Approaches to identify exposure to real-life chemical mixtures in the general population

Abstract

Humans are exposed to large number of chemicals from various sources in their environment. Current risk assessment approaches often focus on single chemicals, sources or routes, not accounting for possible combinations of chemicals and overlooking potential additional risks associated with chemical mixtures. This thesis focusses on how to accurately measure and describe exposure patterns of chemical mixtures in the general population. Measuring exposure to chemical mixtures in real-life situations at a population level is a complex task. The preferably individual measurements can be obtained either externally (e.g. silicone wristbands) or internally (human biomonitoring). In my thesis, chemical mixtures are described as combinations of manufactured chemicals that co-occur within the same individual or sample. My thesis revolves around three different approaches to identify chemical mixtures in the general population. 1) The first approach focusses on describing co-occurrence patterns of chemicals in existing human biomonitoring datasets, combining a graphical correlation network model with a clustering algorithm. This method visualizes and summarizes these co-occurrence patterns by identifying highly correlated groups or clusters of chemicals. We demonstrated that presenting biomonitoring data in networks facilitates the identification of exposure patterns that contribute to the observed exposure levels in the samples. The application of community detection with a clustering algorithm was instrumental in identifying patterns within and between chemical families. 2) The second approach involves measuring chemical mixtures at individual level, employing external and internal measurements. 3) The third approach refers to the analytical measurement of chemical mixtures in urine samples, for which a suspect screening approach based on high resolution mass spectrometry was applied, enabling detection of a broad range of biomarkers in a single sample. The second and third approach were applied on pesticides, as an example of a mixture of chemicals. By utilization of wristbands we were able to detect multiple pesticides over a longer period of time, reflecting highly individual exposure profiles. It was demonstrated that a harmonized pan-European sample collection, combined with suspect screening provided valuable new insights into the presence of pesticide mixture exposure in the European population. Forty pesticide biomarkers corresponding to 29 different pesticides were confidently identified across six countries. Some variation but no consistent pattern in the probability of detection of pesticide biomarkers was observed based on residential location or season of urine sampling. Pesticide mixture patterns in the adult populations of the Netherlands and Switzerland were analyzed by their semi-quantitative levels. Urinary concentrations of the two most frequently detected pesticides (acetamiprid and chlorpropham) showed an inverse association with high organic vegetable and fruit consumption. In both countries, detection rates and co-occurrence of pesticides in the same urine sample were typically low. This thesis highlights the importance of understanding and measuring the mixture of chemicals we encounter. Knowledge of what these mixtures are helps us to figure out the risks to human health. I recommend to incorporate the approaches of this thesis in future studies to learn more about human exposure to chemical mixtures, and that research and sharing data is conducted in a harmonized way.