Biotechnology and Pharmaceuticals · global
AI Finds a New Target—and a Pain Drug Candidate: ISM9528 Aims for Non-Opioid Treatment
Insilico used multi-omics data to screen targets, then applied generative chemistry to design a small molecule capable of entering the brain. Animal studies have shown preliminary analgesic activity, but the target has not yet been disclosed, and evidence of efficacy and safety remains limited to the company’s announced preclinical findings.
Pain treatment has long faced a dilemma: potent opioids may carry risks of dependence, respiratory depression, and misuse, while some non-opioid options may not adequately control neuropathic pain. Insilico Medicine has now selected the oral drug candidate ISM9528, seeking to intervene in pain signaling through a novel mechanism that has not yet been disclosed, with plans to advance it into clinical development.
The program began with target discovery. According to the company, the research team used PandaOmics to analyze multi-omics data, literature, and data from human and animal models, selecting a target provisionally called “Target Z.” It then used the generative chemistry platform Chemistry42 to design and screen molecules, progressively optimizing selectivity, brain penetration, and drug properties including absorption, distribution, metabolism, excretion, and toxicity, ultimately selecting ISM9528 as the preclinical candidate.
ISM9528 is described as an orally available small-molecule inhibitor capable of crossing the blood–brain barrier. In a rat spinal nerve ligation model, the company said it improved mechanical allodynia in a dose-dependent manner; the effect at a medium dose was comparable to pregabalin within six hours after administration. Another acute postoperative pain model showed that the drug began to take effect in about 30 minutes and outperformed the same dose of pregabalin.
These results provide early evidence of efficacy for the candidate, but they cannot yet establish whether it can become a clinical pain medication. Animal models can simulate only part of human pain, while pain assessment itself is also readily influenced by factors such as disease etiology, sensation, and emotion. More importantly, the current data were disclosed by the company, and full study designs, statistical results, or peer-reviewed papers have not yet been made available, making it difficult for outside observers to independently assess the magnitude and reproducibility of the effects.
Keeping the target confidential also leaves a crucial gap. Without knowing the physiological functions of Target Z in the nervous system and other organs, it is impossible to fully assess the adverse effects that long-term inhibition might cause or to compare how ISM9528 differs from other non-opioid analgesic pathways. For a drug capable of entering the brain, central nervous system safety, sedation, cognitive effects, and potential misuse risk will all be important areas of evaluation before formal human trials.
Insilico said it will accelerate completion of the studies needed to support clinical trials, with the goal of advancing to human trials in 2027. If subsequent toxicology and pharmacokinetic data support an application, the first Phase 1 trial will primarily need to demonstrate safety and human exposure, rather than analgesic efficacy. ISM9528 therefore both illustrates how an AI platform can connect target selection with molecular design and serves as a reminder: algorithms can shorten the path to finding a drug candidate, but they cannot replace comprehensive clinical validation.