Introduction

Chronic kidney disease (CKD) is a common disease encountered in clinical practice, which is the consequence of continuous deterioration of renal excretory function caused by various renal diseases. Replacement treatment including dialysis and renal transplant is the ultimate solution. CKD affects approximately 10–15% of the global population (Matsushita et al. 2022), and significantly increases the risk of adverse outcomes, such as stroke, myocardial infarction, and heart failure, constituting the main cause of death from non-communicable diseases (Webster et al. 2017). Data from the Global Burden of Disease Study 2019 revealed that, in 2019 alone, there were 19.0 million incident CKD patients, 697 million prevalent CKD patients, 1.43 million deaths caused by CKD, and 41.5 million disability-adjusted life years (DALYs) attributed to CKD globally. These numbers represent a significant increase compared to the figures reported in 2010, with incident patients, prevalent patients, deaths, and DALYs rising by 33.9%, 24.7%, 28.8%, and 20.1%, respectively (GBD 2019 Diseases and Injuries Collaborators. 2020). In addition, CKD can exacerbate the symptom burden and seriously jeopardize health-related quality of life (Fletcher et al. 2022).

As a reactive epoxide, ethylene oxide (EO) is mass produced in factories and is widely applied in the production of ethylene glycols, glycol ethers, ethanolamine, ethoxylate, and acrylonitrile. Besides, it is utilized for sterilization of food, fragrances, cosmetics, medical devices, etc. (Kirman et al. 2021). Individuals can access EO through a number of ways. For the general population, endogenous production including gut bacterial synthesis and formation of EO in the liver from ethylene is an important pathway of EO exposure (Bogen et al. 2019). Exogenous pathways, including metabolic production of EO following exposure to exogenous ethylene from various sources, such as tobacco smoke and incomplete combustion of fossil fuels in vehicles, forest fires, fruit and vegetable ripening, and volcanic eruptions, as well as direct exposure to environments containing EO (such as smoking and industrial facilities) are also major sources of EO exposure (Kirman and Hays 2017; Schettgen et al. 2002).

EO is a highly reactive organic substance that can alkylate proteins, RNA, and DNA. Hemoglobin adduct N-(2-hydroxyethyl) valine (HbEO) has been identified as a useful biomarker to characterize total EO exposure of all exposure pathways (Boogaard et al. 1999). Although no definite correlation between EO exposure and malignancies has been found in human epidemiological data, EO has been categorized as a Group I human carcinogen by the International Agency for Research on Cancer based on a thorough evaluation of available scientific evidence, including animal experiments, genotoxicity data, and associated mode of action analyses (Lynch et al. 2022). In the toxicological profile for EO, animal studies revealed that high doses of EO exposure caused damages in multiple systems, including respiratory, hematological, neurological, reproductive, and renal systems (Toxicological Profile for Ethylene Oxide. 2022). Recent analyses from NHANES have demonstrated that high levels of EO exposure are independently associated with metabolic and chronic diseases including dyslipidemia (Zhu et al. 2022), diabetes (Guo et al. 2021), hypertension (Wu et al. 2022), cardiovascular disease (Zeng et al. 2021), and asthma (Li et al. 2023). However, data on the association between HbEO and the prevalence and prognosis of CKD is limited. In this study, we set out to evaluate the impact of HbEO on the prevalence and prognosis of CKD among US adults.

Methods

Study population

NHANES is an ongoing project designed to examine the health and nutritional status of US citizens. It recruits approximately 5000 representative individuals nationwide each year, and the National Center for Health Statistics (NCHS) is responsible for the implementation of the project. Data are collected through face-to-face interviews and physical examinations. The interviews were performed by trained researchers, and the physical examinations were conducted at a mobile examination center (MEC). Detailed information about the NHANES project was available on its website. Participants in the project were informed in detail and signed a written informed consent form. The protocol of NHANES project was reviewed and approved by the institutional review board.

