Workshop Report 24

Experiences with higher tiered assessment of persistence

The aim of this session was to describe experiences in the assessment of biodegradation and persistence using higher tiered tests. A series of presentations addressing issues including (i) experiences with the OECD water-sediment transformation test, (ii) the definition and characterisation of non-extractable residues, (iii) strategies to identify transformation products, (iv) the impact of light on degradation in aquatic-sediment environments and (v) the challenges posed by testing complex or difficult substances. This was followed by a panel discussion and two syndicate sessions.

Jon Ericson (Pfizer) described some of the experiences that the pharmaceutical industry has gained with the OECD 308 Sediment-Water Transformation Test (OECD, 2002a) since its introduction into the European Medicines Agency (EMA) environmental risk assessment guidance in 2006 (EMA, 2006). Four pharmaceutical companies (Pfizer, GlaxoSmithKline, AstraZeneca and Novartis) have collated the data for 31 pharmaceutical compounds to determine how the data is used and what can be learnt from the studies. Jon challenged the purpose of the study as it is used under the EMA regulations versus its original design. Within EMA, the OECD 308 test is used to simulate degradation of a down-the-drain chemical such as a pharmaceutical within a river; however the test was originally designed to simulate degradation of plant protection products in shallow ditches exposed via spray drift. The study is also used as a trigger to assess effects on sediment organisms if greater than 10% of the substance is present in the sediment layer at any time point after 14 days.

Jon described and illustrated the OECD 308 test system and summarised the key endpoints that are derived from the study. Some of these endpoints were not always possible to assess for all studies. These included the chemical half-life in sediment and the total 14 C half-life for the total system. Failure to measure these half-lives was attributed to the formation of NER. This was true for all groups of pharmaceuticals (neutral, cationic and anionic), although a greater proportion of NER and lower level of transformation products was formed with cationic pharmaceuticals. This NER effect had an impact on the half-life for the total system leading high variability for cationic and neutral pharmaceuticals. A mean parent half-life for the total system of 56 days ±79 days was observed across all classes of the pharmaceuticals tested. Jon reported that there was no correlation between dissipation rate and K d or K oc within the data set. However, the amount of total parent found in the water and sediment at day 50 and day 100 correlated well to the parent total system half-life. Jon described this observation as a possible case for having a single point screening study rather than a full OECD 308 study.

In summary, Jon concluded that: (i) the high level of NER formed in the studies posed an extraction and interpretation challenge that limits the use of the data in risk assessment unless the NER is considered removed and unavailable, (ii) the data from the OECD 308 are not used to revise the predicted environmental concentration (PEC) for surface waters or sediment and it is unclear how to do this within the EMA guidance should the PEC need revision, (iii) a separate sediment biodegradation test that is relevant to pharmaceutical risk assessments is required, (iv) the total system half-life should be used to refine the PEC within the risk assessment, and (v) the OECD 308 study should not be required at the screening level of a pharmaceutical risk assessment.

Graham Whale (Shell) gave a presentation outlining the challenges with undertaking soil biodegradation assessments on a complex hydrocarbon substance. The current recommended method uses the OECD Guideline 307 for testing chemicals: Aerobic - Anaerobic Transformation in Soil (OECD, 2002b) which was originally designed to provide degradation rate data for crop protection products. However, it is apparent that this test is now being undertaken to provide data for other ‘chemicals’ under the auspices of the EU REACH regulations.

Assessing the fate of complex substances raises many issues and there are scientists who question both the applicability and relevance of the current test guidelines. For example, the current approach recommended by CONCAWE (Conservation of Clean Air and Water in Europe) for complex hydrocarbon substances is to model persistence of the constituent hydrocarbon blocks.

In the case presented by Graham an OECD 307 assessment of a gas to liquid (GTL) fuel was requested by ECHA to determine potential persistent hydrocarbon components of the substance which may warrant further investigation. Consequently, a series of OECD 307 studies were undertaken on GTL fuel and individual alkanes. Difficulties were encountered with these studies that related to analytical constraints, different physico-chemical properties of components and dose rates at which the test can be conducted. All of these factors can complicate interpretation of results and it needs to be recognised that, when using an OECD 307 type soil study to assess the fate of components of complex substances, the objectives will differ to those for ‘standard’ OECD 307 studies.

A key recommendation was the inclusion of sterile (abiotic) dosed soil systems to assess the losses by abiotic factors (e.g. volatilisation and/or non-extractable residues). By incorporating sterile controls the OECD 307 test has potential to improve the understanding of the fate of components of complex substances like GTL fuel in soil. For example, it has potential to give an indication of the disappearance rate of components and identify recalcitrant components which may warrant further investigation. For the case study presented it appeared that the predicted half-lives of alkanes were conservative and, that no additional bioaccumulation assessments of the components of GTL fuel were warranted (based on the premise that even if some remain in soil they will not be bioavailable to soil organisms because they cannot be extracted using acetone and hexane).

