Intra-individual variability of urinary non-persistent chemical concentrations during pregnancy in the ECHO Cohort

Abstract

Studies of prenatal exposure to non-persistent chemicals often assume biomarker measurements from a single urine sample can represent exposure pregnancy. The validity of this assumption depends on biomarker concentration variability across gestation, which is not well-characterized for many chemicals. To assess the intra-individual variability of non-persistent chemical concentrations measured in 3 pregnancy urine samples. We measured 154 analytes from 12 classes in 3 urine samples collected across gestation from 90 participants in 3 Environmental influences on Child Health Outcomes Cohort sites. For each cohort site, we estimated intraclass correlation coefficients (ICCs) of 58 analytes with >60% detection using log2-transformed concentrations of specific gravity (SG)-standardized values. We classified analyte ICCs by the site-specific median relative to the limits of detection (LODs) to evaluate the influence of exposure levels on reliability estimates. As secondary analyses, we calculated kappa statistics for 56 analytes with 20–60% detection to assess agreement of detection frequencies across the 3 timepoints. ICCs (range: −0.12 to 0.91) varied by cohort site, except for benzophenone-1 and propyl paraben (consistently >0.5) and bisphenol F (consistently <0.1). We observed site-specific trends for parabens, benzophenones, antimicrobials, phthalates, polycyclic aromatic hydrocarbons, and aromatic amines. Analytes with median:LOD ratios >50 were more stable throughout pregnancy (e.g., benzophenones, propyl paraben, antimicrobials, and naphthalenes), except for some phthalate metabolites, which may indicate high sporadic exposures. Analytes with median:LOD ratios <10 had lower ICCs, except for benzophenone-8, which may indicate low-level chronic exposure. Kappa statistics of the 56 analytes with 20–60% detection ranged from −0.22 to 0.53. ICCs varied by cohort site. Lower ICCs were generally observed for analytes with lower concentrations. For the majority of non-persistent chemicals, a single pregnancy measure may better represent exposure near the time of sample collection rather than for the entire pregnancy. Understanding the intra-individual variability of urinary non-persistent chemical biomarkers across multiple pregnancy timepoints is important to inform study design, statistical analysis, and interpretation in future studies.