The National Health and Nutrition Examination Survey (NHANES) is a nationwide cross-sectional study administered by the National Center for Health Statistics (NCHS), which is affiliated with the Centers for Disease Control and Prevention (CDC). It utilizes a multistage random probability sampling method. The research ethics were approved by the NHANES review board of the Centers for Disease Control and Prevention. For this study, we drew upon data from three survey cycles of NHANES that took place between 2013 and 2018 to explore the link between PAEs exposure and CircS in adults. Following the screening of comprehensive examination records, laboratory results, and questionnaire information, a total of 2,519 participants were ultimately included in the analysis. Figure 1 shows the flowchart of participant inclusion and exclusion.

In the NHANES 2013–2018 dataset, 13 phthalate metabolites were identified, comprising Mono(carboxyisononyl) phthalate, Mono(carboxyisoctyl) phthalate, Mono-2-ethyl-5-carboxypentyl phthalate (MECP phthalate), Mono-n-butyl phthalate, Mono-(3-carboxypropyl) phthalate, Mono-ethyl phthalate, Mono-(2-ethyl-5-hydroxyhexyl) phthalate (MEHP phthalate), Cyclohexane 1,2-dicarboxylic acid monohydroxy isononyl ester (MHNCH), Mono-(2-ethyl)-hexyl phthalate, Mono-isobutyl phthalate, Mono-isononyl phthalate, Mono-(2-ethyl-5-oxohexyl) phthalate (MEOH phthalate), Mono-benzyl phthalate. Comprehensive details regarding the chemical measurement methodologies are accessible through the NHANES Laboratory Data module (https://wwwn.cdc.gov/nchs/nhanes/Default.aspx). The limits of detection (LOD) varied between 0.2 and 1.2 ng/mL. For the ensuing analyses, only phthalate metabolites with detection frequencies exceeding 60% were incorporated (15), prompting the exclusion of MHNCH, Mono-(2-ethyl)-hexyl phthalate, and Mono-isononyl phthalate (with sample detection rates of 46.2%, 53.7%, and 22.9%, respectively). The detection rates of the remaining phthalate metabolites ranged from 84.2% to 99.88%. Values below the detection threshold were substituted with the lower limit of detection divided by the square root of 2.

Anthropometric and biochemical assessments were performed by qualified staff at the Mobile Examination Center (MEC) and through home visits. The CircS is characterized by the presence of four or more of the following seven components (7): (1) increased waist circumference (≥ 88 cm for females, ≥ 102 cm for males); (2) elevated fasting glucose (≥ 100 mg/dL) or on pharmacological treatment for hyperglycemia; (3) elevated serum triglycerides (≥ 150 mg/dL) or on pharmacological treatment for hypertriglyceridemia; (4) reduced high-density lipoprotein cholesterol (HDL-C < 40 mg/dL for males; HDL-C < 50 mg/dL for females) or patients being treated with medication; (5) elevated blood pressure (≥ 130 mmHg systolic or ≥ 85 mmHg diastolic) or on pharmacological treatment for hypertension; (6) short sleep duration (< 6 h per day); (7) depressive symptoms as indicated by a Patient Health Questionnaire (PHQ-9) score of ≥ 5. The PHQ-9 is a nine-item screening tool for depression, designed to assess the presence of depressive symptoms within the past 2 weeks, with scores categorized as none/mild depression (0–4 points) or moderate/severe depression (≥ 5 points) (16).

Data on participants' age, sex, ethnicity, educational attainment, alcohol consumption habits, cotinine levels, and physical activity were gathered through questionnaires, physical examinations, and laboratory tests. Race was classified into categories including non-Hispanic White, non-Hispanic Black, Mexican American, and other races (including other Hispanic and multiracial individuals). Educational attainment was classified into two groups: “less than high school” (including 12th-grade students without a diploma) and “high school and above”. The assessment of drinking habits was based on the question “Had at least 12 alcoholic drinks in the past year,” which included data from both current drinkers and non-drinkers. Cotinine concentrations were employed to assess the degree of tobacco exposure among individuals; in this study, cotinine levels were divided into two categories: below 0.05 ng/mL (low) and ≥ 0.05 ng/mL (high) (17). Metabolic Equivalent of Task (MET) was utilized to determine physical activity levels, with the calculation formula for physical activity given as: Physical Activity (MET-h/week) = MET value × frequency per week × duration of each activity. The classification criteria for physical activity levels are as follows: 0 MET-h/week represents no activity, below 30 MET-h/week is defined as low activity level, 30 to 60 MET-h/week is defined as moderate activity level, and 60 MET-h/week and above is defined as high activity level (18).

Categorical variables are presented as counts (N) and proportions (%). For continuous variables that do not conform to a normal distribution, the median and interquartile range (IQR) are utilized. Comparisons between groups for categorical variables were conducted using the Pearson chi-square test, whereas differences in continuous variables were examined using the Wilcoxon rank-sum test. A logistic regression model was employed to assess the association between PAEs and CircS, taking into account the complex survey design of the NHANES data by utilizing appropriate sampling weights. PAEs were categorized into four groups based on quartiles (Q1, Q2, Q3, and Q4), with Q1 as the reference group, and the strength of the association was determined through odds ratios (OR) and confidence intervals (CI). The initial model included each PAE without any adjustments, while subsequent models adjusted for age, gender, race, education level, drinking habits, cotinine levels, and physical activity. A restricted cubic spline (RCS) analysis was conducted to evaluate the dose-response relationship between the log₁₀-transformed individual PAEs and CircS, using the 25th percentile as the reference value. The associations between the mixture of 10 log₁₀-transformed PAEs and CircS were analyzed by Bayesian kernel machine regression (BKMR) and quantile g-computation models. The significance threshold for this study was established at 0.05, and all analyses were conducted using R (version 4.4.0). The BKMR and quantile g-computation models were implemented using the R packages “bkmr” and “qgcomp” respectively.

