A research team from the University of California, San Diego has developed a new method for destroying cancer stem cells, which has made a breakthrough in the study of colon cancer. They use artificial intelligence to accurately identify therapeutic targets and "reprogram" cancer stem cells to promote their self destruction. This method only targets cancer cells and does not affect surrounding tissues, making it a safer and more precise alternative to existing treatment methods. The relevant results were published on the 20th in the journal Cell Reporter Medicine. Cancer stem cells, these "chameleon" cells hidden in tumors, have always been a challenge in cancer treatment. They can not only evade existing treatment methods, but also help cancer relapse, spread, and even have strong resistance to the most advanced therapies. To tackle these stubborn cancer cells, the research team created a machine learning tool called Cancer Associated Node Differentiation Targeting (CANDiT). This tool can identify new therapeutic targets based on the unique genome of specific tumors. The working principle of CANDiT is to start with a key gene that is crucial for the normal growth of healthy cells, but is usually missing in invasive cancer cells. Next, CANDiT will conduct in-depth analysis of the gene network associated with this gene and propose potential therapeutic targets that can restore cellular health by regulating these networks. The team chose CDX2, a key gene in colon cancer, as the starting point for their research. Using the CANDiT tool, they conducted genomic scans on over 4600 different tumor samples. Through analysis, the team unexpectedly discovered a new therapeutic target - PRKAB1 protein, which promotes cell survival under stress. Subsequently, they activated PRKAB1 using existing drugs, successfully restoring the function of CDX2 gene in colon cancer stem cells. However, these cancer stem cells seem unable to tolerate the loss of their 'cancer identity' and have not stopped at restoring their normal cellular state, but have chosen to self destruct. To validate the clinical potential of this method, the team extended drug testing to patient derived organoid models. They also developed a gene signature, a method that can predict a patient's response to this treatment based on gene activation patterns. Through simulation analysis of over 2100 patients, it was found that after the restoration of CDX2 gene function, the risk of recurrence and death in colon cancer patients can be reduced by 50%. (New Society)