Exposure to environmental toxicants and obstructive sleep apnea: a cross-sectional study Authors: Weixing Liu, Pei Li, Zhiyuan Wang, Yue Liu, Jia Chen, Hui Liu, Qintai Yang, Jin Ye Published: 2026 ABSTRACT AbstractObjectiveTo evaluate associations between environmental toxicant (ET) exposures and obstructive sleep apnea (OSA) using an exposome-wide approach.MethodsThis cross-sectional analysis included 13,161 participants from two NHANES cycles (2015–2016 and 2017–March 2020). Fifty-six toxicants across six classes (arsenic species, ethylene oxide, heavy metals, nicotine metabolites, VOC metabolites, and thyroid-disrupting compounds) were measured. OSA status was defined using a validated questionnaire based on self-reported symptoms. Exposome-wide association analysis (ExWAS) adjusted for sociodemographic, lifestyle, and clinical covariates, and restricted cubic spline (RCS) analysis characterized nonlinear exposure–response relationships.ResultsThe overall prevalence of OSA was 49% (n = 6,443). Nine ETs showed significant associations with OSA, including blood and urinary cadmium, seven volatile organic compound metabolites (e.g., 2,5-dimethylfuran, ethylbenzene, furan, o-xylene), and three urinary cysteine-derived metabolites (N-acetylcysteine derivatives). Subgroup analyses revealed higher susceptibility among women and individuals aged <65 years. RCS analyses demonstrated nonlinear, nonmonotonic relationships for cadmium and cysteine-derived metabolites with OSA prevalence.ConclusionsNine ETs were significantly associated with OSA prevalence, with evidence of differential susceptibility by sex and age. These findings highlight environmental factors associated with OSA, which may serve as potential targets for future mechanistic research and public health interventions.Keywords: Environmental toxicants, obstructive sleep apnea, NHANES, cross-sectional study, exposome-wide association study Introduction Obstructive sleep apnea (OSA), a serious sleep-related breathing disorder, is experiencing an exponential increase in prevalence [1]. The global prevalence of OSA is estimated at 936 million adults aged 30–69 years, exceeding 50% in some countries [2]. OSA manifests as recurrent pauses in breathing during sleep, resulting in frequent decreases in blood oxygen levels (desaturations), altered sleep architecture (fragmentation), and significant daytime consequences, such as persistent exhaustion and impaired attention [3,4]. These symptoms not only impair quality of life but also increase the risk of traffic accidents, underscoring the public health significance of this condition [5]. In addition, OSA is closely associated with various health issues, including cardiovascular diseases, hypertension, metabolic syndrome, and cognitive dysfunction [6,7]. These complex psychosomatic conditions impose significant economic and social burdens. This makes it clear that identifying modifiable risk factors for OSA is an urgent priority. Obesity, the primary risk factor for OSA, is strongly correlated with the condition, while other factors like gender, smoking, and alcohol consumption also play a role. Together, these physiological, demographic, and behavioral factors are associated with the development and progression of OSA [6,7]. Beyond traditional risk factors, emerging evidence highlights the roles of genetic variants and neurotransmission alterations in OSA pathogenesis [10], as well as its complex interplay with metabolic disorders [11]. Recent years have seen growing attention to potential environmental factors influencing OSA, with environmental toxins (ETs) emerging as a critical area of focus [12,13]. Previous research has accumulated substantial evidence that indicates that a wide range of natural and synthetic toxicants, including volatile organic compounds (VOCs), metals, and cotinine, are linked to OSA in humans [14–16]. However, the research exploring the links between certain ETs and OSA remains sparse. The limitations of previous research include a small sample size, which compromised the precision of the estimates and the