Perspectives for integrating human and environmental exposure assessments

STOTEN-18754; No of Pages 10 Science of the Total Environment xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Perspectives for integrating human and environmental exposure assessments P. Ciffroy a,⁎, A.R.R. Péry b,c, N. Roth d a

Electricité de France (EDF) R&D, National Hydraulic and Environment Laboratory, 6 quai Watier, 78400 Chatou, France AgroParisTech, UMR ECOSYS, 78850 Thiverval-Grignon, France INRA, UMR ECOSYS, 78850 Thiverval-Grignon, France d Swiss Centre for Applied Human Toxicology (SCAHT) Directorate, Regulatory Toxicology Unit, Missionstrasse 64, 4055 Basel, Switzerland b c

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• Opportunities to integrate human and environmental exposure assessment are identified. • Perspectives to harmonize exposure assessment data, models and methods are presented. • Use and sharing of emission and exposure data are a prerequisite for integration. • Developing a common model for exposure assessment is a key point for integration. • This work may serve as an input to develop guidelines for exposure extrapolation.

a r t i c l e

i n f o

Article history: Received 25 September 2015 Received in revised form 17 November 2015 Accepted 17 November 2015 Available online xxxx Editor: D. Barcelo Keywords: Environmental exposure assessment Human exposure assessment Integration

a b s t r a c t Integrated Risk Assessment (IRA) has been defined by the EU FP7 HEROIC Coordination action as “the mutual exploitation of Environmental Risk Assessment for Human Health Risk Assessment and vice versa in order to coherently and more efficiently characterize an overall risk to humans and the environment for better informing the risk analysis process” (Wilks et al., 2015). Since exposure assessment and hazard characterization are the pillars of risk assessment, integrating Environmental Exposure assessment (EEA) and Human Exposure assessment (HEA) is a major component of an IRA framework. EEA and HEA typically pursue different targets, protection goals and timeframe. However, human and wildlife species also share the same environment and they similarly inhale air and ingest water and food through often similar overlapping pathways of exposure. Fate models used in EEA and HEA to predict the chemicals distribution among physical and biological media are essentially based on common properties of chemicals, and internal concentration estimations are largely based on inter-species (i.e. biota-to-human) extrapolations. Also, both EEA and HEA are challenged by increasing scientific complexity and resources constraints. Altogether, these points create the need for a better exploitation of all currently existing data, experimental approaches and modeling tools and it is assumed that a more integrated approach of both EEA and HEA may be part of the solution. Based on the outcome of an Expert Workshop on Extrapolations in Integrated Exposure Assessment organized by the HEROIC project in January 2014, this paper identifies perspectives and recommendations to better harmonize and extrapolate exposure assessment data, models and

⁎ Corresponding author.

http://dx.doi.org/10.1016/j.scitotenv.2015.11.083 0048-9697/© 2015 Published by Elsevier B.V.

Please cite this article as: Ciffroy, P., et al., Perspectives for integrating human and environmental exposure assessments, Sci Total Environ (2015), http://dx.doi.org/10.1016/j.scitotenv.2015.11.083

2

P. Ciffroy et al. / Science of the Total Environment xxx (2015) xxx–xxx

methods between Human Health and Environmental Risk Assessments to support the further development and promotion of the concept of IRA. Ultimately, these recommendations may feed into guidance showing when and how to apply IRA in the regulatory decision-making process for chemicals. © 2015 Published by Elsevier B.V.

1. Introduction The assessment of risks from chemicals to the environment and human health is traditionally based on a four steps common paradigm: hazard identification and dose (effect)-response assessment (i.e. hazard characterization), exposure assessment and risk characterization. Exposure assessment is a central pillar of the risk analysis process, which involves the estimation or measure of the magnitude, frequency and duration of exposure to chemicals, along with the number and characteristics of the target exposed. Yet exposure assessment is generally considered as a weak point in risk assessment; this is due to the fact that exposure assessment is often hampered by a general lack of exposure data and by the complex landscape of different pattern of uses and combined exposure (ranging from single exposure to multiple exposures from a single chemical or from multiple chemicals), the use of the many different exposure assessment models (area, concentration, species, life cycle analysis), and the inherent natural variability in exposure levels, leading to uncertainty in the estimates. Exposure assessment can be directed towards non-human living organisms (called hereafter “biota”) (for further Environmental Risk Assessment — ERA) or humans (for further Human Health Risk Assessment — HHRA). So far, Environmental Exposure Assessment (EEA) and Human Exposure Assessment (HEA) have generally used and developed their own data, methods, scenarios and models in parallel, with poor linkage between them. There is a rationale behind such differences. One the one hand, for historical and practical reasons, the separation of ERA and HHRA is deeply rooted in the culture and practices of many risk assessment or management institutions and organizations at the European Union (EU) level and beyond, which is mainly a consequence of the allocation of the risk assessment of different chemicals categories to distinct regulatory authorities and scientific disciplines. On the other hand, from a scientific standpoint, scenario building must indeed account for different pathways: while ‘on site’ exposure (i.e. local exposure to chemicals that are emitted into the environment under non intentional or controlled conditions) is mainly of concern for biota, exposure to humans can be extended to chemical production (occupational exposure) or application (e.g. pesticides exposure for operators, bystanders and residents), regional and global use of resources (imported products) and intentional and/or non intentional use of products by consumers (e.g. cosmetics). The protection goals in EEA and HEA are also clearly different: except in case of endangered species that must be protected in their own right, environmental protection is expressed in terms of protection of ecosystem structure (biodiversity) and functions (life support) and thus targets populations and their interactions within ecosystems. Human protection instead is targeted towards individuals with the objective of preventing any adverse effect on each human being health. But despite such unavoidable differences, EEA and HEA also overlap in several instances. Fate models used in EEA and HEA for predicting the distribution of chemicals among physical and biological media are essentially based on properties of environmental compartments (soil, plants, etc) and on common properties of chemicals (e.g. partitioning and degradation in environmental media). Given the limited number of species for which experimental data is available, bioconcentration, biodegradation and metabolism data and models used to estimate the internal concentration in biological media are by necessity based on inter-species extrapolations, including biota-to-human extrapolations. Species that are assessed in the frame of EEA can also form part of the human food chain (e.g. fish). Beyond these scientific overlaps, EEA and HEA are also facing common challenges, in particular with regard to

increasing scientific complexity and resources. The different categories and amount of substances for which EEA and HEA are required will continue to increase substantially due to revised legislation (e.g. the REACH [Registration, Evaluation, Authorisation and Restriction of Chemicals, EC/1907/2006; EC, 2006] Regulation, risk assessment for emergent chemicals such as pharmaceuticals, nanomaterials, etc). Finally, the desire to assess the impact of multiple stressors and the exposure to mixtures adds additional complexity in exposure assessment. Altogether, these points create the need for a better exploitation of all currently existing data, experimental approaches and modeling tools and it is assumed that a more integrated approach of both EEA and HEA may be part of the solution. Integration of EEA and HEA was already evaluated in pioneer activities as part of a framework on Integrated Risk Assessment (IRA) developed under the auspices of the International Program on Chemical Safety (IPCS) of the World Health Organization (WHO), the European Commission (EC), the Organization for Economic Cooperation and Development (OECD) and the US Environmental Protection Agency (US EPA) (WHO, 2001). Opportunities for integration were identified in the modeling of chemical transport, fate, and exposure, in particular for environmental exposure models, where concentrations in water, soil, air and different food items must be estimated (Vermeire et al., 2007). Key elements to be integrated in the exposure characterization were outlined, i.e. sources and emissions, distribution pathways, transport and fate models, external and internal exposure models, measures of exposure related parameters (metrics), analytical tools such as methods for sensitivity and uncertainty analysis (WHO, 2001). Critical research recommendations were formulated, including: i) harmonization and improvement of exposure characterization, human health and environmental surveillance methods and exposure models; ii) incorporation of multiple sources and pathways into models of exposure that include both human and wildlife receptors; and iii) integration of monitoring data including measures of exposure and effect (Munns et al., 2003). Building on this legacy, new opportunities to better integrate EEA and HEA were evaluated by the EU FP7 HEROIC (Health and Environmental Risks: Organization, Integration and Cross-fertilization of Scientific Knowledge, Grant Agreement no. 282896, www.heroic-fp7.eu) Coordination action (Péry et al., 2013a), as an input to promote the concept of IRA, as outlined in a recent White paper (Wilks et al., 2015). HEROIC defines IRA as “the mutual exploitation of ERA for HHRA and vice versa in order to coherently and more efficiently characterize an overall risk to humans and the environment for better informing the risk analysis process”. Accordingly, in the frame of the present paper, ‘Integrated Exposure Assessment’ is defined as the possibility of combining information generated for different purposes, in particular from information generated for HEA to information dealing with EEA and vice-versa. Because of the need to develop a common understanding of Integrated Exposure Assessment, and to assess how EEA and HEA can benefit from each other as an input for IRA, the HEROIC project organized on January 21–22, 2014 in Paris, France, a dedicated workshop gathering several experts from academia, regulatory authorities and industry involved in the risk or exposure assessment area. This paper reflects experts' views on current gaps and needs as well as new opportunities and recommendations for extrapolating across human and environmental exposure data, models and methods, to support the further development and promotion of the concept of IRA. Four topics were covered in specific breakout group sessions: i) exposure scenario building, exposure waiving; ii) exposure scenario, temporal and spatial scales; iii) metrics; and iv) toxicokinetics for biota and humans.

