By Jiaqi (Fiona) Wang, PhD researcher in Environmental Science at Radboud University (Nijmegen, The Netherlands)

Pharmaceuticals have been detected as emerging pollutants in global water bodies, potentially affecting aquatic organisms such as fish. Physiologically based kinetic (PBK) models can predict the active pharmaceutical ingredients (API) uptake, distribution, metabolism and excretion of fish and estimate API concentrations over time in fish organs such as brain and liver. To date, fish PBK models have been restricted to a few species (e.g., rainbow trout), where full sets of input parameters (preferably experimental values) were available. Additionally, over 60% of APIs are ionised at physiological pH values, while fish PBK models rarely account for ionisation. Our work aimed to develop a generalised fish PBK model applicable to a broader range of fish and substances. We assessed its performance for five pharmaceuticals (covering both neutral and ionic structures) that are of interest to PREMIER. Our work belongs to Work Packages 2.2 of the IMI PREMIER project.

To realise this aim, we first estimated the required input parameters as functions of fish (e.g. body mass) and chemical properties (e.g. octanol-water distribution coefficient). For instance, we derived statistical regressions of organ weight, cardiac output and the like as a function of body mass and temperature. For biotransformation, we took half-lives (HL) from a QSAR model developed by Arnot and co-workers. We also applied HL measurements after correcting for body mass and temperature differences. Finally, for model evaluation, we collected literature data on API concentrations measured in various fish organs after exposure to water.

Generally, our model is more accurate for ionisable compounds than any of the existing species-specific PBK models. The performance also improved when using experimental biotransformation HLs. Consequently, we will proceed to validate our approach for other APIs and refine estimated parameters with more measured pharmacokinetic data generated by other Work Packages.

Our PBK modelling process becomes (far) less intensive in terms of data requirements since we estimated input parameters mechanistically. Additionally, our in silico approach supports less animal testing and the “Replacement, Reduction, and Refinement” principles related to animal welfare. Moreover the model facilitates its application in chemical risk assessments for different fish, chemicals and aquatic environment conditions (pH and temperature) of concern. The results indicate whether potential significant levels of pharmaceuticals are present in fish (organs) to warrant further investigation.

This work was recently published in ES&T.

Happy read!