Advancing environmental toxicology through chemical dosimetry: External exposures versus tissue residues

Integrated Environmental Assessment and Management — Volume 7, Number 1—pp. 7–27 ß 2010 SETAC

7

Advancing Environmental Toxicology through Chemical Dosimetry: External Exposures Versus Tissue Residues LS McCarty,*y PF Landrum,z SN Luoma,§ JP Meador,k AA Merten,# BK Shephard,yy and AP van Wezelzz y LS McCarty Scientific Research and Consulting, Newmarket, Ontario L3X 3E2, Canada z Great Lakes Environmental Research Laboratory, Ann Arbor, Michigan, USA § USGS, Menlo Park, California, USA k NOAA, Northwest Fisheries Science Center Seattle, Washington, USA #NOAA, Coastal Response Research Center, Seattle, Washington, USA yy USEPA, Seattle, Washington, USA zz KWR Watercycle Research Institute, Nieuwegein, Netherlands

(Submitted 25 December 2009; Returned for Revision 19 February 2010; Accepted 7 May 2010)

This paper represents 1 of 6 review articles generated from a SETAC Pellston Workshop entitled ‘‘The Tissue Residue Approach for Toxicity Assessment (TRA)’’ (June 2007, Leavenworth, Washington, USA). The main workshop objectives were to review and evaluate the science behind using tissue residues as the dose metric for characterizing toxic responses and to explore the utility of the TRA for mixtures, guidelines or criteria, and ecological risk assessment.

ABSTRACT The tissue residue dose concept has been used, although in a limited manner, in environmental toxicology for more than 100 y. This review outlines the history of this approach and the technical background for organic chemicals and metals. Although the toxicity of both can be explained in tissue residue terms, the relationship between external exposure concentration, body and/or tissues dose surrogates, and the effective internal dose at the sites of toxic action tends to be more complex for metals. Various issues and current limitations related to research and regulatory applications are also examined. It is clear that the tissue residue approach (TRA) should be an integral component in future efforts to enhance the generation, understanding, and utility of toxicity testing data, both in the laboratory and in the field. To accomplish these goals, several key areas need to be addressed: 1) development of a risk-based interpretive framework linking toxicology and ecology at multiple levels of biological organization and incorporating organism-based dose metrics; 2) a broadly applicable, generally accepted classification scheme for modes/mechanisms of toxic action with explicit consideration of residue information to improve both single chemical and mixture toxicity data interpretation and regulatory risk assessment; 3) toxicity testing protocols updated to ensure collection of adequate residue information, along with toxicokinetics and toxicodynamics information, based on explicitly defined toxicological models accompanied by toxicological model validation; 4) continued development of residueeffect databases is needed ensure their ongoing utility; and 5) regulatory guidance incorporating residue-based testing and interpretation approaches, essential in various jurisdictions. Integr Environ Assess Manag 2011;7:7–27. ß 2010 SETAC Keywords: Review

Toxicity

Dose-response

Tissue residue

INTRODUCTION Tissue residues in toxicological theory and practice The primary dose metric used in existing environmental toxicity data and in risk assessment and regulatory activities is the exposure concentration, that is, a dose based on the concentration in the external media to which test organisms are exposed. Historically, the concept of dose as an amount or concentration of a substance applied to an organism by oral, dermal, or respiratory routes comes from the medical roots of toxicology. Also, most toxicity testing protocols were developed decades ago and reflect compromises required to address limitations in analytical chemistry methods and equipment at those times. In the aquatic environment, ambient exposure was also considered a useful dose metric * To whom correspondence may be addressed: [email protected] Published online 18 May 2010 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/ieam.98

Risk

Dosimetry

because it could be applied directly to water quality issues that were the immediate concern when such tests were developed. Thus, it is not surprising that dose metrics based on external media concentrations comprise the bulk of available environmental toxicity data. Before delving further into the issues associated with using the body or tissue residue approach (TRA) in environmental toxicology and risk assessment, it is appropriate to examine the nature and extent of its relationship with the current exposure-based approaches. If there is a single seminal contribution to the theoretical and practical aspects of environmental toxicology, it is the work of John Sprague. In a series of papers, Sprague (1969, 1970, 1971, 1973) provided detailed advice on most aspects of conducting aquatic toxicity testing. Of particular relevance was his guidance on the design and interpretation of toxicity tests. His strong advocacy for the use of time-independent toxicity metrics (threshold or incipient toxicity) is a key factor facilitating the use of a simple model as the theoretical construct for interpreting and understanding toxicity tests.

Special Series

EDITOR’S NOTE

8

Within the context of toxicity testing, a simple 1-compartment, first-order kinetics model describes how an organism, idealized as a single compartment with little or no biotransformation, interacts with an exposure medium. Using an aquatic example, with the chemical dissolved in water, the organism will adsorb and/or absorb a fraction of that chemical on and/or in its body. The process takes time, and at some point a steady state between the concentration in water and the concentrations on or in the organism can usually be reached. The characteristics and activities of both the substance and the organism generally result in variable chemical concentrations in the exposure medium and organism throughout the period of exposure. Sprague’s advice on understanding the nature and magnitude of physical, chemical, and biological toxicity-modifying factors on toxicity testing outcomes is directly applicable to the interpretation of toxicity test data within the context of this model. Because toxicity-modifying factors alter the relationship between the surrogate dose metric and the true target or biologically effective dose (BED) somewhere in the body of the organism, quantitative knowledge of such alterations allows for adjustment or correction of case-specific experimental dose surrogate data. Valid toxicological comparisons of substances must be on the basis of a consistent, comparable measure of relative toxicity. This is best achieved by accounting for the effects of various toxicitymodifying factors. Examination of the body and/or tissue residue approach to toxicity test interpretation has been carried out using a simple 1-compartment toxicokinetics model equation to describe the time course of adsorption/absorption of chemical by the organism (e.g., Landrum et al. 1992; McCarty et al. 1992). The equation parameters include water and organism concentrations of the test substance, uptake and elimination rate coefficients for movement of substance between water and organism, and time. This model can be employed in 2 ways in toxicity testing design and interpretation. Either the concentration of test substance in the organism is not known but is used implicitly (i.e., the external media serves as the dose metric), or the organism concentration is used explicitly with estimates or measurements of body residues. Commonly the concentration in the organism is used implicitly, i.e., implied from the concentration in the water. This allows the effort to be minimized as chemical analysis of organism tissue has typically been time-consuming, expensive, and confounded by various factors. Nonetheless, whenever standard toxicity testing protocols are followed, the accumulated test substance on/in the body of exposed organisms is actually considered, either implicitly or explicitly, as an integral part of the underlying toxicological theory. However, neither exposure-based external nor organismbased internal dose metrics are completely correct. Both are surrogate measures because it is generally agreed that the effective dose at the sites of toxic action on and/or in an organism is the ‘‘true’’ measure of dose. Unfortunately, this is very rarely estimated, and in most cases the sites of toxic action are unknown. In practice, the answer to the question ‘‘What is a dose?’’ really reflects selection of a surrogate dose metric that is most efficient and effective for the particular task in question. Complications arise from the poor quantitative understanding of the relationships both among and between various surrogates and the actual effective dose,

Integr Environ Assess Manag 7, 2011—LS McCarty et al.

introducing substantial uncertainty into interpretation and use of toxicological data. Nevertheless, information on organism-based internal dose metrics provides additional key knowledge that is vital for more thorough interpretation and understanding. Temporal aspects are also required for adequate interpretation of toxicity regardless of the unit of the dose. Too often, researchers attempt to make comparisons between toxicity tests that have different durations of exposure. The reason Sprague advocated the use of toxicity determinations at steady state was to ensure that the temporal element reflecting the differences in the toxicokinetics of different compounds would be minimized. Care must always be taken to ensure differentiation between discussions about toxicokinetic/dynamic issues for the case-specific dose surrogate model being used versus the actual effective dose. Thus, toxicity testing is based on a theoretical foundation that can be expressed as a simple model with associated assumptions and limitations inherent to all models. Although detailed guidance for carrying out various media-based toxicity testing protocols is available (e.g., ASTM 2003), and detailed guidance for interpretation can be found in many sources (e.g., Rand et al. 1995), few official protocols and interpretive guidance are available for tissue residue-based toxicity testing. Whereas explicit broad consideration of organism-based dose metrics within a theoretical framework was initiated more than 100 y ago (see Ferguson 1939), the absence of residue-based approaches in most regulatory arenas must be attributed to the historical factors noted above, rather than any lack of scientific validity.