ability to draw meaningful conclusions, as well as the absence of comprehensive measurements of ETs. Therefore, conducting a comprehensive exposome analysis of ETs is imperative to understand their associations with OSA. To our knowledge, no prior exposome-wide study has examined these associations in a large, nationally representative sample, which motivated the present investigation. The aim of this study was to investigate the associations between exposure to a broad panel of environmental toxicants and OSA symptoms in a large, nationally representative sample of US adults using an exposome-wide association approach, thereby laying the groundwork for future policy and preventive interventions. Materials and methods Study design and population NHANES is a nationally representative cross-sectional study conducted by the Centers for Disease Control and Prevention (CDC) and the National Center for Health Statistics (NCHS) to evaluate the health and nutritional status of the US population. For this study, we combined data from two consecutive NHANES cycles (2015–2016 and 2017–March 2020), which initially included 25,531 participants with documented informed consent. All study procedures complied with the ethical principles of the Declaration of Helsinki and were approved by the NCHS Research Ethics Review Board. We excluded participants without diagnostic data for obstructive sleep apnea (OSA) (n = 10,286) or environmental toxicological measurements (n = 2093). The final analytical cohort comprised 13,161 (6,431 males and 6,730 females). Figure 1 illustrates the flow of participant selection. Flowchart of study participants from NHANES 2015–2020. Abbreviations: NHANES, National Health and Nutrition Examination Survey; OSA, obstructive sleep apnea. Outcome definition OSA status was determined using a validated tripartite questionnaire [17]. Participants were classified as OSA-positive if they met one or more of the following criteria: excessive daytime sleepiness: self-reported sleepiness occurs 16–30 days per month despite obtaining 7 hours of sleep per night. Respiratory disturbance: Frequent episodes of gasping, snorting, or cessation of breathing (3 times per week). Frequent Snoring: Snoring episodes reported 3 times per week. Exposure assessment We analyzed a panel of 85 environmental toxicants (ETs) spanning 6 chemical classes (Supplementary Table S1), measured in blood or urine samples from NHANES subsample A. Urinary ET concentrations were adjusted for creatinine using the following formula: Adjusted concentration (ng/mg creatinine) = Measured concentration (ng/mL) / Urine creatinine (mg/dL) × 100. ETs with >90% of samples below the detection limit (BDL) or >40% missing data were excluded. Dichotomization: ETs with values of 30-90% BDL were dichotomized (non-ETs coded 0). Log-transformation: ETs with <30% BDL values were logarithmically transformed for analysis [18]. Missing data for each exposure are reported in Supplementary Table S1. Following rigorous quality control, 56 ETs were retained in 10 categories for primary analysis. Measurement of covariates Potential confounder variables were measured using established epidemiological definitions and standardized protocols. The data were collected through structured questionnaires, physical examinations, and laboratory tests, using standardized questionnaires and protocols from NHANES. Participants were divided into those aged <65 years and those aged 65 years or older. Biological sex was classified as male or female. Body mass index (BMI) was calculated as weight (kg)/height2 (m) and classified into three groups: normal weight (<25 kg/m2), overweight (25–30 kg/m2), and obese (30 kg/m2). Race/ethnicity was grouped into four categories: Mexican American, non-Hispanic white, non-Hispanic Black, and other races. The household income-to-poverty ratio (PIR) was used to assess socioeconomic status, with participants classified as “poor” (PIR <1.0) or “not poor” (PIR ≥1.0) based on federal poverty thresholds. Smoking status was defined based on the question, “Have you smoked at least 