In the present study, data were extracted from NHANES from 2013 to 2016, among which a random selection of study participants who were aged ≥ 6 years was made for the measurement of HbEO, comprising one-third of the total sample (n = 4860). After excluding participants who did not meet the inclusion criteria, 2900 participants were included in the study (as illustrated in Fig. 1). The exclusions were made based on the following criteria: absence of subsample A weights (n = 586), HbEO values below the detectable limit (n = 106), extreme energy intake (defined as daily total energy intake < 500 or > 8000 kcal for males, and < 500 or > 5000 kcal for females (n = 34), pregnant (n = 38), and missing data for covariates (n = 540).

Fig. 1
figure 1

Study population flowchart. BMI, body mass index; CKD, chronic kidney disease; HbEO, hemoglobin adducts of ethylene oxide; NHANES, National Health and Nutrition Examination Survey

Full size image

Measurement of HbEO

The collection, processing, and transportation of specimens were performed according to standard procedures. The Division of Laboratory Sciences at the National Center for Environmental Health (NCEH) was responsible for the measurement of HbEO, and detailed information on the laboratory analyses procedures is illustrated in the relevant website. The modified Edman reaction was utilized to test the HbEO levels in whole blood or erythrocytes, with the high-performance liquid chromatography coupled with tandem mass spectrometry (HPLC-MS/MS) method applied. The lower limit of detection of HbEO is 12.9 pmol/g Hb, and 109 participants with HbEO values below detectable limit were excluded in this analysis. The accuracy and precision of all assays met the NCEH Laboratory Science Division Quality Control and Quality Assurance Standards (CDC 2023).

Definition of CKD and assessment of prognosis of CKD

The DxC800 modular chemistry analyzer was applied to measure serum creatinine and urine creatinine levels by the Jaffe rate method. Urinary albumin concentrations were determined using the method described by Chavers et al., and it was recommended to use solid-phase fluorescent immunoassay for detection (Chavers et al. 1984). The chronic kidney disease epidemiology collaboration equation was used in this study to calculate the estimated glomerular filtration rate (eGFR) (Levey et al. 2009). CKD was defined as having an eGFR < 60 mL/min/1.73 m2 or a urine albumin to creatinine ratio (UACR) ≥ 30 mg/g, in accordance with the 2012 clinical practice guidelines for the evaluation and management of kidney disease (Levin et al. 2013). The prognosis evaluation system initiated by KDIGO was used for the assessment of prognosis of CKD, and four risk grades according to the levels of eGFR and UACR were constructed. The calculation formula of eGFR and the scoring system were detailed in Supplementary file 1.

Assessment of covariates

Demographic information (including age, gender, ethnicity), social-economic status (including marital status, education level, household income), health status (including smoking, alcohol use, self-reported health status, physical activity, comorbidities such as diabetes, hypertension, hyperlipidemia, cardiovascular disease, chronic obstructive pulmonary disease, cancer), and dietary habits (energy intake) were determined through standardized questionnaire home interviews. Measurements of blood pressure, body weight and height were conducted at the MEC using standard procedures. In NHANES, dietary intake was assessed via a first 24-h dietary recalls (Stumbo 2008). Energy deficiency is defined as daily energy intake < 2400 kcal/d for adult males and < 1600 kcal/d for adult females (US Department of Health and Human Services and US Department of Agriculture. 2015). Physical activity was quantified in metabolic equivalent (MET) minutes per week and subdivided into insufficient physical activity (< 600 MET min/week) and adequate physical activity (≥ 600 MET-min/week) (Armstrong and Bull 2006). Urinary cadmium was measured using an inductively coupled plasma-mass spectrometry from a single urine specimen.

Statistical analysis

RStudio software (version 2023.03.0 + 386, Posit Software, PBC) was used to conduct statistical analyses, and a 2-tailed P value < 0.05 was statistically significant. As NHANES is designed as a complex, multi-stage, probability sampling study to select representative US individuals nationwide, sample weights, clustering, and stratification must be considered when conducting statistical analysis (Centers for Disease Control and Prevention. 2006). A 2-year weight (2013–2014, 2015–2016) was generated to obtain weighted percentages adjusted to the US adult population.

