AI Drug Discovery Accelerates with Promising ISM3830 Inhibitor as FDA Agentic AI Streamlines Regulation
- 5 days ago
- 3 min read
The pharmaceutical landscape is experiencing a profound transformation, driven by artificial intelligence (AI) both accelerating the pace of developing novel therapies and modernizing the oversight of regulated products. News this week highlighted a significant preclinical triumph facilitated by generative AI, coupled with a major regulatory modernization effort by the US Food and Drug Administration (FDA).
Insilico Medicine, a clinical-stage company specializing in generative AI Drug Discovery, announced promising preclinical results for its new AI-designed cancer immunotherapy candidate, ISM3830. This compound is designed as a potential best-in-class, orally available, and highly selective CBLB Inhibitor, built on a novel scaffold identified using Insilico’s proprietary AI platform. The target, CBLB (Casitas B-lineage lymphoma-b), functions as an intracellular checkpoint and a key negative regulator of T-cell and natural killer (NK) cell activation.
CBLB is highly expressed in multiple immune cell subsets and various cancers, making it a particularly promising immunotherapy target. By inhibiting CBLB, ISM3830 has demonstrated a strong potential to modulate immune tolerance, enhance T-cell and NK-cell activity, and even restore function in exhausted T cells. Dr. Feng Ren, co-CEO and CSO of Insilico Medicine, emphasized that the fundamental mechanism of CBLB inhibition supports indications with low response or resistance to current immune checkpoint inhibitors—an area of significant unmet need.
The company’s generative AI Drug Discovery platform was crucial to this success, overcoming traditional bottlenecks related to metabolism and absorption that have previously hindered CBLB inhibition therapies. The platform deployed over 40 generative AI models, utilizing Chemistry42 and its ADMET predictor module to design and optimize the candidate compound. Preclinical studies in mouse models revealed robust anti-tumour efficacy and evidence of long-term tumor immunity in CT26 rechallenge experiments. Furthermore, the compound demonstrated favorable druggability and ADME/PK characteristics in vitro and in vivo, alongside low risks of hypotension, gastrointestinal toxicity, and off-target toxicity. This milestone contributes to Insilico's growing pipeline, which has nominated 23 preclinical candidates since 2021, following the June report of the first clinical proof-of-concept for an AI-discovered drug, Rentosertib. This advancement is part of a broader trend, with a drug explicitly "engineered with AI" recently reaching Phase 3 testing, marking a key innovation milestone.
Meanwhile, the US regulator is dramatically expanding its internal use of technology. The FDA announced on December 1, 2025, the deployment of Agentic AI capabilities for agency staff. Agentic AI systems are advanced AI models designed for planning, reasoning, and executing multi-step actions autonomously, following built-in guidelines. This new deployment is intended to help staff create more complex AI workflows and harness AI models to achieve operational efficiency.
While voluntary for staff, the Agentic AI systems will assist with complex functions such as pre-market reviews, post-market surveillance, inspections, compliance, and administrative tasks. Jeremy Walsh, the FDA’s Chief AI Officer, stated that Agentic AI will provide a powerful tool to streamline work and help ensure the safety and efficacy of regulated products. This follows the successful June 2025 launch of Elsa, a generative AI tool (LLM) already used by staff to accelerate clinical protocol reviews and scientific evaluations. Crucially, both the Elsa tool and the new Agentic AI models are securely built within high-security GovCloud environments and are designed not to train on input data or submitted regulatory industry data, thus safeguarding sensitive research. Commissioner Marty Makary, MD, MPH, stressed that this modernization helps the agency "radically improve our ability to accelerate more cures and meaningful treatments". This dual evolution—accelerated AI Drug Discovery and streamlined regulation through Agentic AI—signals a mature integration of technology across the entire healthcare ecosystem.powerful tool to streamline work and help ensure the safety and efficacy of regulated products. This modernization drive is aimed at accelerating "more cures and meaningful treatments".
🔖 Sources
Keywords: AI Drug Discovery
AI Drug Discovery
Comments