Please cite this article as: Ciffroy, P., et al., Perspectives for integrating human and environmental exposure assessments, Sci Total Environ (2015), http://dx.doi.org/10.1016/j.scitotenv.2015.11.083

P. Ciffroy et al. / Science of the Total Environment xxx (2015) xxx–xxx

2. Exposure modeling 2.1. Defining overlapping pathways of exposure As previously indicated, some exposure pathways are specific to humans. These pathways correspond to purely anthropogenic life conditions like occupational exposure, intentional use of products by consumers (e.g. cosmetics), exposure to indoor air, treated drinking water or food packaging contaminants, and urban way of life. Despite these specificities, human and wildlife species also share the same environment and they similarly inhale/ingest air, water and food. The overlapping pathways that are common to humans and biota are called here ‘environmental exposure pathways’. Exposure to chemicals via ingestion of food products can actually be a predominant pathway for HEA in many cases. For example, importance of monitoring bioaccumulation in aquatic organisms is stressed by the EC (Directive 2013/39/EU; EU, 2013) for both ERA and HHRA purposes. It states that for very hydrophobic substances which accumulate in biota and are hardly detectable in water, environmental quality standards (EQS) should be set for biota, as specified in the Technical Guidance for Deriving EQS (EC, 2011). For dioxins and dioxin-like compounds EQS values refer to the concentration in fish, crustaceans, and mollusks (Schäfer et al., 2015). Pesticides provide another example of compounds that are of concern for both EEA and HEA. These compounds are widely used to ensure high crop yields. In 2003, the use of plant protection products (PPPs, i.e. herbicides, insecticides, fungicides and other pesticides) in agriculture amounted to about 220,000 t in the EU15 (Eurostat, 2007). A large proportion of PPPs is applied on crops like fruit and vegetables; for example, fungicides are widely used in vineyards. By nature, pesticides are deliberately released into the environment for controlling undesired organisms such as weeds, fungi and insects. However, pesticides are biologically active compounds and they are also a significant source of diffuse pollution for non-targeted wildlife organisms. In parallel, they might also cause long-term health implications in humans. The EU FP5 PINCHE (Policy Interpretation Network on Children's Health and Environment) Program (Van den Hazel and Zuurbier, 2005; Zuurbier et al., 2007) illustrates this issue through the review of several monitoring programs where pesticide residues were followed: in a monitoring program performed in 1998 in the EU and Norway, pesticide residues were found in 36% of the fruit, vegetable and cereal samples studied and Maximum Residue Limits (MRLs) for chronic ingestion were exceeded in 3.3% of the samples; in a 2003 EU coordinated monitoring program, 14 out of 505 (2.8%) fruit and vegetable samples exceeded the MRLs. Food is also considered to be the major source of human exposure for other types of chemicals like e.g. polyaromatic hydrocarbons (PAHs), as grains, fruits and vegetables can be contaminated from atmospheric deposition and interception. About 90% of the human exposure to dioxins and polychlorinated biphenyls (PCBs) is also estimated to occur through food. From these studies, it can be concluded that for many pollutants, exposure of humans to chemicals is mainly originated from environmental pathways that are common to biota living in similar environments. Integrating information derived from food-web bioaccumulation modeling such as bioconcentration, biomagnification, and biotransformation into the HEA would then enable better risk management. Exposure assessment has to account for all environmental pathways, interacting compartments and physicochemical phenomena involved in the fate of the chemical under investigation. In the Emission-toTarget chain, several steps that are common for humans and biota can be defined. First, environmental transport and fate describe or predict the chemical concentration in physical media such as surface water, groundwater, outdoor air, soils, or sediments. In the specific case of persistent pollutants that can move over long distances, long-range transport models can be used to evaluate the transfer of chemicals to regions far from the emission source. Fate models are essentially based on biodegradation and partitioning properties of chemicals that

3

govern their distribution among physical media. Finally, bioaccumulation, and eventually bioconcentration are used to estimate the internal concentration in biological media that form part of the food chain. All the environmental pathways that are common to humans and biota are summarized in Fig. 1 and form a significant part of EEA and HEA models. 2.2. Developing common modeling frameworks in EEA and HEA To assess exposure of humans and biota with similar parameters/ processes/pathways, a common modeling framework integrating environmental fate of chemicals with both human and environmental targets would be useful. To our knowledge however, most of the models that were developed so far are generally focused towards one of the two targets and/or describe only a limited part of the environment (e.g. models dedicated to freshwaters or soils only). For example, some models were originally developed for assessing human exposures and do not consider non-human living organisms (e.g. CalTox (www.dtsc.ca.gov/AssessingRisk/caltox.cfm)), and vice versa (e.g. AQUATOX (water.epa.gov/scitech/datait/models/aquatox/)); some models are dedicated to a given class of chemicals (e.g. PEARL (www.pearl.pesticidemodels.eu/) and PRZM (http://www.epa.gov/ oppefed1/models/water/#przm), which are dedicated to PPPs); some models are dedicated to a given environmental media (e.g. CLEA (www. gov.uk/government/publications/updated-technical-background-to-theclea-model) and CSOIL (www.rivm.nl/bibliotheek/rapporten/711701054. html), which were developed to estimate the exposure of humans who live in contaminated sites); some models are used for screening assessments rather than for higher realistic assessment tiers (e.g. EUSES (ihcp.jrc.ec.europa.eu/our_activities/publichealth/risk_assessment_of_ Biocides/euses/euses), UseTox (www.usetox.org/), which is used in the domain of Life Cycle Assessment). Those multimedia/pathways exposure models are among the most used internationally at regulatory level to assess human exposure from the environment, but this list is by no mean comprehensive and is given here just for illustrative purpose. Consequently, there is a need for the development of exposure tools integrating both human and wildlife biota targets with common fate modeling in the environment. The modeling tool MERLIN-Expo (http://merlin-expo.4funproject.eu/) is under construction in the frame of the EU FP7 4FUN (http://www.4funproject.eu/) project with such objectives: this exposure tool is composed of a library of fate models dedicated to non biological receptor media (river, lake, soil, air), biological media of concern for humans (cultivated crops, cow milk), wildlife biota (primary producers in rivers, invertebrates, fish) and of Physiologically-Based PharmacoKinetic (PBPK) models; these models can be linked together to create flexible scenarios relevant for both human and wildlife biota exposure and can be a promising way to better integrate EEA and HEA. 3. Exposure scenario building 3.1. Defining tiered approaches/worst case scenarios is a challenge for integration Beyond the availability of models facilitating EEA and HEA integration, another issue to be analyzed is the scenario building. In the regulatory context, stepwise and tiered risk assessment approaches are frequently used, starting with cautious assumptions and simple deterministic models, often using conservative worst-case assumptions. If no potential risk is identified for such worst-case assumptions, regulatory decision can be taken because of low risk concern. Additional data and more realistic models are only needed if the risk concern cannot be concluded from lower tier approaches. A tiered assessment process, including triggers for further assessment, is often included in the regulatory guidance. Defining common worst-case scenario and more generally common tiers for HEA and EEA is however far from being obvious. The worst-case

