Environ. Sci. Technol. 2011, 45, 197–202
Bioaccumulation of Organic Contaminants in Humans: A Multimedia Perspective and the Importance of Biotransformation† M I C H A E L S . M C L A C H L A N , * ,‡ GERTJE CZUB,‡ MATTHEW MACLEOD,§ AND JON A. ARNOT| Department of Applied Environmental Science (ITM), Stockholm University, SE-106 91 Stockholm, Sweden, Swiss Federal Institute of Technology, ETH Zurich, Wolfgang Pauli-Strasse 10, Zurich, Switzerland CH-8093, and Department of Physical & Environmental Sciences, University of Toronto Scarborough, 1265 Military Trail, Toronto, Ontario, Canada, M1C 1A4
Received March 29, 2010. Revised manuscript received July 16, 2010. Accepted July 16, 2010.
Bioaccumulation is an important component of the exposure hazard assessment and risk assessment of organic chemicals. Screening criteria for chemical hazard used in national and international regulations are based on the paradigm that partitioning properties are the primary chemical determinants of bioaccumulation. We use a holistic multimedia perspective to evaluate the partitioning property paradigm with respect to assessing human bioaccumulation. Multimedia bioaccumulation factors (mmBAFs) for humans were modeled for hypothetical chemicals with a wide range of physical-chemical properties. Varying partitioning properties over 12 orders of magnitude (a plausible range for nonionizing organics) resulted in only modest changes in mmBAFs (a factor of ∼10) for all but very volatile or hydrophilic chemicals. In contrast, varying biotransformation rate constants over 6 orders of magnitude resulted in substantial differences in mmBAFs (greater than a factor of 109). Our model results are supported by empirical observations of well characterized pollutants, which demonstrate that chemicals with similar partitioning properties can have very different bioaccumulation behavior. Susceptibility to biotransformation clearly determines bioaccumulation in humans for many chemicals. We conclude that a holistic multimedia perspective for bioaccumulation assessment is necessary to develop regulations, criteria, and policies that are protective of human health and the environment.
Introduction In the 1960s toxic effects observed in upper trophic level fauna, particularly piscivorous birds, were linked to high concentrations of certain organic chemicals (1). This raised the question of what chemical properties favor uptake and retention of organic contaminants from the environment by † Part of the special section “Environmental Policy: Past, Present, and Future”. * Corresponding author e-mail:
[email protected]. ‡ Stockholm University. § Swiss Federal Institute of Technology, ETH Zurich. | University of Toronto Scarborough.
10.1021/es101000w
2011 American Chemical Society
Published on Web 08/11/2010
organisms, a process that was dubbed bioaccumulation. Initial research focused on aquatic environments primarily using fish as a model, which led to the development of aquatic-based bioaccumulation end points such as the bioconcentration factor (BCF) and the bioaccumulation factor (BAF). The BCF and BAF are defined as the ratio of the chemical concentration in the organism’s body (mol kg-1) and the concentration of freely dissolved chemical in water (mol L-1). The BAF includes dietary and aqueous exposure, whereas the BCF includes only aqueous exposure. Strong correlations between the BCF and BAF and the octanol:water partition coefficient (KOW) were found in systems where chemical concentrations in fish and in water were near thermodynamic equilibrium, and thus, the paradigm that bioaccumulation is governed by a chemical’s partitioning properties became an accepted principle of environmental chemistry (2). This occurred despite a recognition that relationships between KOW and bioaccumulation metrics would not hold for chemicals that are biotransformed to a significant extent within the bodies of fish (3). Ostensibly, observations of bioaccumulation of persistent substances and the relative availability of bioaccumulation data for such substances in fish compared to other bioaccumulation data contributed to the development of current regulatory criteria. This is evident in the United Nations Stockholm Convention on Persistent Organic Pollutants (POPs) and regulatory programs in Europe, the United States, and Canada (4-6), which all use aquatic-based screening criteria (KOW, BCF, BAF) for assessing bioaccumulation (7). Conceptually, the use of aquatic-based bioaccumulation end points that relate concentrations in octanol or organisms to concentrations in water is only appropriate for assessing bioaccumulation in aquatic organism that respire water. For plants