Association of Phthalates, Parabens and Phenols Found in Personal Care Products With Pubertal Timing in Girls and Boys

Kim G. Harley; Kimberly P. Berger; Katherine Kogut; Kimberly Parra; Robert H. Lustig; Louise C. Greenspan; Antonia M. Calafat; Xiaoyun Ye; Brenda Eskenazi


Hum Reprod. 2019;34(1):109-117. 

In This Article

Materials and Methods

Study Population

Study participants were mothers and children enrolled in the Center for the Health Assessment of Mothers and Children of Salinas (CHAMACOS), a longitudinal birth cohort examining the effects of in-utero and early life environmental exposures on children's health and development in a predominantly Latino, US farmworking community. Pregnant women were recruited in 1999–2000 from community clinics serving California's Salinas Valley. Eligibility criteria included being at least 18 years of age, <20 weeks gestation, English- or Spanish-speaking, and qualifying for low income health insurance (Medicaid). Of 537 children in the study at birth, 339 were followed through pubertal assessments, conducted between 9 and 13 years of age. We excluded one child who was missing biomarker data, for a final sample of 338 children (159 boys and 179 girls).

Informed consent was obtained from mothers and assent was obtained from children. All study activities were approved by the Institutional Review Board (IRB) of the University of California, Berkeley, with the Centers for Disease Control and Prevention (CDC) IRB deferring to UC Berkeley.

Data Collection

Mothers were interviewed in English or Spanish at two time points during pregnancy (mean: 14.0 and 26.9 weeks' gestation) and when their children were 9 years old. Information collected during pregnancy included maternal age, marital status, race/ethnicity, country of birth, years in the USA, educational attainment, household income and the number of people in the household. Pre-pregnancy BMI was calculated using self-reported pre-pregnancy weight and measured height. Household income was also assessed when children were 9 years old.

Clinical Tanner staging (Tanner, 1986) was used to assess timing of puberty. We examined the children every 9 months between 9 and 13 years of age (i.e. at age 9 (n = 312), 9&frac34; (n = 268), 10&frac12; (n = 300), 11&frac14; (n = 275), 12 (n = 301) and 12&frac34; (n = 264)). Trained research assistants used palpation to assess breast development and visual inspection to evaluate pubic hair development in girls. Girls were asked if they had begun menstruating and the month and year of their first menses was recorded at the first post-menarcheal visit. Boys' stages of genital and pubic hair development were assessed visually and testicular volume was measured in comparison to orchidometer beads. Stage of development for breast (B), genital (G) and pubic hair (PH) development was classified between 1 and 5, with 1 signifying prepubarchal and 5 signifying adult development. Pubertal onset was defined as age when reaching stage B2 (i.e. thelarche) for girls or G2 (i.e. gonadarche) for boys. We also examined age at PH2 (i.e. pubarche) in both girls and boys. A boy was only considered to be in stage G2 if he had both testicular volume >3 cm3 and Tanner genital characteristics consistent with stage 2. Research assistants were trained and supervised by two pediatric endocrinologists (R.L. and L.G.). In on-going training sessions, the research assistants' categorization of whether a child was in stage 2+ versus stage 1 agreed with that of the pediatric endocrinologists 90 and 92% of the time for girls' breast and pubic hair stage, respectively, and 92 and 100% of the time for boys' genital and pubic hair stage.

At each visit, research assistants calculated BMI by weighing and measuring the children without shoes. Children were classified as underweight, normal, overweight or obese at each visit based on CDC percentile guidelines (Whitlock et al., 2005).

Measurement of Urinary Biomarkers

Spot urine samples were collected from mothers at the time of the two pregnancy interviews (prenatal samples) and from the children at the 9-year-old visit (peripubertal samples). Urine samples were aliquoted, frozen and stored at −80°C until shipment on dry ice to the CDC in Atlanta, GA for analysis. Paraben and phenol concentrations were measured in both prenatal and peripubertal urine samples; phthalate metabolites were only quantified in prenatal samples due to budgetary constraints.

Urinary phthalate metabolites were quantified using online solid phase extraction coupled with isotope dilution high performance liquid chromatography–electrospray ionization-tandem mass spectrometry, as described previously (Silva, 2007). The present analysis focused on metabolites of three low molecular weight phthalates known to be used in personal care products: monoethyl phthalate [MEP, a metabolite of DEP]; mono-n-butyl phthalate [MnBP, a metabolite of DnBP]; and mono-isobutyl phthalate [MiBP, a metabolite of (DiBP)]. Associations of high molecular weight phthalates with puberty in this population have been described elsewhere (Berger et al., 2018). Limits of detection (LODs) were 0.6 ng/mL for MEP, 0.4 ng/mL for MBP and 0.2 ng/mL for MiBP. When concentrations were below the LOD, we used values generated by the instrument when available (2% of values for MiBP, 0.5% for MnBP, 0% for MEP); if no signal was detected, we substituted a random value <LOD based on a log-normal probability distribution whose parameters were determined by maximum likelihood estimation (3.6% of values for MiBP, 0.3% for MnBP, 0.2% for MEP) (Lubin et al., 2004).