Please cite this article as: Ciffroy, P., et al., Perspectives for integrating human and environmental exposure assessments, Sci Total Environ (2015), http://dx.doi.org/10.1016/j.scitotenv.2015.11.083

4

P. Ciffroy et al. / Science of the Total Environment xxx (2015) xxx–xxx

Fig. 1. Main environmental pathways to be considered in EEA and HEA — blue lines represent common pathways to EEA and HEA — green lines represent pathways specific to EEA — red lines represent pathways specific to HEA. (1) Partition; (2) dry and wet deposition; (3) absorption/volatilization; (4) deposition/resuspension; (5) diffusion; (6) leaching; (7) runoff and erosion; (8) degradation; (9) direct uptake; (10) food-web transfer; (11) root uptake; (12) interception by leaves and diffusive exchanges; (13) drinking water; (14) food ingestion (e.g. vegetables, fish, meat, milk, eggs); (15) soil particles ingestion (e.g. pica children); (16) inhalation; (17) food web relationships (prey predator relationships). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

scenario can indeed differ according to the investigated process, life span (i.e. human life vs acute assessment) and (eco)toxicological endpoint, and is even more difficult to define in case of multiple targets (e.g. in EEA, several species that interact but that are not sensitive to the same processes have to be assessed together). An example of the difficulty for defining worst-case scenario is the choice of default values for partition coefficient in soil that defines the level of mobility of the investigated chemical (Fig. 2). This parameter can be used both for EEA, because it governs the bioavailability of contaminants for soil organisms, and for HEA, because it governs migration to surface- or groundwater that can potentially be used as drinking water. Low partition coefficients (representative of ‘mobile’ chemicals) can indeed have counter-balanced effects: root transfer (by transpiration stream and/or diffusion) of contaminant is facilitated for weakly adsorbing contaminants, but at the same time leaching towards groundwater and runoff of soil water from watershed to rivers are facilitated; low partition coefficient results in higher transfer to soil organisms, but with shorter time persistency in soil, and in higher groundwater (and subsequent potential drinking water) contamination. On the contrary, high partition coefficients (representative of less mobile chemicals) lead to limited root uptake by crops and wildlife plants, but at the same time contaminated particles ingestion by some organisms is facilitated if contaminants remain at the soil surface. This example shows that biota and/or human exposures can be maximized for different values of environmental parameters according to the target and time scales that are considered. 3.2. Developing tiered probabilistic approaches in EEA and HEA One way to build ‘consistent worst-case’ scenarios (i.e. scenarios that could be considered conservative for different targets) is to select some

predefined percentiles in the distribution of sensitive parameters. Since EEA and HEA share the same methodological questions, the most relevant input data and parameters percentile(s) for building a consistent worstcase scenario, taking into account targeted life span, and data quality and availability, should be defined. To make progress on this issue, it is recommended to develop screening probabilistic approaches that could be commonly used in EEA and HEA. ‘Probabilistic assessment’ is generally associated to the highest tier in EEA and HEA because it requires the description of parameters by Probability Density Functions that are difficult to derive. More simple probabilistic approaches that require only quantiles in parameter values (instead of the exact form of the distributions) are however possible; they are classically known as ‘qualitative’ or ‘screening’ probabilistic methods and are based on One-factor-At-aTime (OAT) screening design. Such methods (e.g. the Morris design, see Campolongo et al., 2007; Ciric et al., 2012) are considered global because the design intends to cover the entire space of the parameters. They are based on a random OAT design in which each parameter is described by uniform uncertainty interval. Such screening sensitivity approaches could help the building of consistent worst-case scenarios for both EEA and HEA scenario because they identify the zone of parameter space that maximizes some targets while remaining poorly data-consuming. It has to be noted that such methods are promoted by WHO in its tiered approach for uncertainty analysis in exposure assessment (WHO, 2008). However, exposure models that were developed so far didn't include such screening sensitivity methods and more generally methods for uncertainty and sensitivity analysis that are in line with the tiered approach proposed by the WHO (2008). Yet, in order to bridge this gap, the modeling tool MERLIN-Expo (aforementioned) contains a set of operational methods for uncertainty/sensitivity analysis (from screening sensitivity approaches to robust variance-based methods)

Please cite this article as: Ciffroy, P., et al., Perspectives for integrating human and environmental exposure assessments, Sci Total Environ (2015), http://dx.doi.org/10.1016/j.scitotenv.2015.11.083

P. Ciffroy et al. / Science of the Total Environment xxx (2015) xxx–xxx

5

Fig. 2. Effect of partition coefficient in soil on human and biota exposures — increasing and decreasing arrows indicate if exposure is increasing or decreasing for low partition coefficients (i.e. high mobility in soil).

that have been designed to comply with the WHO guidance specifications for uncertainty analysis. The availability of such options for uncertainty and sensitivity analysis could facilitate the identification of the important drivers (e.g. parameters, processes) contributing the most to uncertainty, in particular the combination of parameter values maximizing exposure, while improving altogether future decision-making in EEA and HEA. 4. Exposure-based waiving 4.1. Applying the ‘threshold of toxicological concern’ (TTC) concept to build common exposure-based waiving rules in EEA and HEA Another area in the regulatory context that would potentially benefit from a more integrated approach is the concept of Exposure-Based Waiving under REACH (ECHA, 2008). This concept assumes that (eco)toxicological tests may be waived if it can be shown that humans or organisms in the environment are either not or only minimally exposed to the investigated substance; it also requires a comprehensive exposure assessment of all exposure/use scenarios, human exposure routes and environmental pathways throughout the life cycle of a chemical (Vermeire et al., 2010). In contrast, additional testing (including extra experimental animal testing) can be necessary when the exposure is high according to the Exposure-Based Triggering approach. To support the risk prioritization process and screening of a large number of chemicals, e.g. in setting data-generation priorities and promoting the use of non testing data over animal testing-generated data, building common exposure-based waiving rules in EEA and HEA would be useful. One tool for building exposure-based waiving rules is the concept of Threshold of Toxicological Concern (TTC); it is based on the possibility

of establishing common exposure threshold values for all chemicals below which no significant risk to human health is expected to exist. The TTC concept was initially established for human health endpoints in the context of food safety evaluation (Kroes et al., 2004). These threshold values were derived from extensive analysis of available chronic toxicity data of substances, which were subdivided into three chemical classes on the basis of their structure (i.e. Cramer classes I, II or III; Cramer et al., 1978). The TTC concept was designed to be applied for low concentrations of chemicals (except some chemical classes like e.g. proteins, heavy metals, and polyhalogenated dibenzo-p-dioxins) without toxicity data but with known chemical structure allowing the identification of structural alerts (e.g. for genotoxicity and/or carcinogenicity). Such analysis led to the derivation of three thresholds (expressed as maximum daily intakes) that can be commonly applied to all chemicals belonging to Cramer classes I, II or III. Even if initially developed for chemicals in food, the TTC concept was extended or has potential value in the assessment of risks in other exposure scenarios like cosmetics, pesticides, pharmaceuticals impurities and airborne substances (Hennes, 2012). TTC being a tool for exposure-waiving in HHRA, a challenge is to develop a similar concept for ERA; once IRA would be active, waiving could then be used in an homogeneous way for HHRA and ERA purposes. 4.2. Moving from the ‘threshold of toxicological concern’ concept to ‘environmental thresholds of no toxicological concern’ (ETNC) concept: opportunities for IRA While the TTC concept could be a promising approach for facilitating the definition of common exposure-based waiving rules in EEA and HEA, further developments are however needed. First, the TTC concept has been mainly used so far in the regulatory context for direct exposures from consumer products (e.g. food

Please cite this article as: Ciffroy, P., et al., Perspectives for integrating human and environmental exposure assessments, Sci Total Environ (2015), http://dx.doi.org/10.1016/j.scitotenv.2015.11.083

