(T1230-02-08) Reverse Translational Modeling for Early Benchmarking of Clinically Active Doses from In Vitro Cell Killing and T-Cell Activation Data for BCMAxCD3 T-Cell Engagers

Purpose: Implementation of Project Optimus necessitates early dose rationalization strategies. This is challenging for T-cell engaging bispecifics (TCEs) which exhibit a complex mechanism of action. While several strategies have evolved for selecting a safe first-in-human (FiH) starting dose (e.g. minimal anticipated biological effect level (MABEL)), fewer efforts have focused on early benchmarking of active doses for expansion [1]. In this work, we demonstrate a three-step reverse translation strategy leveraging new in vitro data and published ex vivo data with teclistamab, a bispecific targeting B-cell maturation antigen (BCMA) and CD3 for relapsed/refractory multiple myeloma (RRMM). Using in vitro redirected T-cell cytotoxicity (RTCC) data and teclistamab-derived potency scaling factors, we apply our framework to elranatamab and WVT078, two other BCMA bispecifics, and attempt to recapitulate clinically active doses. This work outlines and qualifies a simple model-informed drug development (MIDD) translational strategy for early approximations of clinically relevant dose prediction from in vitro data and clinical data analysis.

Methods: Teclistamab RP2D selection was based on plasma PK achieving maximum EC₉₀ determined using ex vivo RTCC cell killing from healthy donor T cells and MM patient bone marrow (BM) samples [2]. In the absence of such ex vivo BM data with other candidates, we devised a reverse translation strategy using QSP modeling and in vitro RTCC data (Figure 1A). Briefly, in vitro RTCC data for teclistamab, elranatamab, and WVT078 were generated using healthy donor PBMCs and the KMS11-Fluc-mCherry tumor cell line (5:1 E:T ratio) at 48 hours (peak tumor cell count) and 168 hours (peak tumor cell death) by Incucyte. Immunophenotyping to measure T-cell activation (CD25+ CD4+/CD8+) was evaluated by FACS at 168 hours (time of peak activated T cells). We extracted teclistamab T-cell activation and cell killing data from patient samples [2], teclistamab clinical pharmacokinetics (PK) and RP2D [3], WVT078 clinically active doses [4], elranatamab clinical PK parameters [5], and elranatamab RP2D [6] from literature. WVT078 clinical PK parameters were from internal analysis. Bispecific activity depends on ability to bind, activate, and kill. Our analyses account for degree of target engagement (depending on BCMA and CD3 affinity and expression), T-cell activation from target engagement, and tumor cell death sensitivity accordingly. Immunological synapse (“trimer”) modeling was performed according to [7] using published and in-house measured binding kinetics. Using the trimer model, we converted concentration-based EC₉₀ potency to trimer number per T cell (TC₉₀) values under a quasi-equilibrium assumption as described in [8].

Results: The central hypothesis of the reverse translation strategy is that tumor cell killing by activated T cells is bispecific drug-agnostic, supported by in vitro RTCC data with KMS11 cells for teclistamab, elranatamab, and WVT078 (Figure 2A). Based on this hypothesis, we reverse translated the ex vivo clinical potency based RP2D for teclistamab for in vitro scaling (Figure 1A). We first benchmark the number of active T cells required for 90% tumor cell death in vitro (“ActiveT₉₀”). Over a correlation plot of measured active T cell/µL and calculated trimers per cell (TC_x), we identify the in vitro TC₉₀ using the ActiveT₉₀ threshold. We next analyzed teclistamab ex vivo T-cell activation and cell killing data, focusing on data from the most and least sensitive MM patients (5 patients total) for a response range. Over a plot of ex vivo trimers per cell (calculated using published affinity and BM expression data in these MM patients) versus concentration, we benchmark the individual drug TC₉₀ to determine the “expected” clinical EC₉₀ for teclistamab, assuming similar intrinsic sensitivity as KMS11. Scaling factors were derived from the fold-difference between expected EC₉₀ and measured maximum and minimum EC₉₀ from the MM patient samples. Since the contribution of affinity and expression are accounted for in TC calculations and both teclistamab RTCC datasets use healthy donor T cells (assumed to activate similarly), this fold-difference in EC₉₀ was attributed to intrinsic sensitivity differences. With our in vitro RTCC assay data with teclistamab, elranatamab, and WVT078 as a bridge for translation, the scaling factors derived from the above teclistamab analysis were applied to elranatamab and WVT078 to estimate an equivalent predicted clinical EC₉₀ range. We leveraged a minimal physiologically-based PK (mPBPK) model integrated with a target engagement model (Figure 1B) ([9], modified) to identify doses with plasma and bone marrow PK predicted to cover the scaled clinical EC₉₀ range. The resulting dose predictions were well aligned with clinically active doses reported for elranatamab (Figure 2C) and WVT078 (not shown).