BRIEF HISTORICAL PERSPECTIVE ON BODY AND/OR TISSUE RESIDUES IN ECOTOXICOLOGY The field of body and/or tissue residues in ecotoxicology has been one of considerable activity, especially since the 1980s, with the publication of a number of integrative reviews examining theory and existing experimental data began. However, as the old saying goes, ‘‘There is nothing new under the sun.’’ The seminal event in the history of the residue-based dose approached occurred more than 100 y ago. Discussions about narcosis in anesthesia starting in the mid-1800s culminated in the publishing of independent papers near the turn of the 20th century by Meyer and Overton that linked narcosis to the lipid solubility of the substance and pointed to an amount of chemical in target sites in organisms as the effective dose metric. The Meyer– Overton theory of narcosis stimulated much research into both estimating the effective tissue dose and understanding the relationship between external and internal dose for organic chemicals (see Lipnick 1995). Ferguson (1939) reviewed important studies based on this general approach of linking the physical properties of chemicals to toxic effects. He consolidated theoretical and experimental aspects and offered an improvement by advocating the use of measures of chemical activity rather than chemical concentration. He also proposed a mode of action classification scheme, i.e., physical toxicity for narcosis and chemical toxicity for other types of toxic action. This was followed up by a 1948 paper containing new supportive experimental data (Ferguson and Pirie 1948). Statements made by Ferguson about the confounding effects of the differences in the distribution of chemical between the exposure media and the site(s) of toxic action in the

External Exposures Versus Tissue Residues—Integr Environ Assess Manag 7, 2011

organisms, i.e., ‘‘effect of phase distribution,’’ are still relevant today. As stated by Ferguson (1939): It will be shown that the toxic indices so determined for any one organism lie within a relatively narrow range of values, although the measured toxic concentrations may vary very widely from substance to substance.The method adopted is to use, instead of the measured concentrations, the chemical potentials of the substances in a phase in equilibrium with that phase (whatever it may be), which is the seat of toxic action. If a true equilibrium exists, the chemical potential of the toxic substance must be the same in all phases partaking in the equilibrium. Hence the chemical potential of the toxic substance at the actual point of attack is known. Ferguson determined that narcosis occurred within a relatively narrow range of activity, between 0.1 and 1.0 of the ratio between the vapor pressure causing the toxic effect and saturated vapor pressure for the chemical. Although he explained the ‘‘toxicity cutoff’’ reported in toxicity testing for homologous series of chemicals and expected that the improved estimates would further narrow the activity range in which narcotic toxicity occurred, he declined to refine his predictions because of the poor quality of the experimental data available for analysis (Ferguson and Pirie 1948). Mullins (1954) extended the concept of activity by introducing fugacity as a measure of activity and proposed that a constant volume (mole fraction) in a nonaqueous phase of the organism was responsible for narcosis. In a series of 3 papers, McGowan (1951, 1952a, 1952b) continued the theoretical development started by Ferguson. McGowan used molar concentrations, rather than using Ferguson’s chemical activity directly, and further promoted the quantitative structure–activity relationship (QSAR) approach to interpret toxicity test results, employing molar volume/parachor and hydrogen bonding as the molecular descriptors. On the basis of extensive QSAR analysis, McGowan (1951) offered the following equation (1) to explain physical toxicity (narcosis), in which concentrations are in moles/L: log10 Ct ¼ log10 CB þ 0:012P þ EA ;

ð1Þ

where Ct is the toxic concentration in water, CB is a constant for a given manifestation of toxicity and represents the toxic concentration in some biophase, [P] is the parachor of the toxic substance in cubic centimeters, and EA is a correction term for associated substances with a tendency to hydrogen bonding; it is zero for nonassociated substances with no tendency to take part in hydrogen bonding. Exposure time, organism body size, and severity of toxic effect were acknowledged as important toxicity-modifying factors. McGowan also expanded the mode of toxic action classification to 4 classes: chemical toxicants, metabolite analogues, surface-active compounds, and physical toxicants (narcosis). He then used the above equation to separate physical toxicity from the others and examine additivity within this mode of action. He used this approach to examine the general residue-based hypothesis that ‘‘there is some ‘biophase’ in which a given molar concentration of any compound will prove toxic . . .’’ (McGowan 1951, p 120). The results of his analysis displays substantial noise because the data sets varied widely by taxa (bacteria, fungi, plants,

9

invertebrates, and vertebrates), exposure (vapor, water), duration, toxic effect, and chemicals tested. McGowan’s general estimate of a critical biophase concentration for narcosis was on the order of 1 mol/L, and more likely within the range of 0.2 to 0.5 mol/L for acute narcosis based on analysis of experimental data. This finding is consistent with modern estimates. Van Wezel and Opperhuizen (1995) reported effective narcotic concentrations in membrane preparations of 0.04 to 0.16 mol/kg lipid, whereas Di Toro et al. (2000) estimated 0.034 to 0.286 mol/kg octanol, with octanol used as a surrogate for lipid. For small fish with about 5% body lipid, 0.2 mol/L in the target biophase would correspond to approximately 0.01 mol/L (10 mmol/L) on a whole-body basis. This finding agrees with modern estimates on the order of approximately 5 mmol/L for whole-body residues at about 5% body lipid content (McCarty et al. 1992). Although the narcotic target site is not the total hydrophobic or lipid phase of an organism, rather some fraction of it, these approximations provide valuable insights (Escher et al. 2011). Perhaps the most currently neglected aspect of this key historical work is the lack of exploitation of fugacity. As a measure of the chemical activity of the molar concentration of chemicals in an abiotic or biotic phase, it could both provide a basis for a better resolution of the influence of varying lipid content on body and/or tissue residue measurements while providing the opportunity to use an activity-based dose metric for toxicological comparisons. The activity of the molecules of chemical at the sites of toxic action may be even better correlated with adverse effects than the number of molecules present. In addition, fugacity has considerable advantages for environmental fate and exposure modeling (Mackay 2001). Residue–effect relationships for metals and/or inorganics also have a long history in environmental toxicology. The oldest known residue-effects study on an aquatic species is by White and Thomas (1912), who studied the effects of Cu exposure to fish. Whole-body concentration of Cu (6.8 to 81.7 mg/kg wet wt) in mummichog (a killifish) was lethal in 1 to 96 h. Given analytical procedures available in 1912, the results of these early researchers are remarkably similar to those reported in the mummichog study by Eisler and Gardner (1973). In this case, 26.2 mg/kg whole-body wet weight Cu was associated with histologic lesions, erratic swimming, and darker pigmentation; 112 mg/kg wet weight was lethal. Another early residue-effect study with fish appears to be the first to look at tissue residues in multiple tissues, the whole body, and mucus precipitate on gills (Carpenter 1927). Interestingly, mortality caused by Pb exposure in this study was attributed to Pb precipitate in mucus on gills leading to suffocation rather than to adverse effects from Pb accumulated in the tissues of the fish. Thus, it is apparent that the argument that the residuebased dose approach is somehow new, untested, or perhaps theoretically deficient is incorrect. Modern applications differ little conceptually for either organic chemicals or inorganics/metals compared with the pioneering work done in the first half of the 20th century. Fortunately, substantial improvements in both chemistry and toxicity testing have provided greater amounts of higher-quality experimental data. In addition to success with simple body residue relationships for organic chemicals, some successes have been achieved with some metals and inorganics, especially with organotins and organic mercury (e.g., Meador

10

2000). However, metals and/or inorganics as a group continue to be more toxicologically challenging than organics because of the disparate physicochemical characteristics of metals, their environmental presence in multiple forms and/ or states, and the fact that some are essential micronutrients and some are actively controlled by metabolic processes (Vijver et al. 2004). Nevertheless, residue-based approaches for metals can be successful with detailed consideration of metal-specific and species-specific details and determination of the toxicologically active fraction of the total body, organ, and tissue residue (Luoma and Rainbow 2005).