All participants were grouped into three categories according to tertile of HbEO: T1, 13.2–27.3 pmol/g Hb; T2, 27.3–47.0 pmol/g Hb; T3, 47.0–1781.2 pmol/g Hb. One-way analysis of variance was conducted for comparisons of continuous variables, and the Rao-Scott χ2 test was performed for comparisons of categorical variables (Rao and Scott 1984). Survey-weighted generalized linear models and proportional odds models were constructed to estimate the influence of HbEO levels on the prevalence and prognosis of CKD, respectively. For these two analytical methods, three models were constructed, respectively: the first model was adjusted for age, gender, and race; the second model was further adjusted for body mass index (BMI), marital status, education levels, household income, smoking, alcohol use, physical activity, energy intake, health status, and urinary cadmium; and the third model was additionally adjusted for comorbidities including cardiovascular disease, hypertension, diabetes mellitus (DM), hyperlipidemia, chronic obstructive pulmonary disease, and cancer. HbEO was natural logarithm transformed and brought into regression models as a continuous variable for analysis. The Akaike Information Criterion (AIC) values were calculated to estimate the fitting degree of models by using the Rao-Scott approximation to the weighted loglikelihood (Lumley and Scott 2015). Subgroup analyses were performed by age, gender, BMI, race, smoking, and alcohol use. The results of above regression analysis were expressed as odds ratios (ORs) for prevalence of CKD and proportional odds ratios (PORs) for prognosis of CKD, along with their corresponding 95% confidence interval (CIs).

Restricted cubic spline (RCS) analysis was performed to depict the relationship between HbEO levels and CKD. The analysis was based on the third survey-weighted generalized linear models with adjustments for all covariates mentioned above, among which ln-transformed HbEO was included as a continuous variable and CKD was brought into the models as a binary outcome variable. Based on considerations of AIC values, P values for nonlinearity tests, and the interpretability of the resulting curves, 3 knots were chosen to depict the curves. RCS analyses were performed by stratifying participants according to smoking status. Sensitivity analyses were conducted to confirm the reliability of our research results. First, to eliminate the potential impact of cancer and premature death on the association between HbEO and CKD, analyses were performed after excluding individuals who had cancer (n = 269) or died within first year of follow-up (n = 12). In addition, as nicotine metabolites may confound the results, cotinine and hydrocotinine were additionally included in multivariate regression models, to reduce their confounding effects the correlation of HbEO with CKD.

Results

Table 1 illustrates the baseline characteristics of participants. The weighted mean age was 47.7 years, of which 50.3% were male. The weighted median (interquartile range) of HbEO was 31.3 (23.1–60.3) pmol/g Hb. Compared to participants with low HbEO concentrations, those with high HbEO concentrations were younger, more likely to be male, non-Hispanic Black, single, less educated, and had lower income. Besides, they smoked and drank more frequently, and had worse health status.

Table 1 Baseline characteristics of participants according to tertile of HbEOa
Full size table

