BetaEntity Annotation Prototype
← Back to diseases

Annotated abstract

O23 Gut microbiome associations with colorectal neoplasia across stages and subtypes in a large cross-sectional colonoscopy cohort

gutjnl · 2026-06-23 · canonical JSON source

6 visible annotations · policy: published · automated confidence ≥ 75.00%

Document resource

Introduction Gut microbiome alterations are well described in colorectal cancer (CRC), but it remains unclear whether reproducible microbial signatures are present in precursor lesions or across neoplasia subtypes and stages relevant for early detection. Understanding microbial changes in early lesions could inform non-invasive screening and mechanistic studies.Methods We analysed pre-bowel preparation stool shotgun metagenomes from 1,762 adults undergoing colonoscopy in the COLO-COHORT study. Colorectal neoplasia was classified using British Society of Gastroenterology surveillance criteria into non-advanced polyps, advanced polyps, and cancer, and by subtype (adenomatous, serrated, and sessile serrated lesions). Microbial alpha and beta diversity, taxonomic abundances, and functional metabolic pathways were assessed using linear mixed-effects models adjusting for demographic, dietary, lifestyle, and medication covariates. Random forest models were applied to evaluate the predictive value of microbiome features for identifying high-risk neoplasia.Results Overall gut microbiome composition differed statistically between participants with and without neoplasia, but effect sizes were negligible (Bray-Curtis R2: 0.001). Alpha diversity did not differ by neoplasia presence, stage or subtype. A small number of bacterial species, including Mediterraneibacter faecis and Pseudoruminococcus massiliensis, and selected metabolic pathways related to amino acid biosynthesis and cell wall metabolism were modestly associated with neoplasia, particularly non-advanced lesions, though many associations attenuated after comprehensive covariate adjustment. Previously reported CRC-associated taxa were infrequently detected or not consistently associated with precancerous lesions. Predictive models combining microbiome features with clinical and demographic variables showed only modest improvement in classifying high-risk neoplasia compared with clinical variables alone (area under the curve 0.64 vs 0.58).Conclusions In this large, deeply phenotyped colonoscopy cohort, gut microbiome associations with precancerous colorectal neoplasia were weak, inconsistent, and largely explained by lifestyle or clinical factors. No robust or clinically useful microbial signatures were identified. These findings, consistent with our prior systematic review and other emerging evidence, suggest that major microbiome alterations may emerge later in CRC development rather than during early neoplasia. Longitudinal, multi-omic studies are needed to clarify temporal dynamics, causal relevance, and potential utility for early detection.