Abstract 1536: Machine learning guided engineering of nivolumab for superior physicochemical properties for subcutaneous injection

Abstract Nivolumab is a programmed death receptor-1 (PD-1) blocking antibody that is used as a type of cancer immunotherapeutic agent, called a checkpoint inhibitor. Nivolumab is administered by intravenous infusion. Given the recent studies and potential demand for maintenance immunotherapy, the next generation of Nivolumab for subcutaneous injection would be beneficial for patients’ quality of life. We studied whether we could create a Nivolumab variant that has superior physicochemical properties required for a subcutaneous formulation. A machine-learning guided protein engineering technique was used to improve structural stability at little cost of affinity. As a result, a variant showed an increased melting temperature by 10 ºC with a comparable affinity to Nivolumab. Consistent with the result, the variant is resistant to aggregation under heat stress. Nivolumab is seriously aggregated after treatment at 60 ºC, while the variant does not show any aggregation at all. We also present a result of risk assessment of aggregation at high concentrations. Citation Format: Norio Hamamatsu, Naoyuki Kuwabara, Emi Suzuki, Ryo Yamazaki, Misaki Oikawa, Tomonori Tawara, Tomoyuki Ito, Hikaru Nakazawa, Shiro Kataoka, Mitsuo Umetsu. Machine learning guided engineering of nivolumab for superior physicochemical properties for subcutaneous injection [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 1536.