To reiterate the key points although incorporating sterile controls can provide an indication of physical losses, for complex substances like the GTL fuel the OECD 307 test will ultimately determine ‘disappearance’ of components. This raises the issue that for complex substances it is unlikely to be feasible to calculate biodegradation rates per se using the OECD 307 test, and the data from such studies needs to be put into an appropriate context in an overall risk assessment strategy.

Non-extractable residues

Two presentations by Charles Eadsforth (Shell, UK) and Caren Rauert (Umweltbundesamt, Germany) on non-extractable residues (NER) were followed by a lively discussion:

Formation of non-extractable residues is regularly observed in studies on the fate of organic chemicals in soil. NER formation may be interpreted as a specific form of compound persistency – a ‘hidden hazard’ - or as a detoxification step – a ‘safe sink’. Despite the considerable scientific progress made in analysing NER and identifying their binding types, these insights, which are described in the following section, have not yet been incorporated into the regulatory risk assessment framework.

Following the ECETOC workshop on “Significance of bound residues in environmental risk assessment” in 2009, two Task Forces were set up to (1) understand the relationship between extraction technique and bioavailability and (2) develop interim guidance for the inclusion of non-extractable residues in environmental risk assessment. Charles Eadsforth gave a summary of the aims, objectives and outputs of these two ECETOC Task Force activities. The goal of the first Task Force was to address knowledge gaps in the relationship between bioavailability and extraction technique with regards to bound and non-extractable residues with the ultimate goal being the development of a standard framework for intelligent extraction strategies. A number of residue ‘categories’ were defined (dissolved, readily desorbed, slowly desorbed, irreversibly sorbed and incorporated) as well as the terms bioavailable and bioaccessible which were aligned with each type of residue within the framework model. It was decided to differentiate residues termed ‘reversibly bound’ into those ‘readily desorbed’ and ‘slowly desorbed’. This differentiation was based on the solvent strength necessary to extract each type of residue and led to the development of the extraction regime to tie in with the framework model.

To be able to better predict the chemical dynamics once a chemical enters the soil, it is necessary to understand its interaction with the soil matrix. Generally, chemicals which were most strongly associated with the soil (and least bioaccessible) were either covalently bound to the soil, or physically sequestered and trapped in soil pores. Other interactions which were shown to lead to NER or slowly desorbed residues included ionic and ligand exchange. Chemicals were also shown to interact with the soil matrix via Van der Waals forces, hydrophobic partitioning, charge transfer complexes and hydrogen bonds, these interactions are generally thought of as weaker and most likely to lead to desorbable residues. The various interactions studied (and their bond strength ranges) were aligned with the extraction regime and framework model.

One of the major issues of particular concern with regards to environmental risk assessment is the potential for future re-release of NER. For example, there is evidence that physical processes such as freeze-thaw and wet-dry cycling can cause the release of sequestered residues via the breakup of the soil matrix and soil organic matter (SOM). Additionally, chemical and biological processes such as microbial metabolism and pH changes have been found to cause the release of NER. The current literature suggests that the amounts of NER released by such processes do not pose an environmental risk. However, it was agreed that further research is necessary in this area, especially with regards to release caused by physical processes because of the current paucity of studies on this topic.

One of the key components of soil is organic matter and the potential interactions between chemicals and this complex soil constituent are poorly understood and warrant further research. However, some steps have been undertaken to redress this situation and the ECETOC Task Force has developed a framework model and extraction scheme (ECETOC, 2013a). Furthermore, it is expected that research in this area will greatly increase over the coming years as environmental risk assessment of chemicals in soil becomes an increasingly important issue.

In a workshop held at the German Federal Environment Agency (UBA) in June 2010, a slightly different terminology than that used by ECETOC was used to describe NERs. Caren Rauert gave an overview of the key outputs of this UBA Workshop. ‘Type 1’ NER were defined as those substance molecules that may be remobilised, possibly over prolonged times, ‘Type 2’ NER as those that are unlikely to be released in their original structure under environmental conditions. Finally, NER can also be formed via incorporation of single labelled atoms or small fragments from the original substance into biomass. These ‘biogenic’ NER are no longer structurally related to the original substance. While the formation of Type 2 and biogenic NER can be considered a ‘safe sink’, Type 1 NER would constitute a ‘hidden hazard’. Where no information on their nature is available, NER should in principle be assumed to belong to Type 1 (i.e. worst case scenario), unless experimental data is provided to support categorization as Type 2 NER or as biogenic.

