Workshop Report 29

Syndicate Session 1: Full utilisation of the chemical activity concept for non-polar organic chemicals (Log KOW ≥ 2)

Participants
Moderator:    P. Mayer
Moderator:     F. Gobas
Rapporteur:    A. Jahnke
T. Bowmer
J. Appel
T. Henry
D. Knapen
L. McCarty
T. Parkerton
S. Nørgaard Schmidt
F. Smedes
J. Tunkel
D. van den Meent

Introduction
Thermodynamic activity (also frequently referred to as chemical activity) was first proposed by Lewis (1901) and has been successfully applied in many areas of science, engineering and medicine to better understand the behaviour of chemical substances and pharmaceutical drugs. In the field of environmental chemistry and toxicology, the activity approach has also been put forward as a theoretically well founded and practical approach to study and manage environmental contaminants (Di Toro et al., 1991; Reichenberg and Mayer, 2006; Mackay and Arnot, 2011). Chemical activity is a multimedia exposure parameter, which provides a metric for comparing, relating and combining exposure and toxicity data from various sources, which represents a potentially useful tool for environmental risk assessment and management.
The overall objective of WG1 was to explore the practical utilisation of the chemical activity concept for non-polar (i.e., neutral) organic chemicals in environmental risk assessment.  Non-polar organic chemicals with a log KOW ≥ 2, were the focus of the discussions, because the concept of chemical activity has been largely developed for organic chemicals with these properties. Hence, any potential application of the activity approach in environmental assessment and management is most feasible and likely for non-polar organic chemicals with a log KOW ≥ 2, as this represents the current understanding of the domain of applicability. The chemical activity of non-polar organic chemicals can both be measured and calculated. Non-polar organic chemicals  often only exert baseline  toxicity, which has been demonstrated to initiate in a well-defined and narrow range of chemical activities, of between 0.01-0.1, thus providing the potential for a novel approach for conducting risk assessments. This rather narrow activity range related to the initiation of baseline toxicity for non-polar organic chemical also provides a basis for the assessment of mixtures of non-polar organic chemicals, whereby the chemical activity of each of the individual components of a mixture can be summed. The assessment of the effects of chemical mixtures has been a long-term goal in the field of environmental toxicology, and therefore the relationship between mixture effects for non-polar baseline toxicants and chemical activity provides a potentially novel method for interpreting the toxicity of mixtures.
In an effort to help shape the discussion within the workgroup, participants of WG1 presented their own individual research commenting on:
Current or potential applications of thermodynamic activity in their area of research and/or management of the environmental risks of commercial chemicals.
Actual or potential merits of the application of thermodynamic activity in their area of research and/or management.
Possible limitations of the application of thermodynamic activity in their area of research and/or management.
Areas for further research to advance the application of thermodynamic activity in their area of research and/or management.
The purpose of the presentations, questions and subsequent discussions was to address the potential application of the activity approach in environmental risk assessment and management.
Summary of presentations and the resulting discussions:
Each of the presentations given during the WG1 session, and summarised here, led to stimulating discussions, which were useful in challenging the limits of applying the chemical activity approach in risk assessment and management of chemicals, as well as emphasising where the approach is to be most useful.
The role of passive dosing in ecotoxicological test studies for non-polar organic chemicals with log KOW ≥ 2 was presented, in which it was demonstrated how baseline toxicity  is well related to chemical activity, and that the absence or presence of baseline toxicity for solid chemicals depends on their maximum chemical activity. The data presented demonstrated how baseline toxicity of individual PAHs was initiated at chemical activities between 0.01-0.1 (Schmidt et al., 2013b) and further showed that effects caused by mixtures of two to three PAHs could be plotted on a single “sum activity” (∑a) response curve, indicating the same activity range (Schmidt et al., 2013a). Two additional studies were presented that:
demonstrated how mixtures of individually non-toxic PAHs could cause toxic effects (Smith et al., 2013)
assessed the influence of combining stressors, i.e., chemical and drought stress (Schmidt et al., 2014).
Lastly, recent research involving pyrethroids, which were shown to exhibit toxicity at much lower activities, gave strong evidence of the excess toxicity related to the exposure of this group of chemicals. Data mining activities presented at SETAC Salt Lake City were also discussed that partly covered an algal growth inhibition study (Schmidt and Mayer, 2015) and partly included additional work by Christensen et al., 2009.
