Pancreatic cancer has a devastating prognosis, in large part because it is most often diagnosed when it has already spread — metastasized — to other organs. Finding new ways to predict which patients will develop metastatic disease could allow earlier, more targeted interventions.
The gut microbiome — the trillions of bacteria living in the digestive tract — has emerged as an unexpected player in cancer biology. Gut bacteria can influence the immune system, promote blood vessel growth, and even help cancer cells survive in circulation. This study asked whether specific patterns of gut bacteria can distinguish between patients with and without metastatic pancreatic cancer.
53 patients with pancreatic cancer were enrolled at an Italian cancer center — 25 with non-metastatic and 28 with metastatic disease. Fresh stool samples were collected before any cancer treatment began, and bacterial DNA was extracted and sequenced using 16S rRNA gene sequencing to profile the full gut microbial community.
Two complementary analytical approaches were used. First, PELORA (Penalized Logistic Regression Analysis) identified clusters of bacteria whose combined abundance differed between metastatic and non-metastatic patients. Second, iterative Random Forest (iRF) modeled complex, non-linear interactions between microbial species to identify the bacteria with the greatest power to discriminate between the two groups.
A subset of 20 patients also underwent metabolomics analysis — measuring small molecules produced by gut bacteria in fecal samples — to connect the microbial findings to biological activity and potential mechanisms.
Metastatic patients had an overall higher abundance of Gram-negative bacteria compared to non-metastatic patients. At the family level, clusters including Porphyromonadaceae, Fusobacteriaceae, and several others were significantly enriched in metastatic patients. At the phylum level, Tenericutes, Bacteroidetes, and Nitrospinae were all elevated in the metastatic group.
The iterative Random Forest analysis identified specific bacteria with the highest ability to predict metastasis. At the species level, these included Anaerostipes hadrus, Coprobacter secundus, Clostridium sp. 619, and Roseburia inulinivorans. At the genus level, Porphyromonas and Odoribacter were most discriminatory.
Metabolomics data showed that these differentially abundant bacteria correlated with changes in metabolic compounds in the gut, linking the microbial differences to biological mechanisms that could promote metastasis through immune modulation or promotion of blood vessel growth.
Several of the bacteria enriched in metastatic patients are known to produce lipopolysaccharide (LPS) and other compounds that can activate immune pathways, stimulate blood vessel growth, and create a favorable environment for cancer cells arriving at distant organs — the so-called pre-metastatic niche.
The metabolomics correlation confirms that the microbiome's influence is not merely associative but likely functional, with bacterial metabolites serving as molecular messengers that alter the body's immune landscape in ways that help pancreatic cancer cells survive and colonize new sites.
These findings open the possibility of using stool-based gut microbiome profiling as a non-invasive biomarker to predict which pancreatic cancer patients are at high risk of metastasis — potentially guiding more aggressive treatment decisions earlier in the disease course.
Moreover, the specific bacteria identified as metastasis-promoting could themselves become therapeutic targets. Modifying the gut microbiome through diet, probiotics, or targeted antibiotics to reduce harmful bacterial populations might, in the future, help limit pancreatic cancer spread.