Environmental risk assessment and human pharmaceuticals: limitations and future improvements

A cocktail of human pharmaceuticals contaminate surface waters worldwide in the ng-μgl-1 range. Adverse effects on non target organisms including endocrine disruption and alterations in behaviour and growth have been reported. All new pharmaceuticals require an environmental risk assessment (ERA) prior to market authorisation. The aims for this research were to (1) assess the limitations of the current ERA by comparing crude and refined predicted environmental concentrations (PECs) used in ERAs with measured environmental concentrations (MECs) from the literature; (2) interview key people working in the field of risk assessment and ecotoxicology of pharmaceuticals in order to establish expert opinion in the area; (3) to establish whether bioinformatics databases can be used as a potential tool to aid ecotoxicological tests for use in ERAs.

The scientific literature was data mined for environmental concentration data and compared with calculated PECs for ten pharmaceuticals carbamazepine, diclofenac, 17α ethinyl estradiol, fluoxetine, gemfibrozil, ibuprofen, paracetamol, propranolol, tamoxifen and Trimethoprim. An engagement exercise through questionnaire based interviews with representatives of regulatory bodies, water companies and pharmaceutical companies as well as academics involved in ecotoxicology was undertaken to establish experts’ views on pharmaceutical risk assessment and management. A genomic search for human drug target homologues in aquatic species for the ten selected pharmaceuticals was undertaken. Molecular docking experiments on two pharmaceuticals, diclofenac and ibuprofen were carried out for human drug target homologues in Daphnia pulex, (water flea) Oncorhynchus mykiss (rainbow trout), Salmo salar (Atlantic salmon) and Danio rerio (zebra fish).

The current environmental risk assessment may be insufficient to protect the aquatic environment. PECs can underestimate MECs due to the simplicity of the calculations and the assumptions underpinning them. The interviewees regarded the exposure assessment of the ERA including the PEC calculation as inaccurate and recommended using exposure modelling computer software as a potential solution. The bulk of the scientific literature had substantial deficiencies in the reporting of environmental data; setting reporting standards for peer reviewed journals may make such data more useful for regulators and policy makers. Interviewees felt that the current ecotoxicity tests would benefit from a more intelligent approach incorporating the mode of action of the drug. The bioinformatics results show that protein drug targets are highly conserved in some aquatic organisms but not others. The molecular docking results indicate that the cyclooxygenase (COX 2) primary drug target homologues are probably functional in O.mykiss, S.salar and D.rerio but not D.pulex. It appears from this data that bioinformatics and molecular docking indeed may be a useful tool to aid ecotoxicology tests by informing choice on relevant chronic test endpoints and directing sensitive species selection. Such techniques might contribute to more appropriately targeted ecotoxicity testing. Interviewees felt that the 10 ngl-1 action limit was an inappropriate mechanism to trigger ecotoxicological tests. The pharmaceuticals data analysis shows that many existing pharmaceuticals regularly exceed the 10 ngl-1 action limit for ecotoxicological assessment. A system of prioritisation is required to assess the need for retrospective risk assessment of these medicines.

This thesis provides an original analysis of the current environmental risk assessment of human pharmaceuticals and makes recommendations for improvements. A novel application of molecular docking utilizing the mode of action of the pharmaceutical has the potential to aid and direct ecotoxicological tests.