Associations between volatile organic compounds exposure and multiple oxidative damage biomarkers: Method development, human exposure, and application for e-waste pollution prediction

Abstract

Atmospheric monitoring studies reveal substantial health risks from exposure to volatile organic compounds (VOCs) for workers and nearby children in e-waste recycling areas (ERA), yet internal exposure risks are seldom examined. To address this, we developed a method to simultaneously analyze 12 urinary VOC metabolites (mVOCs) and oxidative damage biomarkers (ODBs) in workers and children from ERA and general adults from control areas using ultrahigh performance liquid chromatography coupled with quadrupole/orbitrap high-resolution mass spectrometry (UPLC-Orbitrap-HRMS). The results showed that e-waste workers exhibited significantly higher levels of VOC exposure and ODBs than e-waste children and control adults. Exceeding 91.1 %, 69.1 %, 20.8 %, 19.7 %, and 3.26 % of e-waste workers faced non-carcinogenic risk from exposure to acrolein, acrylonitrile, acrylamide, 1,3-butadiene, and 1,2-dichloroethane, respectively. The weighted quantile sum, quantile g-computation, and Bayesian kernel machine regression models consistently indicated significant positive associations between these VOC mixtures and cholesterol ODB levels (i.e., glycocholic acid, cholic acid, and glycochenodeoxycholic acid), highlighting the necessity for improved protective measures for occupational workers. Interestingly, cholesterol ODBs significantly mediated the association between VOCs exposure and nucleic acid ODBs, accounting for 12.0-26.0 % of the association with 8-hydroxy-2'-deoxyguanosine (an oxidative DNA damage biomarker) and 25.4-53.4 % with 8-hydroxyguanosine (an oxidative RNA damage biomarker). This suggests that cholesterol ODBs potentially serve as better indicators of health risks from VOC exposure than nucleic acid ODBs. Additionally, the combination of mVOCs and ODBs (Mean AUC: 0.906, ACC: 0.821) as exposure fingerprints outperformed either mVOCs (Mean AUC: 0.878, ACC: 0.802) or ODBs (Mean AUC: 0.843, ACC: 0.768) alone in predicting the presence of e-waste pollution, underscoring the importance of integrating exposure and effect biomarker fingerprints to accurately capture e-waste pollution characteristic. Our findings offer a novel approach for screening e-waste pollution in unknow e-waste recycling sites and provide a foundation for developing high-precision prediction models for other polluting industries.

Keywords: Carcinogenic risk; Exposure fingerprints; Machine learning; Oxidative stress; VOCs.