Air pollution and prostate cancer: Unraveling the connection through network toxicology and machine learning

Abstract

Background: In recent years, air pollution has been demonstrated to be associated with the occurrence of various diseases. This study aims to explore the potential association between air pollutants and prostate cancer (PCa) and to identify key genes that may play a critical bridging role in this process. Methods: This study utilized multiple online databases to obtain relevant target genes associated with air pollutants and PCa. Protein-protein interaction (PPI) analysis and visualization were conducted for the intersecting genes, followed by Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) functional enrichment analyses to explore potential mechanisms. Subsequently, the best predictive model was selected through a combination of 108 machine learning algorithms. A prognostic model was constructed using the Random Survival Forest (RSF) model in conjunction with Lasso regression model, and its performance was validated in four external datasets. Finally, molecular docking analysis was conducted to investigate the interaction between key genes and air pollutants. Results: Seven common air pollutants (benzene, SO₂, NO, CO, NO₂, toluene, and O₃) were selected for analysis, and 48 intersecting targets related to PCa were identified. GO and KEGG functional enrichment analyses revealed that these targets are primarily involved in regulating biological processes such as apoptosis, carcinogenesis, and cell proliferation. Based on machine learning algorithm selection, the combination of RSF and Lasso regression was identified as the optimal predictive model, which highlighted five key genes associated with air pollutants and PCa. The model exhibited strong predictive performance across all four independent external datasets. Additionally, molecular docking analysis further confirmed the potential interactions between air pollutants and these core targets. Conclusion: The findings suggest that HDAC6, CDK1, DNMT1, NOS3, and DPP4 play crucial roles in the process by which air pollutants influence PCa. The results offer new insights into the molecular mechanisms linking air pollutants and PCa, highlighting the need for greater public awareness of air pollution issues.

Keywords

• Air pollutant
• Prostate cancer
• Network toxicology
• Machine learning
• Molecular Docking