A 28-microbial-species risk score derived from shotgun metagenomic sequencing of stool samples in 5,997 Dutch participants predicted incident cancer diagnoses over up to 10 years of follow-up with a hazard ratio of 1.62 per one standard-deviation increase (95% CI: 1.44–1.82; P=1.54e-15), independent of established risk factors including smoking, obesity, and diet. The score replicated in an external case-control dataset of 7,220 individuals and showed consistent trends in Finland's FINRISK cohort.
This finding carries significant weight in the cancer-prevention landscape. Most prior microbiome-cancer research has been cross-sectional — capturing microbial signatures after cancer is already present — making causal directionality ambiguous. This prospective design is a meaningful upgrade, suggesting gut microbial composition may reflect or even participate in pre-malignant biology years before clinical detection. The 28-species panel spans a tractable biomarker set that could, in principle, complement existing screening tools like colonoscopy or PSA testing.
Critical caveats apply, however. The score was developed and validated predominantly in Northern European cohorts, limiting generalizability across ancestries and dietary environments. The mechanism linking these species to oncogenesis remains unestablished — correlation does not confirm causation. Additionally, cancer type heterogeneity within the outcome variable warrants deeper subgroup analysis. As a preprint posted on medRxiv and not yet peer-reviewed, these results should be considered preliminary; independent replication and methodological scrutiny are essential before clinical translation. Still, the prospective design and large cohort make this one of the more compelling microbiome-cancer association studies to date.