DOSIMETRY All toxic events can be understood by means of a simple framework describing the 3 components of toxic action: exposure, toxicokinetics, and toxicodynamics, or, more colloquially, exposure, accumulation, and toxicity. For exposure, a number of physical, chemical, and biological toxicity-modifying factors act to determine the effective bioavailability of the substance in the exposure media to the organism. In terms of toxicokinetics, a portion of the bioavailable substance in media is absorbed into the organism, where it undergoes distribution to various phases, organs, and tissues, possible metabolic conversion/degradation, and elimination. As for toxicodynamics, a fraction of the absorbed dose reaches the targets or sites of toxic action, where it initiates a chain of biochemical reactions that lead to the adverse effect. The nature of the adverse effect is determined by the mode/mechanism of toxic action (for definitions, see Meador et al. 2011); the magnitude is related to the potency of the substance for the mode/ mechanism in the exposed organism in question. For both toxicokinetics and toxicodynamics, a number of intra- and interspecies factors may confound precise determination of toxicity (Rand et al. 1995; Meador, McCarty, et al. 2008). However, by selecting internal dose as the metric for toxicity determination, the influence of some inter- and intraspecies kinetic factors can be determined. Although dose is a fundamental concept in toxicological theory, there are operational difficulties. The true effective dose present at the sites of toxic action, and responsible for causing an adverse effect, is almost always unknown; only surrogate dose measurements or modeled target site doses are available. Thus, currently available information on dose is complicated and is confounded by model-design assumptions, limitations, and data requirements. The simplest model is the 1-compartment, first-order kinetics model noted earlier. More sophisticated multicompartment, physiologically based kinetics models have been developed (e.g., Andersen et al. 1995; Lien et al. 2001), but the level of effort (specifically the data required) to use them is much greater and precludes routine application. A residue-based effects model explicitly considers the separation of toxicokinetics (adsorption, distribution, metabolism, and elimination [ADME]) from toxicodynamics (the adverse effect associated with the presence of substance at the sites of toxic action). This approach develops a surrogate dose metric that is more closely connected to the site of toxic action, assuming that the accumulated residue measured in a whole body or a specific tissue is consistently proportional to the concentration at the receptor site(s). Early efforts used simple kinetic models, such as 1compartment toxicokinetics without biotransformation, and/or QSAR to emphasize the importance of residue-based

Integr Environ Assess Manag 7, 2011—LS McCarty et al.

dose metrics and interpret and/or predict the whole-body residue from acute aqueous exposure toxicity data. (e.g., Bobra et al. 1985; Connolly 1985; Friant and Henry 1985; Hermens et al. 1985; Lassiter 1986; McGowan and Mellors 1986; McCarty et al. 1985; McCarty 1986, 1987a, 1987b). Such simple analyses worked best for the case of poorly metabolized organic chemicals causing acute toxicity from narcosis in small aquatic organisms in which the whole-body residue was a reasonable dose surrogate. For example, for small fish with a body lipid content of about 5%, such an analysis demonstrated that much of the more than 100 000fold range of LC50 results was attributable to the combination of a relatively constant body residue (estimated to be
5 mmol/kg body weight) with large hydrophobicity-driven differences in bioaccumulation into the nontarget hydrophobic (lipid) phase (McCarty et al. 1992). Examples such as this demonstrate the strength of the TRA. Large differences in apparent toxicity of a broad group of chemicals, as measured by their aqueous LC50 values, were reduced to a narrow lethal residue (LR50) range that could be approximated by a constant value. This provides a better interpretation and explanation of the experimental data in addition to facilitating improved toxicological interpolation and extrapolation of the data for regulatory activities. Such conclusions represent a modern restatement of the findings of Ferguson, McGowan, and others, as noted earlier. Subsequent studies often employed this simple approach, largely using mortality as the toxicity endpoint. However, sublethal endpoints such as growth and development can be linked to body residues often at concentrations about 10 times lower than acutely toxic residues. The body and/or tissue residue reflects the bioavailable and bioreactive target ‘‘dose’’ (i.e., McGowan’s ‘‘biophase’’) better than the traditional dose that is defined in terms of the effective concentration in the exposure media (e.g., water, air, sediment/soil) or a concentration exposed per unit mass of the organism (e.g., diet). The target dose focuses on the toxicologically relevant portion of the exposed substance and may be organism and substancespecific, although biological similarities between organisms and commonality in modes/mechanisms of toxic action can allow some general groupings to be made (McCarty and Mackay 1993; Russom et al. 1997; Escher and Hermens 2002; McCarty and Borgert 2006a). Although much of the investigative work has been carried out with organic chemicals, linking adverse effects with body and/or tissue residues of metals has been successful, largely in metal-specific and species-specific cases. Therefore, metal body and/or tissue residues may be useful for understanding metal toxicity in some situations (e.g., Borgmann et al. 1991; Spry and Wiener 1991). For cationic metals (e.g., Ag, Cd, Cu, Ni, Pb, Zn), tissue residues exhibit a considerably narrower range of effect concentrations in exposure waters with different pH, alkalinity, hardness, and/or dissolved organic carbon (DOC) concentrations, compared with the wide range of aqueous LC50 values observed in the same exposure waters. A number of examples are noteworthy, including Cd in Hyalella (Borgmann et al. 1991), Cu and Ni in fish (Meyer et al. 1999), and Cu in an aquatic oligochaete (JS Meyer et al. 2002). In fact, establishing a relatively constant median lethal accumulation (LA) for a number of metals was a major key to development and acceptance of the biotic ligand model (BLM) (Di Toro et al. 2001). Essentially, the high variability in aqueous LC50 values of a given metal can be reduced to a relatively constant

External Exposures Versus Tissue Residues—Integr Environ Assess Manag 7, 2011

LA50, whether directly measured in tissue or inferred computationally by fitting the BLM to toxicity data. By assuming a constant LA50, the BLM can predict within a deviation of less than plus or minus 2 for predicted and observed values, the LC50 values of a given cationic metal across a wide range of water chemistry. An extensive review of the influence of water chemistry on the accumulation and toxicity of selected metals in aquatic organisms is presented in a recent SETAC book (Meyer, Clearwater, et al. 2007). For both organic chemicals and metals, the bioavailable and bioactive fraction of the dose that reaches the target sites and initiates a given adverse effect is influenced by a number of things. The relationship between exposure concentration, body and/or tissue residues, and dose at the receptors appears to range from a relatively simple proportionality (neutral organic narcotics) to a varied and complex one (many metals). Major influencing factors include: intake rates, ADME, growth and reproductive processes, and mode and/or mechanism of toxic action. Other factors, some of which will be discussed later, also influence the amount of the exposed substance that reaches the target sites. Because the type, nature, and influence of toxicity-modifying factors differ considerably between organic chemicals and metals/metalloids, they will be discussed separately. What is a dose: organics Other than mode/mechanism of toxic action, the largest factor controlling the toxicity of organic chemicals is hydrophobicity, which influences toxicokinetics both externally and internally. Partitioning between exposure medium and organism is well described by the concept of fugacity, a measure of the activities of organic chemicals in various compartments and/or phases of the environment and organisms (Mackay 2001). Although polarity, ionization, molecular size, and other factors can affect the hydrophobicity-driven distribution of organic chemicals, the sheer magnitude of the hydrophobicity effect, on the order of 105 or more, ensures that it will predominate in water exposures. The simple case noted earlier for acute toxicity of poorly metabolized neutral organics to small fish with relatively constant body lipid content illustrates well the key influence of hydrophobicity-driven partitioning on determining simple narcotic toxicity based on whole-body residues (McCarty et al. 1992). It is clear that organic chemicals are also subdivided within compartments in the organism in a manner related to the degree of hydrophobicity. This can be established for either phases within the organism, such as water, lipid (which can be subdivided into various subphases), protein, and DNA, or by organs and/or tissues. The hydrophobic phase in organisms is composed of the sites of toxic action and a nontarget phase consisting largely of various lipids. It is generally thought that the sites of toxic action for narcosis represent a small proportion of the total hydrophobic phase, whereas the nontarget phase represent a much larger proportion, although both may be variable both within and between organisms and species (e.g., age, sex, condition). Because the target and nontarget hydrophobic phases appear to share broadly similar characteristics, concentrations of absorbed organic chemicals are considered generally similar in both, at least as a first approximation. However, differences in the character of the hydrophobic phase, especially differences in sizes, nature, and hydrophobic capacity of the various subphases, will have a