A total of 491 participants (16.9%) were diagnosed with CKD, and its prevalence rate was notably higher in elevated HbEO groups. Regarding to the prognosis of CKD, 320 participants (11.1%) and 153 participants (5.31%) were identified as being at moderate or high/very high CKD risk, and the proportion of moderate CKD risk was higher among participants with high HbEO levels compared to those with low HbEO levels. In age, sex, and race adjusted generalized linear regression models, compared to the first tertile of HbEO, ORs (95% CIs) for CKD in the second and third tertile were 1.54 (1.08, 2.19) and 1.81 (1.27, 2.57). After adjustment for covariates including BMI, education level, household income, marital status, physical activity, energy intake, smoking, drinking, self-reported health status, chronic diseases, and urinary cadmium, elevated HbEO levels were still notably linked to CKD, and ORs (95% CIs) for CKD in the second and third tertile were 1.46 (0.85, 2.50) and 1.69 (1.00, 2.85). Regarding to prognosis of CKD, after adjustment for covariates in Model 3, elevated HbEO levels were significantly associated with high risk grade, and PORs (95% CIs) in the second and third tertile were 1.37 (0.94, 1.99) and 1.58 (1.10, 2.26), with the first tertile as the reference (Table 2). When ln-transformed HbEO was analyzed as a continuous variable in regression models, the OR (95% CI) for CKD was 1.24 (0.97, 1.58) and the POR (95% CI) for CKD prognosis was 1.22 (1.01, 1.47). Including HbEO in the models can improve the models’ goodness of fit (Supplementary 2). Subgroup analysis indicated that smoking status had a prominent influence on the correlation of high HbEO levels with the prevalence of CKD (P for interaction 0.01). Among never smokers, the adjusted ORs (95% CIs) of CKD across HbEO tertiles were 1.00 (reference), 1.69 (0.95, 2.68) and 2.00 (1.04, 3.87), respectively, while this association lost significance in never smokers. No significant interactions were noticed in subgroups stratified by age, gender, ethnicity, BMI, and alcohol use (Fig. 2).

Table 2 Association between HbEO and CKD and prognosis of CKD in US adults participating in NHANES (2013–2016)
Full size table
Fig. 2
figure 2

Associations between natural logarithm of HbEO and CKD by restricted cubic spline regression models. A Total population; B ever smokers; C never smokers. HbEO, hemoglobin adducts of ethylene oxide

Full size image

The RCS analysis revealed a positive dose-response relationship between HbEO and prevalence of CKD for overall participants (P for nonlinearity < 0.05). In never smokers, elevated HbEO levels were linearly correlated with CKD in a positive way (P for nonlinearity > 0.05), whereas in ever smokers, the risk of CKD initially increased and then decreased with increasing HbEO levels (Fig. 3). In sensitivity analysis, consistent findings were obtained when excluding individuals who had cancer or died during the first follow-up year (Supplementary file 3). After additional adjustment for hydrocotinine and cotinine, the relationship between high HbEO levels and increased prevalence and adverse prognosis of CKD were maintained (Supplementary file 4).

Fig. 3
figure 3

Subgroup analysis of the association between HbEO and CKD. BMI, body mass index; CI, confidence interval; CKD, chronic kidney disease; HbEO, hemoglobin adducts of ethylene oxide; HR, hazard ratio

Full size image

Discussion

The present study explored the impact of HbEO on prevalence and prognosis of CKD in US adults from NHANES 2013–2016. Our findings are as follows. First, participants with high HbEO levels suffered from CKD more frequently, and the proportion of moderate CKD risk was higher among participants with high HbEO levels compared to those with low HbEO levels. After adjustment for covariates, high HbEO levels were significantly associated with elevated risk of CKD and adverse CKD-related prognosis. RCS analysis indicated a positive correlation between HbEO levels and prevalence of CKD. Second, the above findings were largely affected by smoking status, and the association was prominent in never smokers, but lost significance in ever smokers. Last, the sensitivity analysis confirmed the robustness of our above research findings.

At room temperature, EO exists in the form of gas, and human exposure to EO is mainly through inhalation. The primary metabolic pathways of EO are glutathione conjugation and hydrolysis, and both pathways are considered to be detoxifying (Filser and Klein 2018). EO is electrophilic by nature and can react with cellular macromolecules including proteins and DNA, to form adducts. HbEO can reflect the cumulative total exposure of all exposure pathways to EO and has been proven to be the best parameter to assess EO exposure (Boogaard et al. 1999). In NHANES, the modified Edman reaction using HPLC-MS/MS method was utilized to test HbEO levels in human whole blood or erythrocytes, which is a well-established method for the detection of EO and ensures the accuracy of the test results. There is rare data on EO exposure levels in the general populations. In a previous study conducted by Szwiec et al. (2020), it was found that residents living near facilities emitting EO had a geometric HbEO level of 35.0 pmol/g Hb, and nonsmokers had a lower HbEO level of 29.7 pmol/g Hb; besides, HbEO levels in nonsmoking participants living in a neighborhood approximately 0.8 km from one of the facilities were significantly higher compared to those living farther away (Szwiec et al. 2020). Our analysis revealed that the weighted median level of HbEO in US adults was 31.3 pmol/g Hb, and the main source of HbEO in this population was endogenous generation rather than inhalation of EO from background ambient air or breath air near EO use/production facilities, as no specific external NO exposures were identified (Kirman et al. 2021). Furthermore, our analysis demonstrated that ever smokers had notably higher HbEO levels compared to never smokers, which provides additional evidence that smoking is a major contributor to EO exposure.

