Background

iPie_collage-2Human-use active pharmaceutical ingredients (APIs) are generally excreted and emitted into the sewerage system following use. The compounds may then be released into surface waters or enter terrestrial ecosystems when sewage effluent is emitted to river systems or used for irrigation or where sewage sludge is applied as a fertilizer to agricultural land (Kinney et al. 2006; Ternes et al. 2004; Cunningham et al., 2012). Consequently, a variety of APIs have been detected in the natural environment across the world (e.g., Hirsch et. al. 1999; Kolpin et al. 2002; Ramirez et al. 2009). Although reported concentrations are generally low, many APIs have been detected in a variety of hydrological, climatic, and land-use settings and some can persist in the environment for months to years (e.g., Monteiro and Boxall 2009).

 

APIs are biologically active compounds that are designed to interact with specific pathways and processes in target humans and animals. Concerns have therefore been raised about the potential effects of APIs in the environment on human and environmental health.

Over the past 15 years, a substantial amount of work has been done to determine the occurrence, fate, effects, and resulting risks of APIs in the environment. Regulations have also been developed regarding the assessment of environmental risks of APIs (e.g., CDER 1998; EMEA 2006; WHO 2011) and the pharmaceutical industry has done a vast amount of testing to meet these regulatory requirements. Currently, environmental assessments of APIs are typically performed at the end of the development process, i.e. when the API is close to an application for market approval. Concerns have been raised over whether the standard OECD testing methods for examining chronic effects on organisms (CDER 1998; CHMP 2006; CVMP 2000, 2004) will identify ecologically important effects of specifically acting APIs (Boxall et al., 2012; Brooks et al. 2009; ECETOC 2008). The effect of the nonsteroidal, anti-inflammatory compound diclofenac on vulture populations (Oaks et al. 2004) is one example of an affected non-target organism that would not have been predicted from standard studies.

 

The challenge is to identify potential environmental risks of new APIs during the early stages of the development process, such that intelligent and efficient testing strategies can be defined. This can be realized by developing predictive models based on preclinical and clinical pharmacological and toxicological data for existing APIs or by using Quantitative Structure-Activity Relationships for environmental endpoints. For example, knowledge of the presence or absence of API targets across a wide range of taxa could be invaluable in identifying those organisms and life stages of organisms that are most likely to respond to exposure to an API and which should therefore be specifically targeted in the risk assessment process (Ankley et al. 2007; ECETOC 2008; Gunnarsson et al. 2008; Huggett et al. 2003; Seiler 2002; Trudeau et al. 2005). Comparative biochemistry, genomics, and other “omic” technologies also offer potential tools for early identification of APIs of potential concern, as well as the most sensitive and vulnerable species.

In addition to addressing potential environmental risks of APIs at an early stage of the development process, there also is a need to address the environmental risks associated with legacy APIs. More than 3,000 APIs are currently in use and sufficient data to adequately assess environmental risks is only available for a small proportion of these. While it would be beneficial to understand the potential effects of the untested APIs, it would be a mammoth task to experimentally assess the hazards and environmental occurrence of all of these in a timely manner.

 

Prioritisation approaches offer a potential solution to focus monitoring, testing, and research resources and to identify those APIs that are likely to pose the greatest risk in a particular situation. Ideally, these prioritisation approaches should not require extensive experimental testing but exploit a combination of the significant amount of data that companies have already generated on the fate and ecotoxicity of APIs, some of the predictive models described above, and targeted experimentation. For example, a lot of information is available from preclinical and clinical studies on the properties and effects of APIs which could be used to support predictions of the environmental impacts of an API.

 

By accessing the wealth of existing data and combining these with predictive models for environmental fate and hazards, it may be possible to establish whether low levels of a pharmaceutical in the environment constitute a threat to environmental and human health without the need for extensive experimental testing (Ankley et al. 2007; Berninger and Brooks 2010; Huggett et al. 2003; Seiler 2002; Winter et al. 2010). This could be invaluable in the design of testing strategies and for prioritisation of legacy APIs.