关键词: 血清, 非靶向筛查, 特征污染物, 工业场地, 职业人群

Abstract:

Persistent organic pollutants emitted from large-scale industrial activities pose significant health threats to humans, especially to industrial workers. Petroleum extraction, coking production, nonferrous metal smelting, and electronic waste (e-waste) dismantling represent four typical industrial sectors in China that release complex mixtures of industry-specific organic contaminants. Workers in these settings experience chronic exposure through inhalation, dermal contact, and inadvertent ingestion, leading to elevated body burdens. However, the profiles and characteristic pollutants in their serum remain poorly characterized due to the limitations of conventional targeted analytical methods, which are restricted to a predefined list and fail to capture unknown or emerging chemicals. To address this gap, non-targeted screening using high-resolution mass spectrometry (HRMS) has emerged as a powerful approach for comprehensive exposure profiling. Nevertheless, single-platform analysis may miss compounds with disparate polarity and volatility, underscoring the need for complementary dual-platform strategies. In this study, we aimed to develop a robust dual-platform non-target screening method based on gas chromatography quadrupole time-of-flight mass spectrometry (GC-QTOF-MS) and liquid chromatography quadrupole time-of-flight mass spectrometry (LC-QTOF-MS) and to apply it to characterize pollutant exposure profiles and identify site-specific characteristic pollutants in serum from occupational populations across four typical Chinese industrial sites. Serum samples (over 4000 individuals) were collected from workers and surrounding residents at large-scale enterprises in four provinces: Hebei (petroleum extraction), Shanxi (coking), Hubei (nonferrous smelting), and Guangdong (e-waste dismantling) during 2020‒2023. To efficiently capture dominant exposure features, a pooled-sample strategy was adopted. For the petroleum site, every 800 individual samples were combined into one pool; for the other three sites, every 7‒8 samples were pooled, yielding three independent pooled samples per site. After pooling, 500 μL serum aliquots were subjected to a modified QuEChERS extraction using acetonitrile, repeated three times. The combined extract was split into two equal portions: fraction 1 (F1) for GC analysis and fraction 2 (F2) for LC analysis. F1 was purified using dispersive solid-phase extraction (d-SPE) tubes, filtered, evaporated to near dryness, and reconstituted in isooctane containing the internal standard tris(2-chloroisopropyl) phosphate-d₁₈ (TCIPP-d₁₈). F2 was acidified with 0.5% formic acid (volume fraction) in acetonitrile, passed through a Captiva ND Lipids cartridge, evaporated, and reconstituted in methanol:water (v꞉v, 2꞉8) with the same internal standard. GC-QTOF-MS was operated in electron ionization mode with an HP-5MS column and a multi-ramp temperature program; LC-QTOF-MS was operated in electrospray ionization mode in both positive and negative modes using a Poroshell 120 EC-C₁₈ column with a gradient of water (containing ammonium acetate or formic acid) and methanol. GC-MS data were processed with Unknowns Analysis (NIST20, match ≥80); LC-MS data were processed with MS-DIAL (GNPS, MoNA). Procedural blanks (isooctane) and fetal bovine serum were analyzed for quality control, and compounds with signals less than 10 times the blank were excluded; 20 compounds were removed. TCIPP-d₁₈ recovery across samples was 71%‒130%, and duplicate serum analyses showed 91% compound overlap with absolute relative abundance differences of 0‒9, confirming reproducibility. Semi-quantification was performed using the peak area ratio of each compound relative to TCIPP-d₁₈, expressed as relative abundance. To mitigate potential bias from unequal pool sizes, relative abundance comparisons among sites were based on the median of the three pooled samples per site. The dual-platform strategy identified a total of 387 organic pollutants: 179 by GC-QTOF-MS, 251 by LC-QTOF-MS, with 43 detected by both. This coverage substantially exceeded that of either platform alone, demonstrating