Use of a Pharmacokinetic Model and Single Urine Voids to Estimate the Dosing Profiles of Pesticide Applicators

Abstract

P-155 Introduction: Pesticide biomonitoring studies typically collect spot urine samples due to low subject burden and cost compared to total urine collection. However, single voids will bias daily dose estimates due to the changing time course of the analytes in the body and urine after exposure events. A single void may still be able to accurately estimate dose over longer time periods when used in conjunction with a pharmacokinetic (PK) model to establish the relationship between absorbed dose and urinary metabolites as a function of time. Methods: Composite and single-void urine samples from fourteen farmers were measured for concentrations of 2,4-D or the chlorpyrifos metabolite TCP. Composite samples were collected over 24-hour intervals starting 24 hours prior to application through 96 hours post-application. The single voids analyzed were collected on the morning after the application day. Biomarker excretion rates from single-void samples, adjusted for baseline concentration, were entered into an inverse PK model to predict FD, the dermally absorbed dose during the exposure event. A one-compartment model with first-order absorption and elimination rate constants was used and PK parameters were taken from the literature. FD was then entered into a model in a forward manner using STELLA software to estimate elimination of the analytes over time. The urinary output of the analytes (μg) predicted by the models during the first 24-hours post-application were compared to measured values. Results: The TCP and 2,4-D models reduced the bias associated with using a single-void surrogate for daily dose. The average percent error between the measured and modeled TCP 24-hour output estimates was 20%, compared to 80% between measured TCP 24-hour output and single-void output (adjusted upwards using subject-specific daily urine volume). Average percent error between the measured and modeled 2,4-D 24-hour output estimates was 25%, compared to 147% between 2,4-D measured 24-hour output and single-void output (using subject-specific daily urine volume). Discussion and Conclusions: Despite the small sample size, our findings demonstrate that PK models improved the capability of single-void samples to estimate a farmer's daily dose after a known exposure to two non-persistent pesticides. Model parameter uncertainty, inter-individual variability, product formulation differences and urine undercollection likely contributed to the disagreement between modeled and measured output. The described methodology has wide applications in epidemiology, industrial hygiene, and risk assessment where there is currently a lack of practical techniques to measure absorbed dose at reasonable cost.