Conclusion: For TCEs, the EC₉₀ from in vitro RTCC assays can inform preclinical-to-clinical dose translation in several ways. Our reverse translation analysis paired in vitro RTCC data with published clinical data analyses. When target-specific historical clinical data are available, generating in vitro data can enable early, rapid benchmarking of predicted clinically active doses. This context is of significance for TCEs, as the FiH starting dose is often low for safety and requires lengthy escalation.

References: [1] Shastri, P.N., et al. “Industry Perspective on First-in-Human and Clinical Pharmacology Strategies to Support Clinical Development of T-Cell Engaging Bispecific Antibodies for Cancer Therapy”, CPT, 2024.
[2] Pillarisetti, K., et al. “Teclistamab is an active T cell-redirecting bispecific antibody against B-cell maturation antigen for multiple myeloma”, Blood Adv, 2020.
[3] Girgis, S., et al. “Translational Modeling Predicts Efficacious Therapeutic Dosing Range of Teclistamab for Multiple Myeloma”, Target Oncol, 2022.
[4] Raab, M.S., Cohen, Y.C., Schjesvold, F. et al. Preclinical discovery and initial clinical data of WVT078, a BCMA × CD3 bispecific antibody. Leukemia 37, 1349–1360 (2023).
[5] European Medicines Agency. Elrexfio: EPAR - Public Assessment Report. 2023. European Medicines Agency.
[6] Bahlis, N.J., Costello, C.L., Raje, N.S. et al. Elranatamab in relapsed or refractory multiple myeloma: the MagnetisMM-1 phase 1 trial. Nat Med 29, 2570–2576 (2023).
[7] Betts, A., et al. “A Translational Quantitative Systems Pharmacology Model for CD3 Bispecific Molecules: Application to Quantify T Cell-Mediated Tumor Cell Killing by P-Cadherin LP DART®”, AAPS J, 2019.
[8] Schropp, et al. “Target-Mediated Drug Disposition Model for Bispecific Antibodies: Properties, Approximation, and Optimal Dosing Strategy”, CPT Pharmacometrics Syst Pharmacol, 2019.
[9] Jiang, X., et al. "Development of a Minimal Physiologically-Based Pharmacokinetic/Pharmacodynamic Model to Characterize Target Cell Depletion and Cytokine Release for T Cell-Redirecting Bispecific Agents in Humans." Eur J Pharm Sci, 2020.
[10] Panowski, S.H., et al. "Preclinical Efficacy and Safety Comparison of CD3 Bispecific and ADC Modalities Targeting BCMA for the Treatment of Multiple Myeloma." Mol Cancer Ther.

Acknowledgements: All authors are shareholders and/or employees of Novartis.


Figure 1. Reverse translational modeling strategy (A) Schematic of three-step reverse translation framework. Accounting for affinity and expression differences in vitro and ex vivo, teclistamab data are used to calculate EC90 scaling factors for the difference between expected (in vitro) and measured (ex vivo) potency (top row). Potency scaling factors are then applied to in vitro data for elranatamab and WVT078 for early benchmarking of clinically active dose predictions (bottom row). (B) Schematic of minimal physiologically-based pharmacokinetic (mPBPK) model (left) with a target engagement model (right) to simulate plasma and bone marrow PK profiles.


Figure 2. Translating teclistamab in vitro-to-ex vivo EC90 scaling factors for benchmarking to elranatamab RP2D (A) In vitro activated T-cell count and KMS11 tumor cell killing data to estimate the activated T cell density required for 90% cell death (ActiveT90) for teclistamab, elranatamab, and WVT078 (B) Simulated population teclistamab plasma PK for RP2D (black) relative to ex vivo EC90 range (potency values for most and least responsive patient in [2]; yellow) and in vitro EC90 threshold ([2]; red) (C) Simulated population elranatamab plasma PK for RP2D (blue) relative to scaled ex vivo EC90 range (yellow) and in vitro EC90 threshold ([10]; red)