The Reassessment1 addresses exposure in terms of sources, environmental fate, environmental media concentrations, food concentrations, background exposures, and potentially highly exposed populations including important developmental stages. In this chapter, the committee discusses the exposure characterization section provided in the Reassessment, Part III. Part I of the Reassessment has a wealth of supporting information and comprises an executive summary and three volumes: Sources of Dioxin-like Compounds in the United States2; Properties, Environmental Levels, and Background Exposures; and Site-Specific Assessment Procedures.
The comments in this chapter are directed specifically at the use of exposure assessment in the risk assessment provided in Part III of the Reassessment, but the committee consulted the more detailed companion documents in Part I for supporting information.
Similar to the Reassessment, Part III, the chapter here is organized into sections on sources, environmental fate, environmental media and food, background exposures, and potentially highly exposed populations and sensitive populations. This chapter has three major sections: an overview and commentary on all aspects of the dioxin exposure assessment with an effort to point out strengths, limitations, and omissions; the committee’s findings; and specific recommendations.
OVERVIEW AND COMMENTARY ON EPA’S EXPOSURE CHARACTERIZATION
In this section, the committee provides summary and commentary on key issues related to exposure characterization for 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD, also referred to as dioxin), other dioxins, and dioxin-like compounds (DLCs). This information includes sources, environmental fate, environmental media and food concentrations, background exposures, and potentially highly exposed populations and particularly sensitive developmental stages.
For sources and environmental fate, EPA had a clearly articulated stepwise approach that the committee primarily accepted with some commentary. The other steps in the exposure assessment are not as easy to track, summarize, and critique. To comment on these steps, the committee used a format that went beyond the simple narrative.
Summary of the EPA Approach
The type, geographic distribution, and time history of the sources and associated emission magnitudes of TCDD, other dioxins, and DLCs are essential inputs for risk characterization. In Part III of the Reassessment, EPA discusses sources and emissions estimates for 1987 and 1995. More recently, EPA issued a report that includes the year 2000 update on sources and emissions estimates (EPA 2005b). These reports consider emissions of polychlorinated dibenzo-p-dioxin (PCDD) and polychlorinated dibenzofuran (PCDF) compounds and dioxin-like polychlorinated biphenyl (PCB) compounds. PCDDs and PCDFs have never been intentionally produced outside research laboratories. They are released to the environment as unintended by-products from various combustion, industrial, and biological processes. PCBs have been produced commercially in large quantities in the United States and other industrialized countries but are no longer commercially produced in the United States and Europe.
Sources of TCDD, other dioxins, and DLCs considered in the Reassess-
ment include combustion sources; metals smelting, refining, and processing industries; and chemical manufacturing, biological and photochemical processing, and reservoir sources. PCDDs and PCDFs are formed in most combustion systems—waste incineration and burning of coal, wood, and petroleum products; other high-temperature sources (such as cement kilns); and poorly or uncontrolled combustion sources (such as forest fires, building fires, and open burning of wastes). PCDDs and PCDFs can be formed during various types of primary and secondary metals operations, including iron ore sintering, steel production, and scrap metal recovery. PCDDs and PCDFs can be formed as by-products from the manufacture of chlorine-bleached wood pulp, chlorinated phenols (e.g., pentachlorophenol [PCP]), PCBs, phenoxy herbicides (e.g., 2,4,5-trichlorophenoxyacetic acid, or 2,4,5-T), and chlorinated aliphatic compounds. Recent studies suggest that PCDDs and PCDFs can be formed under certain environmental conditions (e.g., composting) from the action of microorganisms on chlorinated phenolic compounds. EPA also reported that PCDDs and PCDFs have formed during photolysis of highly chlorinated phenols.
Reservoir sources of TCDD, other dioxins, and DLCs are materials or places that contain previously formed PCDDs and PCDFs or dioxin-like PCBs and have the potential for redistributing and circulating these compounds into the environment. Potential reservoirs include soils, sediments, biota, water, and some anthropogenic materials. Reservoirs become sources when they release compounds to the surrounding environment.
