Person:

Herrick, Robert

Background: CYP19 and PPARγ are two genes expressed in the placental trophoblast that are important to placental function and are disrupted by phthalate exposure in other cell types. Measurement of the mRNA of these two genes in human placental tissue by quantitative real-time polymerase chain reaction (qPCR) offers a source of potential biomarkers for use in epidemiologic research. We report on methodologic challenges to be considered in study design. Methods: We anonymously collected 10 full-term placentas and, for each, sampled placental villi at 12 sites in the chorionic plate representing the inner (closer to the cord insertion site) and outer regions. Each sample was analyzed for the expression of two candidate genes, aromatase (CYP19) and peroxisome proliferator activated receptor protein gamma (PPARγ) and three potential internal controls: cyclophilin (CYC), 18S rRNA (18S), and total RNA. Between and within placenta variability was estimated using variance component analysis. Associations of expression levels with sampling characteristics were estimated using mixed effects models. Results: We identified large within-placenta variability in both transcripts (>90% of total variance) that was minimized to <20% of total variance by using 18S as an internal control and by modelling the means by inner and outer regions. 18S rRNA was the most appropriate internal control based on within and between placenta variability estimates and low correlations of 18S mRNA with target gene mRNA. Gene expression did not differ significantly by delivery method. We observed decreases in the expression of both transcripts over the 25 minute period after delivery (CYP19 p-value for trend = 0.009 and PPARγ (p-value for trend = 0.002). Using histologic methods, we confirmed that our samples were comprised predominantly of villous tissue of the fetal placenta with minimal contamination of maternally derived cell types. Conclusion: qPCR-derived biomarkers of placental CYP19 and PPARγ gene expression show high within-placental variability. Sampling scheme, selection of an appropriate internal control and the timing of sample collection relative to delivery can be optimized to minimize within-placenta and other sources of underlying, non-etiologic variability.

Background: Although urinary concentrations of phthalate metabolites are frequently used as biomarkers in epidemiologic studies, variability during pregnancy has not been characterized. Methods: We measured phthalate metabolite concentrations in spot urine samples collected from 246 pregnant Dominican and African-American women. Twenty-eight women had repeat urine samples collected over a 6-week period. We also analyzed 48-hr personal air samples (n = 96 women) and repeated indoor air samples (n = 32 homes) for five phthalate diesters. Mixed-effects models were fit to evaluate reproducibility via intraclass correlation coefficients (ICC). We evaluated the sensitivity and specificity of using a single specimen versus repeat samples to classify a woman’s exposure in the low or high category. Results: Phthalates were detected in 85–100% of air and urine samples. ICCs for the unadjusted urinary metabolite concentrations ranged from 0.30 for mono-ethyl phthalate to 0.66 for monobenzyl phthalate. For indoor air, ICCs ranged from 0.48 [di-2-ethylhexyl phthalate (DEHP)] to 0.83 [butylbenzyl phthalate (BBzP)]. Air levels of phthalate diesters correlated with their respective urinary metabolite concentrations for BBzP (r = 0.71), di-isobutyl phthalate (r = 0.44), and diethyl phthalate (DEP; r = 0.39). In women sampled late in pregnancy, specific gravity appeared to be more effective than creatinine in adjusting for urine dilution. Conclusions: Urinary concentrations of DEP and DEHP metabolites in pregnant women showed lower reproducibility than metabolites for di-n-butyl phthalate and BBzP. A single indoor air sample may be sufficient to characterize phthalate exposure in the home, whereas urinary phthalate biomarkers should be sampled longitudinally during pregnancy to minimize exposure misclassification.