A newly identified and rare genetic variant slows the growth of mutated blood stem cells, researchers report, and it reduces the risk of leukemia. The findings offer insight into why some people are naturally more resistant to clonal expansion and age-related blood cancers despite acquiring risky mutations. As tissues age, they quietly accumulate many mutations that can drive cancer. In the blood-forming, or hematopoietic, system, such mutations often appear in otherwise healthy individuals as clonal hematopoiesis (CH), a process in which certain blood stem cell (HSC) clones gain a growth advantage over non-mutated clones, allowing them to expand steadily over time. This condition, also known as CHIP (clonal hematopoiesis of indeterminate potential), is associated with an increased risk of blood cancers as well as other chronic illnesses, including heart disease. However, not all those with CHIP develop disease, and some mutant stem cell clones remain stable or even diminish over time, suggesting that inherited and/or environmental factors can restrain or slow CH.

To investigate this, Gaurav Agarwal and colleagues performed a GWAS meta-analysis on data from more than 640,000 individuals to search for inherited DNA variants that protect against CH. Agarwal et al. identified a noncoding regulatory variant, rs17834140-T, that substantially lowers the risk of CHIP and reduces the likelihood of developing blood cancers. According to the findings, this protective effect traces to a single DNA change that weakens the activity of the musashi RNA binding protein 2 (MSI2) gene – a key factor in stem cell maintenance. Using gene-edited human HSCs, Agarwal et al. discovered that rs17834140-T disrupts a binding site for the endothelial transcription factor GATA-2. This interference reduces MSI2 expression in HSCs, which further suppresses an entire network of genes that mutant stem cells rely on for competitive growth. Notably, the authors also found that this same gene network is unusually active in HSCs carrying high-risk cancer mutations and in children with acute myeloid leukemia, where it was associated with reduced survival. “The ability to predict disease risk at the individual level is a long-standing goal of modern medicine,” write Francisco Caiado and Markus Manz in a related Perspective. “The study of Agarwal et al. supports MSI2 targeting as a potential pan-cancer therapeutic approach, and small-molecule approaches are in preclinical development.”

For reporters interested in research integrity-related themes, author Vijay G. Sankaran told SciPak: “In human genetics and genomics, science integrity has improved substantially through routine data sharing and deposition, especially public release of sequencing datasets and GWAS summary statistics. Our study directly benefited from this ecosystem by leveraging prior GWAS summary statistics to make new discoveries and prioritize targets for functional follow-up. Going forward, I would like to see broader requirements for timely deposition of summary statistics and other data to broaden equitable reuse.”

This scoping review identifies the first wave of randomized controlled trials testing large language models in digestive diseases. Four published and ten ongoing trials suggest that AI tools such as ChatGPT are increasingly used for clinical decision-making and patient education, but robust, international evidence based on real patient outcomes is still urgently needed.