100 cigarettes in your lifetime?,” with affirmative respondents categorized as smokers (current or former) and others as nonsmokers. Alcohol consumption was defined as drinking 12 alcoholic beverages annually, with participants classified as drinkers or nondrinkers. Hypertension was diagnosed if participants met any of the three criteria: an average systolic blood pressure of 140 mmHg or diastolic blood pressure of 90 mmHg (based on three measurements at the Mobile Examination Center), self-reported physician-diagnosed hypertension, or current use of antihypertensive medications. Coronary heart disease (CHD) was identified by self-reported medical histories of medically diagnosed conditions such as angina pectoris, myocardial infarction, coronary artery disease, heart failure, or stroke. Diabetes mellitus was confirmed if the participants met at least one criterion: self-reported clinician diagnosis, active use of glucose-lowering medications, or a fasting blood glucose (FBG) of 7.0 mmol/L or a glycated hemoglobin (HbA1c) of 6.5%. All variables were collected systematically through structured questionnaires, physical examinations, and laboratory tests to ensure consistency with previous cohort studies. Statistical analysis Multiple Imputation by Chained Equations (MICE) was utilized to impute missing data for all variables, including covariates and exposure variables. We implemented an Exposome-wide Association Study (ExWAS) framework to systematically evaluate ETs, analyzing each exposure’s independent effect while controlling for potential confounding. Specifically, separate multivariable logistic regression models were fitted for each environmental toxicant, with OSA status as the binary outcome. The analytical workflow utilized the specialized rexposome package for exposure-wide association testing, supplemented by restricted cubic spline (RCS) regression with four knots to characterize potential nonlinear exposure-response relationships. All analytical models incorporated prespecified covariates based on theoretical relevance to exposure-outcome pathways. All hypothesis tests were two-tailed. To account for multiple comparisons, we applied Benjamini–Hochberg false discovery rate (FDR) correction to the p-values from the ExWAS models, with statistical significance defined as FDR-adjusted q < 0.05. The complete analytical workflow was executed in the statistical environment R (version 4.4.3), using a suite of validated packages: rexposome for exposure-wide analyses, mice for multivariate imputation, and ggplot2 for graphical representations. Core data management and workflow operations were facilitated through foreign, dplyr, and purrr packages. Results Among the 13,161 participants, 6,431 (49%) were men and 6,730 (51%) were women. 3,030 (23%) were 65 years old, while 10,131 (77%) were <65 years old. Racial and ethnic distribution was as follows: 1,798 (14%) were Mexican American, 3,256 (25%) were non-Hispanic black, 4,430 (34%) were non-Hispanic white, and 3,677 (28%) were of other races. The baseline data before handling missing values are shown in Table 1. Table 1.Participant characteristics.VariableOverall N = 13,161aNo OSAN = 6,718aOSAN = 6,443ap-ValuebAge <0.001<6510,131 (77%)5,326 (79%)4,805 (75%) ≥653,030 (23%)1,392 (21%)1,638 (25%) Gender <0.001Male6,431 (49%)3,024 (45%)3,407 (53%) Female6,730 (51%)3,694 (55%)3,036 (47%) Race 0.002Mexican American1,798 (14%)845 (13%)953 (15%) Non-Hispanic White4,430 (34%)2,274 (34%)2,156 (33%) Non-Hispanic Black3,256 (25%)1,673 (25%)1,583 (25%) Other Race3,677 (28%)1,926 (29%)1,751 (27%) Education <0.001Below high school2,663 (20%)1,262 (19%)1,401 (22%) High School or above10,498 (80%)5,456 (81%)5,042 (78%) PIR 0.037<12,696 (20%)1,425 (21%)1,271 (20%) ≥110,465 (80%)5,293 (79%)5,172 (80%) Marital <0.001No5,093 (39%)2,813 (42%)2,280 (35%) Yes8,068 (61%)3,905 (58%)4,163 (65%) BMI <0.001<253,714 (28%)2,556 (38%)1,158 (18%) 25–304,118 (31%)2,113 (31%)2,005 (31%) ≥305,329 (40%)2,049 (31%)3,280 (51%) Drinking 0.4No4,364 (33%)2,204 (33%)2,160 (34%) Yes8,797 (67%)4,514 (67%)4,283 (66%) Smoking <0.001No7,894 (60%)4,334 (65%)3,560 (55%) Yes5,267 (40%)2,384 (35%)2,883 (45%) Diabetes mellitus <0.001No4,097 (31%)2,273 (34%)1,824 (28%) Yes9,064 (69%)4,445 (66%)4,619 (72%) Hypertension <0.001No7,612 (58%)4,439 (66%)3,173 (49%) Yes5,549 (42%)2,279 (34%)3,270 (51%) Coronary heart disease <0.001No11,660 (89%)6,097 (91%)5,563 (86%) Yes1,501 (11%)621 (9.2%)880 (14%) Abbreviations: PIR, ratio of family income to poverty; BMI, body mass index; CHD, coronary heart disease; OSA, obstructive sleep apnea. an (%). bPearson’s Chi-squared test. Participant characteristics. Abbreviations: PIR, ratio of family income to poverty; BMI, body mass index; CHD, coronary heart disease; OSA, obstructive sleep apnea. an (%). bPearson’s Chi-squared test. Among the 85 toxicants, 56 with sufficient data were included. 18 were dichotomously analyzed based on detection limits, and 38 were normalized (Supplementary Table S1). The 56 toxicants exhibited varying degrees of correlation, with coefficients ranging from 0.16 to 0.94 (Figure 2a). ExWAS analysis, accounting for possible confounders, revealed nine ETs across two categories that were positively associated with OSA symptoms (Table 2; Figure 2b). Specifically, these categories included metals (two types) and VOC metabolites (seven types). For heavy metals, both blood cadmium (adjusted OR: 1.13, 95% CI: 1.04–1.23) and urinary cadmium (adjusted OR: 1.45, 95% CI: 1.18–1.77) showed positive relationships. Among volatile organic compound (VOC) metabolites, elevated blood levels of 2,5-dimethylfuran (adjusted OR: 1.25, 95% CI: 1.08–1.45), ethylbenzene (adjusted OR:1.22, 95% CI: 1.08–1.37), furan (adjusted OR: 1.26, 95% CI: 1.08–1.46) and o-xylene (adjusted OR: 1.24, 95% CI: 1.11–1.40) were significantly associated with OSA (all FDR q-value < 0.05). Additionally, three urinary cysteine-derived metabolites: N-acetyl-S- (4-hydroxy-2-methyl-2-butenyl) -L-cysteine, N-acetyl-S- (4-hydroxy-2-butenyl) -L-cysteine, and N-acetyl-S- (N-methylcarbamoyl) -L-cysteine demonstrated significant associations with the risk of OSA. Correlation of the environmental toxicants and associations between environmental toxicants and obstructive sleep apnea OSA. (a) Lower triangle correlation matrix of environmental toxicants. (b)Volcano plot of associations between environmental toxicants and OSA symptoms. Abbreviations: FDR, false discovery rate; OSA, obstructive sleep apnea. The x-axis shows individual toxicants; the y-axis shows –log10(p-value). The horizontal dashed line indicates FDR-corrected significance threshold (q < 0.05). Table 2.Associations between environmental toxicants and obstructive sleep apnea symptoms: an exposome-wide association analysis.Toxicant exposureLabelProcessingModel1 or (95%CI) p Model2 or (95%CI) p Arsenic ArsenicURXUASlog(In)1.001(1.000–1.002)0.341.002(1.000–1.003)0.1Ethylene Oxide Ethylene OxideLBXEOA2 categories1.001(1.000–1.001)0.181.001(1.000–1.001)0.14Metals AntimonyURXUSBlog(In)0.98(0.70–1.39)0.961.03(0.72–1.47)0.9BariumURXUBAlog(In)0.99(0.96–1.01)0.350.99(0.97–1.01)0.64CadmiumLBXBCDlog(In)1.16(1.08–1.24)p < 0.0011.13(1.04–1.23) 0.02 CadmiumURXUCDlog(In)1.32(1.11–1.57)0.011.45(1.18–1.77) 0.01 CesiumURXUCSlog(In)1.01(0.99–1.03)0.371.02(1.004–1.05)0.09ChromiumLBXBCR2 categories0.91(0.81–1.01)0.180.93(0.83–1.05)0.48CobaltURXUCOlog(In)0.98(0.94–1.03)0.570.98(0.94–1.03)0.66CobaltLBXBCOlog(In)0.94(0.87–1.02)0.280.96(0.89–1.04)0.59LeadLBXBPBlog(In)1.06(1.02–1.09)0.0031.01(0.98–1.05)0.61LeadURXUPBlog(In)0.97(0.90–1.06)0.620.97(0.89–1.06)0.67ManganeseLBXBMNlog(In)0.99(0.98–1.000)0.121.001(0.99–1.01)0.89ManganeseURXUMN2 categories1.15(1.02–1.30)0.081.15(1.01–1.31)0.11MercuryLBXTHGlog(In)1.01(0.99–1.02)0.451.02(1.004–1.04)0.09MercuryURXUHGlog(In)1.22(1.09–1.37)0.0031.17(1.03–1.32)0.07MolybdenumURXUMOlog(In)0.999(0.998–1.001)0.411.000(0.998–1.001)0.72SeleniumLBXBSElog(In)1.001(0.999–1.002)0.431.000(0.999–1.002)0.89ThalliumURXUTLlog(In)0.57(0.39–0.83)0.010.84(0.57–1.24)0.61TinURXUSNlog(In)0.99(0.97–1.02)0.660.97(0.95–1.000)0.14TungstenURXUTUlog(In)0.75(0.57–0.997)0.120.79(0.60–1.05)0.25Nicotine metabolites