Harnessing AI for the Advancement of Immunotherapeutic Agents

Recent findings have highlighted the role of kynurenine pathway enzymes in cancer immunity, prompting the development of inhibitors for IDO and TDO. However, clinical outcomes for these inhibitors have been underwhelming due to pathway redundancy and suboptimal pharmacological properties. To address these issues, an artificial intelligence (AI) platform utilizing deep learning was employed to design a novel kynurenine pathway regulator with significant immunotherapeutic potential.

The methodology involved the creation of deep learning models specific to IDO and TDO to predict compound interactions. These models were used to screen a vast chemical library, identifying top candidates that were then experimentally tested for inhibitory activity. A lead compound, STB001, was developed from a core structure that showed activity against both IDO and TDO. STB001 was tested in vivo for its ability to suppress plasma kynurenine levels and was administered orally to mice with CT26 colon cancer. The compound demonstrated a dose-dependent delay in tumor growth and a marked enhancement of adaptive anti-cancer immunity, including increased CD8+ cytotoxic T cells and decreased regulatory T cells.

The optimal dosing regimen for STB001 was established based on its therapeutic efficacy and safety profile. Furthermore, the combination of STB001 with immune checkpoint inhibitors (αPD-1 and/or αCTLA-4) resulted in the complete regression of tumors in the study.

In conclusion, the use of AI and deep learning for drug development has proven to be a viable and effective strategy, as evidenced by the successful identification and development of STB001. This AI-based approach could be extended to other molecular targets to accelerate the development of immuno-oncology drugs.