A Gut Bacterial Toxin Signature Found in 45% of Japanese Colon Cancers

If you've been told a colon cancer diagnosis "runs in the family" or "just happens," a new genomic study complicates that story. Researchers scanning the tumors and gut bacteria of 600 people in Japan found that a toxin made by certain gut bacteria leaves a distinctive mutational fingerprint in nearly half of colorectal cancers, and that fingerprint shows up far more often in people born after 1965 than in older generations. It's the kind of finding that reframes colon cancer less as bad luck and more as a story about decades of bacterial exposure playing out in the gut.

Key takeaways

• A bacterial toxin's mutational signature (SBS88/ID18) turned up in 44.8% of non-hypermutated colon tumors in a 600-person Japanese cohort, most concentrated in patients born after 1965 and diagnosed before 40.
• AI-based clustering sorted tumors into four microbiome-linked subtypes, including one tied to younger patients with an immune-suppressed tumor environment, and another tied to advanced disease and the bacterium Fusobacterium nucleatum.
• The colibactin-cancer link itself is well established; what's new here, a generational rise in exposure and four distinct microbiome subtypes, comes from one study and needs independent confirmation.
• This is a retrospective, associative study; it does not prove when or how the bacterial exposure caused the cancer, only that the mutational trace is there.

Colibactin's Mutational Fingerprint Found in 45% of Tumors

Researchers behind the new Nature Genetics study found that colibactin, a DNA-damaging toxin (a "genotoxin") produced by certain strains of E. coli that carry a specific gene cluster called the pks island, leaves behind a recognizable pattern of mutations in the genome. That pattern, made up of two related mutational signatures called SBS88 and ID18, showed up in 44.8% of non-hypermutated colorectal tumors (tumors without an unusually high overall mutation rate, which are analyzed separately because a different biology drives their mutation counts). In other words, closer to one in two ordinary colon cancers in this cohort carried a bacterial fingerprint in their DNA. This isn't a fringe idea: a landmark 2020 Nature study already showed that exposing human gut organoids to colibactin-producing bacteria directly produces this same mutational pattern, giving the signature a plausible mechanistic backstory rather than just a statistical coincidence.

Why a Birth-Year Pattern Matters for Younger Adults

The stakes here connect to something oncologists have been alarmed about for years: colorectal cancer is climbing in people who are supposed to be too young for it. The American Cancer Society reports that colorectal cancer diagnoses in adults under 50 have risen about 2.9% per year from 2013 to 2022, even as rates in older adults keep falling. This new study adds a specific, testable piece to that puzzle: the colibactin mutational signature was significantly more common in patients born after 1965, and the enrichment was strongest in patients diagnosed under age 40. That's a pattern that points toward a generational shift in exposure to genotoxin-producing bacteria, not simply better detection of a cancer that was always there. For anyone under 50 who has watched colon cancer show up in their peer group, or in themselves, this offers a possible biological thread, though not yet a proven cause, connecting the "why now, why younger" question to something happening in the gut microbiome.

Inside the 600-Person Genomic and Microbiome Cohort

To get here, the researchers combined two kinds of data rather than relying on either alone. They performed whole-genome sequencing on 200 tumors to catalog each cancer's mutational signatures, and ran fecal metagenomic sequencing (a method that sequences all the microbial DNA in a stool sample to profile the entire gut bacterial community) on 395 colorectal cancer cases and 146 healthy controls, drawn from a total study population of 600 people (454 with colorectal cancer, 146 controls). That paired design, genome plus microbiome, on the same population is what let them connect a mutation pattern in the tumor to a bacterial signature in the gut, rather than just noting the two exist separately. They then applied an AI classifier to the microbiome data and identified four distinct microbiome-defined subtypes of colorectal cancer. One subtype, more common in younger patients, combined a high colibactin signature with an immunosuppressive tumor environment rich in regulatory T cells (a type of immune cell that dampens immune attacks, including on tumors). A separate subtype, associated with more advanced disease, was instead dominated by Fusobacterium nucleatum, a bacterium separately linked to colorectal tumors in prior genomic work. The researchers frame colibactin exposure as "a highly prevalent but modifiable risk factor" for colorectal cancer in the Japanese population, language that signals optimism about future prevention without claiming that prevention tools exist yet.

What This Means If You're Managing Colon Cancer Risk

None of this changes today's screening advice, and it does not point to a specific supplement, diet, or antibiotic regimen as protective, the study doesn't establish any intervention as proven. What it does offer is a more concrete way to think about why colon cancer might be trending younger: your gut's long-term bacterial residents may matter as much as inherited genes or classic risk factors like diet and weight. If you're under 50 and have a family history of colorectal cancer, unexplained changes in bowel habits, or blood in your stool, this study is a reason to keep taking those symptoms seriously and to talk to a doctor about screening, not because colibactin testing exists (it doesn't, clinically, yet) but because the broader pattern of younger-onset disease this study documents is real and already reflected in national screening guidelines that now start at 45.

Where the Evidence Is Still Thin

Two things keep this from being settled science. First, the study is retrospective and associative (it looked back at existing tumors and stool samples rather than tracking people forward in time), so it can show that the mutational signature and birth year correlate, not exactly when or how the bacterial exposure translated into a cancer-causing mutation. Second, the four AI-derived microbiome subtypes still need independent, external validation in other cohorts before researchers can be confident they're a real, reproducible biological category rather than a pattern specific to this dataset. It's also worth being precise about what's actually new here: the colibactin-to-mutation link itself is well established in prior research; what's novel, and unreplicated, is this study's specific claim about a birth-cohort surge and its four-way microbiome subtyping.

What Comes Next for Colibactin Research

The logical next step is replication: running the same paired genome-and-microbiome analysis in cohorts outside Japan to see whether the same birth-year pattern and the same four subtypes appear in other populations with different diets, antibiotic histories, and E. coli strains. Longitudinal studies that track people's gut bacteria over years, rather than sampling once at diagnosis, would help pin down when colibactin exposure actually happens relative to when mutations accumulate. Until that replication work lands, this remains a large, well-designed single study rather than a confirmed piece of colorectal cancer biology.

Common Questions

What is colibactin and how would someone be exposed to it? Colibactin is a DNA-damaging toxin made by certain strains of E. coli that carry a specific genetic toolkit called the pks island; these bacteria can be long-term residents of the gut in many people, so exposure is thought to happen gradually over years of colonization rather than through a single infection.

Does this study prove gut bacteria cause colon cancer? It shows a strong association between a bacterial mutational signature and colon tumors, and separate lab work has shown colibactin-producing bacteria can directly cause that same mutation pattern in gut tissue, but this specific study's birth-cohort and subtype findings are associative and still need independent replication.

Is this relevant to people outside Japan? The cohort was entirely Japanese, so whether the same birth-year pattern and microbiome subtypes hold in other populations, with different diets and bacterial strains, is an open question the researchers themselves flag as needing further study.

Should younger adults get tested for colibactin exposure? There is no clinical test for colibactin exposure available today; the practical takeaway is to follow current colorectal cancer screening guidelines (now starting at 45 in the US) and raise any symptoms like rectal bleeding or persistent bowel changes with a doctor regardless of age.

References

1. Satoshi Shiba, Shinichi Yachida, Sayaka Mizutani, et al. (2026). Prevalence and chronology of colibactin-associated mutational processes and their microbiome spectra in Japanese colorectal cancer. Nature Genetics. https://doi.org/10.1038/s41588-026-02692-x
2. Mutational signature in colorectal cancer caused by genotoxic pks+ Escherichia coli. Nature. https://www.nature.com/articles/s41586-020-2080-8