HydroxycotinineLBXHCT2 categories1.09(1.02–1.17)0.051.03(0.95–1.12)0.63CotinineLBXCOT2 categories0.998(0.93–1.07)0.980.98(0.90–1.06)0.68Perchlorate, nitrate and thiocyanate NitrateURXNO3log(In)1.000(1.000–1.000)0.011.000(1.000–1.000)0.7PerchlorateURXUP8log(In)0.99(0.98–1.002)0.30.996(0.99–1.005)0.6ThiocyanateURXSCNlog(In)1.000(1.000–1.000)0.431.000(1.000–1.000)0.16VOC metabolite 1,4-DichlorobenzeneLBXVDB2 categories1.04(0.95–1.14)0.521.04(0.94–1.15)0.642,5-DimethylfuranLBX2DF2 categories1.27(1.12–1.43)0.0021.25(1.08–1.45) 0.02 2-Aminothiazoline-4-carboxylic acidURXATClog(In)1.000(1.000–1.000)0.571.000(1.000–1.000)0.9992-Methylhippuric acidURX2MHlog(In)1.000(1.000-1.000)0.511.000(1.000–1.000)0.643- and 4-Methylhippuric acidURX34Mlog(In)1.000(1.000–1.000)0.491.000(1.000–1.000)0.61BenzeneLBXVBZ2 categories1.04(0.94–1.14)0.571.01(0.91–1.12)0.9BromodichloromethaneLBXVBM2 categories0.92(0.81–1.04)0.320.93(0.82–1.06)0.56ChloroformLBXVCF2 categories0.92(0.84–1.01)0.190.93(0.84–1.02)0.29DibromochloromethaneLBXVCM2 categories1.04(0.90–1.20)0.661.01(0.87–1.18)0.9EthylbenzeneLBXVEB2 categories1.26(1.14–1.40)p < 0.0011.22(1.08–1.37) 0.01 FuranLBXVFN2 categories1.27(1.11–1.44)0.0031.26(1.08–1.46) 0.02 m-/p-XyleneLBXVXY2 categories1.19(1.08–1.31)0.0031.13(1.02–1.25)0.09Mandelic acidURXMADlog(In)1.000(1.000–1.000)0.321.000(1.000–1.000)0.53N-Acetyl-S-(4-hydroxy-2-methyl-2-butenyl)-L-cysteineURXIPM3log(In)1.003(1.001–1.01)0.011.004(1.002–1.01) 0.01 N-Acetyl-S-(1-cyano-2-hydroxyethyl)-L-cysteineURXCYHA2 categories1.19(1.03–1.36)0.051.21(1.02–1.43)0.1N-Acetyl-S-(2-carbamoyl-2-hydroxyethyl)-L-cysteineURXGAM2 categories1.23(1.10–1.38)0.0031.14(1.003–1.29)0.13N-Acetyl-S-(2-carbamoylethyl)-L-cysteineURXAAMlog(In)1.000(0.999–1.000)0.361.000(0.999–1.001)0.89N-Acetyl-S-(2-carboxyethyl)-L-cysteineURXCEMlog(In)1.001(1.000–1.001)0.021.000(1.000–1.001)0.34N-Acetyl-S-(2-cyanoethyl)-L-cysteineURXCYMlog(In)1.000(1.000–1.001)0.511.001(1.000–1.001)0.27N-Acetyl-S-(2-hydroxyethyl)-L-cysteineURXHEM2 categories0.97(0.87–1.09)0.651.01(0.90–1.15)0.89N-Acetyl-S-(2-hydroxypropyl)-L-cysteineURXHP2log(In)1.000(1.000–1.000)0.631.000(1.000–1.001)0.49N-Acetyl-S-(3,4-dihydroxybutyl)-L-cysteineURXDHBlog(In)1.000(1.000–1.000)0.981.000(1.000–1.000)0.64N-Acetyl-S-(3-hydroxypropyl)-L-cysteineURXHPMlog(In)1.000(1.000–1.000)0.171.000(1.000–1.000)0.1N-Acetyl-S-(3-hydroxypropyl-1-methyl)-L-cysteineURXPMMlog(In)1.000(1.000–1.000)0.091.000(1.000–1.000)0.09N-Acetyl-S-(4-hydroxy-2-butenyl)-L-cysteineURXMB3log(In)1.01(1.002–1.01)0.011.01(1.003–1.01) 0.01 N-Acetyl-S-(benzyl)-L-cysteineURXBMAlog(In)0.999(0.997–1.001)0.460.999(0.997–1.001)0.61N-Acetyl-S-(N-methylcarbamoyl)-L-cysteineURXAMClog(In)1.001(1.000–1.001)p < 0.0011.001(1.000–1.001) 0.002 N-Acetyl-S-(n-propyl)-L-cysteineURXBPMlog(In)0.999(0.997–1.001)0.430.999(0.997–1.001)0.62o-XyleneLBXVOX2 categories1.30(1.17–1.44)p < 0.0011.24(1.11–1.40) 0.01 Phenylglyoxylic acidURXPHGlog(In)1.000(1.000–1.000)0.651.000(1.000–1.000)0.89Abbreviations: OR, odds ratio; CI, confidence interval; VOC, volatile organic compound; Model1 = unadjusted; Model2 = adjusted for age, sex, race, PIR, education, marital status, BMI, drinking, smoking, diabetes, hypertension, and coronary heart disease. Bold values indicate statistical significance after FDR correction (q < 0.05) in Model 2. Associations between environmental toxicants and obstructive sleep apnea symptoms: an exposome-wide association analysis. Abbreviations: OR, odds ratio; CI, confidence interval; VOC, volatile organic compound; Model1 = unadjusted; Model2 = adjusted for age, sex, race, PIR, education, marital status, BMI, drinking, smoking, diabetes, hypertension, and coronary heart disease. Bold values indicate statistical significance after FDR correction (q < 0.05) in Model 2. Stratified analysis revealed differential associations between toxicant types and OSA symptoms by sex and age. Specifically, more types showed significant associations in women (6 toxicants) and in individuals under 65 years (13 toxicants) compared to men and those aged 65 or older, respectively (Figure 3). Volcano plot of associations