11

significant influence on the amount of chemical accumulated in each phase and in the organism as a whole. Thus, overall, it is not surprising that toxicity estimates are often correlated with body lipid content and that lipid content is a dominant toxicity-modifying factor for the toxicity of organic chemicals (e.g., Geyer, Scheunert, Rapp, et al. 1993; Geyer, Scheunert, Bruggemann, et al. 1993). However, this general rule on the relationship between whole-body residues and toxicity should only be exploited with caution. Distribution of absorbed chemical within and between the various organism compartments and/or phases will exhibit temporal variation, being especially long term for poorly perfused lipid compartments, compartments with large hydrophobic capacity, and very hydrophobic organic chemicals. Thus, the relationship between whole-body concentration and target concentration can vary temporarily, perhaps substantially, until all phases/compartments reach their respective steady state. Although broad residue–effect relationships will often be improved by correction to a constant lipid level, because partitioning of organics is dominated by hydrophobicity, quantitative evaluation of the influence of lipid content on target to whole-body residue relationships requires that the relative absorption capacity and status of the various phase and/or compartments be known. As noted earlier, the use of chemical activity (i.e., fugacity) remains a valuable but largely underexploited approach that should enhance our ability to determine the nature and extent of partitioning into various phases with differing lipid content and/or hydrophobicity. Although the sites of toxic action for narcosis are often considered to be cell membrane phospholipids, evidence suggests that it may in fact be specific protein-binding sites in cells (Franks and Leib 1982, 2004). In the quest to better establish BEDs in terms of a target-phase concentration, more precise estimation of substance concentrations in various hydrophobic subphases, starting with the broad categories of nonpolar and polar lipids, may be of substantial value. Overall, detailed understanding of partitioning of chemicals into various body subphases is key to identifying and understanding the nature of the target sites and their relationship with whole-body and tissue residues. Myriad factors caused by multiple, sometimes speciesspecific, modes of toxic action, differences in rates of adsorption, distribution, metabolism, and elimination, biotransformation to more or less toxic metabolites, polar or ionizing chemicals, reactive chemicals, etc. (e.g., Rand et al. 1995; Hwang et al. 2004; Schuler et al. 2006; Escher et al. 2011), can influence toxicity testing outcomes for many chemicals, making interpretation complex and difficult. However, the simple case of narcosis (baseline) toxicity of poorly metabolized, neutral organic chemicals to small aquatic organisms with relatively similar body lipid levels provides a strong heuristic base for organic chemicals. As discussed below, metals and organic chemicals differ substantially in a number of important characteristics. Even though internal dose is important for understanding the toxicity of metals, as noted earlier, there is no common, relatively simple case for metals, such as narcosis for organic compounds, to build on. What is a dose: metals For this discussion, we have lumped together metals, metalloids, and organometallics because of many similarities

12

in biophysicochemical behavior. It is often assumed that the broad principles that drive bioaccumulation and effects of organic contaminants can be extrapolated to metals. But important differences also exist. Failure to appreciate the differences contributes to some of the apparent regulatory contradictions with metals. The most important difference is that bioavailability, uptake processes, and factors that determine toxicity are more specific toward individual metals, species, environments, and exposure routes than for organic chemicals. Unlike organics, no single characteristic describes the dominant factor governing bioaccumulation of metals. No single coefficient spans these differences, in the way fugacity facilitates understanding of the behavior of organic contaminants. This results from the fundamental difference in the mechanisms that govern toxicokinetics. Organic contaminants are often accumulated largely via passive transport across membrane barriers whereas metals require active or facilitated transport for uptake and elimination. Distribution within the organism is also different because of different types of binding, i.e., nonpolar interactions and water solubility for organics versus ligand binding and active transport into cells for metals. Thus, for metals as a group, a multifaceted conceptual and mathematical construct is necessary to tie environmental exposure to internal dose and then to toxicity (Luoma and Rainbow 2005). The unifying construct recognizes each metal exists in many forms of differing bioavailability in water, sediment, and organisms. The distribution among forms is driven by diverse environmental conditions, including dissolved organic material, major ion concentrations, redox conditions, and biological productivity. This chemical distribution greatly influences metal bioaccumulation. Chemical activity (e.g., free ion activity) drives bioavailability of some metals (e.g., Cd, Cu, Pb, Zn), whereas redox states drive the bioavailability of another set of metals/metalloids (e.g., Cr, Se, As). Formation of an organometallic complex can increase or decrease bioavailability of metals. The bioaccumulation process is also metal specific. Translocation of metals across the phospholipid membrane depends on specific carriers or pores. The carriers responsible for transport among the metals have important differences. For example, Cu appears to be transported as Cu1þ on Naþ carriers and Cd2þ on Ca2þ carriers and the most likely process for organo-Hg transport is cotransportation with an amino acid (Croteau et al. 2007). The latter may also be important for at least some metals in the gut. Finally, like organic chemicals, hydrophobic organometal complexes appear to be taken up largely via passive diffusion. Thus, metal-specific models are necessary to determine how major ion concentrations in a water body affect uptake rates. Bioaccumulation processes also differ widely among species. For example, both uptake rate and efflux rates of Cd differ by orders of magnitude among species of freshwater aquatic insect larvae (Buchwalter et al. 2007). This is because insects manage major ions through different mechanisms. Species with large numbers of chloride cells, used for osmoregulation, have more rapid uptake rates. The result is a wide interspecific variability, under the same environmental conditions, in metal bioaccumulation. A unifying construct must recognize this diversity of biological and chemical specific uptake characteristics and their implications for the bioaccumulated ‘‘dose.’’ As metal bioaccumulation is a metal-organism-specific process application

Integr Environ Assess Manag 7, 2011—LS McCarty et al.

of internal residues to assessment and management of risks from metals requires applying existing knowledge of organism biology and metal chemistry in ways that take advantage of our understanding without oversimplifying the complexities (Adams et al. 2011). Externally, chemical activity has considerable influence on bioavailability from dissolved sources, as noted above. However, after metals are taken up into the rich organic milieu inside the organism, chemical activity is less relevant as the metals are largely tied up with various ligands and do not exist to any great extent in the free state, although the extent and degree of ligand binding are likely influenced by internal chemical activity. All organisms evolved in the presence of metals and many are essential for physiological processes. Toxic effects ensue when an excess of metal is present internally in a form that is metabolically available (bound to sites where it can disrupt molecular function). Survival of the species required evolution of biochemical and physiological mechanisms to control internal binding and either distribute the metal to useful sites (essential trace metals) or eliminate potential disruption of cellular processes. As a simplified model, metals can be divided into either detoxified or metabolically active forms. In reality, a complicated dynamic series of reactions bring about detoxification of trace metals to different degrees rather than an absolutely black-and-white division into 2 extreme forms (Luoma and Rainbow 2008). Trace-metal detoxification processes may involve binding to soluble ligands (e.g., metallothioneins, glutathione) and the formation of extracellular or intracellular insoluble granules. Detoxification occurs because the intracellular metal is bound by a ligand of such high affinity that it effectively prevents the metal from binding to other molecules or causing toxicity. Detoxification by soluble ligands involves association with metal-specific binding proteins or metallothioneins (MT), which are low-molecular-weight cytosolic proteins that are induced by and complex with certain trace metals (e.g., Zn, Cu, Cd, Ag, Hg) (Roesijadi 1992; Amiard et al. 2006). These MTs are involved in both cellular regulation and detoxification. Hopkin (1989) described 3 types of intracellular granules. Type A consists of concentric layers of calcium and magnesium phosphates that may contain trace metals such as Mn and Zn. Type B granules are more heterogeneous in shape and always contain sulfur in association with metals that include Cu and Zn. Type C granules are often polyhedral with a crystalline form, mainly containing Fe, probably derived from ferritin. Thus, analogous to organic contaminants, biotransformation is important and different species have different biotransformation capabilities and approaches. But the generalizations about detoxification capabilities among groups of organisms are much less developed for metals than for organic contaminants, although much remains to be learned about both. Toxicity of a metal is thus a function of both uptake characteristics of the organism and the species-specific detoxification capabilities. More specifically, toxicity results from the interaction of 3 rate processes: uptake, detoxification, and excretion. Kinetics are clearly important for metal bioaccumulation and toxicity. If the rate of metal uptake into an organism is greater than the combined rate at which it can be excreted or detoxified, then that metal will accumulate internally in metabolically available form that can bind

External Exposures Versus Tissue Residues—Integr Environ Assess Manag 7, 2011

inappropriately to internal molecules and cause toxic effects. If the trace metal continues to accumulate in a metabolically available form, sublethal toxic effects of increasing strength and severity will ensue, followed by lethality. Thus, accumulation of metabolically available metal determines toxicity, not total metal. Conceptually, the metabolically available fraction (also known as the biologically active metal (BAM)) can be defined (see Adams et al. 2011). Methods have been developed to operationally isolate, quantify, and provide a crude estimate of the activity of this fraction of the total metal concentration. Consequently, the BAM varies widely among species and because detoxification is highly variable among metals, the relationship between total burdens in the organism and toxicity is often complex. These are usually species specific and metal specific in addition to being conditional, in that they are influenced by environmental properties. Thus, similar to organic contaminants, the dose of metal depends on the internal burden. Metals are also similar to organics in that they partition between target and nontarget phases. Hence, the relationship between total internal metal burden and the active fraction forming the effective dose is confounded by complex internal partitioning. It is clear that the relationship for metals is sufficiently complex that the concept of a similar simple critical body and/or tissue residue–effect relationship similar to baseline narcosis is not broadly applicable.

