Document resource
Background MDX2004 is a trifunctional antibody-fusion protein that binds and activates T cells through their three main signaling pathways involving CD3, CD28 and 4-1BB. This immunostimulatory molecule was designed to optimize anti-tumor immunity. Its proposed mechanism of action is based upon stimulation through the CD3 arm of the T cell receptor (signal 1), CD28 enhancement of T cell proliferation through its effects on survival, and proliferation (signal 2), and 4-1BBL expansion of stem and memory T cells (signal 3). The relationship between drug exposure and the rate of MDX2004 complexing to T cells is complicated by its dependence on several variables, including local drug concentrations, target binding affinities, receptor densities, and concentrations of T cells. Quantitative systems pharmacology (QSP) modeling is well-suited for integrating MDX2004 data across multiple preclinical study systems to predict pharmacologic activity given its dependency upon multiple interdependent variables. The objective of this study was to inform dose selection for the first-in-human (FIH) study for MDX2004 using a QSP model that captured the major T cells binding events driving the molecule’s core multispecific pharmacology.Methods The QSP model was designed and calibrated using inputs from in vitro binding, T cell activation and cytokine release assays, and data generated in mouse and nonhuman primate (NHP) pharmacokinetic and pharmacodynamic studies. The analysis also included pharmacodynamic data using a surrogate trifunctional molecule that cross-reacts with NHP CD3, CD28 and 4-1BB, MX485. Binding metrics between MDX2004 and receptors on T cells were generated by using the physiological target parameters in humans and preclinical species representing target and same-cell binding attributes. The resulting model generated estimated starting doses based on the minimal anticipated biological effect level (MABEL) as well as pharmacologically active doses. Metrics for proposed doses were based on model-predicted estimates for T cell activation, cytokine release and tumor growth inhibition.Results Based upon effective concentrations for T cell activation, the model predicted a MABEL of 18.5 mg/kg. The model predicted an efficacious dose in the high mg/kg range based upon effective concentrations for both T cell activation and tumor growth inhibition.Conclusions The final QSP model was successful in informing dose selections for the FIH study with MDX2004 in participants with advanced malignancies.