Researchers integrated 6,779 gut microbiome profiles across 27 studies, identifying a consistent microbial signature for colorectal cancer.
Researchers from EMBL, LUMC, and collaborators conducted a meta-analysis of 27 studies involving 6,779 gut microbiome profiles from colorectal cancer patients and controls. The analysis identified a microbial signature associated with colorectal cancer that was consistent across populations, sequencing methods, and age-of-onset groups.
Key Findings
- A machine-learning classifier could distinguish colorectal cancer from non-cancer microbiomes across datasets.
- The colorectal cancer microbiome signature was linked to lower dietary fibre intake and was reduced following fibre-focused dietary interventions.
- Analysis of 906 intestinal tissue samples showed that microbes enriched in tumour tissue were similar to the signature observed in faecal samples.
- In tissue samples, cancer-associated microbes were detectable in early-stage tumours; detection accuracy in stool was lower for early-stage cancers and tumours located further upstream in the colon.
- Pre-cancerous adenomas showed weak and inconsistent microbial changes, indicating limited detectability via stool microbiome profiling.
- Fusobacterium subspecies showed differential enrichment: Fusobacterium nucleatum subsp. animalis was consistently enriched across continents, while other subspecies exhibited geographic variability.
Methods
The researchers developed computational approaches to integrate microbiome datasets generated using different sequencing methods. The machine-learning algorithm outputs a score indicating how 'cancer-like' a microbiome is, applicable to any gut microbiome dataset.
Implications
The study provides a reference for future microbiome-based detection and risk assessment research. However, microbiome-based classifiers did not match the performance of faecal immunochemical tests in comparisons. Larger studies are needed to assess potential complementarity with existing clinical tests.