Important Aspects of EPA’s Approach, Assumptions, and Findings
The key output of the Reassessment regarding sources is provided in Table 4-2 of the Reassessment, Part III, which summarizes an “inventory” of sources for the United States expressed as toxic equivalent quotients (TEQ). In constructing this table, EPA developed a qualitative confidence-rating scheme in which they used qualitative criteria to assign high-, medium-, or low-confidence ratings to the inventory classes. This table and comparisons of the years 1987, 1995, and 2000 are important inputs to EPA’s conclusions about long-term trends in the emissions of TCDD, other dioxins, and DLCs (furans and dioxin-like PCBs). In particular, the committee notes that EPA relied more on emissions estimates than environmental and biological media concentrations as a means of characterizing temporal trends in exposure to TCDD, other dioxins, and DLCs.
EPA’s use of the inventory table represents a “bottom-up” approach. EPA compiled a list of all potentially important source categories and provided an estimate of the probable magnitude of emissions from each of these categories. Summing these emissions by categories then provides an overall estimate of current and historical emissions. As noted by EPA, this
approach comes with large uncertainties in assigning emission values to each category and may exclude an unknown major category or fail to identify a number of minor categories that together provide large emissions. An alternative “top-down approach” would consider levels of PCDD and PCDF compounds in various environmental media (soils, sediments, and so forth) or biological media (vegetation, tree bark, fish tissues, and so forth) and identify the level of emissions required to account for these PCDD and PCDF levels. The top-down approach uses fate modeling and mass-balance analyses applied to environmental samples along with a number of assumptions to determine the extent to which deposition matches emissions (Rappe 1991; Harrad and Jones 1992; Brzuzy and Hites 1995; Eisenberg et al. 1998). Because of discrepancies among estimates of TCDD and related compounds in reservoirs relative to known sources, several researchers using a top-down approach concluded that EPA estimates of historical national emissions might underestimate emissions (Rappe 1991; Harrad and Jones 1992; Brzuzy and Hites 1995; Eisenberg et al. 1998). This suggests the possibility of unknown sources. Although the bottom-up and topdown approaches come with uncertainties, EPA could benefit substantially from using both approaches simultaneously to set plausible bounds on the historical and current trends in emissions. The committee recognizes that each approach has significant limitations. For example, the identification of ball clay as a potential source represents an interesting case, because it represents an identified (and managed) new source. In the absence of any other information, a bottom-up or a top-down approach is unlikely to find a minor contributor, such as ball clay, to overall national-level TEQ.
One of the most important aspects of the EPA analysis emerges in the discussion of the trends over time. With the most recent update to the inventory (EPA 2005b), there were dramatic declines relative to 1995 and 1987 in the emissions of TCDD, other dioxins, and DLCs from identified major sources. Unfortunately, the Reassessment and the background documents do not provide sufficient information for the committee to review the emission inventory table inputs, either the qualitative assessments or the quantitative estimates. In the current organization of the Reassessment, EPA does not clearly lay out the path for derivation of the emissions numbers. The lack of clarity makes a task as basic as checking the calculations and logic difficult.
Summary of the EPA Approach
Part III of the Reassessment provides a summary of key findings about the transport and environmental fate of TCDD, other dioxins, and DLCs.
Another apparent purpose of section 4.2 in Part III of the Reassessment is for EPA to make clear that assessment of environmental fate cannot be based on TEQ but must be based on individual congeners, but in section 4.1, EPA presents estimates of environmental releases as TEQ. They elected to present TEQ in place of mass quantities to better facilitate comparisons across sources. For purposes of environmental fate modeling, however, EPA notes that it is important to use the individual PCDD, PCDF, and PCB congener quantities rather than TEQ, because the physical and chemical properties of individual congeners vary and will behave differently in the environment. This material on the need to address specific congeners appears to have been added to the Reassessment in response to the Science Advisory Board’s comment that the original dioxin reassessment report (EPA 1994) implied that emissions expressed as TEQ could be used as source terms for modeling transport, fate, and exposure in risk assessments.
Important Aspects of EPA’s Approach, Assumptions, and Findings
In its assessment of environmental fate in the Reassessment, EPA makes the following key findings:
TCDD, other dioxins, and DLCs are widely distributed in the environment as a result of a number of physical and biological processes.
Because physical and chemical properties vary substantially among individual congeners, the congeners will behave differently as they are transported through and transformed in the environment. Thus, for purposes of environmental fate modeling, it is important to use the individual PCDD, PCDF, and PCB congener levels rather than TEQ.
