Y-ECCO Literature Review: Ruta Inciuraite

Alterations in gut microbial composition have been increasingly implicated in Inflammatory Bowel Disease (IBD) [1–3]. Previous studies have linked IBD to enrichment of proinflammatory bacteria [4–6] and depletion of short-chain fatty acid producers [7], contributing to intestinal inflammation [8, 9]. Despite accumulating evidence, the diagnostic value of microbiome-derived biomarkers for IBD remains insufficiently characterised. To address this gap, the authors conducted comprehensive metagenomic analyses to evaluate the predictive capacity of disease-associated bacterial species. They subsequently constructed diagnostic models and developed a multiplex droplet digital PCR (m-ddPCR) panel targeting multiple bacterial biomarkers to establish a microbiome-based diagnostic tool for IBD.

The study analysed 5979 faecal metagenomic samples from multi-ethnic cohorts across eight countries, including 1884 samples generated in-house and 4095 samples obtained from public datasets. These cohorts comprised patients with Ulcerative Colitis (UC), Crohn’s Disease (CD), healthy controls and additional non-IBD disease groups to test model specificity. A Hong Kong discovery cohort, along with independent in-house and public datasets from Australia, the United States, Europe and mainland China, was used to identify and validate microbial signatures.

The authors identified 1175 microbial groups in total, of which 788 were classified at the species level. IBD was characterised by reduced microbial diversity, a depletion of Firmicutes, and enrichment of Proteobacteria, while CD additionally showed a reduction in Bacteroidetes. The authors reported distinct disease-specific microbial profiles, with UC characterised by altered abundance of 63 species and CD by 58 species, highlighting both shared and unique taxa associated with each condition. The study further identified novel associations; notably, Actinomyces sp. oral taxon 181 was enriched in both UC and CD.

The researchers subsequently developed diagnostic models using general linear models followed by random forest algorithms. As optimal discriminators, the authors selected panels of ten species for UC (four enriched: Gemella morbillorum, Blautia hansenii, Actinomyces sp. oral taxon 181 and Clostridium spiroforme; six depleted: Clostridium leptum, Fusicatenibacter saccharivorans, Gemmiger formicilis, Ruminococcus torques, Odoribacter splanchnicus and Bilophila wadsworthia) and nine species for CD (three enriched: Bacteroides fragilis, Escherichia coli and Actinomyces sp. oral taxon 181; six depleted: Roseburia inulinivorans, Blautia obeum, Lawsonibacter asaccharolyticus, Roseburia intestinalis, Dorea formicigenerans and Eubacterium sp. CAG: 274) to differentiate patients from controls. These models achieved strong classification accuracy, with AUCs approaching 0.95 in both training and test datasets. Key reported discriminative species included F. saccharivorans, C. leptum and G. formicilis for UC and B. obeum, R. inulinivorans, L. asaccharolyticus, Actinomyces sp. oral taxon 181 and E. coli for CD.

Beyond compositional changes, the authors identified 545 dysregulated metabolic pathways – 244 in UC and 315 in CD. Analyses showed associations of disease-enriched taxa with enrichment of pathways involved in amine and polyamine degradation, as well as fatty acid and lipid biosynthesis. Depletion of beneficial species corresponded to reduced contributions to amino acid biosynthesis, reflecting impaired microbial metabolic capacity in IBD.

Furthermore, the authors showed that the multibacteria biomarker panel achieved robust performance in distinguishing IBD patients from controls, independent of disease activity or treatment, with AUCs approaching 0.90. Diagnostic accuracy was maintained across different geographic regions and ethnicities for UC and CD, with consistent results in independent international cohorts (with AUCs varying from 0.73 to 0.97). Importantly, the panel proposed by this study demonstrated disease specificity, effectively differentiating IBD from various non-IBD gastrointestinal and systemic conditions (with AUCs of 0.77 for IBD, 0.78 for UC and 0.72 for CD).

To translate metagenomic findings into a clinically applicable tool, the authors developed a multiplex m-ddPCR assay to quantify selected bacterial species in faecal samples. The assay demonstrated strong concordance with metagenomic sequencing and reliably differentiated patients with UC and CD from controls, achieving AUCs of 0.88 and 0.87, respectively, in discovery cohorts and comparable values in validation cohorts. Importantly, when compared with faecal calprotectin, the m-ddPCR panel performed consistently better in UC and was similar or superior in CD, particularly in distinguishing inactive disease (either UC or CD) from controls, underscoring its potential as an affordable and robust microbiome-based diagnostic approach for IBD.