6

P. Ciffroy et al. / Science of the Total Environment xxx (2015) xxx–xxx

additives) and not for environmental exposure pathways, and it is expressed as the chronic intake per person per day (i.e. in mg ingested · kg−1 · d−1). On the basis of generic realistic intake scenarios from environmental products (i.e. drinking water, air, crops), a relationship between the TTC and the possible concentrations and distributions of the chemical in the environmental media (used e.g. as food supply) should be established. This unit conversion would facilitate the comparison with threshold values applicable for environmental targets. Second, TTC applicable to biota should be derived according to a similar approach as those followed for human exposure. De Wolf et al. (2005) proposed a transposition of the TTC concept into an ‘Environmental Thresholds of No toxicological Concern’ concept (ETNC). Such ETNC were derived for freshwater systems and for organic chemicals from an extended analysis of ecotoxicological data to support aquatic ERA (in a very comparable way to what was done for human toxicology). Similarly to the Cramer classes, ETNC were derived with data stratification based on Verhaar's categorization system of mode of action (MOA) (i.e. including four classes: MOA 1, inert chemicals; MOA 2, less-inert chemicals; MOA 3, reactive chemicals; and MOA 4, specifically acting chemicals; Verhaar et al., 1992), and with application of appropriate safety factors. The derived ETNC values for MOA 1–3 were 0.1 μg/L. Sokull-Klüttgen (2007) also reported that from an analysis of more than 1000 chemicals for which lethal concentration (LC)/effect concentration (EC50) toxicity results were available for fish, Daphnia, and algae, less than 5% exceeded an ETNC of 0.1 μg/L when an assessment factor of 1000 was applied. Caldwell et al. (2008) showed that the Predicted No Effect Concentration (PNEC) derived for ethinylestradiol, with a receptor-mediated MOA, was nearly identical to the ETNC for MOA 4 chemicals, and he suggested that the ETNC concept is a conservative estimate of chronic thresholds in data-poor situations where empirical ecotoxicological data are lacking (Caldwell et al., 2014). Gross et al. (2010) also discussed the use and limits of TTC concept in the frame of EEA of aquatic systems for endocrine active substances with an estrogenic MOA. The ETNC concept was used by Hollander et al. (2011) to define an Exposure-Based Waiving system in ERA, based on a tiered approach strategy; in tier 1, three dimensional tabular data derived from physicochemical properties of the chemical (octanol-water partitioning coefficient Kow, air-water partitioning coefficient Kaw and degradation rate) and emission data were built to identify whether the environmental exposure levels are smaller than the conservative De Wolf's ETNC. However, the concept of ETNC is limited to freshwater organisms because sufficient data on organic chemicals for the sediment, marine or soil environments are not available. In conclusion, commonalities exist in the development of the TTC concept in the field of food safety and environmental safety, and the TTC concept could be a promising perspective for defining common exposure-based waiving rules for EEA and HEA. The TTC concept is a pragmatic approach that supports the Intelligent Testing Strategies (ITS) scheme under REACH and the regulatory decision-making of chemicals by integrating all existing toxicological information, and which offers perspectives in the screening and setting of datageneration priorities of a very large number of chemicals, therefore contributing to minimize the use of animals (viz. the ‘3R’ concept1) (van Leuwen et al., 2007). Application of such a scheme would facilitate the setting of common exposure threshold values for all chemicals below which no significant risk is expected, help setting datageneration priorities, and support the ‘3R’ efforts. Yet, more research to extend the field of application of the TTC concept and gain further validation, as well as increased harmonization efforts are needed. As outlined in the HEROIC White paper (Wilks et al., 2015), progress in the area of Adverse Outcome Pathways (AOPs)/MOA is expected to be 1 “Replace, Reduce, Refine”. The term “3R” was coined by Russell and Burch (1959) who set ethical aspects and laid down the basis for the development and progress of humane procedures in the laboratory.

a major driver in the further development of integrated hazard and exposure assessment. Hollander et al. (2011) anticipated that once more MOA-specific TTCs have been developed, these values could be used as cutoff criteria in their tiered exposure-based waiving strategy to support the ERA of chemicals. For Caldwell et al. (2014), progress in AOPs will contribute to make more chronic ecotoxicity data available with a plausible link to MOAs to advance ERA of other MOA 4 chemicals. 5. Data collection and harmonization Human and environmental exposure assessments face common challenges, limitations and shortcomings in term of data quality, availability, accessibility, sharing and harmonization. Data are stored in heterogenous forms and databases. Accordingly, experts recognized: (i) the need to better exploit, collect and share existing exposure data, in particular environmental or human (bio)monitoring data; (ii) the lack of integrated databases; (iii) the need for harmonization and sharing of sampling design; (iv) the need to better use metrics. 5.1. Data collection and sharing A central question for Integrated Exposure Assessment is the crossfertilization of exposure data across the human and environmental disciplines. Exploitation of data used in EEA could be of high interest for HEA and vice-versa. Yet some shortcomings and differences in term of data quality exist. Data used for exposure assessment (e.g. contamination levels in environmental media, population distribution) can exhibit significant temporal and spatial variability at various scales depending on the exposure target, and may differ between EEA and HEA. For example, this variability can be expressed from the millimetric scale, such as observed for contamination levels in sediment depth, to continental scale variability, such as observed in industrial emissions and long-range transport. The temporal and spatial scales that are relevant for exposure assessment also highly depend on the exposure target. Concerning temporal variability, for example, seasonal and/or tidal fluctuations can be of concern for aquatic organisms that are sensitive to physicochemical variables like temperature or salinity. For other organisms that show longer life cycles, averaged values over long periods of time can be sufficient without taking into account intra-annual fluctuations. For humans, temporal scales can be quite different with respect to biota when exposure over the entire lifetime is considered. The same issue arises for spatial variability. Huge differences can be observed within very fine spatial grid for some media like sediments or soils and that may affect exposure assessment of organisms living in such media. Such variability can be ignored for other targets, where averaged values in surface soils can be used as input data, e.g. like in the case of pica children (“pica” disorder is the persistent behavioral pattern of eating non food items such as dirt, soil, paper, etc). In practical terms the spatial and temporal scales should be designed according to the risk assessment objectives, and should be coherent with the effect assessment, ensuring the optimal resolution of the spatial and temporal coverage of the risk characterization output. Despite these unavoidable differences, there is no doubt that data collected for EEA purposes could be of high concern for HEA and viceversa. However, there is a lack of integrated databases, which can result from historical or management reasons since environmental, socioeconomic (e.g. diet composition), biomonitoring and health status followup are generally not managed by the same institutions. Actually, data allowing a better exploitation of exposure models sometimes exist, but not always in a format allowing them to be readily collected in a homogenous form. A way to better integrate monitoring data that were initially generated for EEA and HEA purposes could be to merge such data, even if sporadically available, with modeling results and/or expert judgment through e.g. Bayesian techniques. Exposure modeling results, eventually in conjunction with other existing knowledge gained e.g. from expert judgment, can form prior estimates of exposure. Such

Please cite this article as: Ciffroy, P., et al., Perspectives for integrating human and environmental exposure assessments, Sci Total Environ (2015), http://dx.doi.org/10.1016/j.scitotenv.2015.11.083