ADDRESSING KEY ISSUES WITH THE TISSUE RESIDUE APPROACH Exposure-based toxicity testing data that explicitly use exposure-media concentrations as the reportable dose metric, while implicitly using body and/or tissue residues, are the most abundant type of environmental toxicity data. Although this information has served environmental toxicology well, increasing sophistication in technical aspects of toxicology and chemistry and increasing complexity in risk assessment goals and processes are testing the limits of these traditional approaches. Accepting the status quo entrenches limitations and constrains the valuable contributions novel approaches might make in addressing these new challenges. Quantitative knowledge of the relationships between and within the continuum of dose surrogates from external, media-based surrogates, through body, organ, and tissue surrogates, to the actual site(s) of toxic action should substantially improve toxicological information providing the basis for improving toxicological understanding and enhancing regulatory activities (McElroy et al. 2011). Brief discussions of key issues are presented below. Exposure For a chemical to elicit an adverse response, an organism must be exposed to it for a sufficient time to allow the substance to accumulate and reach at the sites of toxic action. Also, the concentrations at the sites of action must be at adequate levels for a sufficient duration for the effect to be initiated. In conventional exposure-based approaches, the dose is based on an external exposure duration and an exposure medium concentration, little or nothing is known of the actual internal exposure duration or concentration inside the organism. When tissue concentrations are not considered, the determination of bioavailability from the exposure medium is precluded. Because the amount of chemical and levels

13

within the body are unknown, it is also difficult to establish clear causality and toxic potency relationships. For example, metabolism is a well-known modifier of toxicity that usually reduces exposure to the parent compound. This makes the compound appear less toxic, because additional amounts must be added to the exposure medium to compensate for that degraded by the organism, making it unavailable to induce toxicity. Alternatively, the toxic agent can be a metabolite or conversion product of the original substance. This phenomenon is difficult to determine without at least knowledge of body and/or tissue concentrations of the parent compound and preferably residue information for both parent and metabolites. The problem also extends to evaluation of body and/or tissue residues in field-collected organisms. Measured tissue concentrations in field samples reflect the interactions and contributions of usually poorly quantified exposure characteristics such as sources, duration, and magnitude. Although not always diagnostic of the presence or absence of exposure and/ or toxic effects in the field, measured body and/or tissue concentrations of a given substance can be helpful in confirming and/or evaluating exposure. Furthermore, tissue residue toxicity data can help improve estimates for risk based on the field-measured data. Bioavailability and routes of exposure. Body and/or tissue residues reflect the bioavailability of accumulated chemical from the exposure media and can be used experimentally to establish and quantify routes of exposure. The TRA for organic contaminants promises to help resolve issues in which the toxicity of compounds becomes complicated by the exposure scenario. Sediment toxicity is a prime example where issues of bioavailability complicate the exposure– response relationship. Recent work demonstrates that, for organic contaminants, the bioavailable fraction can be determined as the fraction that is rapidly desorbed (Cornelissen et al. 2001; Kraaij et al. 2001). Similarly, the difference between extracted metal and acid volatile sulfide (SEM/AVS) is employed in some circumstances to estimate metal bioavailability from sediments (Ankley et al. 1996). However, SEM/AVS does not capture the full complexity of metal exposure from contaminated sediments. In all cases, exposure is not only a function of concentration in interstitial water but also reflects ingestion of contaminated particles (including living organisms). Functional ecology is also important in determining bioavailability as, for example, how and where an organism feeds can alter the effective bioavailability, and the dominant route of exposure may vary with species and circumstances. Chemical methods can capture such complexities for organic contaminants, with infaunal species in which biotransformation is not an issue (Landrum et al. 2007) but may not be as successful for those with more limited connections to sediment or greater ability to biotransform compounds. However, residue-based dose estimates allow for better accounting for the influence of bioavailability and toxicokinetics issues. For example, the residue-toxicity data for DDT, DDD, and DDE were generated for 10-d exposures providing body residue values and showing differences in the potency of the compounds (DDT > DDD > DDE) in Hyalella azteca independent of the exposure scenario (Lotufo et al. 2000). These residue-effects data can then be applied to the observed toxicity of H. azteca in sediments containing DDT, where the

14

composition of the exposure was complicated by degradation and biotransformation. The toxicity response was well explained on a residue basis using a toxic unit approach to account for the differences in the potency of the degradation products compared with the parent DDT (Lotufo et al. 2001). Interpretation of mixture toxicity on a media basis would have been complicated, because different equimolar mixtures of the compounds would have produced different toxic responses. Thus, the potential differences in bioavailability, rates of degradation, and biotransformation were integrated; the ability to interpret the contribution of mixture components, their relative potencies, and potential biotransformation within the organism were accomplished through a residue-based approach. For metals, controversy about exposure routes has long held back interpretation of toxicity in natural settings. In the last decades, models have become available that explain the relative importance of exposure for metal accumulation in many species (Wang and Fisher 1999; Luoma and Rainbow 2005). Within the past 5 y, experiments also show that toxicity via dietary exposure can occur at far lower water concentrations and substantially reduced body residues, in at least some cases, than toxicity via water-borne exposure (Hook and Fisher 2002). Appropriate residue-based approaches provide a new avenue by which to interpret such heretofore ignored complexities. Toxicokinetics. Exposure-based toxicity testing tends to treat the organism as a ‘‘black box.’’ In contrast, residue-based approaches force explicit consideration of the characteristics of the interpretive model being used along with its assumptions, including quantification of model parameters. Toxicokinetics: exposure time. Before steady state between the external medium and the target site is reached, increases in exposure duration usually cause increased concentrations at the target site. Adverse effects can be initiated any time a critical target concentration is achieved; steady state for either the body and/or tissue or the target is not required. However, steady state is a common requirement for interpreting many ambient exposure-based toxicity tests. It represents the situation in which the external media and the organism/target are in a fixed but not usually quantified relationship, largely independent of time, that may be used to estimate relative toxicity/potency and provide an endpoint for some models. When measured, external concentrations are related to the body exposure and duration through a combination of both toxicokinetics (ADME) and toxicodynamics to yield an effect. When body and/or tissue information is available it becomes possible to separate the kinetics of absorption, distribution, metabolism, and excretion. These processes provide information on the sequence of events that allow quantification of accumulation from the exposure medium and delivery to the sites of toxic action in addition to the time course for toxicant accumulation and response by the organism. This separation is imperfect, and rarely, if ever, is the actual dose at the target sites known. However, as many toxicity-modifying factors affect primarily toxicokinetics, any reduction in uncertainty, even imperfect separation, is vital to improving understanding of the toxicological process. The extent of knowledge about the relationship of exposure time and estimated absorbed dose (whole body, tissue, or target) also determines the level of sophistication of toxicity modeling that can be conducted.

Integr Environ Assess Manag 7, 2011—LS McCarty et al.