Additional experimental evidence of baseline toxicity occurring between 0.01 and 0.1 for PAHs and chlorinated benzenes was also presented and discussed (Tcaciuc, 2015). In an effort to explore the use of the chemical activity concept for complex real-world mixtures, seven contaminated sediments were equilibrated with polyethylene for use in further toxicity experiments. For these seven sediments, the extracts from the equilibrated polyethylene were run on a GCxGC-FID to separate the compounds based on boiling point and polarity. Using the unidentified chemicals’ positions in GCxGC space, estimates of partitioning into lipids (and therefore the potential for baseline toxicity) were calculated and this information was then used for predictions of baseline toxicity of mixtures. Passive dosing experiments using Daphnia magna confirmed that this method predicted baseline toxicity for six of the sediments with the seventh sediment showing signs of excess toxicity (40% immobilisation of D. magna instead of the <5% predicted) (Tcaciuc, 2015).
As a further illustration of the application of the chemical activity concept, its use in interpreting data from various environmental media was presented and discussed. Environmental media that could be characterised as being rich in “transporter agents”, and which effectively facilitate equilibration, i.e., sediment rich in organic carbon (Mäenpää et al., 2011; Jahnke et al., 2012) and lipid-rich biota (Jahnke et al., 2009), enabled the measurement of equilibrium partitioning data and comparison of measured activity ratios using polymer “chemometers” (Jahnke et al., 2014). Conversely, passive sampling in environmental media such as water, typically requires an extrapolation to equilibrium, which can be done based on the loss of performance reference compounds spiked before sampling is initiated (Booij and Smedes, 2010; Rusina et al., 2010).
The relationship between chemical activity and the adverse outcome pathway for baseline toxicity was also discussed.  Due to non-specific interactions in relation to the molecular initiating event, a complete adverse outcome pathway (AOP) for baseline toxicity with causal relationships of high scientific confidence has been difficult to delineate. At least two different putative pathways, mainly focusing on apical endpoints, have been described in the literature. An important topic for debate focused on which cellular membranes are being targeted by baseline toxicants. Given the fact that membranes are dynamic structures incorporating many different regulatory processes, it is reasonable to assume that different membrane types are capable of reacting differently to the presence of baseline toxicants, leading to potential specific effects on membrane-bound processes (e.g., electron transport chain). While the chemical activity range of 0.1 to 0.01 for acute baseline toxicity (i.e., mortality) has been established with a relatively high level of confidence, a high level of uncertainty remains with respect to sub lethal effects. Since organisms are capable of activating compensatory mechanisms at lower exposure levels – before the onset of system breakdown and failure – organelle membrane-specific processes could possibly cause sub lethal effects to be significantly different among organelles, tissues and biological species, at different chemical activities.  The relationship between baseline toxicants and chemical activity for chronic effects thus represents a key data gap that needs to be addressed in order to better assess the utility of the chemical activity in risk assessing chemicals where exposure is characterised by a steady-state emission at low concentrations.
The group briefly discussed reactive chemicals and noted that this could be an area where the chemical activity concept could help in the development of better predictive tools.  Discussion was supported by reference to a study reporting on the toxicity of a series of alkyllactates to daphnia, algae and fish (Bowmer et al., 1998), which discussed potential modes of action for this group of chemicals across different species. It was emphasised that the ecotoxicity profiles of many chemical groups with a specific mode of action may differ greatly, e.g. triazine herbicides which inhibit photosynthesis have severe effects on plants including green algae – while unsurprisingly for fish and crustaceans their effects are in line with baseline toxicity (see Solomon and Cooper, 2008) but there are more complex examples. Esters such as the alkyllactates hydrolyse readily to a polar alcohol,  and a reactive lactic acid molecule in the cells of an organism.  However, different toxicity responses across algae, crustaceans and plants are known to occur as a result of exposure to this class of chemicals. Mode of action is thus not so much a  property of the chemical but of its (complex) interaction with a species. In the regulatory context, there is a role for novel tools in predicting aquatic toxicity, and it was suggested that chemical activity may have a complementary role to play. However, it was pointed out that such tools needed to be robust, validated, and well packaged prior to being implemented into existing hazard and risk assessment frameworks, as further emphasised below.