P. Ciffroy et al. / Science of the Total Environment xxx (2015) xxx–xxx

approaches were commonly used in occupational exposure assessment where prior expert judgments about workers exposure are merged with exposure modeling (Tielemans et al., 2007; Ramachandran, 2008; Sottas et al., 2009). In this perspective, results derived for EEA could be used as prior knowledge for HEA and vice versa. Similarly, reverse modeling approaches from environmental or human biomonitoring data respectively could be used for generating consistent input data for EEA and HEA, respectively. For example, Ulaszewska et al. (2012) used biomonitoring data of PCBs levels in Italian women breast milk, and PBPK models to determine the most probable scenario of exposure: for each congener, the authors determined the most probable long-term history of PCBs emission in air, as well as concentrations in environmental receptor media and food, and estimated the time evolution of the daily intakes over the lifetime. As a result, they were able to reconstruct accurately the exposure and filled in data gaps on environmental concentrations over decades. Such reverse modeling that uses human biomonitoring data as input reference for HHRA can generate also consistent environmental input data for ERA. 5.2. Optimization and harmonization of sampling designs Another important issue regarding harmonization of data collection regards sampling strategy. Spatial variability requires suitable sampling designs that consider source location, relation to vulnerable/valuable areas/targets, gradients, sample number in relation to precision and cost efficiency. Indeed, several tools were proposed to optimize sampling designs for EEA and HEA purposes, for weighting estimates and median values and for sub-dividing large contamination areas into smaller geographic units; one can cite in particular: (i) Kriging/ geostatistics (e.g. Carlon et al., 2001; Critto et al., 2003); (ii) Multivariate statistics (e.g. Principal Component Analysis — PCA); (iii) Random sample designs versus more complex multistage designs; (iv) Geographical Information Systems (GIS) (e.g. Nuckols et al., 2004); and (v) Congener ratio approach (e.g. Needham et al., 2007). Such methods can be equally useful for EEA and HEA and can help to merge or at least harmonize sampling designs. 5.3. Harmonization of metrics Finally, an issue that should be considered for harmonizing data collection in EEA and HEA is the metrics issue. In general, the metrics are common between EEA and HEA, with only a few exceptions (for instance, the use of Gray for EEA and Sievert for HEA in radiological exposure assessment) but the choice will be context-dependent. Several metrics can indeed be used to characterize the environmental state on a contamination point of view: concentration levels can be expressed in dry or wet weight, in total concentrations or pore water concentrations (e.g. in soil and sediments), etc. The relevance of such metrics is highly dependent on the investigated target. For EEA, different targeted organisms may require different metrics. For example, for soil and/or sediment organisms, when the contaminant uptake is mainly via body surface (diffusion), soil or sediment porewater concentration is probably a relevant metric, whereas total dry weight concentration can be a more relevant metric when dietborne exposure pathway is predominant (or for small pica children). Sampling tools like passive samplers that collect specific fractions in environmental media can be more or less suitable for exposure assessment according to the organism and life stage. Similarly, concentrations in animals can be expressed in different metrics if it is considered in an environmental or a human exposure (through feeding) context: the contaminant concentration in fish that is of concern for EEA is either the total wet weight concentration (as requested in the context of REACH) or the concentration in organs relevant to certain type of adverse effects (endocrine disruption for instance; Watanabe et al., 2009; Li et al., 2011; Péry et al., 2014), whereas human dietary exposure is governed by concentration in edible organs of fish (i.e. fish muscle). For instance, in the context of human food

7

safety, toxicokinetic models will focus on fish filet, and not much on whole body (Berntssen et al., 2013) (except for specific ethnic cohorts that have unique food preferences such as cooking and eating entire animals). McCarty et al. (2013) report an example of coordination between ERA and HHRA. A methylmercury fish tissue criterion of 0.3 μg/g wet weight was developed for human health after consumption of fish. The question was to assess if this threshold would also be protective of wildlife species. It was shown that this fish tissue criterion was below any identified critical body residue for nearly all fish species. However, as shown with a wildlife-ingested dose model linked to a food-web bioaccumulation model, it may adversely affect the bald eagle, preying on fish. It appears that what would be required for cross-fertilization between EEA and HEA are more tools to relate metrics and exploit the data in parallel than harmonization between metrics. More precisely, it is crucial: (i) to keep the original information clear, detailed and accessible (for instance, if fish whole concentration is deduced from measurements in different compartments, these measured values should be kept); and (ii) to develop methods able to predict notmeasured metrics from measured ones. For instance, muscle concentration could be deduced from whole body concentrations provided the latter concentration is well predicted based on Kow and lipid contents and vice versa. In conclusion, new scientific methods and tools are at hand to better exploit and structure available exposure information between EEA and HEA, e.g. by merging human and environmental monitoring data or sampling designs, or at least harmonizing them. Indeed, harmonization is a major driver for integration. While the development of shared, common integrated databases is also limited by data accessibility (due e.g. to proprietary reasons), harmonization efforts will support a better integration of exposure data between EEA and HEA. As noted by Suter et al. (2003), harmonization of the principles and methodologies used to characterize human and environmental risks is relevant to all forms of integration. HEROIC considers harmonization approaches in exposure (and hazard) assessment as being a necessary prerequisite for EEA and HEA (and IRA in general); it is anticipated that IRA will foster harmonization efforts, and in this respect, IRA can also be considered an incentive for harmonization (Wilks et al., 2015). 6. Internal concentrations Another area of potential cross-fertilization between EEA and HEA identified by experts is the use of internal dose in the exposure assessment, complementary to exposure concentrations. In data-poor situations where exposure data are lacking in a given species, opportunities exist to extrapolate internal concentrations from another species within the same taxa or across taxa, based on toxicokinetics data for human and biota. 6.1. Relevance of internal concentrations for exposure assessment Internal concentrations, i.e. concentrations in fluids (urine, blood) or organs inside the body, are relevant to reflect the actual exposure of the species and may thus be a better option than exposure concentrations for exposure-based waiving. Indeed, internal concentrations aggregate the contributions of many routes of exposure (ingestion, inhalation, etc), and account for the bioavailability of the chemicals. This is also a regulatory requirement, especially in EEA; for instance, REACH and the ‘CLP’ Regulation (EC/1272/2008; EC, 2008) on classification, labeling and packaging of substances and mixtures (which implemented in the EU the United Nations Globally Harmonized System — GHS) require for the EEA that the bioconcentration factor (BCF), preferably with fish data, is determined (Lombardo et al., 2010). Internal concentrations are also more relevant than external concentrations to provide a link between exposure and effects. They account for the concentration to which target organs or systems are actually

Please cite this article as: Ciffroy, P., et al., Perspectives for integrating human and environmental exposure assessments, Sci Total Environ (2015), http://dx.doi.org/10.1016/j.scitotenv.2015.11.083

8

P. Ciffroy et al. / Science of the Total Environment xxx (2015) xxx–xxx

exposed to, and as such, permit a more relevant extrapolation between species, between levels of biological organizations or between laboratory tests and field monitoring. For instance, Devillers and Devillers (2009) showed that introducing a parameter accounting roughly for bioaccumulation (the 1-octanol/water partition coefficient) improved the quality of the prediction of rodent acute toxicity based on ecotoxicity data. Hendriks et al. (2005) linked quantitative structure-activity relationships (QSARs) for response concentrations in water (LC50) to critical concentrations in organisms through a generic model of accumulation in different phases covering many substances and species (including aquatic species and mammals). The model was able to account for 60% of the variability of the critical internal concentrations associated with the LC50 collected in a validation set of data. In support of extrapolation from laboratory to field, Salazar and Salazar (1998) showed that tissues residues of tributyltin in field mussels were consistent with critical body residues determined in laboratory conditions. In the context of human predictive toxicology, there is an increase of the use of models coupling human toxicokinetics modeling and dose–response models relating internal concentration and effects at target level, as shown, for instance by Péry et al. (2013b) to predict acetaminophen hepatotoxicity in humans from effects measured in vitro on hepatic cell lines. Internal concentrations are thus relevant metrics for exposure (and hazard) assessment (for species occupying similar niches in the environment and having similar uptakes), but as there are many possible metrics for ‘internal concentration’, there is a question of what should be the internal concentration to look at. Usually, the whole-body concentration is used as a surrogate of the concentration at the site of toxic action, assuming proportionality between both concentrations (Meador, 2005). As explained by Meador (2005), using tissue-residue approach for toxicity assessment has many advantages, among which: (i) low variability among species for many contaminants; (ii) skipping the variability due to toxicokinetics and focusing on interspecies variability in terms of toxicodynamics, which can still be large (McCarty et al., 2013); and (iii) a causal relationship between tissue residues and adverse outcome. However, some authors are in favor of going a step further with internal concentrations by considering the free fraction only. For instance, total accumulation may be a poor predictor of metal toxicity for Chironomus riparius exposed to contaminated field sediments. As an alternative, relating effects on Chironomus growth with cytosolic metal accumulation permits to predict successfully the effects of mixtures of cadmium, zinc, and copper on the growth of larvae exposed to spiked sediments, as well as to field sediments (Péry et al., 2008). However, there seems to be no universal dose metric. In toxicology, Groothuis et al. (2015) proposed a chart to select the appropriate in vitro dose metric (i.e. nominal concentration, free available concentration, total concentration, etc), as a function of analytical facilities, physicochemical properties, MOA and assay setup. 6.2. Extrapolating internal concentrations: PBTK models vs generic models When information is lacking, internal concentrations for a given species can be deduced from information obtained in another species, in different ways. First, parameters required to calibrate toxicokinetic models can be extrapolated between species, especially when these models are physiologically-based ones. The so-called Physiologically-Based ToxicoKinetic models (PBTK) consist of a series of mathematical equations with parameters based on the specific physiology of an organism and on the physicochemical properties of a substance, which are able to describe the absorption, distribution, metabolism and elimination (ADME) of the compound within this organism. The solution of these equations provides the time-course of the parent compound and possibly some of its metabolites in the organs and allows for a sound mechanistic description of the kinetic processes including accumulation in tissues. Substance-dependent parameters like partition coefficients