and animals in the terrestrial environment, air is a more appropriate reference exposure medium since it is the usual source of contaminant entry at the base of the terrestrial food web and a potential pathway of direct exposure and elimination via respiration. The octanol:air partition coefficient (KOA) is strongly correlated with the partition coefficient between air and leaves (or foliage) for many organic chemicals (8, 9). Biomagnification of organochlorines in terrestrial food webs has been shown to correlate better with KOA than KOW (10). Thus, recognizing that aquatic-based bioaccumulation assessment end points are not adequate for all organisms, alternative criteria that include KOA as well as KOW (10-13) and other bioaccumulation end points such as the biomagnification factor (BMF) and the trophic magnification factor (TMF) have been proposed (14). Recently, a multimedia approach that considers the whole environment was proposed to avoid the “reference phase dilemma”, i.e., whether air or water should be selected as the reference phase for assessing bioaccumulation in organisms, like humans, that consume food originating in both the aquatic and terrestrial environment (12). The environmental bioaccumulation potential (EBAP) was defined as the ratio of the body burden of a perfectly persistent substance in an organism (mol organism-1) to the quantity of the chemical present in the environment (including air, water, soil, and sediment) normalized to the land area (mol m-2) (12). The EBAP has units of m2 organism-1 and can be viewed as the land area or spatial extent of the environment containing the same amount of chemical as the organism. This can be compared with the BAF units of L kg-1, which is the volume of water containing the same amount of (freely dissolved) chemical as 1 kg of the organism. A high EBAP implies that VOL. 45, NO. 1, 2011 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
197
the organism has concentrated chemical from a large area of the environment. By normalizing the levels in biota to levels in the whole environment, the multimedia approach provides a different perspective on bioaccumulation. Although less well suited for understanding the underlying mechanisms of bioaccumulation than the single media approach, this perspective is more relevant for chemical regulation and screening, as it allows one to directly address the fundamental question at the root of bioaccumulation: What fraction of the chemical in the environment is accumulating in the organism of interest? The multimedia bioaccumulation approach was used to explore the sensitivity of bioaccumulation of perfectly persistent chemicals in a northern European woman to phase partitioning properties (12). One finding from this work was that, for a large range of partitioning properties of organic chemicals, the EBAP is nearly constant indicating that partitioning properties are not the primary determinant of a chemical’s bioaccumulation potential for humans. The present analysis further explores the holistic multimedia approach to identify key parameters responsible for the bioaccumulation of organic chemicals in humans. We critically examine the dominant paradigm that bioaccumulation is primarily determined by chemical partitioning properties. To this end, we test an alternative hypothesis: That biotransformation in the organism and its food web is a more significant determinant of bioaccumulation than partitioning properties. We do this using the multimedia bioaccumulation factor (mmBAF; m2 organism-1), the quotient of the body burden of a chemical in an organism to the quantity of the chemical present in 1 m2 of the multimedia environment. Although we use body burden as the measure of levels in the organism for this study, other metrics such as the chemical concentration can be used (which would give the mmBAF units of m2 kg-1). The mmBAF is equivalent to the EBAP but more broadly defined to include nonpersistent chemicals. We favor the use of the mmBAF over EBAP because it clearly communicates the multimedia nature of this bioaccumulation metric, and we use it for both persistent and nonpersistent chemicals in the following. We calculate the mmBAF for a 30 year old female human using evaluative fate and bioaccumulation mass balance models for hypothetical chemicals with a range of partitioning properties and biotransformation rates that encompass plausible properties of chemical pollutants and corroborate our model results with empirical observations from studies of bioaccumulation in aquatic and terrestrial systems. Our findings are explained using the model, and the scientific and regulatory implications are discussed.
