Improvements in Estimating Bioaccumulation Metrics in the Light of Toxicokinetic Models and Bayesian Inference.

Abstract

The surveillance of chemical substances in the scope of Environmental Risk Assessment (ERA) is classically performed through bio-assays from which data are collected and then analysed and/or modelled. Some analysis are based on the fitting of toxicokinetic (TK) models to assess the bioaccumulation capacity of chemical substances via the estimation of bioaccumulation metrics as required by regulatory documents. Given that bio-assays are particularly expensive and time consuming, it is of crucial importance to deeply benefit from all information contained in the data. By revisiting the calculation of bioaccumulation metrics under a Bayesian framework, this paper suggests changes in the way of characterising the bioaccumulation capacity of chemical substances. For this purpose, a meta-analysis of a data-rich TK database was performed, considering uncertainties around bioaccumulation metrics. Our results were statistically robust enough to suggest an additional criterion to the single median estimate of bioaccumulation metrics to assign a chemical substance to a given bioaccumulation capacity. Our proposal is to use the 75th percentile of the uncertainty interval of the bioaccumulation metrics, which revealed an appropriate complement for the classification of chemical substances (e.g. PBT (persistent, bioaccumulative and toxic) and vPvB (very persistent and very bioaccumulative) under the EU chemicals legislation). The 75% quantile proved its efficiency, similarly classifying 90% of the chemical substances as the conventional method.