Mayo Clinic researchers have identified 341 bacterial species associated with five cancer groups after analyzing pre-treatment stool samples from 1,364 cancer patients, in a study published in Cell that also flags distinct gut microbiome patterns in adults aged 50 or younger with colorectal and breast cancers.

The team, led by Purna Kashyap, MBBS, director of the Mayo Clinic Microbiome Program, and Ruben Mars, PhD, a microbiome researcher at Mayo Clinic in Minnesota, drew on the Mayo Clinic Cancer Microbiome (MCCM) cohort — a real-world study recruiting patients at Mayo sites in Arizona, Florida and Minnesota, representing 40 states. Samples were collected before treatment began and linked to detailed clinical data on side effects and outcomes. The investigators compared the cancer patients against 287 people without cancer, then performed cross-cancer comparisons, adjusting for other health conditions, to isolate disease-specific signals.

The paper, titled "Microbiome signatures linked to cancer and treatment adverse events in a real-world cohort," lands at a moment when early-onset cancer rates are climbing. Colorectal cancer incidence is rising roughly 3% per year among adults aged 20 to 49, and breast cancer incidence is up 1.4% annually in women under 50, according to figures the authors cite from the American Cancer Society.

What the data show — and what they do not

The findings are associations, not established causes, and the authors state this plainly. In early-onset colorectal cancer, younger patients had higher lactate levels and more Veillonella parvula, a gut bacterium that feeds on lactate, which tumors can produce in abundance. Whether that relationship drives early-onset disease remains unknown.

Early-onset breast cancer showed shifts across 64 bacterial species and lower levels of primary bile acids. One of the 64 was Clostridium scindens, a bacterium involved in bile acid and steroid metabolism. Brain cancer, the third cancer in the early-onset analysis, showed no comparable age-related microbiome differences — a useful negative control that tempers broad claims.

Among the cross-cancer signals: neuroendocrine tumors showed a broad loss of common health-associated gut bacteria; liver and intrahepatic bile duct cancers carried higher levels of several bacteria including Enterococcus faecalis; esophageal cancer showed elevated levels of six species, including Streptococcus; and distinct links appeared in lymphoid leukemia, multiple myeloma and related plasma cell cancers.

The team also found survival associations. In liver and intrahepatic bile duct cancer, Bifidobacterium longum correlated with longer survival and Blautia A massiliensis with shorter survival, with additional gut bacteria tied to outcomes in colorectal, ovarian and prostate cancers and melanoma.

A mechanism-level finding on 5-FU toxicity

The most actionable result for oncology R&D concerns chemotherapy-induced diarrhea. Among patients receiving 5-fluorouracil (5-FU), those who later developed diarrhea had lower levels of bacterial genes capable of breaking down the drug — much of that function traced to Anaerostipes hadrus, a common gut bacterium. The signal did not appear in patients on carboplatin, indicating specificity to 5-FU rather than a general diarrhea signature.

"This example underscores the potential of leveraging our real-world mixed-cancer, mixed-treatment cohort to identify microbiome-driven mechanisms underlying cancer treatment-related adverse effects," the authors wrote.

Kashyap framed the translational stakes: "The microbiome is not the sole driver of cancer or treatment outcomes, but it is an underappreciated component that has not traditionally been considered in therapeutic approaches. This gives us a proof of concept that we can begin to understand why some patients experience a particular side effect and identify a target that could potentially be acted upon."

Mars emphasized prioritization: "We can now narrow the search to those microbial changes that are most specific to individual cancers. Those are the signals we need to understand first if we want to determine whether the microbiome plays a causal role in cancer and ultimately develop targeted interventions."

Portfolio implications

For microbiome-focused programs, the study provides both a target shortlist and a caution. The cohort design — mixed cancers, mixed treatments, pre-treatment sampling at scale — complements traditional case-control setups and could de-risk candidate selection for microbial interventions or predictive biomarkers. But causality remains unproven, and the authors' own framing — "as the field shifts from association to causality, a key challenge is prioritizing microbes and pathways most likely to drive specific cancers" — signals that interventional validation is still ahead.

Next steps for the group: testing whether cancer-specific microbial signals play a causal role in disease, and validating the microbiome's power to predict treatment side effects in larger patient groups. The authors describe MCCM as "a foundational resource for the discovery of cancer-specific microbiome signatures and predictive biomarkers."

For now, the 5-FU/A. hadrus finding offers the clearest near-term path from microbiome data to a measurable clinical decision — patient stratification or bacterial augmentation before chemotherapy.