A ‘seesaw’ pattern in gut bacteria may flag colon cancer risk early

If you've ever wondered whether your gut bacteria could tip you off to colon cancer risk before a colonoscopy does, a new study gets closer to answering that than almost any before it. Researchers sequenced stool from nearly 3,800 people being screened for colorectal cancer and found that certain gut bacteria don't act as isolated culprits, they organize into two opposing teams that rise and fall together like a seesaw. The balance between those teams tracked how far someone had progressed from a healthy colon toward adenomas and cancer, and the pattern held up again when the team checked it against outside data from seven other countries.

Key takeaways:

• Gut bacteria split into two opposing networks that behave like a seesaw: one group of 14 species rises with colorectal cancer risk, a separate group of 37 falls.
• A composite "topology score" built from that balance increased step by step from healthy participants to adenomas to colorectal cancer, and the pattern replicated across cohorts in eight countries.
• Based only on which of those species were present, a model sorted colorectal cancer cases from healthy people with 66% to 87% accuracy (AUC) in outside datasets, better than chance but not diagnostic-grade.
• The study is observational, the authors call the cancer-detection use "exploratory," and the pattern held up less well in people eating a Mediterranean-style diet.

A seesaw pattern links gut bacteria to colorectal cancer risk

The study, published in Advanced Science, analyzed 3,807 stool metagenomes (a metagenome is the full set of genetic material sequenced directly from a stool sample, capturing every microbe at once instead of growing bacteria one at a time in a lab) from participants in TARGET-C, a colorectal cancer screening trial run across five provinces in China. Comparing people with healthy colons to those with precancerous growths called adenomas and to a small group with confirmed colorectal cancer, the researchers found two clusters of bacteria that moved in opposite directions. One cluster, made up of 14 species including Clostridium symbiosum and Hungatella hathewayi, both previously tied to colorectal tumors, became more common as colorectal neoplasia (abnormal tissue growth, covering both adenomas and cancer) advanced. A second cluster of 37 species, including Roseburia and Lachnospira, two genera known for producing short-chain fatty acids that help keep the gut lining healthy, became correspondingly less common. Within each group the species tended to appear together, but across the two groups they excluded one another, a pattern the authors describe as "seesaw-like."

Why this matters to anyone facing a colorectal cancer screening

Colorectal cancer was the third most commonly diagnosed cancer worldwide in 2022, according to global cancer registry data, and catching it or the adenomas that precede it early is one of the few interventions proven to cut deaths from the disease. That matters most for people already inside a screening program, those over 50 or with a family history of colorectal cancer, since a reliable microbial signal could eventually supplement colonoscopy rather than replace the guesswork between scheduled scopes. Specific bacterial species, including certain strains of Fusobacterium nucleatum, have already been linked to colorectal tumors in prior research; what's new here is evidence that the surrounding bacterial community, not just a single suspect species, shifts in a coordinated, measurable way.

Inside the 3,807-person TARGET-C stool sequencing study

The researchers split their main cohort, 2,725 healthy controls, 759 people with non-advanced adenomas, 297 with advanced adenomas, and 26 with colorectal cancer, roughly in half, using one half to identify the seesaw pattern and the other to check it. They then tested whether the same two bacterial networks turned up in eight independent, already-published datasets from China, India, Japan, the United States, Germany, France, Italy, and Austria, plus a separate 460-person colorectal cancer cohort from Guangzhou. Despite differences in sequencing labs and populations, the same two clusters of bacteria reappeared in the outside data, the kind of cross-cohort consistency often missing from single-species microbiome findings.

What a rising "topology score" could mean down the line

To turn the pattern into something measurable, the team built a "topology score" from the proportion of cancer-associated versus protective species present in a person's stool. Each small increase in that score corresponded to higher odds of having adenomas or cancer, and the score rose in a stepwise way, lowest in healthy participants, higher in those with adenomas, and highest in those with colorectal cancer. Using only which of these species were present or absent, without factoring in how abundant each one was, a model sorted colorectal cancer cases from healthy controls with an accuracy (AUC, where 0.50 means no better than a coin flip and 1.0 means perfect separation) ranging from 0.66 to 0.87 across the outside datasets, a range similar to other gut-microbiome screening models built with machine learning. The researchers also found that lower blood levels of a microbial byproduct called indole-3-propionic acid tracked with higher adenoma risk, people with below-average levels had about 45% higher odds of having an adenoma, hinting at a possible biological pathway linking the bacterial shift to metabolites the body absorbs, though the study did not test whether raising that metabolite would lower risk.

The biggest caveat: correlation, not proof of cause

This is an observational study, it shows that the microbial pattern and colorectal neoplasia track together, not that the bacteria cause the cancer or that correcting the balance would prevent it. The authors call the model's use for spotting colorectal cancer itself "exploratory," partly because their own cohort included only 26 cancer cases, too few to draw firm conclusions, and their statistical tests found no single metabolite that explained the link, suggesting the real mechanism, if there is one, runs through a more complex web of microbes and metabolites than this score captures. The score also performed noticeably worse, failing to significantly separate adenoma cases from healthy people, in validation cohorts from Japan, Austria, France, and Italy, with the authors specifically pointing to the Mediterranean-style diet common in the French and Italian cohorts as a likely reason the pattern didn't transfer as cleanly.

What researchers want to study next in larger cohorts

The team says the next step is testing the topology score in larger, cancer-specific cohorts, since their own supply of confirmed cancer cases was thin, and in populations with different diets and genetics to see how much recalibration a test like this would need before clinical use. They also point to strain-level sequencing, distinguishing closely related bacterial strains rather than lumping them at the species level, and direct experiments on the metabolites involved, as ways to move from an interesting pattern toward a validated biological mechanism. No timeline for a clinical or commercial test was given.

Common Questions

Is this something I can get tested for right now?

No. The topology score described here is a research tool built for this study, not a commercially available stool test, and the authors themselves call its use for detecting cancer "exploratory."

Does this prove certain gut bacteria cause colon cancer?

No. The study is observational: it shows the bacterial pattern and disease status move together across a person's screening journey, but it doesn't establish that the bacteria themselves are driving the cancer.

Which bacteria does the study flag as concerning versus protective?

Species like Clostridium symbiosum and Hungatella hathewayi rose alongside precancerous and cancerous tissue, while Roseburia and Lachnospira, both linked to a healthy gut lining, fell.

Why did the pattern work better in some countries than others?

The authors suspect diet is a major factor: the score was weaker in cohorts from France and Italy, where Mediterranean-style eating patterns are common, suggesting the bacterial balance itself may shift with diet in ways this score doesn't yet account for.

References

1. Kai Song, Youwen Qin, Jiahui Luo, et al. (2026). Gut Microbial Topology and Metabolic Signatures Associated With Colorectal Neoplasia. Advanced Science. https://doi.org/10.1002/advs.77991
2. Freddie Bray, Mathieu Laversanne, Hyuna Sung, et al. (2024). Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA: A Cancer Journal for Clinicians. https://doi.org/10.3322/caac.21834
3. A Type of Fusobacterium nucleatum Linked to Colorectal Cancer. cancer.gov. https://www.cancer.gov/news-events/cancer-currents-blog/2024/colorectal-cancer-fna-c2-bacteria
4. Yi-Jian Tsai, Wei-Ni Lyu, Nai-Shun Liao, Pei-Chun Chen, Mong-Hsun Tsai, Eric Y. Chuang (2025). Gut microbiome-based machine learning model for early colorectal cancer and adenoma screening. Gut Pathogens. https://doi.org/10.1186/s13099-025-00750-z