Genetic relationships between the gut microbiota and prostate cancer: Mendelian randomization combined with bioinformatics analysis.

MR analysis identified 16 gut bacterial taxa with causal links to prostate cancer, including 7 risk-increasing and 9 protective taxa.

• Genetic instruments related to gut microbiota were screened and paired with PCa genome-wide association study (GWAS) data for Mendelian randomization analysis.
• Seven gut bacterial taxa were identified as causally associated with increased risk of PCa.
• Nine gut bacterial taxa were identified as causally associated with decreased risk (protective) of PCa.
• The 16 causally linked gut bacteria were associated with 144 related genes.

Five feature genes (PLCL1, VSNL1, ROR2, NRXN3, and TEAD1) were identified for constructing a nomogram to predict prostate cancer risk.

• Differentially expressed associated genes (DEAGs) were identified using GEO dataset differential expression analysis.
• Importance scores of DEAGs were determined through four machine learning models.
• The five feature genes were selected from the DEAGs based on their importance scores across the machine learning models.
• A nomogram was constructed based on these five feature genes to provide quantitative prediction of PCa onset risk.
• The nomogram was validated in a separate GEO dataset.

Colocalization analysis was performed on positive Mendelian randomization findings to strengthen causal inference.

• Positive MR findings were subjected to colocalization analysis as a follow-up step.
• Colocalization analysis was used to assess whether gut microbiota genetic variants and PCa share the same causal variant at a locus.
• This approach was used to reduce the potential for confounding from linkage disequilibrium in the MR findings.

Feature genes identified in this study may affect prostate cancer occurrence by inhibiting epithelial-mesenchymal transition, proliferation, migration, and invasion of cells.

• The five feature genes PLCL1, VSNL1, ROR2, NRXN3, and TEAD1 were implicated in biological processes related to PCa development.
• The proposed mechanism includes inhibition of epithelial-mesenchymal transition (EMT).
• Additional mechanisms proposed include inhibition of cell proliferation, migration, and invasion.

The study combined Mendelian randomization with bioinformatics analysis using GEO datasets to identify and validate genetic relationships between gut microbiota and prostate cancer.

• The analytical pipeline integrated MR analysis, colocalization analysis, differential expression analysis, and machine learning-based feature selection.
• Four separate machine learning models were used to determine importance scores of DEAGs, providing a multi-model approach to feature gene identification.
• Gene Expression Omnibus (GEO) datasets were used for both discovery and validation of the nomogram.
• The study identified 144 genes related to the 16 causally linked gut bacteria as a starting point for bioinformatics analysis.