Formation of Type 1 NER should have implications on the environmental risk and hazard assessment. In particular, their potential for substance remobilisation will impact groundwater risk assessment and persistence assessment. Existing trigger values and decision criteria for NER formation were deemed inappropriate for addressing those concerns; hence, a need for developing new criteria was identified.

In a recent publication (Kaestner et al, 2013) a classification scheme has been proposed for differentiating type I NER (xenobiotic, sequestered), type II NER (xenobiotic, covalently bound), and type III NER (biogenic), respectively, with decreasing environmental risk in the order I, II, and III NER. The further development of extraction schemes for separating type 1 NER and type 2 NER (and type 3 NER) was suggested, although it seems necessary to distinguish different schemes for different substance classes. Another way forward could be the identification and quantification of the fraction of biogenic residues within the NER (Nowak et al, 2013).

Identification of degradation products and their risks

Kathrin Fenner (EAWAG, Switzerland) presented strategies to identify degradation products and their risks. At the higher tiers of chemical risk assessment, regulatory guidance typically recommends the performance of simulation-type transformation studies to identify major transformation products (TPs). However, most risk assessment guidelines fall short of providing guidance on how the risk of identified TPs should ultimately be assessed.

In this presentation two possible approaches to identify risk-relevant TPs were presented and contrasted in terms of their advantages and disadvantages. This was based on earlier published work (Escher and Fenner, 2011). The default approach recommended in most regulatory risk assessment frameworks is exposure-driven, i.e. chemical-analytical identification of major TPs followed up by their synthesis and subsequent effect testing. Recent approaches to speed up TP structure identification (see Helbling et al, 2010) such as high-resolution mass spectrometry combined with high-throughput data analysis tools were discussed in this context. An effect-driven approach based on toxicity was presented as an alternative, potentially more direct way of identifying toxicologically relevant TPs. In this approach, samples from simulation studies are not only subjected to chemical analysis, but are also analysed with one or more bioassays to follow the development of toxicity over the course of the experiment. Comparison of parent compound concentration and toxicity development over time then indicates whether any toxicologically relevant TPs are formed.

Both of the above-mentioned experimental approaches are laborious and time consuming suggesting that there is a role for models for prioritisation of TPs which warrant further investigation. A model to estimate relative concentrations of pesticide TPs in surface waters was presented and its performance assessed relative to measured field data. Further, a model for estimating plausible ranges of toxic effects of TPs relative to their parent compounds was discussed. A combination of such models could potentially help to estimate the contribution of TPs to overall environmental risk caused by the release of a given parent compound.

Experiences with higher tier study designs to investigate the fate and behaviour of chemicals in the environment

Robin Oliver (Syngenta, UK) presented his experiences with higher tier study designs to investigate the fate and behaviour of chemicals in the environment. Many regulatory risk assessments for chemicals are based on laboratory studies in which the key processes of sorption, hydrolysis, photolysis and microbial degradation are evaluated separately in simple, standardised systems, in accordance with the appropriate guidelines. These studies provide information on the fate and behaviour of the chemical in soil, sludge, sediment and water.

Over recent years Syngenta has developed test systems to investigate the potential significance of degradation resulting from indirect photolysis and metabolism by phototrophic organisms in soil and sediment / water systems. A semi-field aquatic test system has also been developed to enable the determination of the rate and route of degradation, when multiple processes are acting together.

Study results showed that the overall rate of photodegradation in natural waters is a combination of direct and indirect photolysis and, in some cases, light scavenging by constituents of the water can reduce the rate of direct photolysis to a greater extent than is compensated for by indirect photolysis. These findings suggest that this will only be significant for compounds where direct photolysis is very rapid and the overall photodegradation rate in natural waters will still be fast. For those compounds that are not degraded very rapidly by direct photolysis, photodegradation in natural water is likely to be significantly faster than that observed in a sterile buffer.

In his talk, Robin Oliver had described aquatic simulation studies in which algae had been included in an attempt to more realistically mimic environmentally relevant conditions. It was thought that generation of hydroxyl and other oxygen radical species by virtue of the algae’s presence might contribute to the increased degradation seen in these studies, although this mechanism was not described in detail.

In conclusion there are studies indicating that under more realistic environmental conditions other factors such as light and presence of algae or soil surface dwelling phototrophs can increased the relevance and realism of higher tiered exposure studies as well as leading to more rapid degradation of chemicals. However, there is currently no clear understanding of how this could be incorporated into the determination of environmental persistence for regulatory risk assessments.