Table 1 presents the baseline characteristics of the study subjects. The average age of the participants was 48.85 ± 0.64 years, comprising 1,249 females and 1,270 males.

The characteristics of ten PAEs metabolites are detailed in Supplementary Table S1. The average concentrations of PAEs ranged from 1.20 to 29.90 ng/mL. Mono-ethyl phthalate exhibited the highest concentration, followed by Mono-n-butyl phthalate and MECP phthalate, with average concentrations of 29.90 ng/mL, 9.70 ng/mL, and 8.60 ng/mL, respectively. The exposure level of MECP phthalate in CircS patients was significantly higher than that in individuals without CircS (9.00 vs. 8.30 ng/mL, P < 0.05).

Figure 2 summarizes the relationship between phthalate metabolites exposure and CircS. In univariate logistic regression analysis, higher exposure levels of MECP phthalate (Q2 vs. Q1: OR = 1.457, 95% CI: 1.086, 1.954; Q3 vs. Q1: OR = 1.401, 95% CI: 1.017, 1.930; Q4 vs. Q1: OR = 1.395, 95% CI: 1.064, 1.829) and MEHP phthalate (Q2 vs. Q1: OR = 1.732, 95% CI: 1.337, 2.245; Q3 vs. Q1: OR = 1.553, 95% CI: 1.162, 2.078; Q4 vs. Q1: OR = 1.411, 95% CI: 1.063, 1.873) were significantly associated with an increased risk of CircS compared to the Q1 group. After adjusting for age, gender, educational attainment, race, alcohol consumption, physical activity, and cotinine, the elevated exposure levels of MECP phthalate (Q2 vs. Q1: OR = 1.386, 95% CI: 1.013, 1.894; Q3 vs. Q1: OR = 1.630, 95% CI: 1.127, 2.357; Q4 vs. Q1: OR = 1.632, 95% CI: 1.159, 2.300) and MEHP phthalate (Q2 vs. Q1: OR = 1.800, 95% CI: 1.300, 2.490; Q3 vs. Q1: OR = 1.712, 95% CI: 1.198, 2.446; Q4 vs. Q1: OR = 1.830, 95% CI: 1.301, 2.573) remained significantly associated with an increased risk of CircS. Additionally, higher exposure levels of MEOH phthalate (Q4 vs. Q1: OR = 1.56, 95% CI: 1.065, 2.279) and Mono-benzyl phthalate (Q4 vs. Q1: OR = 1.699, 95% CI: 1.156, 2.496) also significantly increased the risk of CircS compared to the Q1 group. Supplementary Figures S1–S7 summarize the relationship between phthalate metabolites exposure and the components of CircS. Participants in the fourth quartile of exposure to MECP phthalate, MEHP phthalate, MEOH phthalate, and mono-benzyl phthalate exhibited a significantly increased risk of central obesity and elevated glucose compared with those in the first quartile of exposure (Supplementary Figures S1, S2). Additionally, participants in the fourth quartile of MECP phthalate exposure had a higher risk of low HDL-C and depression than those in the first quartile (Supplementary Figures S4, S7). Similarly, participants in the fourth quartile of MEHP phthalate exposure had a higher risk of low HDL-C than those in the first quartile (Supplementary Figure S4). Participants in the fourth quartile of mono-benzyl phthalate exposure also had an increased risk of depression compared to those in the first quartile (Supplementary Figure 7).

Figure 3 depicts the linear and nonlinear relationships between individual PAE exposures in urine and CircS based on RCS regression. The analysis revealed a linear positive correlation between Mono(carboxynonyl) phthalate, MECP phthalate, MEHP phthalate, and Mono-benzyl phthalate with CircS (P_overall < 0.05, P_nonlinearity > 0.05). Conversely, Mono(carboxyoctyl) phthalate exhibited a nonlinear relationship with CircS (P_overall < 0.05, P_nonlinearity < 0.05).

Using the BKMR model, an overall association analysis of the mixtures revealed a joint association between PAE mixtures and CircS. When all metal concentrations were above the 25th percentile, the metal mixture demonstrated a significant positive correlation with CircS levels. Conversely, when all metal concentrations were below the 25th percentile, a significant negative correlation was observed compared to when all metal concentrations were at the 25th percentile (Figure 4). Moreover, the BKMR model was employed to explore the potential exposure-response relationships while maintaining other PAEs levels at their corresponding 25th percentiles. The findings from the BKMR model aligned with the results derived from the RCS model, indicating a linear positive correlation between Mono(carboxynonyl) phthalate, MECP phthalate, MEHP phthalate, and Mono-benzyl phthalate with CircS (Supplementary Figure S8).

In the quantile g-computation model, the combined exposure to PAEs was significantly positively correlated with CircS (Estimate = 0.123, P < 0.05). The weights of individual PAEs are illustrated in Figure 5, where darker bars on the right side (positive) indicate that the overall effect is positive, with the highest positive weight attributed to MECP phthalate, followed by Mono-benzyl phthalate and Mono(carboxynonyl) phthalate.