Analyzing per- and polyfluoroalkyl substances (PFAS) in triple-negative breast cancer (TNBC): an integrative transcriptomic and computational study

Per- and polyfluoroalkyl substances (PFAS), including perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS), are persistent environmental contaminants. Epidemiological studies have associated PFAS exposure with several cancers, but TNBC-specific molecular evidence remains limited. We integrated computational target prediction, network analysis, bulk transcriptomic and microarray cohorts, single-cell RNA sequencing, spatial transcriptomics, prognostic modeling, molecular docking, molecular dynamics simulation, and expression assessment in 15 paired TNBC and adjacent tissues. The public patient cohorts and tissue specimens did not contain measured PFAS exposure, so all PFAS-related interpretations were prespecified as associative and hypothesis-generating. Results The intersection of 654 predicted PFAS targets, 3,944 TNBC-associated genes, and 27,529 TNBC differentially expressed genes yielded 197 candidates. A hypergeometric test indicated that this overlap exceeded chance expectation (observed/expected = 3.22, P = 3.65 × 10 − 52), although the permissive DEG set and heterogeneous disease-gene sources remain important limitations. Six candidate genes (CDK9, PF4, VEGFC, FABP7, NR2F6, and PLAU) formed the prognostic signature. The 2-year training-set AUC of 0.999 was interpreted as probable optimism rather than clinical-level performance. RT-qPCR showed significant differences for CDK9, PF4, VEGFC, FABP7, and NR2F6, whereas PLAU was directionally increased but not significant. Docking and simulation supported computational compatibility for selected protein-PFAS complexes but did not demonstrate cellular target engagement. The findings define a TNBC gene signature that overlaps with predicted PFAS targets and provide a testable computational hypothesis. Causal inference requires PFAS-stratified patient cohorts, direct exposure transcriptomics, dose–response experiments, and independent prospective validation.