Dynamic in vitro PKPD assessment to improve pharmacological response profiling of T-cell bispecifics
Miro J. Eigenmann (1), Sylvia Herter (2), Sarah Diggelmann (2), Florian Limani (2), Jitka Somandin (2), Nicolas Frances (1), Martin Lechmann (1), Anneliese Schneider (2), Marina Bacac (2), Antje-Christine Walz (1)
(1) Roche Pharmaceutical Research and Early Development, Roche Innovation Center Basel, Switzerland, (2) Roche Pharmaceutical Research & Early Development, Roche Innovation Center Zürich, Switzerland
Objectives: The approach for First-in-Human dose selection for cancer immunotherapy is typically based on minimally-anticipated biological effect level (MABEL) using in vitro data. A recent review by the FDA highlights that these in vitro experiments, conducted as single time measurements result into broad ranges of EC50 values and can vary under different assay conditions [1]. Response to immuno-oncology treatment is complex and includes series of events such as tumour-cell killing, cytokine release and T-cell activation which can be related to efficacy and safety of the drug. In such a complex network, these effects, however, often occur on different time scales which may lead to a bias when two readouts are quantitatively compared based on a static readout. This subsequently leads to uncertainty and inaccuracy in a derived MABEL dose prediction. Therefore more mechanistic means are needed for a better translation from in vitro to human. Here, we compare the performance of a static and dynamic in vitro PKPD assessment for CEA-TCB, a T-Cell-Bispecific Monoclonal Antibody targeting the carcinoembryonic antigen [2]. We outline how experiments and subsequent data analysis can be performed in order to get a more robust assessment of the drug’s potency on various pharmacological readouts which are relevant for human dose selection.
Methods: CEA expressing tumour cell lines MKN45 (CEA high) and CX1 (CEA low) were co-cultured with human PBMCs (peripheral blood mononuclear cells) at different drug concentrations. A dynamic in vitro assay was conducted where tumour cell killing, immuno-phenotyping and cytokine release was assessed over time at 24, 48, 72, 96 and 168h. Tumour cell killing was measured by FACS and LDH release. Immuno-phenotyping of CD4 and CD8 T-cells are performed by FACS while cytokine release (IL2/6/10, IFNμ, TNFα, …) was assessed by cytometric bead array. PKPD analysis was conducted using Phoenix WinNonlin. An Emax model was fitted first to the static data where effect over concentration was evaluated at each time point. In a second step the AUCE (Area under the curve of the effect) was calculated for each concentration level and an Emax model was fitted based on the AUCE over concentration profiles for each PD readout. Derived EC50s, time course of the different PD readouts and results of the static vs. dynamic approach were compared. Potency comparison on tumour cell killing and IL6 release was proposed to explore the therapeutic index of the drug.
Results: CD4 and CD8 T-cells expand in a timely delayed manner around 48h after drug is added to the suspension. Combined with tumour cell killing over time this leads to an increase of the effector-/target cell ratio (E/T-ratio) over time. EC50 for IL2 secretion is lowest when estimated at 24h but is not identifiable after 48h as IL2 has been consumed in the system by that time. Lowest EC50 values for IL6 release are found at later times, after 96 hours. Also for tumour cell killing EC50 estimates are lower at later times, whereas they are consistently lower than EC50s for IL6 at all times. As readout for the therapeutic index, we computed the ratio of EC50 of IL6 as a potential marker for safety over the EC50 of tumour killing efficacy readout. Here, dynamic in vitro PKPD predicts a therapeutic index of 150, while this assessment with the static analysis leads to indices ranging from 5 to 130. These findings demonstrate that estimating EC50s based on a static in vitro readout is variable across different time points and appears unreliable.
Conclusions: Our results indicate that a comparison of the drug’s effect on different PD readouts occurring at different time scales is not meaningful. This can be circumvented when considering the full time course in the PKPD analysis such as relating the integral of the time course of the effect (AUCE) versus concentration to derive the potency. The relevant time span thereby depends on the time when the effect comes into play during the complex cascade of immune response and T-cell mediated tumour killing. Further integrating such dynamic in vitro data into a systems pharmacology framework will help to better understand the mechanics behind tumour lysis and immune response upon treatment with TCBs as function for target expression and drug-target interaction. Progressing in the understanding of this complex system will eventually enable a refined MABEL approach and a better in vitro-in vivo extrapolation.
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] Bacac, M., et al., A Novel Carcinoembryonic Antigen T-Cell Bispecific Antibody (CEA TCB) for the Treatment of Solid Tumors. Clin Cancer Res, 2016. 22(13): p. 3286-97.