between blood and organs, uptake, excretion or metabolism rates can be extrapolated from one species based on allometric factors (Campbell et al., 2012). However, such allometric extrapolations have known limitations. For instance, differences in renal resorption explain the large interspecies differences in terms of elimination of perfluoroalkylacids (Andersen et al., 2006). Relative to metabolism, interspecies differences may be even more substantial. When comparing metabolism activity by trouts and humans, Connors et al. (2013) showed substance-dependent Cytochrome P450 (CYP) 3A4-like activity in trout, absence of CYP2C9 activity, and presence of CYP1A-like activity, all accounting for phase I (e.g. oxidation or reduction reactions) metabolism. This suggests that some pharmaceuticals may more accumulate in fish than anticipated. In the same way, Péry et al. (2014) suggest that CYP2B6-like activity may be reduced in fish, but that phase II (e.g. conjugation reactions) metabolism is present. A second way to extrapolate information relative to toxicokinetics between species is to use generic models that cover many species and predict total internal concentration or critical body residues based on log Kow and contents of proteins and lipids. The model proposed by Hendriks et al. (2005) could be applied to algae, fish, and rats to estimate critical organism concentrations based on octanol-water partitioning, organism composition and a QSAR model. The fact that the model performs well for one species would support the use of this model for other species, especially those for which data are scarce. At this stage, we recommend the compilation of available information relative to uptake, metabolism and excretion rates, volumes of distribution and partition coefficients to assess both the relevance of available models targeting the predictions of these properties and the interspecies differences. Penalty scores or statistical distributions (if a probabilistic approach is used) would be derived in a next step based on this information. Homology between receptors/enzymes and measures of taxonomic distances may also contribute to the setup of these scores or distributions. Third, toxicokinetic models can also be used in a reverse way, i.e. deducing exposure concentrations from measured internal ones. This is the basis of dose reconstruction based on human biomonitoring (Clewell et al., 2008), usually based on urinary measurements, but not only. For instance, Ulaszewska et al. (2012) could reconstruct the dynamics of exposure of Italian women to PCBs based on measurements in breast milk and deduce the most likely profile of emissions and atmospheric PCB concentrations throughout time, from the 1940s till nowadays. Similarly, data collected in animals in the environment can be used to fill data gaps in HEA. As an example, we can cite the works performed on the Rhône River in France to reconstruct spatial and temporal trends of PCBs in sediments from the late 1960s to present (Desmet et al., 2012; Mourier et al., 2014), and estimate the transfer of PCBs from bottom sediment to freshwater river fish based on a food-web bioaccumulation model using a Bayesian inference approach (Lopes et al., 2012), as an input for risk assessment of fish consumption and to develop sediment quality guidelines protective of human health. Once exposure has been reconstructed based either on human or animal biomonitoring, it can be used for both EEA and HEA purposes. In conclusion, internal concentrations for a given species can be deduced from information in another species, given that: (i) parameters required to calibrate toxicokinetic (e.g. PBTK) models can be extrapolated between species; (ii) if a generic toxicokinetic model performs well for one species, this would support the use of this model for other species covered by the model, especially if these are data-poor. Reverse modeling can also be used as an input for both EEA and HEA to deduce exposure concentrations from measured internal ones, e.g. to reconstruct exposure based either on human or animal biomonitoring. 7. Conclusions and perspectives We evaluated the availability of theoretical assumptions, experimental datasets and/or models and operational tools that allow

Please cite this article as: Ciffroy, P., et al., Perspectives for integrating human and environmental exposure assessments, Sci Total Environ (2015), http://dx.doi.org/10.1016/j.scitotenv.2015.11.083

P. Ciffroy et al. / Science of the Total Environment xxx (2015) xxx–xxx

examining the respective contribution of external exposure pathways to body burden across species. Key points for integration across the human and environmental disciplines is the move from environmental fate and exposure estimations to the internal dose in the exposure assessment, which will be facilitated by the common use and sharing of emission and exposure data, i.e. distribution, fate and exposure models, monitoring data on concentrations in environmental media and food, PBPK and PBTK models, dose–response models and assessment of the variability for the critical effect across and within species, and the development of a common model for exposure assessment. Promoting the use of internal concentrations will provide a link between exposure and effects for a more relevant extrapolation between species, levels of biological organizations and field monitoring. Based on the experts' discussions in the frame of the Expert Workshop on Extrapolations in Integrated Exposure Assessment organized by the EU FP7 HEROIC Coordination action and complementary bibliographic research, several potential opportunities for the crossfertilization of EEA and HEA data can be outlined, as an input to develop an Integrated Exposure Assessment, namely: i) Identifying overlapping pathways of exposure based on the many existing commonalities between human and wildlife species; ii) Building of common exposure scenarios based on a tiered approach using cautious assumptions and simple deterministic models; iii) Defining consistent worst-case scenarios based on the development of screening probabilistic approaches common to EEA and HEA, taking that the two types of risk assessment pursue different targets, protection goals and timeframe; iv) Building of exposure-based waiving rules using tools such as the TTC concept and the ETNC concept; v) Merging of EEA data with (reverse) modeling results and/or expert judgment through e.g. Bayesian techniques as an input for HEA and vice-versa; vi) Developing tools to support the harmonization and sharing of EEA and HEA data and sampling designs; vii) Developing tools to better relate metrics and methods able to predict not-measured metrics from measured ones and/or information to derive the latter; viii) Promoting the use of internal dose to provide a link between exposure and effects for a more relevant extrapolation between species, levels of biological organizations or laboratory tests and field monitoring.

This paper has identified research recommendations to support the development of an Integrated Exposure Assessment and eventually of a guidance for exposure extrapolation, as an input to promote the concept of IRA. The aim of such a guidance would be to show where, when and how to apply IRA in the various existing European regulatory frameworks for chemicals, as outlined in the roadmap recently proposed by the EU FP7 HEROIC Coordination action (Wilks et al., 2015).

Acknowledgments The authors would like to thank the participants of the workshop organized by the HEROIC Coordination action, which was funded under the European Commission's 7th Framework Program (grant number 282896), and dedicated to exposure assessment: S. Andres (INERIS), M. Babut (IRSTEA), K. Beaugelin (IRSN), J. Bessems (JRC), S. Beulke (FERA), F. Brulle (INERIS), C. Brochot (INERIS), P. Ciffroy (EDF), A. Charistou (BPI), A. Dalzell (FERA), J. Devillers (CTIS), M. Ismert (INERIS), K. Machera (BPI), S. Martin (BfR), A. Péry (INERIS), M. Simon-Cornu (IRSN), and R. Smolders (VITO). The authors wish also to thank Martin Wilks,

9

coordinator of the HEROIC project, for his comments that improved this manuscript.

