Biomonitoring data is becoming an important resource for many areas of research and public policy, including pesticide exposure and regulatory risk assessment. However, the complexities of biomonitoring for rapidly metabolized chemicals have often been ignored, with inappropriate steady-state assumptions used in place of more realistic time-dependent models. In this paper we present an integrated Bayesian exposure and pharmacokinetic model that combines dietary recall data individually matched to biomonitoring data, with pesticide residue information to predict exposures to lambda-cyhalothrin that are calibrated to urinary pyrethroid metabolite 3-phenoxybenzoic acid (3PBA) measurements. Our one compartment pharmacokinetic model modified to include metabolism and bladder storage delay closely replicated measured urinary 3PBA results over time from an independent lambda-cyhalothrin controlled dosing study. Additionally, the model predicted urinary 3PBA concentrations similar to the actual measurements recorded in the National Health and Nutrition Examination Survey (NHANES), a nationally representative sample. Using Approximate Bayesian Computing (ABC) to accept or reject Monte Carlo exposure candidates, we accepted only those exposure estimates resulting in predicted urinary 3PBA concentrations within thresholds of either an order of magnitude or half an order of magnitude from their respective measured values. The correlation between the modeled 3PBA measurements and the NHANES measurements was 29% and 65% for the two ABC thresholds, with acceptance rates of 76% and 43%, respectively. The median measured urinary 3PBA concentration was 0.38 μg/g creatinine, whereas the median value of the unfiltered Monte Carlo predicted urinary 3PBA concentrations was 1.0 μg/g. With ABC, the median predicted 3PBA concentration decreased to 0.89 and 0.76 μg/g respectively for the two thresholds. Restricting the Monte Carlo predictions using ABC brought the median predicted urinary 3PBA concentration closer to that of the NHANES measured values, while increasing the correlation between the predictions and measured values, making the final accepted collection of Monte Carlo exposure estimates more realistic. Bayesian integration of exposure and pharmacokinetic modeling with measured biomarkers is a powerful tool for producing more realistic exposure and risk assessments.

Bayesian integration of monte carlo exposure and pharmacokinetic modeling for lambda-cyhalothrin with urinary 3PBA measurements

Fall 2022 abstracts, posters, and presentations will launch in mid-September.

ACS Fall 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.