Identification of Key Indicators in the Adverse Outcome Pathway for Genetic Damage Induced by Chromate Exposure: A Machine Learning-Based Cross-Sectional Study.

How chronic inflammation and immune dysregulation drive genetic damage upon environmental exposure is a pivotal question in toxicology. Focusing on hexavalent chromium [Cr-(VI)], a human carcinogen, we dissected this link using a machine-learning approach in an occupational population. We discovered that specific immune checkpoint molecules (PD-L1, PD-1, LAG-3) and regulatory patterns mediated the effect of blood Cr on key immune populations, including PD-1-(+) B cells and CD8-(+) T cells. Furthermore, we defined a core set of four inflammatory-immune indicators (NLRP3, positive/negative immune patterns, CD8-(+) T cells) that robustly explain genotoxicity. From this, we conjectured a coherent cascade involving the NLRP3 inflammasome, immune regulatory patterns, and effector cell subsets, culminating in genetic damage. This study establishes a novel paradigm for understanding the immunoinflammatory mechanisms of environmental carcinogens by employing, for the first time, machine learning within an adverse outcome pathway (AOP) framework to systematically identify and validate NLRP3, positive/negative immune regulatory patterns, and CD8-(+) T cells as central players in Cr-(VI)-induced genetic damage.