Methods The methodological procedure was analogous to that used to study the sensitivity of the EBAP to partitioning properties (12). A Level I Unit World model of chemical fate in the physical environment (15) was linked to the bioaccumulation model ACC-HUMAN that describes chemical transfer through the aquatic and agricultural food chains to humans (16). In numerical experiments, chemicals with a fixed set of partitioning properties but varying biotransformation rate constants were discharged into the unit world, and the degree to which they were transferred to humans was compared. The chemical applicability domain of the models consists of nonionic organic chemicals subject mainly to nonspecific sorption in the environment and biota. Physical Model. The fugacity-based Level I Unit World model consisted of the four compartments air, water, soil, and sediment (15) and employed the parametrization recommended in ref 17. One tenth of the Unit World was covered by water and the remainder by soil. The depths of 198
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 45, NO. 1, 2011
the air, water, soil, and sediment compartments were 1000 m, 20 m, 10 cm, and 1 cm, respectively. The organic carbon contents of soil and sediment were set to 2 and 4%. Phase partitioning was modeled according to ref 18, and partitioning equilibrium in the physical environment was assumed. Bioaccumulation Model. ACC-HUMAN is a nonsteady state, mechanistic, fugacity-based model (16). It focuses on dietary sources of human exposure to environmental pollutants via fish, dairy products, and beef. It additionally considers uptake from air and drinking water. The model is subdivided into an agricultural food chain represented by grass, milk cows, and beef cattle and an aquatic food chain represented by zooplankton, planktivorous fish, and piscivorous fish. The human is linked to both systems as the top predator. Contaminants enter the food chains via uptake from air, water, and soil and are transferred to higher trophic levels via predator-prey interactions. Growth and the temporal variation of physiological and environmental parameters such as the ingestion rate and temperature are taken into account. The model was parametrized according to environmental and agricultural conditions in southern Sweden as described in ref 16. The dominant commercial fish species in the Baltic Sea, herring and cod, were chosen as model organisms for planktivorous and piscivorous fish, respectively. In a model validation exercise using this parametrization, good agreement was found between predicted and measured concentrations of polychlorinated biphenyls for cows’ milk, beef, herring, cod, and human tissue in southern Sweden. The dietary ingestion was based on German dietary habits as described in ref 12. Model Experiments. To explore the influence of the biotransformation rate on bioaccumulation in humans, sets of hypothetical chemicals with constant partitioning properties and varying combinations of first order rate constants for whole body elimination via biotransformation kM(fish), applied to both fish species, and kM(mammal), applied to all mammals (beef cattle, milk cows, and humans) were defined, whereby each rate constant was varied between 1 h-1 (a readily biotransformed substance) to 3 × 10-7 h-1 (a substance very resistant to biotransformation). A fixed amount of each hypothetical chemical was distributed in the Level I Unit World. These simulations were repeated for three discrete sets of chemicals with constant partitioning properties: set A) log KOW of 7 and a log KOA of 8.5 (the partitioning properties that gave the maximum mmBAF in the assessment of the partitioning property sensitivity, Figure 1); set B) log KOW of 3.5 and a log KOA of 8.5 (a property combination that lies toward the hydrophilic edge of the partitioning space found to have high mmBAF values); and set C) log KOW of 7 and a log KOA of 6 (a property combination at the volatile edge of this space). In all 3 cases the heats of air-water, octanol-air, and octanol-water phase transfer were set to 60, -80, and -20 kJ mol-1, respectively, in accordance with the modeling done to produce Figure 1 (12). To ensure that the bioaccumulation model reached steady state before the simulation of the human started, the model was run for 40 years before the human was born. The lifetime accumulation of the chemicals in a girl/woman was then simulated. The woman’s body burden during nursing of her first child at 30 years of age was used as the basis of comparison for the bioaccumulation behavior.
