Model-informed drug development for immuno-oncology agonistic anti-GITR antibody GWN323: Dose selection based on MABEL and biologically active dose.

GWN323, an agonistic human anti-GITR (glucocorticoid-induced TNFR-related protein) IgG1 antibody, was studied clinically as an immuno-oncology therapeutic agent. A model-based minimum anticipated biological effect level (MABEL) approach integrating in vitro and in vivo data informed dose selection for the first-in-human (FIH) study. Data evaluated included pharmacokinetics (PK) of DTA-1.mIgG2a (mouse surrogate GITR antibody for GWN323), target-engagement pharmacodynamic (PD) marker soluble GITR (sGITR), tumor shrinkage in Colon26 syngeneic mice administered with DTA-1.mIgG2a, cytokine release of GWN323 in human peripheral blood mononuclear cells, and GITR binding affinity. A PK model was developed to describe DTA-1.mIgG2a PK, and its relationship with sGITR was also modeled. Human GWN323 PK was predicted by allometric scaling of mouse PK. Based on the totality of PK/PD modeling and in vitro and in vivo pharmacology and toxicology data, MABEL was estimated to be 3-10 mg once every 3 weeks (Q3W), which informed the starting dose selection of the FIH study. Based on tumor kinetic PK/PD modeling of tumor inhibition by DTA-1.mIgG2a in Colon26 mice and the predicted human PK of GWN323, the biologically active dose of GWN323 was predicted to be 350 mg Q3W, which informed the dose escalation of the FIH study. GWN323 PK from the FIH study was described by a population PK model; the relationship with ex vivo interleukin-2 release, a target-engagement marker, was also modeled. The clinical PK/PD modeling data supported the biological active dose projected from the translational PK/PD modeling in a "learn and confirm" paradigm of model-informed drug development of GWN323.