Variability of urinary pesticide metabolite concentrations during pregnancy in the MARBLES Study

Variability of short-lived urinary pesticide metabolites during pregnancy raises challenges for exposure assessment.

For urinary metabolite concentrations 3-phenoxybenzoic acid (3-PBA) and 3,5,6-trichloro-2-pyridinol (TCPy), we assessed: (1) temporal variability; (2) variation of two urine specimens within a trimester; (3) reliability for pesticide concentrations from a single urine specimen to classify participants into exposure tertiles; and (4) seasonal or year variations.

Pregnant mothers (N = 166) in the MARBLES (Markers of Autism Risk in Babies-Learning Early Signs) Study provided urine specimens (n = 528). First morning void (FMV), pooled, and 24-h specimens were analyzed for 3-PBA and TCPy. For 9 mothers (n = 88 specimens), each urine specimen was analyzed separately (not pooled) to estimate within- and between-person variance components expressed as intraclass correlation coefficients (ICC). Pesticide concentrations from two specimens within a trimester were also assessed using ICC's. Agreement for exposure classifications was assessed with weighted Cohen's kappa statistics. Longitudinal mixed effect models were used to assess seasonal or year variations.

Urinary pesticide metabolites were detected in ≥ 93% of specimens analyzed. The highest ICC from repeated individual specimens was from specific gravity-corrected FMV specimens for 3-PBA (ICC=0.13). Despite high within-person variability, the median concentrations did not differ across trimesters. Concentrations from pooled specimens had substantial agreement predicting exposure categories for TCPy (K = 0.67, 95% CI (0.59, 0.76)) and moderate agreement for 3-PBA (K = 0.59, 95% CI (0.49, 0.69)). TCPy concentrations significantly decreased from 2007 to 2014.

Pooled specimens may improve exposure classification and reduce laboratory costs for compounds with short biological half-lives in epidemiological studies.