Urinary concentrations of three parabens (methyl-, propyl- and butyl paraben) and four phenols (triclosan, benzophenone-3, and 2,4- and 2,5-dichlorophenol) were quantified using online solid phase extraction-high performance liquid chromatography-isotope dilution tandem mass spectrometry (Ye et al., 2005, 2006). The LOD was 1.0 ng/mL for methyl paraben, 0.2 ng/mL for propyl- and butyl paraben, 2.3 ng/mL for triclosan, 0.4 ng/mL for benzophenone-3, and 0.2 ng/mL for 2,4- and 2,5-dichlorophenol. Because the analysis was originally conducted for quantification of bisphenol A only, rather than personal care product chemicals, a number of samples had biomarkers with concentrations above the highest calibration, including prenatal samples for 128 mothers (n = 39 for methyl paraben, n= 41 for propyl paraben, n = 18 for triclosan, n = 66 for benzophenone-3, n = 34 for 2,4-dichlorophenol, n = 111 for 2,5-dichlorophenol) and peripubertal samples for 17 children (n = 3 for methyl paraben, n = 0 for propyl paraben, n = 3 for triclosan, n = 8 for benzophenone-3, n = 1 for 2,4-dichlorophenol, n = 3 for 2,5-dichlorophenol). For concentrations above the highest standard, we substituted with the highest calibrator used (100 ng/mL for 2,4-dichlorophenol; 1000 ng/mL for all other phenols and parabens). Concentrations below the LOD were substituted with the instrument-read value (14.7% of values for triclosan, 1.8% for propyl paraben, 1.4% for 2,4-dichlorophenol, <0.5 for all other phenols) or a random value <LOD (13.6% of values for triclosan, 1.3% for 2,4- and 2,5-dichlorophenol, <0.5 for all other phenols) in the same manner as for phthalate metabolites.

Urinary creatinine concentrations were determined using a commercially available diagnostic enzyme method (Vitros CREA slides; Ortho Clinical Diagnostics, Raritan, NJ). Urinary biomarker concentrations were divided by creatinine concentrations to correct for urinary dilution. Urine specific gravity was also measured for the prenatal, but not peripubertal, samples using a hand-held refractometer (National Instrument Company Inc., Baltimore, MD). Urinary biomarkers were corrected for specific gravity using the equation: analyte concentration * [(1.024 – 1)/(SGsample – 1)], where 1.024 is the reference urinary SG for pregnant women (Mahalingaiah et al., 2008). Although the main analyses used creatinine-corrected biomarker concentrations, we also examined specific-gravity-corrected prenatal concentrations in sensitivity analyses.

Data Analysis

We examined prenatal and peripubertal biomarker concentrations as continuous (log 2-transformed) and categorical (quartiles) measures of exposure. For prenatal exposure, we used the average of the creatinine-corrected concentrations in the two pregnancy urine samples. For peripubertal exposure, we used the single creatinine-corrected concentration quantified in children's urine at 9 years of age.

We conducted parametric accelerated failure time (AFT) models assuming a 2-parameter Weibull distribution to determine the association of urinary biomarker concentrations with timing of pubertal onset using the Stata intcens program, which allowed for interval-censoring (i.e. pubertal onset occurring at an unknown time between two study visits). We converted time ratios into mean shift (in months) at pubertal onset by multiplying by median age at onset. Median ages at thelarche, pubarche and gonadarche for the study population were calculated using an unadjusted AFT model.

Separate models were developed for prenatal and peripubertal concentrations. Prenatal and peripubertal concentrations were included in the same model in sensitivity analyses. Covariates were selected a priori using directed acyclic graphs (see Supplementary Figure S1) and included maternal education, maternal years in the USA, family income and maternal pre-pregnancy BMI as categorical variables, categorized as shown in Table I. Obesity is a risk factor for early puberty, particularly in girls (Jasik and Lustig, 2008), and may be on the causal pathway between early life exposure and timing of puberty. Thus, child BMI at age 9 was not included as a covariate in main models but was included in sensitivity analyses (categorized as underweight/normal weight versus overweight/obese). We further examined child overweight/obesity as a possible mediator using the Paramed package in Stata (VanderWeele and Vansteelandt, 2009; Emsley and Liu, 2013) with binary chemical exposure variables and binary overweight/obese status as the mediator in the association with age at puberty.

Supplementary Figure S1 .

Directed acyclic graphs (DAGs) of role of covariates in hypothesized association of urinary phthalate and phenol concentrations and timing of puberty in girls (A) and boys (B).