effective polarity complementarity: GC primarily captured nonpolar and weakly polar volatile or semivolatile compounds (e.g., polycyclic aromatic hydrocarbons, PAHs; methylsiloxanes, MSs; organophosphate esters, OPEs), whereas LC extended coverage to polar and moderately polar substances (e.g., fatty acids, pharmaceuticals, certain phenols). Distinct pollutant composition profiles emerged for each industrial site. At the petroleum extraction site, phenols (n=20), PAHs (n=8), and MSs (n=4) were dominant. Notably, both parent PAHs (naphthalene, phenanthrene, pyrene) and their alkylated homologues were detected, a hallmark fingerprint of petrogenic sources. The four cyclic siloxanes (octamethyl cyclotetrasiloxane, D₄; decamethyl cyclopentasiloxane, D₅; dodecamethyl cyclohexasiloxane, D₆; and tetradecamethyl cycloheptasiloxane, D₇) are associated with defoamers and dehydrating agents. In e-waste dismantling workers, 274 pollutants were identified, including phenols (n=21), OPEs (n=6), PAHs (n=4, only parent), MSs (n=4), and per- and polyfluoroalkyl substances (PFASs, n=4). These chemical classes closely match additives (plasticizers, flame retardants, insulators) in electronic components released during crude dismantling and heating. Coking site serum contained phenols (n=18), benzenes (n=5, the highest across all sites), and PFASs (n=6), with PAHs exclusively as parent compounds, which is consistent with high-temperature coal carbonization (900‒1100 ℃) where alkyl side chains are cleaved. The nonferrous metal smelting site showed phenol (n=15), benzene (n=7), PAH (n=4), and PFAS (n=4) congener patterns, but uniquely featured five organic salt compounds, likely originating from electrolytic processes in metal refining. Semi-quantitative relative abundance analysis revealed marked site-specific disparities. At the petroleum extraction site, total relative abundances of phenols and MSs were 370 and 340, respectively, with 2,5-di-tert-butylphenol (190), D₅ (170), and D₄ (140) as the most abundant single compounds. The coking site exhibited the lowest overall abundance; phenols totaled 110, and 2,5-dimethylresorcinol (38) was predominant, reflecting exposure to crude phenols from coke oven emissions. E-waste dismantling workers bore the highest overall pollution burden: phenols reached 670 and MSs 260, with extreme levels of 4-tert-butyl-2,6-diisopropylphenol (540) and D₅ (130), originating from insulating materials and thermal decomposition. The nonferrous smelting site showed the highest phenol abundance (850) and benzene abundance (330) among all sites. 2,5-Di-tert-butylphenol accounted for most of the phenol load (750), and 1,3,3-trimethyl-1,2-dihydroindene was the dominant benzene compound, implying combined exposure from flotation agents, antioxidants, and pyrolysis byproducts. These findings highlight that the dual-platform approach effectively captured distinct pollutant fingerprints in human serum closely corresponding to industrial processes. Phenols and MSs were common markers, but specific homologues and co-occurring compound classes (alkylated and parent PAHs) provided site-specific signatures. The identified potential characteristic pollutants (2,5-di-tert-butylphenol, D₄, D₅ for petroleum extraction; 2,5-dimethylresorcinol for coking; 4-tert-butyl-2,6-diisopropylphenol and D₅ for e-waste; and 2,5-di-tert-butylphenol for smelting) can serve as priority candidates for future targeted biomonitoring and biomarker development. Given the pooled-sample and semi-quantitative nature, the observed differences indicate relative trends; individual-level validation with stable-isotope dilution quantification is warranted to confirm these markers and establish exposure-health relationships. In conclusion, this first systematic dual-platform non-target screening of serum from four Chinese industrial occupational cohorts reveals differentiated pollutant exposure profiles and potential site-specific characteristic contaminants, providing a scientific basis for targeted exposure assessment in these high-risk industries.

Key words: serum, non-target screening, characteristic pollutants, occupational population