The study subjects consisted of the hospital-based birth cohort of the Ewha Birth and Growth Cohort Study, which enrolled pregnant women in the second trimester (24–28 weeks) who visited the Department of Obstetrics and Gynecology at Ewha Womans University Mokdong Hospital between 2001 and 2006, and whose offspring were followed-up (n = 940). Follow-up started in 2005 (see details in [18]). Of the 940 cohort members, the follow-up rates of subjects who visited at least once during the follow-up period were 67% and 59%, respectively. Consent was obtained from all participants and their parents through a written consent form, and the protocol was approved by the Institutional Review Board (IRB) of Ewha Womans University Hospital (IRB No. SEUMC 2020–07-016).
Blood and urine samples and body measurements were collected from each participant through follow-up observations in each period, and demographic and socioeconomic information were collected through structured questionnaires. Venous blood was collected from the children in the morning after fasting for 8–12 h using a vacuum tube containing ethylenediaminetetraacetic acid (EDTA). The collected samples were stored at –80 °C in a freezer, and were transported in the frozen state for examination.
In total, 164 participants with measurable phthalate in urine samples participated in the follow-up for two exposure periods (3–5 and 7–9 years of age). Among them, 126 participated in further follow-up examinations at 10–15 years of age, and were included in the analysis.
Exposure assessment: measurement of urinary phthalate metabolite concentrations
Urinary concentrations of phthalate metabolites in the study subjects were repeatedly measured during both exposure periods (at 3–5 and 7–9 years of age). The measured metabolites were mono-benzyl phthalate (MBzP), mono(2-ethyl-5-carboxypentyl) phthalate (MECPP), mono(2-ethyl-5hydroxyhexyl) phthalate (MEHHP), mono(2-ethylhexyl) phthalate (MEHP), mono(2-ethyl-5-oxohexyl phthalate (MEOHP), monoethyl phthalate (MEP), monoisobutyl phthalate (MiBP), monoisononyl phthalate (MiNP), and mono-n-butyl phthalate (MnBP). ΣDEHP was calculated as the molar sum of MECPP, MEHHP, MEHP, and MEOHP, and ΣDBP was calculated as the molar sum of MnBP and MiBP. The molar sum was calculated by dividing each metabolite concentration by its molar mass and then summing the individual metabolite concentrations (μmol/L).
Stored spot urine samples (≥ 1 mL) were sent to a specialized diagnostic laboratory (Lab Frontier, Korea) to measure phthalate metabolites. After adding 50 µL internal standard solution (200 ng/mL, nine types of phthalate metabolites-13C12), 1 mL 2 M ammonium acetate buffer, and 20 µL β-glucuronidase, the urine samples were incubated overnight at 37 °C. Thereafter, 4 mL ethyl acetate was added and extracted twice. The extracted samples were gently shaken several times and centrifuged at 4,000 rpm for 15 min to separate the non-polar fat layer and the organic layer. The upper organic layer, which did not contain non-polar fat, was separated and transferred to a glass tube. The extract was evaporated and dried at 40 °C under a gentle flow of nitrogen, and then re-extracted with 100 µL acetonitrile.
The analysis was performed using high-performance liquid chromatography (HPLC). For HPLC tandem mass spectrometry (MS/MS), Agilent 1200 series and Agilent 6430 Triple Quad liquid chromatograph mass spectrometers (Agilent Technologies Inc., Santa Clara, CA, USA) were used. A CAPCELL PAK C18 MG II column (3.0 × 150 mm, 3 μm) (Shiseido Co., Ltd., Tokyo, Japan) was used for chromatographic separation. The mobile phase was carried out at a flow rate of 0.3 mL/min using 0.1% acetic acid in distilled water and 0.1% acetic acid in acetonitrile. MS/MS was analyzed in electrospray negative ion multiple reaction monitoring mode. The accuracy of the analysis was 70–120%, and the coefficients of variation (CVs) were 0.77% (MBzP), 2.27% (MECPP), 1.21% (MEHHP), 2.41% (MEHP), 1.17% (MEOHP), 3.09% (MEP), 0.89% (MiBP), 1.34% (MiNP), and 1.88% (MnBP). In addition, the limits of detection (LODs) were 0.27 μg/L (MBzP), 0.16 μg/L (MECPP), 0.19 μg/L (MEHHP), 0.24 μg/L (MEHP), 0.22 μg/L (MEOHP), 1.0 μg/L (MEP), 0.20 μg/L (MiBP), 0.17 μg/L (MiNP), and 0.26 μg/L (MnBP). To calculate creatinine-adjusted phthalate concentrations, creatinine was measured by the Jaffé method [19], using Olympus AU 680 (Beckman Coulter, Inc., Brea, CA, USA) equipment.