HbEO covalently binds with nucleophilic groups in DNA, forming reaction products that have the potential to induce genetic mutations and contribute to health issues. The carcinogenic effects of EO have been established mainly through animal experiments. In rodents, EO exposure can significantly increase the incidence of numerous tumors in a concentration-dependent manner, including adenocarcinomas, lung neoplasms, brain tumors, lymphomas, and peritoneal mesothelioma (Zhu et al. 2022). However, epidemiological studies involving 11 independent populations from 7 countries, comprising individuals in various occupations with EO exposure, did not find a significant correlation between cumulative EO exposure and an increased incidence of breast cancer and lymphohematopoietic tumors (Lynch et al. 2022). Dose differences may be the main reason for this discrepancy as dose levels employed in animal studies were much higher than that in humans, even in individuals with occupational exposures. Apart from EO-induced carcinogenicity, animal experiments have provided evidence that high doses of EO exposure can cause damages in multiple systems, including respiratory, hematological, neurological, and reproductive systems. Regarding the renal system, in the study conducted by Hollingsworth et al. (1956), rats and guinea pigs exposed to EO vapor at 841 ppm exhibited kidney abnormalities such as enlargement, slight congestion, and cloudy swelling of convoluted tubules. Similarly, a study involving mice exposed to EO vapor for 14 weeks showed that renal tubular degeneration occurred in a dose-dependent manner, with higher concentrations of EO exposure (400 and 600 ppm) leading to increased incidence of renal necrosis (Toxicological Profile for Ethylene Oxide. 2022). The impact of HbEO on the prevalence and prognosis of CKD rarely has been evaluated. The present study is the first comprehensive analysis to explore the impact of HbEO on the prevalence and prognosis of CKD in US adults. Our analysis, which included 2900 participants from NHANES 2013–2016, revealed that high HbEO levels were significantly associated with the prevalence of CKD and worse prognosis in terms of CKD progression. Our study provides valuable insights into the potential role of EO in the pathogenesis and progression of CKD in the US population.

Our subgroup analysis revealed that the association between HbEO levels and the prevalence of CKD was significantly influenced by smoking status. Among individuals who had never smoked, we observed a linear positive correlation between HbEO levels and CKD prevalence. However, this correlation was not observed among individuals who had ever smoked or were current smokers, despite that they had higher HbEO levels compared to never smokers. These findings are consistent with a previous study that investigated the impact of EO exposure on the risk of DM. In that study, elevated HbEO levels were associated with a higher prevalence of DM, and this correlation was more pronounced in nonsmokers than in smokers (Guo et al. 2021). This finding may be explained by the followings. Firstly, smokers are exposed to a wide range of pollutants released during tobacco combustion, which may also have significant implications for the development of CKD. Hu et al. (2023) explored the impact of exposure to multiple serum metals on the risk of CKD in the elderly and found that concentrations of lead, cadmium, cobalt, and manganese were significantly associated with the risk of CKD. A recent study using data from the NHANES database revealed that elevated cadmium levels are associated with increased all-cause mortality among US adults with CKD, and smoking was the mainly source of cadmium exposure (Zhang et al. 2023). In the present study, urinary cadmium was included in the regression models, and the results showed that the association of HbEO and CKD was not affected. Nevertheless, the potential influence of other pollutants on the association cannot be completely removed due to a lack of relevant data collection. Additionally, co-exposure to other chemicals, both from smoking and other sources, can interact with EO exposure and potentially modify the risk of CKD. Besides, many behavioral or environmental factors that may be associated with increased risk of CKD were not fully adjusted and may contribute to the association between HBEO and CKD observed in never smokers. Furthermore, individual susceptibility may also play a vital role in determining the health effects of EO exposure. As a cross-sectional designed study, we cannot establish a cause-and-effect relationship between HbEO and CKD, and to verify this hypothesis, prospective studies investigating the influence of EO exposure on the incidence of CKD in both smokers and nonsmokers are warranted.

