Tumor microbes, immune cells and metabolites track CEA levels in colorectal cancer

Patients with colorectal cancer can carry very different amounts of carcinoembryonic antigen, or CEA, in their blood before surgery, even when the disease looks similar on scans. A new study argues that part of the answer may lie inside the tumor itself, in an interplay between resident bacteria, immune cells and small metabolic molecules. The work, published in BMC Microbiology, integrated three layers of molecular data from tumor tissue to map what the authors call a microbiota-immune-metabolism axis linked to preoperative serum CEA status.

CEA is one of the most widely used serum markers in the clinical management of colorectal cancer, yet the intratumoral ecological basis of why some patients have high levels and others low levels has remained unclear. To probe this, researchers enrolled 63 patients with primary colorectal cancer who had received no antitumor therapy before surgery and underwent radical resection. Using the conventional preoperative serum CEA cut-off of 5 ng/mL, they divided the cohort into a high-CEA group of 27 patients and a low-CEA group of 36 patients.

On each tumor tissue sample the team ran three complementary measurements: 16S rRNA gene sequencing to profile the bacteria present, RNA sequencing to capture host gene expression, and untargeted metabolomics to survey small molecules. They then compared intratumoral microbiota structure, transcriptional profiles, immune cell infiltration and metabolite differences between the two groups, and followed up with cross-omics correlation analyses, mediation analyses and machine learning modeling.

The microbial findings were nuanced. Several alpha-diversity richness indices, including Observed, Chao1 and ACE, were elevated in the high-CEA group, indicating more bacterial taxa in those tumors. However, the Shannon, Simpson and Coverage indices showed no significant differences, and PERMANOVA tests based on Bray-Curtis, Jaccard and both weighted and unweighted UniFrac distances found no significant overall separation in community structure between the groups. In other words, high-CEA tumors tended to be richer in bacterial types without being wholesale different ecosystems.

Where structure did not separate, individual taxa did. Linear Discriminant Analysis Effect Size analysis identified 39 differential taxonomic units associated with CEA status, 33 of which were enriched in the high-CEA group.

The transcriptomic and immune analyses pointed toward a myeloid inflammatory signature. High-CEA tumors showed increased infiltration by activated mast cells and neutrophils, along with enhanced transcriptional features related to inflammation, chemotaxis, endothelial activation and immunomodulation. Pyramidobacter-related taxa were associated with neutrophil infiltration and with the genes PTGS1, S100A8 and SELE, as well as multiple chemokines, sketching a possible route by which specific bacteria could shape an inflammatory tumor microenvironment.

Metabolomics screened five candidate differential metabolites between the groups. Exploratory mediation analyses then suggested that one of them, magnoline, may partially mediate the association between Synergistaceae/Pyramidobacter and Oscillospiraceae-related taxa and CEA status, with mediation ratios of 57.4 percent and 53.2 percent respectively. A random forest classifier built from the differential microbial taxa showed exploratory discriminative performance in internal cross-validation.

The authors are careful about the limits of these numbers. The classifier estimates are internally derived and may be optimistic, and the study is a single-center observational cohort. Correlations, mediating effects and model performance all need validation in independent cohorts, spatial localization within tumors and mechanistic experiments before any causal story can be claimed. The mediation ratios in particular describe statistical decomposition of associations, not proven biological pathways.

What the study offers is a covariance network rather than a mechanism: a picture of how intratumoral bacterial richness, specific anaerobic taxa, myeloid immune infiltration, inflammatory and chemotactic gene programs and a candidate metabolite move together with serum CEA. The authors propose the Pyramidobacter-myeloid inflammatory immunity axis and the microbiota-magnoline metabolism axis as clues to candidate mechanisms worth testing. If independent cohorts confirm the pattern, intratumoral microbial ecology could eventually help explain a familiar but poorly understood source of variation in one of colorectal cancer’s most routine blood markers.

Subject of Research: Intratumoral microbiome, immune and metabolite features associated with preoperative serum CEA status in colorectal cancer

Article Title: Multi-omics profiling reveals the intratumoral microbiota-immune-metabolism axis associated with preoperative serum CEA status in colorectal cancer

Article References: Qin, X.-Q., Liu, S.-H., Wu, Y.-Z., Qin, M.-J., Zhong, R.-M., Huang, Y.-Q., Huang, Z.-G., Liu, J., Qin, L.-Y., Wu, P.-D., Zhang, W.-S., Huang, X.-L., Tang, W.-Z., Liu, J.-G., & Long, C.-Y. (2026). Multi-omics profiling reveals the intratumoral microbiota-immune-metabolism axis associated with preoperative serum CEA status in colorectal cancer. BMC Microbiology. https://doi.org/10.1186/s12866-026-05840-9

Image Credits: AI Generated

DOI: 10.1186/s12866-026-05840-9

Keywords: colorectal cancer, CEA, intratumoral microbiota, 16S rRNA sequencing, RNA-seq, metabolomics, neutrophils, Pyramidobacter, magnoline, random forest, tumor microenvironment, biomarkers