Physiologically based kinetic (PBK) modelling and human biomonitoring data for mixture risk assessment

Human biomonitoring (HBM) data can provide insight into co-exposure patterns resulting from exposure to multiple chemicals from various sources and over time. Therefore, such data are particularly valuable for assessing potential risks from combined exposure to multiple chemicals. One way to interpret HBM data is establishing safe levels in blood or urine, called Biomonitoring Equivalents (BE) or HBM health based guidance values (HBM-HBGV). These can be derived by converting established external reference values, such as tolerable daily intake (TDI) values. HBM-HBGV or BE values are so far agreed only for a very limited number of chemicals. These values can be established using physiologically based kinetic (PBK) modelling, usually requiring substance specific models and the collection of many input parameters which are often not available or difficult to find in the literature. The aim of this study was to investigate the suitability and limitations of generic PBK models in deriving BE values for several compounds with a view to facilitating the use of HBM data in the assessment of chemical mixtures at a screening level. The focus was on testing the methodology with two generic models, the IndusChemFate tool and High-Throughput Toxicokinetics package, for two different classes of compounds, phenols and phthalates. HBM data on Danish children and on Norwegian mothers and children were used to evaluate the quality of the predictions and to illustrate, by means of a case study, the overall approach of applying PBK models to chemical classes with HBM data in the context of chemical mixture risk assessment. Application of PBK models provides a better understanding and interpretation of HBM data. However, the study shows that establishing safety threshold levels in urine is a difficult and complex task. The approach might be more straightforward for more persistent chemicals that are analysed as parent compounds in blood but high uncertainties have to be considered around simulated metabolite concentrations in urine. Refining the models may reduce these uncertainties and improve predictions. Based on the experience gained with this study, the performance of the models for other chemicals could be investigated, to improve the accuracy of the simulations.