Preclinical pharmacology characterization of HX009, a novel PD1 x CD47 Bi-specific antibody

Abstract

Abstract Certain immune-checkpoint inhibitors have a narrow therapeutic window (TW) as cancer therapeutics, and engineered dual-/multi-targeting agents could potentially widen the TW to bring true clinical benefits. We report a new rationally-designed bispecific-antibody (BsAb), HX009, simultaneously targeting PD1 and CD47 to improve both the efficacy and safety over the respective single-targeting agents by grafting the extracellular domain of SIRPα onto the parental anti-PD1-monoclonal antibody, HX008. This resulted in an IgG4-based “2 × 2” symmetric structure but with an intentionally-reduced CD47-binding affinity, suggesting a novel candidate cancer immunotherapy. Specifically, HX009 has binding affinity constant of 8.951 × 10–9 M for human PD1 and 2.557 × 10–8 M for human CD47, respectively, where the CD47 binding is significantly weaker as compared to the binding affinity of HX008 to PD1 as well as the binding affinity of SIRPα-Fc to CD47, leading to little binding to RBCs and platelets and is contrasting to many CD47-agents in development. However, HX009 effectively and simultaneously binds to the PD1 and CD47 on PD1+CD47+ T-cells via cis-binding and elicits enhanced T cell activation compared to the parental HX008. HX009 caused little cytokine-release in human peripheral blood mononuclear cells. HX009 cross-species binds to cynomolgus monkey PD1/CD47 but not to rodents, making cynomolgus monkeys the choice of species to investigate the pharmacokinetics (PK) and toxicology of HX009. HX009’s anti-tumor activities were confirmed in several humanized preclinical mouse models by determining either its anti-PD1 (humanized hu-CD47-MC38 models) or anti-CD47 (HuT-102 lymphoma CDX and three PDX-AML models) functions, although limited available humanized models have hindered broadly demonstration of enhanced anti-tumor activities contributed from the dual targeting of the BsAb. The expanded DLBCL-PDX trial data suggested that both EBV-status and OX40 expression could potentially be two positive predictors for response to HX009. An intravenous (IV) infusion PK study in cynomolgus monkey revealed its largely vasculature distribution, terminal half-life (T1/2) of ~ 50 h, and dose-proportional exposure without accumulation. The anti-drug antibody (ADA) was observed in all monkeys as expected, affecting the PK parameters of repeated administration. The IV single-dose toxicology study with a 14-day observation revealed a maximum tolerated dose of 150 mg/kg, while the repeated-dose (once weekly for 4 weeks, 5 doses in total) study showed a highest non-severely toxic dose (HNSTD) of 15 mg/kg. The desired preclinical PK and safety profiles, along with its antitumor activity, support HX009’s candidacy for its clinical development.

Keywords

• Innate/adaptive ICIs
• BsAb
• Xenograft and humanized syngeneic tumor models
• Immunotherapy
• Toxicology
• Pharmacokinetics