A Quantitative Systems Pharmacology Modeling Framework for Evaluation of Cytokine Release Mediated By Intravenous Odronextamab Monotherapy in Patients with B-Cell Non-Hodgkin Lymphoma

Background Odronextamab is a hinge-stabilized, human CD20×CD3 IgG4-based bispecific antibody that binds CD20-expressing cells and CD3 on T cells, targeting CD20+ cells via T-cell-mediated cytotoxicity, independent of T-cell receptor-mediated recognition. T-cell activation following odronextamab administration can cause a temporary increase in the levels of circulating cytokines, and this has been associated with a risk of cytokine release syndrome (CRS). CRS is a recognized potential safety concern with all T-cell engaging bispecific antibodies, which, when observed, mostly occurs during the first cycle of treatment. This study was performed to evaluate cytokine profiles, especially interleukin (IL)-6 (a surrogate cytokine associated with CRS) in patients with B-cell non-Hodgkin lymphoma (B-NHL) receiving intravenous (IV) odronextamab monotherapy. Quantitative systems pharmacology (QSP) modeling is an emerging mathematical modeling approach that is based on and integrates our understanding of disease biology and the mechanism of action of the drug of interest. A range of questions arising from drug research and clinical development may be evaluated and addressed using the QSP approach. The objective of this work was to optimize the step-up dosing regimen for achieving improved safety profiles using a QSP model. Methods Odronextamab pharmacokinetics (PK) data, cellular dynamics of T cells and B cells, and disease characteristics of patients with B-NHL were used to develop the QSP model. In order to describe the IL-6 profiles, a published semi-mechanistic model was adapted and modified (Toroghi MK. ACoP11 2020. Poster TUE-054). A time-variant negative feedback loop was incorporated to account for the observed attenuation of IL-6 release following repeated doses. The model parameters were created with in vivo and in vitro preclinical data and relevant literature information and calibrated with clinically observed IL-6 data from patients treated with odronextamab monotherapy, observed odronextamab concentrations in serum, B-cell counts, CD8+ T-cell counts, and tumor size data obtained from the ELM-1 study (NCT02290951). The calibrated model was used to assess time profiles of IL-6 release associated with different odronextamab dosing regimens. These assessments were performed using a virtual population of patients with B-NHL (n=300 per dosing regimen). Results The calibrated QSP model was used to simulate IL-6 profiles with different proposed step-up dosing scenarios, including testing different dose levels in Weeks 1-4, with or without splitting the dose in each week, the ratio of splitting the dose over 2 days in each week, and the step-up dosing period from 2-4 weeks prior to giving a full dose of the treatment. Simulations suggested that splitting the dose at Week 1, Week 2, and Week 3 reduces the likelihood of IL-6 release in the early weeks of treatment. The highest IL-6 peak is expected to occur in Week 1 and an uneven split ratio (0.2/0.5 mg) of a 0.7 mg dose over 2 days of Week 1 would mediate the highest IL-6 release on Day 1 while maintaining a similar or lower IL-6 release on Day 2. As IL-6 release in Week 2 could be high, a low dose of 4 (2/2 split) mg is predicted to release less IL-6 than doses >4 mg. After comparison of simulation results for various dosing scenarios, an optimized step-up dosing regimen with 0.7 (0.2/0.5) mg at Week 1, 4 (2/2) mg at Week 2, and 20 (10/10) mg at Week 3 was identified for testing in clinical trials. In addition, the QSP model included variability in PK parameters of odronextamab and could predict PK profiles of odronextamab in the virtual B-NHL population, which showed comparable exposures with observed odronextamab concentrations following the previously tested step-up dosing regimen of 1 mg (Week 1), 20 mg (Week 2), and 160 mg (Week 3). The model was also able to predict the CD8+ T-cell and B-cell profiles over time following IV odronextamab split-dosing regimen. Conclusion This QSP model was developed to address the safety concern related to CRS following treatment with odronextamab. The work demonstrated that the QSP modeling is a powerful tool that enabled optimization of odronextamab step-up dosing to minimize the risk of higher grade (i.e. Grade 2 and 3) CRS in lymphoma patients treated with odronextamab.