Validation of a clinically relevant humanized mouse model for the safety assessment of 4-1BB agonists utomilumab and urelumab.

Abstract

e14602 Background: Immunotherapy is proven to be powerful and effective in treating cancer. One of the strategies for the therapy is to boost a patient’s immune system by stimulating tumor necrosis factor receptors via the activation of 4-1BB (CD137) costimulatory molecule. Urelumab was the first 4-1BB agonist investigated in clinical trials and showed clinical promise but doses were limited by severe liver toxicity. Preclinical assessment of toxicity failed to predict the clinical safety outcome. Methods: To address the critical needs of in vivo evaluation of toxicity in a preclinical setting, we have developed an animal model using PBMC humanized mice to assess the safety of immunotherapy. In this model, agents targeting human immune system including monoclonal and bispecific antibodies can induce an acute response of cytokine release within hours of treatment and with some therapies a systemic response can manifest in tissue damage and lethality of mice days later. Results: The potential clinical relevance of the PBMC humanized mouse model was demonstrated using 4-1BB agonists urelumab and utomilumab. The antibody-treated PBMC humanized mice were evaluated daily for clinical score and assessed for clinical chemistry and liver histopathology at study terminus. Animals dosed with urelumab at 10 mg/kg exhibited body weight loss and underwent early euthanasia, while animals dosed with utomilumab survived to scheduled euthanasia with minimum body weight loss. Urelumab showed marked liver toxicity relative to utomilumab and controls with more necrosis in the tissue and higher mean ALT, AST and GLDH activities, and the tolerability of urelumab was increased by lowering the dose to 1 mg/kg. The differences captured by the humanized mouse model presented a safety profile similar to the findings about urelumab and utomilumab from the clinical trials. Utomilumab was well tolerated at the dose that caused severe toxicity for urelumab in patients, and the toxicity of urelumab could be reduced by lowering the dose. Conclusions: Our data suggest that the PBMC humanized mouse model could identify human 4-1BB agonists that caused potentially severe liver toxicity. Preclinical safety assessment using humanized mouse models as used for these studies could be an important step in the course of the development of novel immunotherapy for the safety of patients as well as mitigating drug development cost.