Results The variability of mmBAF of hypothetical perfectly persistent chemicals for a woman as a function of partitioning properties is plotted on a logarithmic scale in Figure 1. This figure is adapted from ref 12 and shows that the modeled mmBAF varies only between 10 and 120 for a large range of partitioning properties (3 < log KOW < 11 and 6 < log KOA < 12). As shown
FIGURE 1. The multimedia bioaccumulation factor mmBAF (in m2 person-1) for perfectly persistent chemicals plotted as a function of the partition coefficients KOW and KOA. The logarithmic scale of mmBAF is marked on the lines separating the colored fields. This figure is adapted from results presented in ref 12. The letters A, B, and C indicate the partitioning properties of the sets of hypothetical chemicals shown in Figure 2. in ref 12, many chemicals of environmental concern fall within this partitioning space of low sensitivity. A second modeling study of bioaccumulation through the marine food web to an Inuit woman showed a similar lack of sensitivity to partitioning properties in a large portion of the partitioning space (19). Even though a Level I (equilibrium, steady state) environmental fate model was used in the first study and a Level IV (nonequilibrium, nonsteady state) in the second, both identified a similar region of the partitioning space as being insensitive. The letters A, B, and C in Figure 1 indicate the partitioning properties of the three discrete sets of hypothetical chemicals that we selected to illustrate the variability of mmBAF as a function of biotransformation in Figure 2. The variability of mmBAF of set A, hypothetical chemicals with a log KOW of 7 and a log KOA of 8.5, as a function of the biotransformation rate constant in mammals, kM(mammal), and fish, kM(fish), is plotted on a logarithmic scale in Figure 2A. This combination of partitioning properties is similar to the highly bioaccumulative PCB 153 and is identified in Figure 1 as leading to the maximum mmBAF for perfectly persistent chemicals. The modeled mmBAF of the set of chemicals in Figure 2A ranges over 9 orders of magnitude, from a value
of 120 m2 person-1 for a hypothetical chemical that is very resistant to biotransformation, to 3.7 × 10-7 m2 person-1 for a chemical that is readily biotransformed in both fish and mammals. Overall, biotransformation in mammals including humans has a stronger influence on the mmBAF than kM(fish); for a given kM(fish), kM(mammal) causes mmBAF to vary over at least 5 orders of magnitude, while for a given kM(mammal), kM(fish) causes mmBAF to vary less than 2 orders of magnitude except for the highest values of kM(mammal) (Figure 2A). Figure 2B,C presents corresponding modeled mmBAFs for a more hydrophilic set of chemicals, set B, with partitioning properties similar to the herbicide atrazine (log KOW of 3.5 and a log KOA of 8.5), and a more volatile set of chemicals, set C, with partitioning properties similar to cyclic siloxanes that are used in personal care products and as a raw material in the production of silicone polymers (log KOW of 7 and a log KOA of 6). Like for set A, the modeled mmBAF for chemicals in set B and set C ranges widely (over 6-7 orders of magnitude) as a function of kM(mammal) and kM(fish). Biotransformation in mammals has a stronger influence on the mmBAF than biotransformation in fish for set C, similar to chemicals in set A. The variability in the mmBAF of set B is almost entirely attributable to kM(mammal). Relative Importance of Biotransformation and Partitioning. Our hypothesis that biotransformation is a more significant determinant of human bioaccumulation than partitioning properties can be evaluated by comparing the sensitivity of mmBAF to biotransformation rate constants with the sensitivity to partitioning properties. As seen in Figure 2, a realistic range of biotransformation rate constants (from 1 h-1 to recalcitrant) causes the mmBAF of chemicals with fixed partitioning properties to vary by between 6 and 9 orders of magnitude. On the other hand, varying the partitioning properties of a set of chemicals with fixed biotransformation rate constants by 8 orders of magnitude of KOW (103-1011) and 6 orders of magnitude of KOA (106-1012) results in variability of mmBAF of just 1 order of magnitude (Figure 1). It is thus clear that for a large part of the chemical partitioning space the biotransformation rates are more important determinants of bioaccumulation to humans than partitioning properties. Figure 2 indicates that biotransformation in mammals is a more important factor determining mmBAF than biotransformation in fish. This is because biotransformation in mammals includes biotransformation in the organism of interest, humans. Biotransformation in fish will only influence the fraction of total dietary exposure that comes from fish, whereas kM(mammal) directly influences the chemical’s
