Principal Scientist Genentech, Inc., United States
Objectives: T cell-dependent bispecific antibodies (TCBs) represent a promising modality for treating various cancers by facilitating an immune synapse between cancer cells and T cells to induce cytotoxicity. The objective of this work was to develop a minimal physiologically-based pharmacokinetic (mPBPK) model that incorporates both T cells (including trafficking between tissues) and target cells to accurately describe the pharmacokinetics (PK) and target engagement of TCBs using publicly available preclinical PK datasets. Because the model is generalizable for new targets, the model can be further developed and validated into a platform to capture PK of TCBs across heme and solid tumor indications using a consistent set of parameter values.
Methods: A mPBPK model1 was constructed to simulate the dynamics of four antibodies: anti-gD/CD3H (high affinity), anti-gD/CD3L (low affinity),2 Obinutuzumab3 (anti-CD20), and the anti-CD20/CD3, Mosunetuzumab.4 The anti-gD/CD3 molecules and Obinutuzumab were used to calibrate the CD3 and CD20 expression in the model, respectively. The anti-CD20/CD3 molecule was then simulated based on the predicted CD3 and CD20 capacity. Briefly, the model includes nonspecific clearance from the central compartment, as well as target-mediated drug disposition from relevant compartments where CD3 and/or target are expressed. The model also includes T cell trafficking through tissue to understand how this mechanism may impact PK. Sensitivity analyses were performed to assess the impact of varying CD3 and CD20 expression levels as well as the antibody binding affinities on the PK and target engagement.
Results: The mPBPK model was calibrated to 1 mg/kg single dose PK data in cynomolgus monkeys for the anti-gD/CD3H and anti-gD/CD3L molecule. The model captured the difference in exposure of these two molecules due only to the difference in CD3 binding affinity. After developing confidence in CD3 target expression levels that explain the anti-gD/CD3 data, we used a similar approach to capture the CD20 capacity using the anti-CD20 Obinutuzumab cynomolgus monkey PK data at dose levels 1 and 10 mg/kg. Using these predicted CD3 and CD20 capacities, we captured the PK of the anti-CD20/CD3 molecule, Mosunetuzumab, in cynomolgus monkeys across dose levels 0.01 - 1 mg/kg Q1W. The model predicts target engagement in both the serum and healthy tissue. Sensitivity analyses of the model demonstrate how variability in PK profiles can be captured by including reasonable variability in CD3 expression levels. The model also predicts the trimeric (drug-CD20-CD3) complex amount formed in circulation.
Conclusions: The mPBPK model developed in this study, initially calibrated using anti-gD/CD3 and anti-CD20 data, provides a robust platform for understanding the PK behavior of TCBs like Mosunetuzumab. By incorporating T cell trafficking and drug binding dynamics, the model offers valuable insights into trimer formation in various tissues, enhancing its applicability for translation, and thus for predicting both first-in-human dose levels associated with cytokine release syndrome as well as efficacious dose levels, paving the way for more informed and effective strategies for TCB development.
Citations: [1] Cao Y et al. JPKPD, 2013. [2] Yadav et al. Pharmaceutics, 2022. [3] Grimm et al. Journal of Pharmaceutical Sciences, 2019. [4] Ferl G et al, CTS, 2018.
Keywords: T cell-dependent bispecific antibodies, target-mediated drug disposition
