Big Data Analytics Reveals Pyrethrins’ Breast Cancer Risks: A Deep Learning-Enhanced Study Combining Mendelian Randomization and Molecular Dynamics

Abstract

Pyrethrins, a class of broad-spectrum insecticides, have garnered extensive utilization in agricultural, public health, and environmental sectors. However, emerging concerns have arisen regarding their potential chronic carcinogenic risks. This study employed an integrative big data approach combining network toxicology, deep learning, mendelian randomization, molecular docking, and dynamics simulations to systematically evaluate pyrethrin's breast cancer-related targets. Computational screening identified 16 high-affinity targets with binding energies < -7.5 kcal/mol, indicative of stable interactions. Molecular dynamics simulations further validated the structural stability of pyrethrin-target complexes. MM/PBSA analyses revealed that both pyrethrin I and II exhibit thermodynamically spontaneous interactions with diverse targets, demonstrating binding free energies ranging from -9.53 to -27.37 kcal/mol. Complex interactions among targets were constructed. Gene Ontology and Kyoto Encyclopedia of Genes and Genomes enrichment analyses revealed significant associations between these targets and breast cancer pathways. Our multi-omics big data evidence positions pyrethrins as a risky toxicant with carcinogenic potential. These findings provide insights for regulatory reevaluation of pyrethrin's safety profile and underscore the need for longitudinal biomonitoring studies to assess its population-level health impacts.