FIGURE 2. The multimedia bioaccumulation factor mmBAF (in m2 person-1) plotted as a function of biotransformation rate constants kM (h-1) in mammals and fish for sets of hypothetical chemicals with A) log KOW of 7 and a log KOA of 8.5, B) log KOW of 3.5 and a log KOA of 8.5, and C) log KOW of 7 and a log KOA of 6. The logarithmic scale of mmBAF is marked on the lines separating the colored fields. VOL. 45, NO. 1, 2011 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
199
FIGURE 3. GC/ECD chromatograms of several PCB congeners from the analysis of extracts of sediment and gray seal blubber from the Baltic Sea. The y-axis scales were adjusted to give the same intensity for the PCB 99 signal in both matrices. elimination in humans as well as dietary exposure from agricultural food sources. For the low KOW set of chemicals (Figure 2B), the sensitivity to kM(fish) is less pronounced than for the high KOW sets A and C. This is because the modeled human exposure to chemicals in set B via consumption of fish is low when compared to exposure via the agricultural food chain. For all chemicals, the fraction of dietary exposure contributed by fish increases when the biotransformation rate in mammalian food sources (most notably, cows) increases, which is reflected in the stronger influence of kM(fish) in the right-hand portion of Figure 2A-C. The importance of biotransformation indicated by our model analysis is clearly confirmed by chromatograms from the environmental analysis of complex mixtures of organic contaminants. For example, in Figure 3 chromatograms of two closely eluting PCB congeners are shown. The first eluting peak is PCB 101 and the second is PCB 99. These two PCB congeners have very similar partitioning properties (20). The red chromatogram shows the PCB pattern in surface sediment from the Baltic Sea, while the blue chromatogram shows the pattern in a gray seal from the same region. In the sediment, which contains the bulk of the PCB inventory in the Baltic Sea, the PCB 101 peak is about 7 times larger than the PCB 99 peak. However, in the seal the pattern is reversed; the PCB 101 peak is only one-sixth of the PCB 99 peak. Since the partitioning properties of these two chemicals are almost identical, it follows that the >40 fold decrease in the concentration of PCB 101 relative to PCB 99 must be due to biotransformation in the food web between the sediments and the seal. PCB 101 has indeed been shown to be more readily biotransformed than PCB 99 in seals and other marine mammals (21-23) and fish (24). Similarly, PCB 101 has been shown to be readily biotransformed in cows (25) and humans (26), while PCB 99 is resistant to biotransformation in these organisms as well. The data and calculations in Table 1 provide an empirical example of the small influence of partitioning properties on the multimedia bioaccumulation of hydrophobic chemicals that are not readily biotransformed. Here the mmBAF of two such chemicals, β-HCH and PCB 180, are calculated from measured concentrations in human milk and the major environmental reservoir compartments in southern Sweden (soil and Baltic Sea water). Although the KOW of the chemicals differs by a factor of 1800 and the KOA by a factor of 25, the mmBAFs of the two compounds are nearly the same. They are also similar to the values predicted for chemicals with these partitioning properties in Figure 1, which lends credibility to the predictive power of the models. Note that the single media BAF (human with respect to water, BAFwater in Table 1) is more than 4 orders of magnitude greater for PCB180 than for β-HCH. This parallels the difference in KOW 200
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 45, NO. 1, 2011
TABLE 1. Multimedia Bioaccumulation Factors (mmBAF) for β-HCH and PCB 180 in Southern Swedend units log KOW log KOA C(woman) inventory(woman) C(water) C(soil) inventory(env) log BAFwater mmBAF
ng g-1 lipid µg person-1 ng L-1 ng g-1 OC µg m-2 L kg-1 lipid m2 person-1
β-HCH
ref
PCB 180
ref
3.91 8.74 11 220 0.70 NAb 4.8 4.20 46
33 33 34
7.16 10.16 41 820 0.00012 8.84 16 8.53 50
20 20 34
35a
36 37c
a Mean of the mean concentrations in Baltic Sea surface water for the years 1996, 1997, 1998, and 1999. b No data available. Total inventory estimated from the inventory in water according to the distribution between water and organic carbon predicted for this chemical by the Level I unit world model employed in ref 12. c Median organic matter normalized concentration from this reference, converted to an organic carbon normalized concentration using the factor 1.7 employed in the same reference. d The data presented include partitioning properties as well as concentrations and inventories in humans and major environmental reservoirs (water for β-HCH and soil for PCB 180). The inventories were calculated from the concentrations using the human lipid mass of a 30 year old woman and the organic carbon (OC) mass in the environment employed in ref 12.
