Darian Schirr 1,2, Andreas Reichel 1
1 Bayer AG (Berlin, Germany), 2 Charité - Universitätsmedizin Berlin (Berlin, Germany)
Introduction
Targeted covalent inhibitors (TCIs) are a therapeutic modality which currently attracts much attention, especially in oncology drug discovery and development. Their prolonged target residence time that results in sustained target inhibition is expected to outlast the plasma concentration-time profile of the administered drug.1,2 This creates a beneficial PK/PD (pharmacokinetic/pharmacodynamic) disconnect and enables inhibition of targets previously considered undruggable.
The PK/PD disconnect arises from the different time scales of drug kinetics and target turnover, i.e., the continuous degradation and resynthesis of the protein.3 To fully comprehend these complex processes, a mechanistic PK/PD model is required to quantitatively link PK to the irreversibility of target binding and to PD of the target. Because of the irreversible binding mode, IC50 values are time-dependent and therefore poorly informative for PK/PD modeling.1,4,5 Time-related IC50 values can, however, be used in the lead generation phase as a proxy for covalent potency since they approach a linear relation with the ratio kinact/KI, the appropriate potency parameter for TCIs, in functional inhibition assays routinely performed in drug discovery.6
Objectives
1. Establish a mechanistic PK/PD model to quantitatively account for the irreversible target binding mode of TCIs.
2. Apply the inferred insights to model-informed drug discovery across the interface between research and preclinical development.
3. Employ the PK/PD modeling framework for real-world drug discovery projects to inform study design, understand in vivo data, and guide and decision-making.
Methods
Our case study uses the allosteric covalent WRN helicase inhibitor VVD-133214. Plasma concentrations, target engagement (TE) in tumor, and tumor growth inhibition studies in xenograft mice are taken from Baltgavis et al. (2024)7 and Kikuchi et al. (2025).8 We fit a two-compartment PK model to observed plasma concentration data and link a target turnover model to the concentration in the central compartment to capture TE in tumor tissue. We then couple TE to the Simeoni tumor growth model9 via an Emax function to explain the observed tumor growth dynamics.
Because preclinical data are often sparse and collected sequentially, early indicators of TE are preferred. Strelow (2017)1 estimates TE directly from the unbound AUC and potency (kinact/KI):
% target engagement = 100 * (1 – exp(-kinact/KI * AUCu))
Because Strelow’s equation omits target turnover, we compare its TE estimation with our PK/TE/PD model to determine scenarios in which target turnover can be neglected. To account for target turnover, we use ordinary differential equations in our full model that additionally include the rate of target degradation.10
To identify conditions where target turnover is critical, we examine the ratio of target half-life to the dosing interval. We simulate the TE difference between our turnover model and the Strelow equation (ΔTE) across wide ranges of dose, fraction unbound in plasma and potency (kinact/KI). Using the upper plateau of our TE-Emax model, we define a ΔTE threshold at which potential overprediction by Strelow still yields full efficacy.
Additionally, we explored whether TMDD models, which also capture time- and concentration-dependent PK/PD effects, can be used to describe TCIs. However, like most small molecules, TCIs are typically given at doses that exceed TMDD-relevant ranges.
Results
We observe that across all tested dose, fraction unbound and potency scenarios the ΔTE is below the defined ΔTE threshold when the half-life of the target protein is equal to or greater than seven times the dosing interval. In these cases, the Strelow equation offers a quantitative measure of TE with negligible in vivo efficacy loss. However, the full target turnover model is required to accurately capture TE when the dosing interval approaches the target turnover time, as compounds with high TE predicted by Strelow will not achieve high in vivo TE.
Conclusion
Model-based guidance is essential in early TCI projects for both understanding and successfully translating the characteristics of this special modality. When aiming to leverage the prolonged activity due to irreversible binding, it is necessary to consider target turnover. Strelow’s equation is a useful starting point for ranking in lead generation but should be replaced by a full turnover model. Our modeling framework (i) guides TCI design and optimization, (ii) enables initial TE predictions from combined in vitro and in vivo PK data, and (iii) informs efficacy study design to reduce animal use while increasing the likelihood of success.
References:
1. Strelow, J. M. A Perspective on the Kinetics of Covalent and Irreversible Inhibition. SLAS Discov. 22, 3–20 (2017).
2. Bhattachar, S. N., Morrison, J. S., Mudra, D. R. & Bender, D. M. Translating Molecules into Medicines. vol. 25 (Springer International Publishing, Cham, 2017).
3. Gabrielsson, J. & Hjorth, S. Turn On, Tune In, Turnover! Target Biology Impacts In Vivo Potency, Efficacy, and Clearance. Pharmacol Rev. 75, 416–462 (2023).
4. Mader, L. K. & Keillor, J. W. Methods for kinetic evaluation of reversible covalent inhibitors from time-dependent IC50 data. RSC Med Chem. 16, 2517–2531 (2025).
5. Krippendorff, B.-F., Neuhaus, R., Lienau, P., Reichel, A. & Huisinga, W. Mechanism-Based Inhibition: Deriving KI and kinact Directly from Time-Dependent IC50 Values. SLAS Discov. 14, 913–923 (2009).
6. Jeon, J., Kholodar, S. A., Tran, B. H., et al. A practical method for determining the rate of covalent modification of fragments and leads. Nat Commun. 16, 11234 (2025).
7. Baltgalvis, K. A., Lamb, K. N., Symons, K. T., et al. Chemoproteomic discovery of a covalent allosteric inhibitor of WRN helicase. Nature. 629, 435–442 (2024).
8. Kikuchi, S., Green, J. C., Rogness, D. C., et al. Identification of VVD-214/RO7589831, a Clinical-Stage, Covalent Allosteric Inhibitor of WRN Helicase for the Treatment of MSI-High Cancers. J Med Chem. 68, 25912–25938 (2025).
9. Simeoni, M., Magni, P., Cammia, C., et al. Predictive Pharmacokinetic-Pharmacodynamic Modeling of Tumor Growth Kinetics in Xenograft Models after Administration of Anticancer Agents. Cancer Res. 64, 1094–1101 (2004).
10. Yang, Z. Achieving a low human dose for targeted covalent drugs: Pharmacokinetic and pharmacodynamic considerations on target characteristics and drug attributes. Biopharm Drug Dispos. 42, 150–159 (2021).
Reference: PAGE 34 (2026) Abstr 11936 [www.page-meeting.org/?abstract=11936]
Poster: Methodology - New Modelling Approaches