References Andersen, M.E., Clewell III, H.J., Tan, Y.-M., Butenhoff, J.L., Olsen, G.W., 2006. Pharmacokinetic modeling of saturable, renal resorption of perfluoroalkylacids in monkeysprobing the determinant of long plasma half-lifes. Toxicology 227, 156–164. Berntssen, M.H.G., Lock, E.J., Zeilmaker, M.J., Van Eijkeren, J.C.H., 2013. Toxicokinetic model assessment on the dechlorination of dietary toxaphene CHB-62 into CHB-44 in Atlantic salmo (Salmo salar L.). Food Addit. Contam. Part A Chem. Anal. Control Expo. Risk Assess. 30, 1581–1589. Caldwell, D.J., Mastrocco, F., Hutchinson, T.H., Länge, R., Heijerick, D., Janssen, C., Anderson, P.D., Sumpter, J.P., 2008. Derivation of an aquatic predicted no-effect concentration for the synthetic hormone, 17 alpha-ethinyl estradiol. Environ. Sci. Technol. 42, 7046–7054. Caldwell, D.J., Mastrocco, F., Margiotta-Casaluci, L., Brooks, B.W., 2014. An integrated approach for prioritizing pharmaceuticals found in the environment for risk assessment, monitoring and advanced research. Chemosphere 115, 4–12. Campbell, J.L., Clewell, R.A., Robinan Gentry, P., Andersen, M.A., Clewell III, H.J., 2012. Physiologically based pharmacokinetic/toxicokinetic modeling. In: Reisfeld, B., Mayeno, A.N. (Eds.), Computational Toxicology: Volume I, Methods in Molecular Biology vol. 929. Humana Press, New York City, pp. 439–499. Campolongo, F., Cariboni, J., Saltelli, A., 2007. An effective screening design for sensitivity analysis of large models. Environ. Model Softw. 22, 1509–1518. Carlon, C., Critto, A., Marcomini, A., Nathanail, P., 2001. Risk based characterisation of contaminated industrial site using multivariate and geostatistical tools. Environ. Pollut. 111, 417–427. Ciric, C., Ciffroy, P., Charles, S., 2012. Use of sensitivity analysis to identify influential and non-influential parameters within an aquatic ecosystem model. Ecol. Model. 246, 119–130. Clewell, H.J., Tan, Y.M., Campbell, J.L., Andersen, M.E., 2008. Quantitative interpretation of human biomonitoring data. Toxicol. Appl. Pharmacol. 231, 122–133. Connors, K.A., Du, B., Fitzsimmons, P.N., Hoffman, A.D., Chambliss, C.K., Nichols, J.W., Brooks, B.W., 2013. Comparative pharmaceutical metabolism by rainbow trout (Onchorynchus mykiss) liver S9 fractions. Environ. Toxicol. Chem. 32, 1810–1818. Cramer, G.M., Ford, R.A., Hall, R.L., 1978. Estimation of toxic hazard — a decision tree approach. Food Cosmet. Toxicol. 16, 255–276. Critto, A., Carlon, C., Marcomini, A., 2003. Characterization of contaminated soil and groundwater surrounding an illegal landfill (S. Giuliano, Venice, Italy) by principal component analysis and kriging. Environ. Pollut. 122, 235–244. De Wolf, W., Siebel-Sauer, A., Lecloux, A., Koch, V., Holt, M., Feijtel, T., Comber, M., Boeije, G., 2005. Mode of action and aquatic exposure thresholds of no concern. Environ. Toxicol. Chem. 24, 479–485. Desmet, M., Mourier, B., Mahler, B.J., Van Metre, P.C., Roux, G., Persat, H., Lefèvre, I., Peretti, A., Chapron, E., Simonneau, A., Miège, C., Babut, M., 2012. Spatial and temporal trends in PCBs in sediment along the lower Rhône River, France. Sci. Total Environ. 433, 189–197. Devillers, J., Devillers, H., 2009. Prediction of acute mammalian toxicity from QSARs and interspecies correlations. SAR QSAR Environ. Res. 20, 467–500. EC (European Commission), 2006. Regulation (EC) no 1907/2006 of the European Parliament and of the Council of 18 December 2006 concerning the registration, evaluation, authorisation and restriction of chemicals (REACH), establishing a European Chemicals Agency, amending directive 1999/45/EC and repealing council regulation (EEC) no 793/93 and commission regulation (EC) no 1488/94 as well as council directive 76/769/EEC and commission directives 91/155/EEC, 93/67/EEC, 93/105/EC and 2000/21/EC. Off. J. Eur. Union L 396, 1–850. EC (European Commission), 2008. Regulation (EC) no 1272/2008 of the European Parliament and of the Council of 16 December 2008 on classification, labelling and packaging of substances and mixtures, amending and repealing directives 67/548/ EEC and 1999/45/EC, and amending regulation (EC) no 1907/2006. Off. J. EurUnion L 353, 1–1355. EC (European Commission), 2011. Common implementation strategy for the water framework directive (2000/60/EC). Guidance Document No. 27: Technical Guidance for Deriving Environmental Quality Standards. ECHA. European Chemicals Agency, 2008. Guidance on information requirements and chemical safety assessment. Adaptation of Information Requirements. Chapter R.5. ECHA, Helsinki, Finland. EU (European Union), 2013. Directive 2013/39/EU of the European Parliament and of the Council of 12 August 2013 amending directives 2000/60/EC and 2008/105/EC as regards priority substances in the field of water policy. Off. J. Eur. Union L 226, 1–17. Eurostat, 2007. The use of plant protection products in the European Union. Data 1992– 2003 available at http://ec.europa.eu/eurostat/documents/3217494/5611788/ KS-76-06-669-EN.PDF/36c156f1-9fa9-4243-9bd3-f4c7c3c8286a?version=1.0 (last accessed September 4, 2015). Groothuis, F.A., Heringa, M.B., Nicol, B., Hermens, J.L., Blaauboer, B.J., Kramer, N.I., 2015. Dose metric considerations in in vitro assays to improve quantitative in vitro-in vivo dose extrapolations. Toxicology 332, 30–40. Gross, M., Daginnus, K., Deviller, G., de Wolf, W., Dungey, S., Galli, C., Gourmelon, A., Jacobs, M., Matthiessen, P., Micheletti, C., Nestmann, E., Pavan, M., Paya-Perez, A., Ratte, H.-T., Safford, B., Sokull-Klüttgen, B., Stock, F., Stolzenberg, H.-C., Wheeler, J., Willuhn, M., Worth, A., Zaldivar Comenges, J.M., Crane, M., 2010. Thresholds of toxicological concern for endocrine active substances in the aquatic environment. Integr. Environ. Assess. Manag. 6, 2–11.

Please cite this article as: Ciffroy, P., et al., Perspectives for integrating human and environmental exposure assessments, Sci Total Environ (2015), http://dx.doi.org/10.1016/j.scitotenv.2015.11.083

