Abstract
<title>Abstract</title> <p>Background The gut bacteria–pancreatic cancer link is widely reported, but mechanisms remain unclear. This study assessed causality between gut microbiota and pancreatic cancer and mediation by circulating metabolites and inflammatory mediators. Methods A two-sample, two-step Mendelian randomization (MR) approach was used. Microbiome data were from MiBioGen (n = 18,340). Discovery used FinnGen R10 (218.8k, 1,018 cases); validation used Biobank Japan (179.5k, 598 cases). Mediators included 14 circulating metabolites and inflammatory mediators. Inverse variance weighting was the main analysis, with sensitivity analyses including weighted median, MR-Egger, leave-one-out, and funnel plots. Mediation effects were estimated via the product coefficient method and tested by the delta method. Results After Bonferroni correction, 19 gut microbiota were negatively associated with pancreatic cancer risk, cross-validated in Biobank Japan (consistent direction, P < 0.05). Only the GCST90012005 locus met thresholds for microbiota→mediator (P < 0.05) and mediator→pancreatic cancer (β=-0.082, P = 0.034). Functional annotation confirmed GCST90012005 corresponds to IL-6. The proportion mediated for 10 microbiota ranged from 11.3% to 65.0% (mean 38.2%), negatively correlated with total effect (r=-0.82, P = 0.004). Sensitivity analyses supported robustness, with no significant pleiotropy. Conclusions This study provides genetic evidence that higher abundance of specific gut bacteria may be linked to lower pancreatic cancer risk, partly mediated through IL-6-related pathways, though extent varies by taxon. Colocalization did not support a shared causal variant (PP.H4 = 0.011), so results are exploratory. Functional studies are warranted to confirm IL-6 mediation and elucidate mechanisms.</p>