Health outcomes: measurement of blood liver enzyme concentrations
The stored blood samples (≥ 1 mL) of the 126 children who completed the follow-up at age 10–15 years were sent to a specialized diagnostic laboratory (Seegene Medical Institute, Korea) for analysis. AST, ALT, and γ-GTP were measured as indicators of liver enzyme levels in the blood. These were measured using Cobas 8000 C702 (Roche, Germany) equipment by the enzymatic method. ASTL, ALTL, and GGT Gen.2 (Roche) reagents were used for AST, ALT, and γ-GTP, respectively. In all of the measured indicators, the CV of accuracy was < 5%.
Covariates
Covariates were considered based on a literature review and directed acyclic graph (DAG) (Figure S1); variables associated with liver enzymes at least once based on p < 0.2 were considered for analysis. All analyses were performed after adjusting for confounding factors such as sex, age, BMI, mother’s education level, household income level, secondhand smoke, and physical inactivity. The BMI was calculated using body measurements obtained during follow-up as weight (kg) divided by the square of height (m2). Height and weight were measured in 0.1-cm and 0.1-kg units using an automatic height and weight measurement system (DS-102; Dong Sahn Jenix Co., Ltd., Seoul, Korea), without shoes and in light clothing.
The mothers’ education levels were divided into two categories (graduated from high school or lower and graduated from college or higher) based on the questionnaire responses (graduated from elementary/middle school, high school, college, or above graduate school). Household income was divided into three categories (˂3 million won, 3–5 million won, and ≥ 5 million won) based on the questionnaire responses about monthly average household income (˂1 million won, 1–2 million won, 2–3 million won, 3–5 million won, or ≥ 5 million won). A questionnaire item was used to assess each child’s exposure to secondhand smoke on follow-up at the age of 10–15 years. The level of physical inactivity at age 10–15 years was divided into three categories (< 1 h, ≥ 1 h, and ≥ 2 h) based on the questionnaire items regarding the average time spent sitting in leisure activities such as watching TV and playing games (< 1 h, ≥ 1 h, ≥ 2 h, and almost none).
Statistical analysis
When the concentration of phthalate metabolites in urine (μg/L) was less than the LOD, LOD/√2 was used for analysis [20]. When a subject visited more than once at the age of 10–15 years, liver enzyme levels at the time of the earliest visit were preferentially used for analysis. Continuous and normally distributed data are presented as means and standard deviations, and categorical data are presented as frequencies and percentages (%).
Analysis of covariance (ANCOVA) was performed to determine the differences in liver enzyme levels (ALT, AST, and γ-GTP) between groups by dividing phthalate exposure into tertiles at two time points (when the children were 3–5 and 7–9 years of age). To assess the dose–response relationship, a trend analysis was performed based on exposure levels by applying an ordinal variable. We used a restrictive cubic spline (RCS) model to evaluate nonlinear relationships. In addition, a sensitivity analysis was performed to exclude the effects of BMI as a mediator, on determining the results.
Exposure levels were redefined for high and low exposure groups based on the medians for the 3–5 and 7–9 years age groups, to assess the differences between exposure levels at the two exposure periods: low exposure levels in both periods (LL); high at 3–5 years, but low at 7–9 years (HL); low at 3–5 years, but high at 7–9 years (LH); and high in both periods (HH). This was to assess whether there were any differences in the mean liver enzyme levels between groups. If a difference was significant, Bonferroni’s post hoc test was performed.
The relationships between phthalate metabolites and liver enzymes were evaluated using linear regression. Furthermore, the interaction effects of BMI with urinary phthalate according to sex were evaluated using linear regression.
All statistical analyses were performed using SAS ver. 9.4 (SAS Institute, Cary, NC, USA) and R Statistical Software (ver. 3.6.2; R Core Team 2019). Statistical significance was tested based on α = 0.05 using two-sided tests.