Researchers use AI to identify cancer drug target with lower toxicity risk
Scientists at St. Jude Children's Research Hospital identified IRS4 as a promising drug target across multiple solid tumors by combining genetic data with artificial intelligence to predict which therapies would cause fewer side effects.
The work, published today in Science Advances, addresses a persistent problem in cancer drug development: an estimated 85%-97% of potential cancer therapeutics that enter phase 1 clinical trials fail, partly because toxicity emerges late in the development process.
For pediatric cancer patients, this matters more than for adults. Therapy-related side effects can cause chronic health problems decades after treatment ends.
A three-step screening approach
The research team developed a method to evaluate toxicity risk early. They started with the Dependency Map portal, a database of genes cancer cells need to survive, then filtered for targets similar to FDA-approved cancer therapies. This narrowed the initial pool to 346 candidates.
Next, they used AI to search scientific literature for people whose natural loss of these genes caused either no problems or only manageable disruptions to normal life. This step reduced the list to 25 candidates.
IRS4 stood out because it had a potential binding pocket for a drug to target-but that pocket wasn't necessary for the protein's cancer-promoting effects. The protein is rarely expressed in normal adult tissues, and people lacking the gene are generally healthy, with only manageable thyroid-related effects.
How IRS4 functions as a therapeutic target
When researchers removed IRS4 or degraded the protein, cancer cells that expressed it could not grow. The protein acts as an on-off switch: cancer cells need it to survive, but normal cells can function without it.
Testing showed IRS4 could serve as a therapeutic target in pediatric malignant rhabdoid tumors, osteosarcomas, and some brain tumors. It also appeared relevant in adult cancers including breast, lung, uterine, and stomach cancers.
The approach provides a proof of principle for evaluating toxicity early in drug discovery rather than waiting until late-stage trials.
The study was supported by grants from the National Health and Medical Research Council of Australia, the National Cancer Institute, and the American Lebanese Syrian Associated Charities.
Your membership also unlocks:
