Sofia Guzzetti (1), Elisa Fevola (1)
(1) Early Oncology DMPK, AstraZeneca, 1 Francis Crick Avenue, Cambridge, Cambridgeshire, CB20AA, UK
Objectives: PROTACs [1] are novel small molecules which violate not only the optimal, well-understood physicochemical design space [2], but also the well-established modelling principles for traditional small molecules. The three-body binding mechanism introduces layers of complexity which make rational compound optimization and translational PKPD prediction extremely challenging. Notably, because potency (DC50) and maximal degradation (Dmax) can be highly sensitive to endogenous ligase and target baselines [3], (i) the criteria for compound screening based on activity and/or selectivity may heavily depend on the choice of cell lines; (ii) degradation in vitro does not necessarily predict degradation in vivo; (iii) traditional metrics for establishing a PK/Efficacy (E) relationship, such as plasma exposure over a target concentration, are inadequate because the same exposure may not correspond to equivalent degradation if Dmax differs; (iv) degradation may not translate across species if protein baselines differ. Moreover, due to the physicochemical properties of PROTACs, in vitro data may be confounded by non-specific binding, and lack of measurable fraction unbound (fu) makes it impossible to relate in vitro free concentrations to in vivo unbound exposures. This work demonstrates how to tackle such major challenges by applying the pragmatic PKPD modelling approach based on [3] to a project-based case study, enriched by bespoke translational modelling of in vitro/in vivo (IVIV) efficacy data. Such framework is currently being used within AstraZeneca to enable acceleration of drug discovery programs and increase probability of success in the clinic.
Methods: In vitro degradation data at multiple timepoints across different cell lines (IF [4]) was modelled with the approach described in [3], which employs a turnover model with a bi-sigmoidal degradation stimulus where Dmax and DC50 are statistical functions of the ligase/target baseline ratio. Such model was then used to predict the kinetics of degradation in vivo using plasma or tumor PK as a driver. In vitro antiproliferation data at different timepoints (CTG [5]) as well as tumor volume in vivo were modelled by Growth Rate inhibition index (GR [6]). The predictivity of in vitro PK/PD and PK/E models was validated with in vivo PD and efficacy data on several compounds. Whenever the fu was unmeasurable, total plasma concentrations were used and in vitro potencies were corrected based on the assumption that the fu scales linearly with the amount of protein in the incubation, and that binding in mouse plasma is comparable to that in 100% foetal calf serum [7]. Finally, for a selected number of compounds a bespoke semi-mechanistic PKPD/E model based on [3] and [8] was built and used to inform study designs exploring intermittent scheduling.
Results: The bespoke methods resulted in a successful translation of the in vitro PKPD relationship, with degradation time courses in vivo being accurately predicted from in vitro data for multiple compounds at different doses. Remarkably, despite the well-known challenge of IVIV translation of efficacy due to the complex physiology and biology of the tumor micro-environment which may not be fully captured by in vitro systems, the GR dose-responses in vitro were consistent with PK/E relationship in vivo based on tumor volume. As a result, also the PD/E relationship translated from in vitro to in vivo, showing that ~80% degradation is consistently required for stasis. Overall, these results also highlight how efficacy metrics based on growth rates (vs. single timepoint data) are mechanistically more robust and increase the likelihood of IVIV translation. Building on top of in vivo PKPD modelling, the PD-driven efficacy model was able not only to capture the tumor growth kinetics upon continuous dosing, but also to predict tumor growth kinetics upon intermittent dosing.
Conclusions: In this work we demonstrate how the many challenges related to the IVIV translation of PROTACs PKPD/E relationship can be overcome by a pragmatic modelling approach which accounts for the sensitivity of degradation to endogenous conditions (i.e. ligase/target baselines), as well as mechanistic efficacy metrics based on cell/tumor growth kinetics instead of single timepoint data (such as cell viability or TGI). Such milestone enables acceleration of drug discovery programs and the development of bespoke PKPD/E models that can be used for optimal study design and prediction.
References:
[1] Sakamoto, K.M., Kim, K.B., Kumagai, A., Mercurio, F., Crews, C.M. and Deshaies, R.J., 2001. Protacs: Chimeric molecules that target proteins to the Skp1–Cullin–F box complex for ubiquitination and degradation. Proceedings of the National Academy of Sciences, 98(15), pp.8554-8559.
[2] Pike, A., Williamson, B., Harlfinger, S., Martin, S. and McGinnity, D.F., 2020. Optimising proteolysis-targeting chimeras (PROTACs) for oral drug delivery: a drug metabolism and pharmacokinetics perspective. Drug Discovery Today, 25(10), pp.1793-1800.
[3] Guzzetti, S. and Morentin Gutierrez, P., 2023. An integrated modelling approach for targeted degradation: insights on optimization, data requirements and PKPD predictions from semi-or fully-mechanistic models and exact steady state solutions. Journal of Pharmacokinetics and Pharmacodynamics, pp.1-23.
[4] Garvey, C.M., Spiller, E., Lindsay, D., Chiang, C.T., Choi, N.C., Agus, D.B., Mallick, P., Foo, J. and Mumenthaler, S.M., 2016. A high-content image-based method for quantitatively studying context-dependent cell population dynamics. Scientific reports, 6(1), p.29752.
[5] Hannah, R., Beck, M., Moravec, R. and Riss, T., 2001. CellTiter-Glo™ Luminescent cell viability assay: a sensitive and rapid method for determining cell viability. Promega Cell Notes, 2, pp.11-13.
[6] Hafner, M., Niepel, M., Chung, M. and Sorger, P.K., 2016. Growth rate inhibition metrics correct for confounders in measuring sensitivity to cancer drugs. Nature methods, 13(6), pp.521-527.
[7] Kalvass, J.C., Phipps, C., Jenkins, G.J., Stuart, P., Zhang, X., Heinle, L., Nijsen, M.J. and Fischer, V., 2018. Mathematical and experimental validation of flux dialysis method: an improved approach to measure unbound fraction for compounds with high protein binding and other challenging properties. Drug Metabolism and Disposition, 46(4), pp.458-469.
[8] Simeoni, M., Magni, P., Cammia, C., De Nicolao, G., Croci, V., Pesenti, E., Germani, M., Poggesi, I. and Rocchetti, M., 2004. Predictive pharmacokinetic-pharmacodynamic modeling of tumor growth kinetics in xenograft models after administration of anticancer agents. Cancer research, 64(3), pp.1094-1101.
Reference: PAGE 32 (2024) Abstr 11109 [www.page-meeting.org/?abstract=11109]
Poster: Methodology - New Modelling Approaches