Toxicokinetics: body size and chemical characteristics. Body size can have substantial influence on the kinetics of accumulation. Generally, with other things being largely equivalent, it will take a larger organism longer to reach steady state than a smaller organism because of the greater surface area-tovolume ratio. In the freshwater aquatic environment, where size of organisms within and between species varies widely, size differences of 1000 or 10 000 may increase time to steady state of organic chemicals by factors on the order of 5 to 10 and 18 to 45 times, respectively, depending on the nature of the allometric relationship (Rand et al. 1995). Toxicokinetics also influence toxicity testing outcomes. For example, organic chemicals often exhibit a toxicity ‘‘cutoff’’ in the toxicityhydrophobicity relationship. In the case of 96 h acute tests with fathead minnows the cutoff begins at a log KOW of about 3.5 to 4.0 (Veith et al. 1983). The explanation is simple. Given the fixed duration of the test, the decreasing water solubility of more hydrophobic chemicals, and the relatively similar critical internal accumulation of chemical needed to cause mortality (largely by narcosis in this example), there is insufficient time for high log KOW chemicals to accumulate in the organisms to a level where toxic effects occur. The influence of body size on toxicokinetics also largely explains why the toxicity cutoff phenomenon can be seen at lower chemical hydrophobicity for larger organisms and higher chemical hydrophobicity for smaller organisms. This general size-scaling phenomenon can be observed in an examination of toxicity testing data for both organic and inorganic chemicals. (Hendriks et al. 2001; Hendriks and Heikens 2001). However, for metals that alter major-cation balance, such allometric influences are probably related to turnover rate of the cation by organisms (i.e., smaller organisms have a higher cation turnover rate than occurs in larger organisms). Grosell et al. (2002) present a good example for impairment of Na homeostasis by Ag and Cu. In aquatic organisms, ventilation rates vary considerably and can influence toxicokinetics for both organics and metals and thereby confound some allometric relationships. Chemical characteristics also affect the kinetics of accumulation. Although the bioconcentration of hydrophobic organic chemicals by aquatic organisms increases by several orders of magnitude as log KOW increases from about 1.5 to 6, this is largely due to lowered elimination. Hydrophilic organics (log KOW <
1.5) partition into the water phase of the organism rather than the lipid-like phase and therefore do not bioconcentrate. Bioconcentration of superhydrophobic organics (log KOW >
6) tends to decline from the peaks seen at about log KOW of 6. This decline can result from one or more factors such as insufficient exposure time to reach steady state, uptake limiting factors such as molecular size or shape (i.e., steric hindrance), bioavailability, and the growth dilution effect. The time to reach steady-state bioconcentration generally increases with log KOW because the time to reach steady state is inversely related to the elimination rate, which is usually inversely related to KOW (Gobas and Mackay 1987; Spacie et al. 1995). For chemicals with relatively slow kinetics, residues can integrate variable exposure in time. For example, field studies with variable exposures, tissue residues, as well as sediment/ soil concentrations or polymer tissue surrogates such as solidphase microextraction (SPME), may better reflect integrated exposure compared with a water concentration at a given point in time.

External Exposures Versus Tissue Residues—Integr Environ Assess Manag 7, 2011

This general trend may e confounded by some factors. Metabolic degradation of absorbed chemical will obviously affect net accumulation (see below). Also, the body size effect may be less important for some inorganics and/or metals where a metabolically active detoxification process is more common than the largely passive internal partitioning process found with many organics. Toxicokinetics: metabolism (biotransformation). Metabolism, including biotransformation and detoxification, are critical toxicokinetic processes that are more readily investigated with residue-based approaches. Metabolism is a modifying factor that directly affects the accumulation and resulting toxicity of chemicals that is often poorly quantified. As noted elsewhere, the TRA helps partition the actual potencies of metabolites from the potency of the parent compound. Because metabolites can be nontoxic, equitoxic, or more toxic than the parent compound (and may even cause toxicity by a different mode of action), a better understanding of metabolism will improve the interpretation of toxicity test data and the use of those data regulatory applications. Even when the actual metabolic profile is unknown, the TRA can identify the conditions where metabolites are contributing to the toxic response. For example, strong differences in the toxicity of fluoranthene were observed in exposure experiments with 3 invertebrates, H. azteca, Chironomus tentans, and Diporeia spp. (Schuler et al. 2004). By using a residue-based approach, the factors that contributed to species differences could be unraveled. One of the contributing factors included the relative ability to bioaccumulate the compounds as a function of lipid content of the organism. This was most pronounced with Diporeia, in which the lipid content was high and strongly affected the observed toxicokinetics. For the other 2 species, lipid content was similar, but the observed toxicity among species, as based on a total accumulated residue, differed markedly with C. tentans being more sensitive. Metabolic data from this experiment demonstrated that H. azteca produced more water-soluble phase II metabolites and some phase I polar metabolites, whereas C. tentans exhibited relatively low biotransformation capability; however, the metabolism resulted primarily in a phase I polar metabolite. When the toxicity was corrected for biotransformation by excluding water-soluble and residue-bound metabolites, it became clear that C. tentans remained the most sensitive. In addition, studies with pentachlorobenzene, which is not biotransformed, showed that all of the organisms were equally sensitive to a compound acting by baseline narcosis toxicity (Schuler et al. 2006). Thus, the TRA was able to demonstrate that the toxic response of C. tentans to fluoranthene was likely caused by a biotransformation product even though the exact metabolic product was not identified. Demonstrating the reasons behind the sensitivity of C. tentans and the importance of metabolites cannot be determined by media-based dose models because toxicity is confounded with toxicokinetics, which residue-based studies explicitly attempt to factor out, even to the point of addressing the role of biotransformation. Similarly, detoxification of metals confounds the relationship between total-body concentrations and toxicity. Detoxification capabilities vary widely among species and greatly influence the relative sensitivity of species when toxicity is based on total metal concentrations. Recognition of sensitivity from this perspective can complement direct toxicity test

15

information. When results from toxicity testing differ substantially or even appear to contradict observations in nature, mechanistic knowledge of detoxification (along with other toxicokinetic processes) in the different species can help explain such influences. For example, certain aquatic insect larvae that can be readily cultured and tested in the laboratory are not an especially sensitive group in traditional toxicity tests. However, in the field, the larvae of some species are the first to disappear in the presence of metal contamination (Winner et al. 1980; Clements 2000). Toxicokinetics and detoxification capabilities vary by orders of magnitude among aquatic insect larvae, explaining why some groups are so sensitive to metals while others (those most amenable to toxicity testing) are quite resilient. By considering detoxification and toxicokinetics the authors were able to explain the ranking of sensitivity for many field-toxicity observations determined for the Clark Fork River, Montana, USA (Cain et al. 2004; Buchwalter et al. 2007).

Effects Toxicodynamics. Even though time is a well-established, important toxicity modifier, it is usually difficult to separate toxicokinetics from toxicodynamics as the time course of body accumulation is not usually monitored in standard toxicity testing protocols, despite suggestions to change this (e.g., McCarty 1991). Residue-based approaches provide a clearer means for tracking the toxic response process, because they allow some separation of toxicokinetics from toxicodynamics, although in practice the separation is constrained by the nature of the understanding of the relationship between the operational organism-based dose metric, the true target sites dose, and the assumptions of the model being employed. Thus, advances can be made in several areas, and relative toxic potency comparisons and risk evaluations can be carried out with greater confidence when residue-informed toxicity data are used (Mackay et al. 2001). Toxicodynamics: causality. Although not usually completely diagnostic, information associating tissue and/or body residues with observed adverse effects is a vital first step in evaluating the nature and extent of effective toxicological exposure and possible causality. Incorporating time dependence and biotransformation in interpreting toxicity test data can give more insight on the response causality. Results of time varying body residues for mortality in fish (Chaisuksant et al. 1997; Yu et al. 1999) and invertebrates (Lee et al. 2002a, 2002b) can be used in more complex models. The Damage Assessment Model (Landrum et al. 2004, 2005; Schuler et al. 2004; Lee and Landrum 2006a, 2006b) is based on the assumed time required for the toxicant at the site of action to produce the observed response. Because toxicity is a function of concentration at the receptor site and time, higher receptor concentrations are required to produce a critical amount of damage in a shorter time frame. This model has been demonstrated for several compounds and species. Furthermore, biotransformation leads to variation in the body residue response. Recent models for both organics and metals that incorporate time dependence and biotransformation better describe the characteristics of exposure and response for pulsed exposures (Ashauer et al. 2006, 2007; Butcher et al. 2006; Meyer, Boese, Morris, et al. 2007).