A compilation of available acute toxicity data for zebrafish for hydrocarbons, alcohols, ethers, ketones and chlorinated solvents was also presented and discussed. LC50 data were expressed in terms of activity and found to occur at a value of > 0.01, further supporting the relationship between chemical activity and non-polar organic chemicals that act as acute baseline toxicants. Recent toxicity test results obtained using passive dosing yielded Ea50 data that were less variable than earlier literature data. Available chronic toxicity data across a range of fish species focusing largely on embryo-larval survival and larval growth effect endpoints when expressed in terms of activity fell above 0.001. A positive trend was also noted between the activity corresponding to chronic effects in fish and the log KOW of the test substance. An increasing number of substances with log KOW > 6 were observed not to be chronically toxic at activities > 0.01. Furthermore, experiments were shown to assess the potential extension of the chemical activity concept to complex petroleum substances. The results reported chemical activities for acute effects data for 11 organisms including fish, invertebrate and algal test species ranging from 0.06 to 0.39 and three chronic studies with daphnia, trout and algae approximately an order of magnitude lower falling between an estimated activity of 0.01 and 0.05. These results suggested that no. 2 fuel oil exhibits chronic effects at chemical activities that are generally consistent with that observed for acute baseline toxicity.
The group discussed whether chemical activities should preferably be measured in various media and toxicity tests conducted at controlled activities, or whether it would be sufficient to translate existing monitoring and toxicity data into the chemical activity space. There was consensus that while directly measured data generally are preferable, a translation of existing data can also lead to enhanced understanding on the thermodynamic controls of the environmental fate and toxicity of environmental pollutants.
Discussions based on the following questions which had been distributed in advance to the workshop participants are outlined below.
Can 0.01 (i.e., 1 % of liquid solubility) be used as a chemical activity benchmark to distinguish baseline toxicity and excess toxicity?
Could the observation of non-toxicity at chemical activity of 1 (100 % of liquid solubility) be used for categorising a chemical as being non-toxic?
Is it possible and meaningful to set a general predicted no-effect activity (PNEA) for baseline toxicity?
Is it possible and meaningful to set a general predicted no-effect activity (PNEA) for mixtures with regards to baseline toxicity?
Is it scientifically correct to assess the sorptive capacity or “solubility” of neutral hydrophobic organic chemicals in non-aqueous phases as the product of the non-aqueous-water partition coefficient and the aqueous solubility?
What are the inherent assumptions in a comparison of activities of neutral organic chemicals among various environmental media?
Is it possible and meaningful to include an activity framework in AOP analysis?
What are the low-hanging fruits for the application of the activity approach?

Can 0.01 (i.e., 1 % of liquid solubility) be used as a chemical activity benchmark to distinguish baseline toxicity and excess toxicity?
The group agreed that the available evidence in the scientific literature with respect to this question provides compelling evidence that this is indeed the case. While the approach in itself is valid, the exact number may have to be adjusted, e.g., for instance better characterisation is needed between acute versus chronic baseline toxicity. For chronic toxicity, the threshold is expected to be lower than for acute toxicity. Establishing and using a chemical activity threshold, such as at 0.01, is seen as a good starting point.  For instance, in relation to assessing the potential risks to the exposure of complex mixtures in the environment, baseline toxicity may be the most critical issue due to the additivity of the activity of hundreds of compounds present at low concentrations. Setting an activity of 0.01 is a useful way to distinguish between baseline and excess toxicity, i.e., if effects occur at activities below the threshold, other modes of toxic action may be implied. The use of a chemical activity threshold value does not, however, allow for conclusions about which mode of action other than baseline toxicity occurs.
From a regulatory perspective, the suggested threshold was considered useful for screening and prioritisation of chemicals that are subject to risk assessment. Another advantage mentioned was that the activity concept offered a way to avoid or reduce animal testing. However, the group emphasised the importance of defining careful rules regarding the applicability domain of the chemical activity concept, since the approach is currently limited to chemical effects associated with baseline toxicity of parent compounds and their mixtures, while it does not include various types of excess toxicity nor the effect of metabolites and other degradation products.
Could the observation of non-toxicity at chemical activity of 1 (100 % of liquid solubility) be used for categorising a chemical as being non-toxic?
The group agreed that this kind of concept already was contained in regulatory documents. However, it needs to be more carefully framed, e.g., rather than ”non-toxic” discussing ”apparently non-toxic” chemicals. While the concept in itself might be useful, the question of how to apply it was discussed.
