The research team, led by investigators at Daping Hospital of Army Medical University in Chongqing, China, recruited 178 patients with inflammatory bowel disease, comprising 104 with ulcerative colitis and 74 with Crohn’s disease, alongside 71 healthy individuals. Each participant provided faecal samples that were subjected to metagenomic sequencing, which reads the genetic material of the entire microbial community, and untargeted metabolomic profiling, which captures the small molecules those microbes and the host jointly produce. Rather than assuming that all IBD patients share the same form of dysbiosis, the researchers let the data speak for itself, applying a computational technique called non-negative matrix factorisation consensus clustering to sort patients into groups based on their microbial composition.

The analysis revealed two robust enterotypes within the IBD cohort. The first, designated GET-B, was enriched for bacteria of the family Bacteroidaceae and contained 112 patients. The second, GET-E, was enriched for Enterobacteriaceae and contained 66 patients. This distinction matters because the two groups did not merely differ in which bacteria were present; they differed in how sick they were. GET-B patients carried microbiotas that resembled those of the healthy controls more closely, while GET-E patients showed markedly lower alpha diversity, a standard measure of how many different species coexist in the gut and in what proportions. Loss of diversity is a well-recognised hallmark of microbial disturbance, and in this cohort it tracked with worse clinical outcomes.

The clinical contrasts were striking. Patients in the GET-E group scored higher on the Gastrointestinal Symptom Rating Scale, a validated questionnaire capturing abdominal pain, reflux, indigestion, diarrhoea, and constipation, and they showed greater overall disease severity. More provocatively, they also scored higher on standard psychiatric instruments: the Hamilton Anxiety Rating Scale and the Hamilton Depression Rating Scale or the Patient Health Questionnaire-9. The link between gut inflammation and psychological distress has long been noted in IBD, with psychiatric comorbidity affecting a large fraction of patients, but the microbial and metabolic mediators of that gut-brain connection have remained murky. This study offers a concrete candidate mechanism rooted in the chemistry of the gut.

That mechanism emerges most clearly from the metabolomic data. GET-B patients, the Bacteroidaceae-dominated group, were enriched in a suite of metabolites widely regarded as protective. Among them were secondary bile acids such as tauroursodeoxycholic acid, or TUDCA, a molecule that has attracted attention for its anti-inflammatory and cytoprotective properties. They also carried higher levels of tryptophan derivatives, including indole-3-propionic acid and indole-2-carboxylic acid. Indole-3-propionic acid in particular has been implicated in strengthening the intestinal barrier and in modulating neural pathways, making it a plausible chemical courier along the gut-brain axis. Short-chain fatty acids were represented too: butyrate, the principal energy source for colonocytes and a well-documented anti-inflammatory agent, was enriched in GET-B, as was L-dopa, a precursor of the neurotransmitter dopamine.

GET-E patients told a very different chemical story. Their metabolomic profiles showed elevated tyramine, an amine produced by certain bacteria through the decarboxylation of tyrosine. Tyramine is biologically active in the host and has been associated in other contexts with sympathetic nervous system effects, so its accumulation in the Enterobacteriaceae-enriched group adds another thread to the picture of a metabolically hostile gut environment. Taken together, the two enterotypes represent not just different bacterial censuses but opposing metabolic economies: one producing molecules that nourish the gut lining and calm inflammation, the other depleting those molecules and accumulating potentially harmful substitutes.

The integrative network analysis sharpened these associations further. Bacteroidetes species correlated positively with the protective metabolites and negatively with disease severity and psychiatric scores, while Proteobacteria species showed the mirror-image pattern, correlating negatively with protective metabolites and positively with severity and psychological distress. Two Proteobacteria members stood out: Escherichia coli and Klebsiella pneumoniae, both of which are recognised as opportunistic pathogens capable of flourishing when the healthy microbial ecosystem is disrupted. Their expansion in GET-E patients fits a broader narrative in IBD research, in which a bloom of facultative anaerobes from the phylum Proteobacteria displaces the obligate anaerobes that normally maintain the gut’s chemical balance.

Beyond describing these patterns, the researchers translated them into a potential diagnostic tool. Using a random forest machine learning model trained on the relative abundances of nine bacterial species, they were able to classify patients into the two enterotypes with high accuracy, achieving an area under the receiver operating characteristic curve of 0.964, with a 95 percent confidence interval spanning 0.919 to 1.000. An AUC close to one indicates near-perfect discrimination, suggesting that a modest panel of microbial markers could, in principle, sort IBD patients into these clinically meaningful groups from a single stool sample. The model’s optimal probability cutoff was determined by maximising Youden’s J index, a standard approach for balancing sensitivity and specificity, and the confusion matrix on the held-out test set confirmed robust performance.

The implications for precision medicine are considerable. If the GET-E enterotype identifies patients at higher risk of severe symptoms and psychiatric comorbidity, clinicians could in future use a simple microbial test to stratify patients at diagnosis, tailoring the intensity of monitoring, the aggressiveness of therapy, and the inclusion of mental health support accordingly. The metabolomic findings also point toward candidate interventions: restoring butyrate-producing bacteria, supplementing secondary bile acids or indole derivatives, or targeting Proteobacteria overgrowth are all avenues that this stratification framework makes testable in properly designed trials. The authors are careful to frame these as associations rather than proven causal pathways, and the cross-sectional design of the study cannot establish whether the GET-E microbiome drives worse outcomes or merely accompanies them.

Even with those caveats, the study represents a methodological step forward for the field. By combining metagenomics, untargeted metabolomics, machine learning, and detailed clinical phenotyping in a single cohort of 249 individuals, it demonstrates that the blunt category of inflammatory bowel disease conceals at least two microbiologically and chemically distinct patient populations with measurably different burdens of disease and distress. The work was approved by the Ethics Committee of the Army Medical Center of the PLA, registered in the Chinese Clinical Trial Registry, and conducted under the Declaration of Helsinki with informed consent from all participants. As microbiome science matures, studies of this kind are shifting the central question from whether the gut microbiota matters in IBD to how specific microbial configurations translate into specific clinical realities, and how those configurations might be deliberately reshaped to change them.

Subject of Research: Gut microbiome enterotypes and metabolomic signatures in inflammatory bowel disease

Article Title: Gut microbiome enterotypes associate with distinct metabolomic landscapes and clinical outcomes in inflammatory bowel disease: a multi‑omics dissection

Article References: Zhao, X., Chen, M., Li, T., Ruan, G., Cheng, Y., Xiao, Z., Chen, L., Yi, A., Tian, Y., Li, N., Zhan, D., Chen, D., & Wei, Y. (2026). Gut microbiome enterotypes associate with distinct metabolomic landscapes and clinical outcomes in inflammatory bowel disease: a multi‑omics dissection. Gut Pathogens. https://doi.org/10.1186/s13099-026-00885-7

Image Credits: AI Generated

DOI: 10.1186/s13099-026-00885-7

Keywords: inflammatory bowel disease, gut microbiome, enterotypes, metabolomics, metagenomics, ulcerative colitis, Crohn's disease, gut-brain axis, bile acids, tryptophan metabolism, Proteobacteria, precision medicine