Nam-Huy Leduc

Project title: A study of volatile organic compound metabolites in human and canine urine as a biomarker and the relationship to proximity to hydraulic fracturing and natural gas drilling wells

Degree: MPH | Program: One Health (ONE) | Project type: Thesis/Dissertation
Completed in: 2020 | Faculty advisor: Peter Rabinowitz


Background: Hydraulic fracturing (fracking) as a process for natural gas extraction has potential to expose nearby residents to environmental hazards, but the extent of the hazard remains poorly understood. Some studies have suggested that animals can function as sentinels for human exposures. We evaluated urinary biomarkers of volatile organic compounds (VOCs) of humans and their canine pets in households in southwestern Pennsylvania as a follow-up from a pilot study that had found an association between proximity to drilling well pads and reported health symptoms, including respiratory symptoms and dermal issues. Methods: Participants consented to a questionnaire survey and provided urine samples for one human and one dog (if present) for each household. This data set includes 109 human subjects and 34 dogs, using information on potential VOC exposures in the past 48 hours of sample collection. Covariates such as burning fuels, using gas-powered equipment, or smoking are some of the factors considered in performing multivariate stepwise regression models. We developed a Z-score Index as a normalizing tool to standardize the wide range of varying metabolite concentration levels analyzed from the urine samples. We also examined the 31 households that shared both a human and a dog subject using Pearson’s correlation method to better understand the relationship between metabolites, with the suggested hypothesis that animals are at a greater susceptibility given higher and more frequent contact to environmental mediums. The primary software used is in R programming for statistical analysis. Results: Our findings indicate that smoking has a significant effect on most metabolite levels in the study. Correlation matrices between dog and human metabolites did not show direct same-metabolite associations, but offered other possible correlations to similarly grouped metabolites of VOC compounds, such as the BTEX group (Benzene, Toluene, Ethylbenzene, Xylene). Additionally, our stepwise regression models generated significant exposures to VOCs that are greater contributors to metabolite concentration levels than our hypothesized distance from the nearest gas well. By stratifying each metabolite, we modeled all covariates to the metabolites and identified the most significant covariates to each species group. Certain metabolites are shown to have higher presence in the dogs than compared to the humans in this examination. URI