Scientists at St. Jude Children’s Research Hospital have identified IRS4 as a promising drug target in multiple solid tumors using an AI-assisted approach. Their study, published in Science Advances, combines genetic cancer dependency data, AI, and naturally occurring mutations to prioritize targets with high efficacy and low toxicity.
Challenges in cancer drug development
Of the potential cancer therapeutics that enter phase 1 clinical trials, an estimated 85%-97% do not become FDA-approved, partly due to toxicity in normal tissues. This high failure rate is a significant challenge in cancer drug development. Furthermore, the long-term effects of cancer treatments are particularly concerning for pediatric cancers. Therapy-related late effects can cause chronic health problems even decades after treatment, including secondary cancers, cardiovascular disease, and neurocognitive issues. Historically, toxicity has often been evaluated late in the drug development process, typically during phase I or II clinical trials. This delay can lead to valuable time and resources being invested in drugs that ultimately prove too toxic for widespread use.
AI-assisted approach identifies IRS4
To address these challenges, the St. Jude team developed a multi-step screening approach. They began by filtering genes in the Dependency Map portal that cancer cells need for survival. This portal, maintained by the Broad Institute, provides a full resource of genetic vulnerabilities across various cancer types. By analyzing this data, the team identified genes that, when inhibited, selectively suppressed the growth of cancer cells while sparing normal cells.
Next, they employed an AI-driven literature search to find instances where people or animals had naturally lost the function of these genes without significant health consequences. This step leveraged the power of AI to sift through vast amounts of scientific literature, identifying humans or mice that carried loss-of-function mutations in these genes. This process narrowed down the list of candidates to 25, including IRS4, which stood out due to its potential druggable binding pocket and its rarity in normal adult tissues.
Related: New drug wipes out lingering leukemia cells
IRS4 was found to be a potential dependency across multiple tumor types. The study demonstrates that IRS4 could serve as a therapeutic target in multiple cancer types, including pediatric malignant rhabdoid tumors, osteosarcomas and some brain tumors, as well as adult cancers such as breast, lung, uterine and stomach cancers.
The work provides a proof of principle for evaluating potential toxicity early in the search for novel therapeutics. By prioritizing targets with a high therapeutic index – those that are highly effective at low doses and have minimal side effects – researchers can focus only on promising, lower-toxicity targets like IRS4. This approach could significantly improve the success rate of cancer drug development and help reduce the long-term side effects of cancer treatments, particularly in pediatric patients.
“Our work shows that we may be able to predict toxicity at the outset of drug discovery, allowing us to focus only on the promising lower-toxicity targets,” said corresponding author Samuel Brady, PhD, St. Jude Department of Pharmacy & Pharmaceutical Sciences. “Historically, we’ve usually prioritized efficacy when searching for drug targets, while toxicity has often been considered later. By reversing this order, we can potentially develop therapies that are both effective and safer for patients.”
The study, titled “IRS4 is a PI3K-activating cancer dependency up-regulated through DNA rearrangements or epigenetic mechanisms in multiple solid tumors,” was published in the journal Science Advances. It was led by first author Banu K and corresponding author Samuel W. Brady, PhD.
