This study aims to bridge toxicological target prediction and bladder cancer transcriptomics by systematically identifying molecular signatures and pathways that converge between PFOA-associated toxicological effects and bladder cancer biology, using an integrative multi-cohort computational framework. Bioinformatics strategies were applied to assemble 9,591 putative molecular targets associated with perfluorooctanoic acid (PFOA) from five independent databases. Differential expression analysis and weighted gene co-expression network analysis were then conducted across five bladder cancer cohorts to delineate tumor-related genes. Functional enrichment analyses, machine learning–based modeling with SHAP interpretation, and molecular docking were subsequently employed to explore affected biological pathways and evaluate predictive performance. A total of 69 candidate genes, predominantly associated with cell cycle control and DNA damage–related processes, were delineated. From these, a nine-gene classifier yielded excellent discriminatory performance, achieving an AUC of 0.986 in the training set and maintaining robust accuracy across multiple external cohorts (AUCs: 0.944–1.000). SHAP-based interpretability analyses identified MCM7 as the most influential contributor to bladder cancer classification. In silico docking further suggested a strong predicted interaction between PFOA and IGFBP2, with a binding energy of -13.0 kcal/mol. By integrating toxicological target prediction with large-scale bladder cancer transcriptomic analyses, this study provides a computational bridge between environmental chemical exposure and cancer-related molecular programs. The resulting nine-gene classifier demonstrates strong and consistent performance across independent cohorts and captures transcriptional features that intersect with PFOA-associated toxicological pathways, offering a systems-level, hypothesis-generating perspective on transcriptional programs that overlap between predicted PFOA-associated targets and bladder cancer biology.
Liu et al. (Tue,) studied this question.