An assessment of measurement error when using maternal urinary phthalates as proxies for placental tissue levels in the estimation of the association of prenatal phthalates and infant anogenital distance.

内容摘要

Direct measurement of in utero phthalate exposure in placental and fetal tissues is generally not possible. Maternal urinary levels serve as proxies and may introduce measurement error, biasing health effect estimates. We adjusted for measurement error when using maternal urinary phthalates as proxies for placental-fetal exposure and to evaluate the impact of this correction on the association between prenatal phthalate exposure and anogenital distance (AGD). We biopsied and analyzed 68 placentas from the TIDES study (San Francisco site: n = 204, 2010-2012). Phthalate metabolite concentrations were measured in three placental tissue types: chorion smooth (CS), chorion frondosum (CF), and basal plate (BP). Phthalate concentrations were standardized to remove pre-processing variation. Regression calibration (RC) and multiple imputation for measurement error (MIME) were used for correction. Associations between phthalates and AGD were estimated by generalized linear models. Bias was quantified by calculating the percent change in the beta coefficient from the gold standard (placental phthalate) to the proxy (urinary phthalate), after correction for measurement error. Phthalate metabolites were detected in over 70% of placental tissues. Monoethylhexyl (MEHP) phthalate was the most abundant metabolite in CF and in CS. Phthalate concentrations were lowest in BP and varied across placental tissues. Weak associations were found between urinary and placental phthalates. MIME outperformed RC, reducing bias in the average phthalate effect on AGD in the male by 30% for AGD-penis and 69% for AGD-scrotum, and in the female by 32% for AGD-clitoris and 45% for AGD-fourchette. Placental phthalate concentrations varied by tissue type and showed poor correlation with maternal urinary levels. MIME outperformed RC in adjusting for measurement error in this setting. Findings suggest maternal and placental exposures are distinct constructs, highlighting the need for direct placental measurement in phthalate toxicity studies. Further research can improve phthalate exposure assessment and knowledge of maternal-placental-fetal transfer. This study highlights a key gap in environmental health research, where maternal urinary phthalates are often used as proxies for placental and fetal exposure. Using two correction methods, we found that maternal urinary and placental phthalates represent distinct exposure constructs and are not interchangeable.