Highly active human pharmaceuticals in aquatic systems: A concept for their identification based on their mode of action.

Widespread occurrence of traces of pharmaceuticals (ng/L to microg/L) has been reported in aquatic systems. However, their effects on the environment and their environmental risks remain elusive. Generally, the acute toxicity towards non-target organisms has been assessed in laboratory experiments, but chronic toxicity studies have been performed only rarely. The guideline issued by the European Medicines Agency in 2006 is aimed at estimating the potential environmental risks of human pharmaceuticals by a tiered approach. The predicted environmental concentration (PEC) of a compound is estimated in phase I, and pharmaceuticals having a PEC above or equal 10ng/L undergo phase II testing. Otherwise they are not expected to pose a risk to the environment. Because some highly active compounds (HC) such as 17-alpha-ethinylestradiol, equine estrogens, trenbolone and the progestin levonorgestrel display adverse effects at concentrations below 10ng/L the question arises, whether additional HC compounds exist, and how they can be identified for undergoing environmental risk assessment. We addressed this question by searching for HC in the literature, and by developing a concept for identification of HC. The suggested mode of action concept is based on (i) the mode of action of the pharmaceutical taking the available toxicological information into account, (ii) the degree of sequence homology between the human drug target and the potential target in aquatic organisms and (iii) the importance of pathways affected by the pharmaceutical. We evaluated the mode of action concept by comparison to existing approaches, the fish plasma model (Huggett et al., 2003) and a QSAR model, called VirtualTox Lab (www.biograf.ch). All concepts result in similar classifications of the selected pharmaceuticals. However, there are some differences not only in the model assumptions, but also in its results. Our study leads to the conclusion that the mode of action concept is most suitable for the identification of HC. A refinement can be achieved by complementing this concept by the QSAR model (VirtualTox Lab), whereas the fish plasma model seemed to be less suitable due to the necessity of environmental concentration above 10ng/L for the identification of a risk. 2009 Elsevier B.V. All rights reserved.