Diversity of Iodinated Water Disinfection Byproducts and Their Formation Dynamics Discovered Using Exposome-Scale Nontargeted Analysis and Machine Learning

Water disinfection is critical to minimizing microbial risk, but unintentionally produces disinfection byproducts (DBPs). Exposure to the currently regulated chloro- and bromo-DBPs in water cannot fully account for the health effects observed from epidemiological studies. Iodinated disinfection byproducts (I-DBPs) are among the most toxic DBPs, but they have not been fully characterized because of significant technical challenges. We report here the first exposome-scale study of iodinated disinfection byproducts formed under environmentally relevant conditions. Here, “exposome-scale” refers to broad-scope, nontargeted profiling of exposure-relevant environmental contaminants. The machine-learning-assisted nontargeted analysis of chloramine-treated water enabled detection and characterization of 8551 and 1541 high-confidence iodinated molecular features, generated under the negative and positive ionization modes, respectively. Sixty-two new iodinated disinfection byproducts were identified, with 11 I-DBPs confirmed using authentic standards. Across four Environmental Protection Agency toxicity prediction end points (fathead minnow, Daphnia magna, Tetrahymena pyriformis, and rat), the 11 I-DBPs consistently showed higher or comparable toxicities relative to both the regulated DBPs and the four known I-DBPs, with predicted toxicities reaching up to 2–3 orders of magnitude greater than benchmark compounds such as chloroform and iodoacetic acid. This study demonstrates improved understanding of the diversity and formation dynamics of highly toxic I-DBPs, enabled by advances in integrating exposomics with machine learning techniques.