Scientists have developed an artificial intelligence (AI) system that analyzes complex gene expression signatures to predict the likelihood that a cancer tumor will spread, SciTechDaily reports.
Why do some tumors spread throughout the body, while others remain “confined” to their place of origin? Scientists do not yet fully understand the processes that cause cancer cells to metastasize. However, finding the answer to this question is essential to improving the way patients are treated.
Researchers from the University of Geneva (UNIGE) investigated this issue using cells taken from colon cancers and identified specific factors that influence the likelihood that a tumor will spread.
The team also discovered genetic signatures that help predict the risk of metastasis. Based on these findings, the researchers developed an artificial intelligence tool called MangroveGS, which turns this biological information into predictions for numerous types of cancer with remarkable reliability. The study, published in the journal Cell Reports, could lead to more personalized medical care and help scientists discover new therapeutic targets.
“The origin of cancer is often attributed to 'rebel cells'”, says Ariel Ruiz i Altaba, professor in the Department of Genetic Medicine of the Faculty of Medicine of UNIGE, who led the study. “However, cancer should be understood rather as a distorted form of development,” he argues.
In this sense, cancer does not occur randomly, but follows an organized biological process. “The challenge is, therefore, to find the keys to understand the logic and form of this process. And, in the case of metastases, to identify the characteristics of the cells that will break away from the tumor to create another one in another part of the body”, says Altaba.
Identification of metastatic cancer cells
Metastases are responsible for the majority of cancer deaths, particularly in colon, breast, and lung cancers. Currently, the earliest detectable sign of metastasis is the presence of circulating tumor cells in the blood or lymphatic system. However, by the time these cells can be detected, they have likely started to spread throughout the body.
Scientists have learned a lot about the genetic mutations that lead to the formation of primary tumors. However, researchers have not identified a single genetic change that explains why some cancer cells leave the original tumor while others stay put.
“The difficulty lies in the fact that we have to determine the complete molecular identity of a cell – an analysis that destroys it – while observing its function requires it to remain alive,” explains Professor Ruiz i Altaba.
The research team led by him measured the activity of several hundred genes in approximately thirty cloned cells derived from two primary colon tumors. Their analysis revealed clear genetic gradients that showed a strong correlation with how easily the cells were able to migrate.
The results also suggest that metastatic risk cannot be determined by studying a single cell in isolation. Instead, it depends on the collective interactions between groups of related cancer cells within a tumor.
Researchers have created a highly reliable prediction algorithm
The research team integrated these genetic signatures into an artificial intelligence model developed in Geneva.
“The great novelty of our tool, called Mangrove Gene Signatures (MangroveGS), is that it exploits dozens, even hundreds of genetic signatures. This makes it particularly robust to individual variation,” explains Aravind Srinivasan, a PhD student at the Department of Genetic Medicine of the Faculty of Medicine of UNIGE and co-lead author of the study.
After training, the system predicted metastasis and recurrence in colon cancer with nearly 80% accuracy, significantly outperforming existing prediction tools.
The scientists also found that genetic signatures identified in colon cancer can help predict metastatic potential in other cancers as well, including gastric, lung and breast cancer.
“This information will prevent the overtreatment of low-risk patients, thus reducing side effects and unnecessary costs, while intensifying the monitoring and treatment of those at high risk,” points out Ariel Ruiz i Altaba.
