Collaboration with Pfizer - Presented at AAPS 2019 PharmSci 360
Abstract
Purpose: T cell engaging bispecific molecules (TCEs) are protein therapeutics that can bind to both T cells, via CD3 interactions, and target cells through target specific/overexpressed molecules. The TCEs bring cytolytic T cells to kill tumor cells through the formation of the trimolecular complex (CD3:TCE:Target). While it offers a potentially very potent approach to treat cancer, it also has a complex relationship between binding potency, exposure and efficacy. These complex relationships bring challenges to drug discovery and development such as identifying optimal Kds for the drug candidate, predicting efficacious dose and selecting dosing regimen in clinical study.
Methods: A quantitative systems pharmacology model was developed for solitomab, a bispecific T cell engager binding to CD3 and EpCAM. The model includes solitomab binding to both CD3 and EpCAM in both tumor and normal tissue, different E:T ratios for tumor and normal tissue, turnover of CD3 and EpCAM proteins as well as pharmacokinetics (PK) and tissue distribution of the drug. Literature data such as binding Kds, in vitro cell killing and PK were used to inform model development.
Results: Using the calibrated QSP model, we demonstrated that the trimolecular complex per T cells required for in vitro T cell killing are ~200 molecules, which is similar to the number predicted at the maximal tolerated dose observed in the clinical study despite many differences between in vivo and in vitro conditions such as E:T ratios. The therapeutic index (TI) for solitomab was predicted to be close to 1 based on the trimolecular complex per T cells in tumor and in normal tissue. This results is consistent with clinical observation. Using the developed model, we explored the effects of varying binding Kds to EpCAM and CD3, varying half-life of the drug and varying target expression levels on TI.
Conclusion: Current work demonstrated the possibility of using a QSP model integrating in vitro data, human PK and target biology to predict efficacious dose and TI for treating cancer. The model also demonstrated the complex relationship between Kds, target expression level, drug half-life and TI for TCEs. This model can be used to guide future drug candidate selection and clinical study design.