16

Toxicodynamics: mode and/or mechanism of toxic action. True potency differences within and between modes of toxic action should be better reflected in residue-based dose metrics than in exposure-based dose metrics because the influence of many confounding toxicity-modifying factors can be estimated. It is axiomatic that various substances may have substantially different toxicities. This may be caused by differences in toxicokinetics, toxicodynamics, and/or mode and/or mechanism of action. However, once the effects of toxicokinetics are addressed, the remaining differences in toxicity are due primarily to differences in both the character and magnitude of adverse effects. Differences in adverse effects are commonly ascribed to differences in modes or mechanisms of toxic action (McCarty and Borgert 2006b). The definition of mechanism indicates a specific and precise process while the more general definition of mode could include a group of related mechanisms (for definitions, see Meador et al. 2011). For example, several mechanisms that block a biochemical pathway at different points, thereby causing the same adverse effect, could be classed as a common mode of action. Although it is clear that the primary interest of toxicology should be focused on mechanisms of action, the lack of data on the effective dose has necessitated the more general modeof-action approach, which is based on other available information. It can be seen that for organic chemicals there is a reasonable relationship between tissue residues and modes of toxic action (McCarty and Mackay 1993; Barron et al. 2002; Traas et al. 2004; Escher et al. 2011). Current thinking is that this relationship between tissue residues and modes of toxic action is the result of a similar underlying relationship between mechanisms of toxic action and chemical residues as the sites of toxic action in the organism. Similar mode– residue relationships may exist for some metals, but, as noted elsewhere, fewer data are available, and the relationships are more complex. Such relationships may not always be distinct because substances may have multiple modes of action, which may or may not be dose and/or time dependent. For example, lowdose DDE is a reproductive toxin, whereas high-dose DDE is a nonpolar narcotic (Hwang et al. 2004). Furthermore, the simple additivity found in many complex mixtures of organic chemicals may be explained by mode-of-action dose dependency. Below the range of about 0.3 to 0.02 of the threshold LC50, organic chemicals with specific modes of action tend not to express their specific mode-of-action toxicity, at least in shorter term exposures Additionally, because they are organic chemicals, they may contribute to nonspecific toxicity (McCarty and Mackay 1993). For some mechanisms of action, such as reactive compounds, there may be multiple sites of toxic action (Hendricks et al. 2005). Clearly, tissue and/or body residue dose metrics can provide a sound approach for advancing our understanding of the toxic action of chemicals. An example is the case of polycyclic aromatic hydrocarbons (PAHs). For PAHs, multiple modes of toxicity complicate the use of body and/or tissue residues, especially in relation to fish species with enzymatic pathways that quickly metabolize the parent compound. PAHs exert toxicity through multiple mechanisms, including nonspecific, reversible nonpolar narcosis, and through enzyme-mediated pathways that lead to developmental abnormalities, and mutagenesis/carcinogenesis. The former is generally linked to mortality as an endpoint (van Wezel and Opperhuizen 1995), whereas the latter results from enzy-

Integr Environ Assess Manag 7, 2011—LS McCarty et al.

matic biotransformation of parent PAHs into metabolites that initiate a cascade of effects such as developmental and reproductive impacts, impaired growth, formation of DNAadducts, and cancers (Johnson et al. 2002; JN Meyer et al. 2002; Meyers et al. 2003; Van der Oost et al. 2003; Incardona et al. 2004; Meador et al. 2006). Narcosis and enzyme-mediated effects associated with PAHs are often assessed separately, hindering assessment of cumulative costs incurred to fish in chronically PAHcontaminated environments. At sublethal exposure concentrations, narcosis and enzyme-mediated mechanisms interact and contribute to metabolic costs. The time scale and dose determine whether one mechanism dominates over another. With longer time scales, the costs likely related to multiple mechanisms cannot be easily partitioned among multiple stresses. Therefore, an internal dose surrogate provides a clearer picture of the potential mechanisms involved. Toxicodynamics: acute-to-chronic relationships. The relationship between acute and chronic responses and the various endpoint types is not readily understood from media-based exposures. Part of the confusion is due to the common use of the terms acute and chronic to describe both exposure duration and the nature of the adverse effects. As discussed in Rand et al. (1995), this practice should be avoided by using shortterm and long-term for exposure and lethal and nonlethal or sublethal for effects. By allowing toxicokinetics and toxicodynamics to be estimated separately, residue-based approaches can enable better estimates of the toxicity tolerance curves and allow insights about acute-to-chronic relationships, in terms of both duration and effects, to be obtained. As discussed in McCarty and Mackay (1993), the relationship between acute and chronic toxicity within a given mode of action can be related to the accumulated residue. As residues increase, the percentage mortality increases. The chronic response typically measured in standard toxicity tests corresponds to the lower tail of the acutely lethal distribution, essentially where little or no mortality occurs. In this case, acute and chronic mortality results for baseline toxicant are often about an order of magnitude different. Obviously, much greater ranges of acute to chronic effects have been reported, but this can be explained if the measured data are for chemicals that have more than one mode of action in addition to baseline narcosis. This is the case for many organic chemicals that exhibit specific toxicity at a 1dose range, yet can cause baseline narcosis under appropriate conditions, e.g., DDE (Hwang et al. 2004). Tissue residue information can be helpful in understanding otherwise problematic situations exemplified by the wide range of acute-chronic ratios (Kenaga 1982). Thus, the use of tissue residues allows development of dose-response spectra; that is, the range of residues associated with a range of responses graded by severity (e.g., Hwang et al. 2001, 2003, 2004; Schuler et al. 2007a, 2007b). Mixtures. A general theoretical basis for the modes/mechanisms of interactions in mixture toxicity is lacking (McCarty and Borgert 2006a, 2006b) in part due to a poor understanding of the relationship between exposure-based and tissue-based dose metrics that confounds test data interpretation. Without the clearer understanding of effective organism-based doses offered by the residue-based approach, interpreting, explaining, and predicting mixture toxicity is

External Exposures Versus Tissue Residues—Integr Environ Assess Manag 7, 2011

17

problematic. Although a number of schemes for mechanisms of toxic action have been developed (see Escher et al. 2011), a comprehensive, generally accepted and broadly applicable classification scheme for mechanisms of toxic action is still needed to ensure a foundation for mixture-related regulatory activity and for iterative hypothesis development and experimental testing. Nevertheless, mixtures are the norm rather than the exception in environmental studies. The residue based approach enables a better understanding of the contribution of mixture components (Dyer et al. 2011). Within a mode of toxic action such a residue-based approach can provide considerable insight and enhancement in development of regulatory guidance (Steevens et al. 2005; Dyer et al. 2011; Sappington et al. 2011). With exposure-based studies, the interaction of compounds to produce a toxic response is complicated by the toxicokinetics, which dictate the amount and distribution of the compound with organisms. As described above, the explicit incorporation of metabolite potencies for DDT allowed clear interpretation of sediment exposures for H. azteca (Lotufo et al. 2001) that would not have been clear from media-based studies. This included the relative potencies of the metabolites to the organism. The residue approach has also allowed confirmation of the current PAH additivity model (Di Toro et al. 2000). The PAH additivity model assumed that toxicity occurred from additive concentration of compounds on an equimolar basis. Measures of PAH residue effects data in Diporia, which are not known to biotransform these compounds, clearly showed that exposure-based toxicity measures ranged at least an order of magnitude with aqueous exposures. In contrast, residue-based toxicity was less than a factor of 2 across the same range of aqueous exposure concentrations (Landrum et al. 2003) among compounds and were nearly identical with the toxicity for a mixture of 26 PAHs to Diporeia exposed to sediment (Landrum et al. 1991). Thus, the TRA could be applied to describe the observed toxicity of mixtures for compound classes that act by the same, or potentially different, modes and mechanisms of toxic action, without regard for route of exposure.

1.5–6) exhibited a range of more than 100 000 times, whereas the body residue estimates were within a factor of 10 (McCarty 1986; McCarty et al. 1992). Clearly, toxicokinetics dominated the expression of toxicity in this case. Demonstrating the reasons behind the sensitivity within and between different species to both metals and organic contaminants, and the importance of biotransformation/ biodegradation and metabolites, is difficult when only media-based dose data are available. A large part of the variance in effects in exposure-based toxicity measures, such as EC/LC50, or ED/LD50, is explained by variance in bioavailability and toxicokinetics. Thus, as noted above, there are major challenges to thoroughly evaluating toxicological risk with only exposure-based data. Much of the remaining variance is related to mode/mechanism of toxic action and toxicodynamics. Table 1 illustrates some major categories of variability in different types of toxicity measurements. Reduction in sources of variability as measurement moves closer to target site(s) of toxicity is evident. Toxicodynamics in the whole organism, and at the target site, explain a possible time-dependency of critical tissue residues. Examples are second-order partitioning over different tissues within the organism, biotransformation, and chemical and/or target site interaction. Toxicodynamics may lead to a different mechanism of toxic action, e.g., after formation of more toxic metabolites or after reaching a critical accumulation or binding at a target site, which can be explored with long-term tissue residue experiments using tissue dose metrics. The variability in toxic responses cannot be completely explained by the above factors. Because the influence of various organism-related factors such as sex, age, and condition will also contribute to differences in sensitivity to any given target-site dose, residue-based information can facilitate partitioning the contributions to variability into source categories. Consequently, the contributions to variability are likely to vary both within and between source categories in difference circumstances, therefore such information should be of great utility in improving understanding of toxicity testing results and refining applications where such knowledge is used.

