fidler M., nlmixr2: Nonlinear Mixed Effects Models in Population PK/PD. R package version 5.0.0, https://nlmixr2.org/.
[5] Milo et al. Nucl. Acids Res. (2010) 38 (suppl 1): D750-D753.
[6] Germovsek et al., 2021, mAbs, 13:1, 1964935
Reference: PAGE 34 (2026) Abstr 12124 [www.page-meeting.org/?abstract=12124]
Poster: Methodology - New Modelling Approaches