10

P. Ciffroy et al. / Science of the Total Environment xxx (2015) xxx–xxx

Hendriks, A.J., Traas, T.P., Huijbregts, M.A.J., 2005. Critical body residues linked to octanolwater partitioning, organism composition, and LC50 QSARs: meta-analysis and model. Environ. Sci. Technol. 39, 3226–3236. Hennes, E.C., 2012. An overview of values for the threshold of toxicological concern. Toxicol. Lett. 211, 296–303. Hollander, A., Bruinen de Bruin, Y., Arie Vonk, J., Vermeire, T., van de Meent, D., 2011. Exposure based waiving in environmental risk assessment: a tiered approach. Hum. Ecol. Risk. Assess. 17, 1173–1192. Kroes, R., Renwick, A.G., Cheeseman, M., Kleiner, J., Mangelsdorf, I., Piersma, A., Schilter, B., Schlatter, J., van Schothorst, F., Vos, J.G., Würtzen, G., 2004. Structure-based thresholds of toxicological concern (TTC): guidance for application to substances present at low levels in the diet. Food Chem. Toxicol. 42, 65–83. Li, Z., Kroll, K.J., Jensen, K.M., Villeneuve, D.L., Ankley, G.T., Brian, J.V., Sepulveda, M.S., Orlando, E.F., Lazorchak, J.M., Kostich, M., Armstrong, B., Denslow, N.D., Watanabe, K.H., 2011. A computational model of the hypothalamic-pituitary-gonadal axis in female fathead minnows (Pimephales promelas) exposed to 17α-ethinylestradiol and 17β-trenbolone. BMC Syst. Biol. 5, 63. Lombardo, A., Roncaglioni, A., Boriani, E., Milan, C., Benfenati, E., 2010. Assessment and validation of the CAESAR predictive model for bioconcentration factor (BCF) in fish. Chem. Cent. J. 4 (Suppl. 1), S1. Lopes, C., Persat, H., Babut, M., 2012. Transfer of PCBs from bottom sediment to freshwater river fish: a food-web modelling approach in the Rhône River (France) in support of sediment management. Ecotoxicol. Environ. Saf. 81, 17–26. McCarty, L.S., Arnot, J.A., Mackay, D., 2013. Evaluation of critical body residue data for acute narcosis in aquatic organisms. Environ. Toxicol. Chem. 32, 2301–2314. Meador, J., 2005. Rationale and procedures for using the tissue-residue approach for toxicity assessment and determination of tissue, water, and sediment quality guidelines for aquatic organisms. Hum. Ecol. Risk. Assess. 12, 1018–1073. Mourier, B., Desmet, M., Van Metre, P.C., Mahler, B.J., Perrodin, Y., Roux, G., Bedell, J.P., Lefèvre, I., Babut, M., 2014. Historical records, sources, and spatial trends of PCBs along the Rhône River (France). Sci. Total Environ. 476-477, 568–576. Munns Jr., W.R., Suter II, G.W., Damstra, T., Kroes, R., Reiter, W., Marafante, E., 2003. Integrated risk assessment — results of an international workshop. Hum. Ecol. Risk. Assess. 9, 379–386. Needham, L.L., Naiman, D.Q., Patterson Jr., D.G., LaKind, J.S., 2007. Assigning concentration values for dioxin and furan congeners in human serum when measurements are below limits of detection: an observational approach. Chemosphere 67, 439–447. Nuckols, J.R., Ward, M.H., Jarup, L., 2004. Using geographic information systems for exposure assessment in environmental epidemiology studies. Environ. Health Perspect. 112, 1007–1015. Péry, A.R.R., Geffard, A., Conrad, A., Mons, R., Garric, J., 2008. Assessing the risk of metal mixtures in contaminated sediments on Chironomus riparius based on cytosolic accumulation. Ecotoxicol. Environ. Saf. 71, 869–873. Péry, A.R.R., Schüürmann, G., Ciffroy, P., Faust, M., Backhaus, T., Aicher, L., Mombelli, E., Tebby, C., Cronin, M.T., Tissot, S., Andres, S., Brignon, J.M., Frewer, L., Georgiou, S., Mattas, K., Vergnaud, J.C., Peijnenburg, W., Capri, E., Marchis, A., Wilks, M.F., 2013a. Perspectives for integrating human and environmental risk assessment and synergies with socio-economic analysis. Sci. Total Environ. 456-457, 307–316. Péry, A.R.R., Brochot, C., Zeman, F.A., Mombelli, E., Desmots, S., Pavan, M., Fioravanzo, E., Zaldívar, J.M., 2013b. Prediction of dose-hepatotoxic response in humans based on toxicokinetic/toxicodynamic modeling with or without in vivo data: a case study with acetaminophen. Toxicol. Lett. 220, 26–34. Péry, A.R.R., Devillers, J., Brochot, C., Mombelli, E., Palluel, O., Piccini, B., Brion, F., Beaudouin, R., 2014. A physiologically based toxicokinetic model for the zebrafish Danio rerio. Environ. Sci. Technol. 48, 781–790. Ramachandran, G., 2008. Toward better exposure assessment strategies—the new NIOSH initiative. Ann. Occup. Hyg. 52, 297–301.

Russell, W.M.S., Burch, R.L., 1959. The Principles of Humane Experimental Technique. Methuen, London. Salazar, M.H., Salazar, S.M., 1998. Using caged bivalves as part of an exposure-dose– response triad to support and integrated risk assessment strategy. In: de Pesyster, A., Day, K. (Eds.), Proceedings-Ecological Risk Assessment: a Meeting of Policy and Science. SETAC Press, Pensacola, FL, USA, pp. 167–192. Schäfer, S., Buchmeier, G., Claus, E., Duester, L., Heininger, P., Körner, A., Mayer, P., Paschke, A., Rauert, C., Reifferscheid, G., Rüdel, H., Schlechtriem, C., SchröterKermani, C., Schudoma, D., Smedes, F., Steffen, D., Vietoris, F., 2015. Bioaccumulation in aquatic systems: methodological approaches, monitoring and assessment. Environ. Sci. Eur. 27, 5. Sokull-Klüttgen, B., 2007. Screening of the European Union new chemicals database concerning an aquatic exposure threshold of no concern. Environ. Toxicol. Chem. 26, 191–192. Sottas, P.-E., Lavoué, J., Bruzzi, R., Vernez, D., Charrière, N., Droz, P.-O., 2009. An empirical hierarchical Bayesian unification of occupational exposure assessment methods. Stat. Med. 28, 75–93. Suter II, G.W., Munns Jr., W.R., Sekizawa, J., 2003. Types of integration in risk assessment and management, and why they are needed. Hum. Ecol. Risk. Assess. 9, 273–279. Tielemans, E., Warren, N., Schneider, T., Tischer, M., Ritchie, P., Goede, H., Kromhout, H., Van Hemmen, J., Cherrie, J.W., 2007. Tools for regulatory assessment of occupational exposure: development and challenges. J. Expo. Sci. Environ. Epidemiol. 17, S72–S80. Ulaszewska, M.M., Ciffroy, P., Tahraoui, F., Zeman, F.A., Capri, E., Brochot, C., 2012. Interpreting PCB levels in breast milk using a physiologically based pharmacokinetic model to reconstruct the dynamic exposure of Italian women. J. Expo. Sci. Environ. Epidemiol. 22, 601–609. Van den Hazel, P., Zuurbier, M. (Eds.), 2005. PINCHE Project: Final Report WP1 Exposure Assessment. Public Health Services Gelderland Midden, Arnhem, The Netherlands. Van Leuwen, C.J., Patlewicz, G.Y., Worth, A.P., 2007. Intelligent testing strategies. In: van Leeuwen, C.J., Vermeire, T.G. (Eds.), Risk Assessment of Chemicals: an Introduction, Second ed. Springer, Dordrecht, The Netherlands, pp. 467–509. Verhaar, H.J.M., van Leeuwen, C.J., Hermens, J.L.M., 1992. Classifying environmental pollutants. 1. Structure-activity relationships for prediction of aquatic toxicity. Chemosphere 25, 471–491. Vermeire, T., Munns Jr., W.R., Sekizawa, J., Suter II, G.W., Van der Kraak, G., 2007. An assessment of integrated risk assessment. Hum. Ecol. Risk. Assess. 13, 339–354. Vermeire, T., van de Bovenkamp, M., de Bruin, Y.B., Delmaar, C., van Engelen, J., Escher, S., Marquart, H., Meijster, T., 2010. Exposure-based waiving under REACH. Regul. Toxicol. Pharmacol. 58, 408–420. Watanabe, K.H., Li, Z., Kroll, K.J., Villeneuve, D.L., Garcia-Reyero, N., Orlando, E.F., Sepulveda, M.S., Collette, T.W., Ekman, D.R., Ankley, G.T., Denslow, N.D., 2009. A computational model of the hypothalamic-pituitary-gonadal axis in male fathead minnows exposed to 17α-ethinylestradiol and 17β-estradiol. Toxicol. Sci. 109, 180–192. WHO, 2001. Integrated Risk Assessment. Report prepared for the WHO/UNEP/ILO International Programme on Chemical Safety (IPCS). World Health Organization, International Programme on Chemical Safety, Geneva, Switzerland (WHO/IPCS/IRA/01/12). WHO, 2008. Uncertainty and Data Quality in Exposure Assessment, Part I: Guidance Document on Characterizing and Communicating Uncertainty in Exposure Assessment, Harmonization Project Document No 6. WHO/IPCS. Wilks, M.F., Roth, N., Aicher, L., Faust, M., Papadaki, P., Marchis, A., Calliera, M., Ginebreda, A., Andres, S., Kühne, R., Schüürmann, G., HEROIC consortium, 2015. White paper on the promotion of an integrated risk assessment concept in European regulatory frameworks for chemicals. Sci. Total Environ. 521-522, 211–218. Zuurbier, M., Lundqvist, C., Salines, G., Stansfeld, S., Hanke, W., Babisch, W., Bistrup, M.L., van den Hazel, P., Moshammer, H., 2007. The environmental health of children: priorities in Europe. Int. J. Occup. Med. Environ. Health 20, 291–307.

Please cite this article as: Ciffroy, P., et al., Perspectives for integrating human and environmental exposure assessments, Sci Total Environ (2015), http://dx.doi.org/10.1016/j.scitotenv.2015.11.083