for the two chemicals and is consistent with the current paradigm about the importance of partitioning properties for bioaccumulation. The Hydrophobicity Paradigm - a Limited Tool for the Identification of Bioaccumulative Chemicals. The relatively weak influence of partitioning properties in determining mmBAF that is evident in our modeling and the empirically based mmBAF values in Table 1 is, at first sight, diametrically opposed to the current paradigms in the field (as exemplified by the BAFwater values in Table 1). The results presented and discussed to this point refer to the human food web. However, it has been previously shown that bioaccumulation in fish is similarly insensitive to partitioning properties when viewed from a multimedia perspective (12). It is important to build a conceptual understanding as to why partitioning is relatively unimportant when bioaccumulation is viewed from a multimedia perspective. This can be understood by considering the two phenomena that together contribute to the bioaccumulation of chemicals: bioconcentration and biomagnification. Bioconcentration refers to increases in chemical concentration in an organism as a result of thermodynamic partitioning from the surrounding environment. When
FIGURE 4. Volume normalized woman: whole environment partition coefficient KH/Env (m3 environment/m3 woman) plotted as a function of KOW and KOA. The logarithmic scale of KH/Env is marked on the lines separating the colored fields. viewed from a multimedia perspective, it is clear that both the environment and biota are mixtures of phases including water, organic material, and (frequently) air. While partitioning between different phases in an organism or in the environment (e.g., between blood/adipose or soil/water) or across organism/environment (e.g., water/fish muscle) does indeed vary greatly between chemicals, partitioning between whole environments and whole organisms does not vary nearly as greatly because they have similar phase composition. The ratio of the whole woman/whole environment volume normalized partition coefficient (KH/Env) calculated by the model is illustrated for hypothetical chemicals with varying combinations of KOW and KOA in Figure 4. The ratio KH/Env is essentially constant for chemicals with log KOW > 3 and log KOA > 5. For these chemicals partitioning is dominated by sorption to organic material in both the woman and the environment, and the magnitude of KH/Env is determined by the ratio of the volume fractions of organic material (i.e., lipid equivalents) in the woman and the environment. For more volatile (lower KOA) and more hydrophilic (lower KOW) chemicals KH/Env decreases. For chemicals with these partitioning properties, air and water, respectively, make an important contribution to the overall sorption capacity of the woman and/or the environment. The decrease in KH/Env reflects the greater volume fractions in the environment of these phases relative to humans, when compared to the organic phase. For highly hydrophilic (low KOW) chemicals, a new plateau in KH/Env is reached at log KOW < 1 and log KOA > 2 that reflects the ratio of the woman/environment volume fraction for water; for highly volatile (low KOA) chemicals on the other hand KH/Env decreases continuously (data not shown). There is good qualitative agreement between the influence of partitioning properties on KH/Env (Figure 4) and the influence of partitioning properties on the mmBAF of persistent chemicals (Figure 1). The second phenomenon that contributes to bioaccumulation is biomagnification, which refers to increases in chemical fugacity in an organism relative to its food or environment (27). Biomagnification, which has been most extensively studied in fish, is commonly believed to occur only for high KOW chemicals, so the relatively constant bioaccumulation over the wide range of partitioning properties in Figure 1 is at first counterintuitive. There are several explanations for this observation. One is that the biomagnification in fish, which occurs primarily at 5 < log KOW < 8, is just one of the biomagnification mechanisms in the food chain. Biomagnification also occurs in plants as a result of the uptake of water-soluble chemicals via the transpiration stream in the roots, which then do not volatilize from the
foliage and hence accumulate there (28). This form of biomagnification occurs largely for low log KOW compounds (