Variability and uncertainty. The use of tissue residue information can help reduce uncertainty and variability in toxicity data by aiding the estimation of the influence of various toxicitymodifying factors as well as helping to explain the contribution of toxicokinetics and toxicodynamics. For example, the exposure-based acute toxicity data (96-h LC50) for small fish exposed to a variety of narcotic organic chemicals (log KOW


APPLICATIONS Laboratory-to-field linkages The tissue residue-effects approach provides an improved method for linking laboratory measurements of exposure and effects to the field, and vice versa. In the field, tissue residues may be used in forensic investigations into both specific acute

Table 1. Some sources of variance within toxicity metrics Exposure-based

Whole-body and/or tissue-based

Target tissue-based

Environmental bioavailability

Toxicological bioavailability

—

Toxicokinetics

Toxicodynamics in organism

—

Toxicodynamics in organism

—

Proportion of target tissue

Proportion of target tissue

Toxicodynamics at target site

Toxicodynamics at target site

Mechanism of toxic action

Mechanism of toxic action

— Toxicodynamics at target site Mechanism of toxic action

18

(short-term) episodes and longer-term effects. For short-term episodes (e.g., spills) it may not always be sufficient to allow unequivocal attribution of causality of death the field, but it may be able to exonerate some suspected toxic agents. By directly examining residues and community change the relationship between tissue residues and adverse effects in the field may be tested without many of the ambiguities associated with interpreting single-species toxicity tests. Questions about causation for field correlations may be further reduced if the organisms affected in a contaminated environment can also shown to be sensitive by residue-based toxicokinetics and toxicodynamics (e.g., Cain et al. 2004; Buchwalter et al. 2007). In the field, animals that die decay rapidly and may not be readily available for measurement of residues. Thus, residue monitoring will usually be conducted on live animals. Measures of concentrations in live and dead organisms provide equivalent estimates of the toxic response on a residue basis (Landrum et al. 2004) as long as the organisms are near steady state. Otherwise, organisms that do not die will continue to accumulate residue above the concentrations that can result in the death of sensitive organisms (Landrum et al. 2005). Also, when monitoring in the field, sensitive species may be missing; therefore, no tissue will be available for comparison with residue benchmarks. This may be overcome by measuring residues in less sensitive species, as long as the relationships between their toxicokinetic and toxicodynamic potentials are known. Certainly issues of interspecies biotransformation would need to be considered as well. This issue is minimized when considering sublethal effects, because portions of the population will be available for residue measurement. Changes in community structure in response to contamination usually reflect the disappearance or reduced abundances of species. Understanding bioaccumulation and detoxification in different species can translate into forecasts for how communities might change. Tissue residues associated with effects in growth and reproductive parameters can be used as inputs into population modeling (Spromberg and Meador 2005) and ecological risk assessments (Sappington et al. 2011). These predictions can be tested against community data with less ambiguity than in the absence of such supporting knowledge. Ecologically important tissue residues can be validated in the field with field-collected residues. This information can then be used as the basis for environmentally realistic laboratory experiments to study the consequences of toxic tissue residues on growth and reproductive parameters. Several examples (below) illustrate the advantages. Munns et al. (1997) developed a stage-based population model to evaluate changes in population growth based on dioxin- and PCB-induced effects on individual reproduction rates and survivorship probabilities measured in the laboratory and in the field for Fundulus heteroclitus. This required using field-measured tissue residues to set the parameters for the dosing strategy in the laboratory. For Fundulus, the dominant route of exposure of hydrophobic organics is through the diet (Black et al. 1998, Gutjahr-Gobell et al. 1999; Matta et al. 2001). Through dietary exposures reflective of field-based dietary exposures, fish accumulated PCBs (whole-body and liver residues) to the point of observable sublethal responses. The laboratory measured residues-effects parameters were incorporated into a population model to evaluate contaminant impacts on population growth rates relative to natural rates, and demonstrated a

Integr Environ Assess Manag 7, 2011—LS McCarty et al.

declining rate. Spromberg and Meador (2005) also used this approach to evaluate effects of sublethal exposures on ageclass distributions and other population-level characteristics for Chinook salmon (Oncorhynchus tshawytscha). PAHs and PCBs present classic examples of chemicals that are effective at exerting baseline toxicity through the narcotic mode of action (Van Wezel et al. 1995; Di Toro et al. 2000). However, these chemicals have multiple modes of action, including concentration- and time-dependent-specific toxicities. For example, a well-studied mechanism of PAH metabolism is the induction of the cytochrome P-450 1A system. This is a precursor process for carcinogenic or other developmental modes of action associated with PAH exposures in fish species (Meador et al. 1995; Johnson et al. 2002; Incardona et al. 2004). Thus, the toxicity of these secondary modes of action may be altered by certain metabolic activities in fish. As another example, Meador and colleagues exposed juvenile Chinook salmon to a dietary mixture of PAHs that related to PAH compositions and concentrations of field prey items (Meador et al. 2006; Meador, Buzitis, Bravo et al. 2008a). The external dose was administered as micrograms (mg) total PAH/g-fish/day via the diet. However, the internal effects concentration was expressed in terms of phenanthrene metabolites (biliary fluorescent aromatic compounds [FACs]) that were correlated with changes to growth rates. Thus, the internal ‘‘tissue residue’’ was linked to an ecological translatable parameter. This type of technique can be applied in the field where biliary FACs can be measured, correlated to a change in growth, and then assessed on whether the chemically induced change in growth on the individual level can impact the population growth. Additionally, Merten (2005) used slowly metabolized compounds (PCBs) as a robust surrogate measure of tissue residue for PAH exposure and then linked dietary PAH exposure to changes in standard metabolic rates in F. heteroclitus. PAHs and PCBs of similar log KOW values are assimilated into tissue via similar routes of uptake. Sites that pose chemical risk to fish often are dominated by PAHs, but also have PCB concentrations at background concentrations (e.g., the creosote-contaminated site in the Elizabeth River, VA). In this study, the standard metabolic rate was used to assess PAH exposure while PCB accumulation was used as a tracer for change in bioaccumulation due to PAH exposure. Thus, a TRA was linked to an energetic parameter that could be translated into a cost to the organism in terms of lost energy for growth and reproduction. This approach of using tissue residues for one class of compounds to help understand exposure and toxicity for other classes (e.g., PCBs and PAHs) can be very informative (Van Veld et al. 1992; Meador et al. 1997; van der Oost et al. 2003). The laboratory-to-field linkage of population-level effects and tissue residues is not limited to organics. Metals/inorganic concentrations are readily measurable in organisms from nature as analytical protocols are well developed (Phillips and Rainbow 1994). Thus, metals and/or inorganics can also be investigated with tissue residue–effect relationships. For example, Van Kirk and Hill (2007) used a population model to examine alterations to population demographics in cutthroat trout in Idaho and Wyoming as a function of selenium tissue residues and effects (growth and fecundity). Validation of a gradient approach to lethal and sublethal effects of Cu in the field can be found in the work on bivalves

External Exposures Versus Tissue Residues—Integr Environ Assess Manag 7, 2011

(Grout and Levings 2001; Hornberger et al. 2000; Salazar and Salazar 1995, 2007). Some of the first reliable estimates for critical concentrations of TBT in water and tissues were determined using caged mussels in a series of transplant experiments (Salazar and Salazar 1991, 1996). It was based on this body of work that the exposure–dose–response triad was developed (Salazar and Salazar 1998). Comparisons were also made showing the similarity in critical body residues of Cu for both marine and freshwater bivalves. This led to the refinement of a conceptual framework for linking bioaccumulation and biological effects to chemicals in water and sediment (Salazar and Salazar 2007). The utility of bivalve mollusks (freshwater and marine) in both applying and advancing the TRA must be acknowledged. They facilitate the simultaneous collection of tissue residue and biological effects data in the same organism at the same time (ASTM 2001). This approach of characterizing exposure and effects is consistent with both environmental risk assessments (ERA) and TRA paradigms. Bivalves can be more useful than other animal groups because they combine some advantages for characterizing the internal chemical dose and associated biological effects. They concentrate and integrate chemicals in their tissues and exhibit a limited ability to metabolize many chemicals compared with other species. Bivalves also provide tissue residue data that can be used to estimate chemical exposure from water or sediment. Finally, they facilitate ‘‘controlled experiments’’ in the field with large sample sizes because they are easy to collect, cage, and measure. For these reasons, bivalves have been used in residue-based ERA more than any other animal group. Not surprisingly, bivalve mollusks are the only group with a consensus-based standardized protocol for conducting insitu field bioassays (ASTM 2001). The experimental control afforded by this approach can be used to place a large number of animals of a known size distribution, exposure history, and genetic background in specific areas of concern to quantify exposure and effects over space and time with a clearly defined exposure period. Chemical exposure can be estimated with survival, growth, and other sublethal endpoints (Salazar and Salazar 1996). Bivalves caged in, on, or above bottom sediment have been used from depths of