between environmental toxicants and OSA symptoms adjusted for confounders: stratified analysis by age and sex. (a) Stratified analysis by age (<65 vs. ≥65 years). (b) Stratified analysis by sex (female vs. male). Abbreviations: FDR, false discovery rate; OSA, obstructive sleep apnea. The horizontal dashed line indicates the FDR-corrected significance threshold (q < 0.05). RCS curves were used to investigate the relationships between five ETs and OSA, adjusting for all relevant covariates. The results indicated non-linear, nonmonotonic relationships between blood or urine cadmium levels and the three cysteine-derived metabolites, as well as the prevalence of OSA (Supplementary Table S2). Discussion This cross-sectional analysis found nine environmental toxicants (ETs) to be significantly linked to an increased risk of OSA symptoms. In particular, our findings indicate that women and people under 65 years of age may be more susceptible to the effects of ETs compared to men and older adults. These results underscore the importance of increased exposure to ETs throughout the population as a strategy to mitigate the risk of OSA. In our comprehensive cross-sectional analysis of six categories of environmental toxicants, substantial evidence revealed that metal and VOCs metabolites are significantly associated with an increased risk of OSA. Our discovery that blood cadmium exposure is associated with OSA symptoms aligns with existing scholarly findings [15,19], underscoring the need for more in-depth research into the role of environmental toxicants in OSA etiology. A growing body of evidence shows that exposure to cadmium is associated with sleep disturbances [20]. For example, Kim et al. recently indicated that inhaling cadmium-laden particles can negatively impact Pittsburgh Sleep Quality Index scores [21]. Given its ubiquity in the environment, cadmium poses a significant public health threat [22]. The general population is exposed primarily to cadmium through tobacco smoke, diet, and workplace settings [23,24]. In particular, smoking is the most substantial source of exposure to cadmium in humans, with smokers exhibiting higher concentrations in their blood and kidneys [23]. This is attributed to tobacco that contains various metal impurities, particularly cadmium, which tend to accumulate within the body [25]. The severity of cadmium poisoning symptoms depends on blood cadmium levels, potentially leading to both acute and chronic poisoning, beyond its well-documented neurotoxic properties [26]. Cadmium is classified as a type I carcinogen. It also acts as an inducer of pro-inflammatory cytokines, triggering chronic inflammatory responses [27]. Epidemiological and animal studies suggest that exposure to cadmium promotes oxidative stress and cytokine production [28]. From a public health perspective, monitoring blood and urinary cadmium levels is important. To the best of our knowledge, our study is the first to explore the relationship between VOCs metabolites and OSA symptoms comprehensively. This relationship may be indicative of metabolic processes within the body and the generation of specific toxic by-products. Our results indicate that blood 2,5-dimethylfuran, ethylbenzene, furan, and three urinary cysteine-derived metabolites are associated with OSA symptoms, with the association of ethylbenzene consistent with previous research [29]. A recent study has emphasized the potential of VOCs as biomarkers to detect OSA, showing their correlation with the AHI index and suggesting their use as a noninvasive screening tool to improve clinical management [14]. VOCs are pervasive in the environment, originating from a broad spectrum of natural processes and human activities [30]. They are released primarily by industrial activities, vehicle exhaust emissions, waste incineration, combustion systems, fuels, and chemical manufacturing processes [31]. Inhalation of VOCs can irritate the mucous membranes of the respiratory tract, potentially leading to respiratory diseases [32], and is known to have various detrimental effects on human