The exact mechanism by which elevated EO exposure increases the risk of CKD is not yet sufficiently understood. Apart from the tumor-promoting effects, data from animal studies have revealed that high-dose EO exposure can cause multi-system damages. However, there is currently no further in-depth mechanism research. Recent analyses based on NHANES data have shown that high levels of EO exposure are significantly associated with various metabolic and chronic diseases including dyslipidemia (Zhu et al. 2022), DM (Guo et al. 2021), hypertension (Wu et al. 2022), CVD (Zeng et al. 2021), and asthma (Li et al. 2023), and it is speculated that inflammatory activation caused by EO might be a common mechanism in mediating these associations. Inflammation is recognized as an important contributor to the incidence of CKD, and oxidative stress, proinflammatory cytokines, acidosis, chronic infections, and altered metabolism of adipose tissue play important roles in the development of CKD (Mihai et al. 2018). The correlation of high HbEO levels with increased prevalence and worse prognosis of CKD might be partially explained by activated inflammation and oxidative stress. Our present study only provides evidence of an association of HbEO and CKD, but cannot establish a causal relationship, besides, we also cannot provide further information on the underlying mechanisms behind this association. Further prospective cohort studies are still needed to clarify the relationship between EO exposure and the occurrence and prognosis of CKD, and in-depth research is required to unravel the specific mechanisms through which EO exposure may lead to renal injury and development of CKD.

One notable strength of this study is that it is the first prospective evaluation of the impact of EO exposure on CKD in a large US adult population. The data collected in the NHANES program was meticulously gathered by trained researchers, adding to the credibility and reliability of the results. The utilization of the well-established modified Edman reaction method for HbEO level testing by HPLC-MS/MS further enhances the reliability of the test results. Several limitations of the study should be acknowledged. Firstly, the cross-sectional design hinders the establishment of a causal association between high HbEO levels and increased CKD risk. To obtain conclusive results, it is crucial to conduct well-designed prospective cohort studies. Secondly, individuals are commonly exposed to various pollutants that may also be linked to the development of CKD (Liu et al. 2023). However, due to the lack of data on these specific substances, it is unavailable to adjust them in the regression models, which may cause significant impacts on current findings. Besides, we did not adjust for all other pollutants released during tobacco combustion, which may introduce confounding effects in subgroup analysis for individuals who were current or former smokers. Further research is needed to examine mixture exposure and determine the specific contribution of EO exposure to the risk of CKD. Thirdly, in our sensitivity analysis, although cotinine was used as a marker of smoking and included in regression models for adjustment, it is important to note that a significant number of individuals with high cotinine levels may be vapers rather than smokers. This distinction could blur the influence of toxicological exposures relevant to CKD. Fourthly, as the RCS model has high degrees of freedom when dealing with binary dependent variables, estimates near the boundary values may not be very reliable. Furthermore, measurement errors may exist in the process of data collection and were not accounted for in our analysis, which may also affect the accuracy of our results to some extent. Lastly, as the study was conducted based on the US population, caution should be exercised when generalizing the conclusions to other populations.

Conclusions

In conclusion, the present study reveals a significant association between high HbEO levels and increased risk of CKD, as well as a worse prognosis of CKD. The impact of smoking status on this correlation was significant, which remained significant in never smokers, but lost significance in smokers. Further studies are needed to confirm the impact of EO exposure on CKD and clarify the underlying mechanisms.