WG1 participants agreed that the statement was generally acceptable, but that it needs to be qualified. Issues that were identified in this context that need particular awareness were the following:
kinetics, in particular in acute toxicity testing
diversity in physicochemical properties
that the concept is limited to the parent compound toxicity
levels above solubility that might occur in the environment
Another aspect that was addressed in the context of this question was how the chemical activity concept provides a stronger scientific approach for interpreting chemicals that may apparently be non-toxic, which represents an improvement relative to the use of an arbitrary toxicity cut-off at log KOW of 5.5, as used in current practice.
Is it possible and meaningful to set a general predicted no-effect activity (PNEA) for baseline toxicity?
The group agreed that a related PNEA value is meaningful, as a fraction of saturation and in case of lethal effects. A particular advantage is that no additional toxicity data are needed to set this limit, potentially (i) reducing the need for additional testing and use of laboratory animals, (ii) providing less uncertainty in the risk assessment process, and (iii) reducing the need to use assessment factors currently used in the derivation of PNEC values. However, the same restrictions and uncertainties discussed above apply regarding the applicability domain. Specifically, the use of PNEA values in risk assessment would only be useful for non-polar organics with log KOW > 2 and for which no excess toxicity is possible, i.e. that the only mode of action for the chemical is baseline toxicity.  It was briefly discussed by which factor below the range where baseline toxicity initiates (i.e., 0.01-0.1) this threshold should be defined, but the value itself needs to be set by the regulatory community, which most likely would benefit from the input of additional data.
Is it possible and meaningful to set a general predicted no-effect activity (PNEA) for mixtures with regards to baseline toxicity?
This question triggered much discussion. The group agreed that the conclusions of Q3 could be extended to the baseline toxicity of mixtures, meaning that a PNEA on a ∑a basis is sound and feasible, where the mode of action is limited to baseline toxicity. In these instances, the concept of chemical activity can help to identify the main drivers of toxicity while at the same time determining the baseline toxic potential of the mixture. The concept was considered particularly useful for site-specific risk assessment and for monitoring the reduction of toxicity during remediation of contaminated sites. However, as above, the limitations to the applicability domain need to be taken into account.
The group, however, could not form a consensus on whether it is meaningful and reasonable to set a chemical activity limit for individual compounds to constrain their contribution to the baseline toxic potential of environmental mixtures (e.g., 1 or 10 %) (see, e.g., Schmidt and Mayer, 2015). Some group members found this difficult to apply without knowing the composition of the environmental mixtures, whereas other members argued that such an approach was needed in order to account for the multitude of chemical emissions and the large number of chemicals present in the environment. The group did not reach any consensus on this question.
Is it scientifically correct to assess the sorptive capacity or “solubility” of neutral hydrophobic organic chemicals in non-aqueous phases as the product of the non-aqueous-water partition coefficient and the aqueous solubility?
The work group stressed the importance of taking into account the uncertainties. As an initial step, good quality non-aqueous/water distribution coefficients are needed, and a discussion was initiated whether the quality of the calculated values was sufficient, without coming to a conclusion. Hence, a substantial knowledge gap was identified for which more research is needed.

What are the assumptions in a comparison of activities of neutral organic chemicals among various environmental media?
A: For measured activities
Activity measurements require equilibrium partitioning between the sampler and the medium and also that the depletion of the medium is kept at a negligible level (i.e., below 5 %) during the sampling. For the issue of measuring activities in sediment either in the field (in situ) or with samples brought to the laboratory (ex situ), data were reported that showed higher activity measurements for ex situ compared to in situ sampling. Hence, the extent to which the activity can be conserved when bringing samples into the lab was discussed.
B: For calculated activities
When no measured activities are available, total concentration data, e.g., from monitoring programs, can be converted to activities. This approach is followed under the assumption of equilibrium partitioning and a certain capacity for one compartment, and the quality of the translation is a function of scaling to log KOW. In general, the work group recommends using good modelling practice, i.e., clearly describing the methods and assumptions. In particular, uncertainty analysis needs to be taken into account.
Is it possible and meaningful to include an activity framework in AOP analysis?
Using activity has clearly been identified as an alternative exposure basis and is expected to be a useful tool for exploring AOPs.
In addition to the first 7 questions addressed above, the group defined an additional question (below) to trigger brainstorming.
What are the low-hanging fruits for the application of the activity approach?
See Chapter 5 for details of opportunities to apply the activity approach.