Within-person variability in urinary phthalate metabolite concentrations: measurements from specimens after long-term frozen storage

Abstract

Laboratory studies show that exposure to phthalates during development can cause adverse effects, especially for males. Studies in humans would be facilitated by collection of urine during pregnancy, long-term storage, and measurement of phthalate metabolites at the time that offspring health is assessed. Our aims were to measure urinary phthalate metabolites after long-term freezer storage, to use those measurements to evaluate within-woman variability over 2- and 4-week intervals, and to determine whether the phases of the menstrual cycle affect metabolite levels. Samples were selected from daily first-morning urine specimens collected by 60 women and stored frozen since 1983-1985. Three specimens per woman were selected at approximately 2-week intervals to include both follicular and luteal phase samples. Seven metabolites of five phthalates were measured by mass spectrometry. Statistical analyses were conducted with correlation, mixed model regression, and the Wilcoxon signed rank test. Creatinine-corrected urinary phthalate metabolite concentrations measured in samples after long-term storage tended to have a similar right-skewed distribution, though with somewhat higher concentrations than those reported for recently collected US samples. The concentrations of three metabolites of di(2-ethylhexyl)phthalate in the same specimen were very highly correlated (Pearson r=0.85-0.97). Reproducibility over a 4-week interval was moderate for the metabolites of diethyl phthalate and benzylbutyl phthalate (intraclass correlation coefficients, ICCs, 0.48 and 0.53, respectively), whereas five other metabolites had lower ICCs (0.21-0.37). Menstrual phase was not related to metabolite concentrations. Although the same samples have not been measured both before and after long-term storage, results suggest that the measurement of phthalate metabolites after long-term sample storage yield generally similar distributions and temporal reliability as those reported for recently collected specimens. These findings support the use of stored urine specimens collected during the relevant stage of human pregnancy to investigate the influence of phthalate exposures on later outcomes.