health, including significant impacts on inflammation [33,34]. Given that oxidative stress and subsequent inflammation, induced by ETs, are pivotal contributors to the pathogenesis of numerous human diseases [35], and considering that OSA is associated with chronic inflammation and immune activation [36,37]. It is plausible that inflammation serves as an intermediary linking environmental chemical exposure to the development of OSA. This study leveraged a large-scale, nationally representative sample of US adults from the National Health and Nutrition Examination Survey (NHANES) with extensive covariate adjustments, thereby supporting the robustness of the conclusions. However, several limitations should be acknowledged. First, the cross-sectional nature of NHANES data precludes the determination of causality between ETs and OSA risk. Second, OSA status in this study was determined using a self-reported questionnaire rather than objective polysomnography. This approach likely introduces significant misclassification bias, as symptom-based definitions have imperfect sensitivity and specificity. Consequently, the observed 49% prevalence should be interpreted as identifying individuals at high risk for OSA rather than confirmed disease. We acknowledge this as a major limitation of the present study. Third, although covariates were selected based on prior literature, we acknowledge the lack of formal justification and absence of sensitivity analyses to test robustness against high exposure correlations. These constitute additional limitations of the present cross-sectional exposome-wide study. From a preventive perspective, reducing population exposure to identified toxicants—particularly cadmium and VOCs—through tobacco control, industrial emission reduction, and improved ventilation could potentially lower OSA risk, although causal inference requires longitudinal studies. Regional differences in environmental exposures should also be considered when generalizing our findings to non‑US settings. Conclusions In this cross-sectional study, we found that metal and VOC metabolites are associated with OSA symptoms. Susceptibility to ETs varies across populations, particularly among women and individuals under 65 years of age, compared with men and older adults. Our research emphasizes the need for further investigation into key environmental toxicants and offers new perspectives for future prevention and treatment research. Supplementary Material Supplementary material.xlsx STROBE_Checklist.pdf Acknowledgments DeepSeek (DeepSeek-V3.2, 2025) was used solely for language polishing and formatting assistance; it was not used for data analysis, interpretation, or scientific conclusions. The authors take full responsibility for the content of the manuscript. Funding Statement This research was funded by the Science and Technology Program of Guangzhou [201903010024]. Ethical approval All study procedures complied with the ethical principles of the Declaration of Helsinki and were approved by the NCHS Research Ethics Review Board. Authors contributions W.X.L.: Conceptualization, Writing—original draft, Formal analysis. P.L.: Methodology, Data curation. Z.Y.W.: Software, Validation. Y.L.: Investigation. J.C.: Investigation. H.L.: Supervision, Writing—review and editing. Q.T.Y.: Supervision, Writing—review and editing. J.Y.: Funding acquisition, Supervision, Writing—review and editing. All authors read and approved the final manuscript. Informed consent All participants provided written informed consent (NCHS IRB/ERB Protocol Number: Protocol #2011-17, Protocol #2018-01). Disclosure statement The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties. Data availability statement The data supporting this article are included in the Supplementary Information. Supplementary information: Tables S1 and S2. References Associated Data Supplementary Materials Supplementary material.xlsx STROBE_Checklist.pdf Data Availability Statement The data supporting this article are included in the Supplementary Information. Supplementary information: Tables S1 and S2.