A domestic research team presented a new microbiome-based diagnostic technology that can distinguish stomach and colorectal cancers by analyzing how much oral microorganisms move into the gut. Because cancer-related signals can be identified with a simple oral specimen, it is expected to complement existing testing methods.
The National Research Foundation of Korea (NRF) said on the 21st that a research team led by Kim Ji-hyeon, a professor in the Department of Systems Biology at Yonsei University, developed an "oral–credit entry microbial transfer index" by comparing microbial information in the mouth and credit entry, and built a stomach and colorectal cancer classification model using it. The findings were published online the same day in the international journal Cell Host & Microbe.
In healthy people, the gastrointestinal tract prevents oral microbes from settling in the gut through gastric acid and the immune system, keeping the microbial ecosystems of the mouth and the gut distinct. The researchers noted that when diseases such as cancer occur, these defense systems can weaken, increasing the influx of oral microbes into the gut.
The researchers analyzed 1,010 oral and credit entry samples from 507 individuals, including healthy people, patients with metabolic diseases, and patients with stomach and colorectal cancers, who visited Severance Hospital and Severance Health Checkup. As a result, the "oral–credit entry microbial transfer index," which represents the proportion of oral-derived microbes found in credit entry, was significantly higher in patients with stomach and colorectal cancers than in healthy people.
They also confirmed that microbial migration patterns vary by cancer type and stage, suggesting the index could be used as an indicator reflecting cancer occurrence and progression.
Using the random forest artificial intelligence (AI) algorithm, the researchers developed a model to distinguish stomach and colorectal cancers and verified the model's reproducibility and applicability with 2,031 samples from seven overseas clinical cohorts. In particular, because cancer-related signals can be identified with oral samples alone, it could be used in the future as a simple oral collection–based cancer screening test.
Professor Kim Ji-hyeon said, "The fact that cancer-related signals can be identified with only a mouth-rinse specimen is meaningful for improving access to screening," and added, "We plan to develop it into a model that predicts individual disease risk by integrating genetic and clinical information."
References
Cell Host & Microbe (2026), DOI: https://doi.org/10.1016/j.chom.2026.07.007