Arthur Van De Vyver (1,3), Miro Eigenmann (1), Melanie Knobloch (2), Tina Weinzierl (2), Thorsten Lehr (3), Antje-Christine Walz (1)
(1) Roche Pharma Research and Early Development, Pharmaceutical Sciences, Roche Innovation Center Basel, Switzerland; (2) Roche Pharma Research and Early Development, Cancer Immunotherapy department 2, Roche Innovation Center Zürich Switzerland; (3) Saarland University, Department of Clinical Pharmacy, Saarbrücken, Germany
Objectives: Minimally Anticipated Biological Effect Level (MABEL) is an often used approach for First-in-Human (FIH) dose selection in cancer immunotherapy. This approach warrants a holistic view on the available non-clinical data in order to make an informed decision on the FIH dose. Therapeutic responses to immune-oncology treatment encompass many different biological actors and processes, including T cell activation, cytokine release, and tumour cell lysis. Biological responses take place at different time scales, which can be studied in vitro and in vivo. However, it may be misleading to compare different read-outs that are kinetically distinct at one single time point as is often done with in vitro assays, which may result in a bias in the analysis as highlighted before [1]. We previously proposed dynamic PK/PD analysis as an alternative approach [2]. In this work, we compare the performance of both static and dynamic assessment of in vitro activity of FolR1-TCB, a T Cell Bispecific antibody (TCB) that targets Folate Receptor 1 (FolR1) across a broad concentration range in high and low target expressing tumour cells and with different ratios of effector to tumour cells (E:T ratio).
Methods: HEK cells stably transfected with high (M5) and low (L2) FolR1 expression were co-cultured with hPBMCs at different TCB concentrations. Both a high and low affinity variant of the FolR1-TCB were used in the experiments. Additionally, cell line M5 was incubated with high affinity FolR1-TCB and co-cultured with 4 different E:T ratios. A dynamic in vitro assay was performed where tumour cell counts were continuously monitored over one week with incuCyte and cytokines were measured at 18h, 44h, 68h, & 92h with CBA. PK/PD analysis was conducted in Phoenix WinNonlin. Dynamic data was compiled by performing NCA to derive the AUCE (Area Under the Curve for Effect) for each TCB concentration. The static readouts and the AUCE values were fit with an Emax model. The EC50s derived from the Emax model were compared between the static and dynamic approach.
Results: Maximal cytokine release is highest for M5 targeted by high-affinity FolR1-TCB. The TCB acting on L2 cells elicits almost no cytokine release. IL6 release is highest at 92h and about 2-fold higher for the high- versus low-affinity TCB. When performing a static analysis, the EC50 values appear to increase over time for both compounds. At 20h, EC50s for IL6 release only differ 30% between the TCBs. At later time points this turns into a multi-fold difference. Eventually, the dynamic IL6 potency is 4.5-fold higher for the high-affinity compound.
Static potency of IL6 release increases with time across the tested E:T ratios and is maximal for the highest E:T. Also, the highest E:T ratio leads to the earliest and lowest peak in IL2 levels. Presumably, this is due to rapid IL2 consumption by higher T cell concentrations. This difference is also reflected in static potency read-outs, with no uniform pattern over time.
As a metric for a therapeutic index (TI) we propose the EC50 ratio between IL6 release as a safety readout and tumour cell killing as efficacy readout. With the static analysis, the EC50 values for tumour killing appear to decrease with time and are lower than the corresponding EC50 values for IL6 release, and yield TI values that vary between time points from 2- up to 248-fold, depending on the experimental condition. Furthermore, the kinetics of cytokine release vary among the PBMC donors, which may impact the potency readouts at each time point. A dynamic analysis of the full time profile may therefore give a better overview of potency differences between donors and experiments.
Conclusions: We showed the superiority of a dynamic PK/PD approach in comparison to static analyses in assessing and comparing in vitro TCB drug effects. Determining a therapeutic index based on in vitro readouts in a time-independent manner provides a more robust insight into the pharmacological response. Our results indicate that a static analysis of the TI, which is the ratio of two kinetically different readouts, is not meaningful to compare across conditions or compounds. The time-independent analysis as presented in this work, enables a robust comparison between compounds and assays. A dynamic PK/PD analysis leverages the available experimental data to its full extent, thereby ensuring a holistic approach for MABEL determination and facilitating comparison between multiple experiments.
References:
[1] Saber, H., et al., An FDA oncology analysis of CD3 bispecific constructs and first-in-human dose selection. Regul Toxicol Pharmacol, 2017. 90: p. 144-152
[2] Eigenmann, M., et al., PAGE 27 (2018) Abstr 8621 [www.page-meeting.org/?abstract=8621].
Reference: PAGE () Abstr 9532 [www.page-meeting.org/?abstract=9532]
Poster: Drug/Disease Modelling - Oncology