Atmospheric transport and deposition of TCDD, other dioxins, and DLCs are the primary means of their dispersal throughout the environment.
The two primary pathways for TCDD, other dioxins, and DLCs to enter the ecological food chains and human diet are air-to-plant-to-animal and water-and-sediment-to-fish pathways.
In reviewing these findings, the committee notes that they are supported by the source and exposure information provided in the Reassessment. The committee further notes that EPA missed an opportunity to use data on individual congeners to assess how TEQ changes in time and space. Moreover, many EPA findings on sources, fate, and exposure tend to be drawn from temporal and spatial trends in emissions. EPA did not make full use of exposure media concentration data, particularly food concentration data, to confirm that the space and time trends are reflected in exposure media. EPA missed the opportunity to use emissions data for individual congeners combined with fate modeling to assess the persistence of
individual congeners to estimate the persistence of TEQ and the spatial distribution of TEQ. Another issue of interest to the committee is how the reliability of the TEQ estimate becomes more uncertain with time. Because of uncertainty about toxic equivalency factors (TEFs) for the more persistent congeners, such as the hexa, hepta, and octa chlorinated congeners, that tend to dominate the TEQ, the reliability of the TEQ characterization degrades with the resulting accumulation of the more persistent congeners. As a result of not considering this issue, EPA does not yet have the ability to determine when reservoir sources will become significant relative to all anthropogenic sources in characterizing the TEQ of TCDD, other dioxins, and DLCs.
Environmental Media and Food Concentrations
EPA developed estimates of concentrations of TCDD, other dioxins, and DLCs in various environmental media, including foods, using only those studies from locations that they considered as representing background levels of these compounds. The extent to which regions with high exposures were either captured or excluded is not clear in the Reassessment. Moreover, because background has a continuum of low to high concentrations, it is also not clear where the line was drawn to distinguish background from “not background.”
Although TCDD, other dioxins, and DLCs in food and environmental media have been declining over the last three decades, the presence of these compounds in foods (primarily in animal fats and oils) now represent 90% or more of human exposure (IOM 2003). However, there are significant uncertainties inherent in calculating with accuracy or precision dietary exposure because of the limited analyses of individual foods; the methodological improvements over time with corresponding lowering of the limits of detection; the limited information on the congener composition of various foods; the values assigned to “nondetects”; the alterations in concentrations of TCDD, other dioxins, and DLCs due to methods of preparation and cooking; the wide diversity of human dietary composition and consumption patterns; and the inherent inaccuracies of the instruments used to assess dietary intake in humans (IOM 2003). The Reassessment extensively details the information available at the time (Part I, Volume 2, Chapters 3 and 4) and briefly mentions the Institute of Medicine (IOM) report (Part III, section 4.3). EPA acknowledges that, in general, the available food data come from “studies that were not designed to estimate national background means” and that “it is not known whether these estimates adequately capture the full national variability” (Part III, section 4.3).
Since the Reassessment, additional studies have estimated human dietary intake of TCDD, other dioxins, and DLCs. In the Netherlands, “the
estimated median life-long-averaged intake of the sum of PCDDs, PCDFs, and dioxin-like PCBs in the population is 1.2 pg WHO [World Health Organization] TEQ per kg of body weight per day” (Baars et al. 2004). The estimated median is below the WHO tolerable daily intake of 2 pg TEQ/kg of body weight, and the authors estimated that approximately 8% of the Dutch population have life-long averaged intakes above the WHO tolerable intake level. Charnley and Doull (2005) and the U.S. Food and Drug Administration (FDA) (CFSAN 2005a,b) estimated human food exposures to TCDD, other dioxins, and DLCs between 1999 and 2003 with data derived from the FDA total diet study. These studies provided intake estimates since 2001 in which the average daily intake for all age groups fall below the WHO tolerable daily intake level of 2 pg TEQ/kg of body weight. However the estimates do not include breast-fed infants.
Charnley and Doull (2005) noted that when assessors represent exposure media concentrations of TCDD, other dioxins, and DLCs—primarily in food—below the limit of detection (LOD) by one-half the detection limit, “approximately 5% of the intake estimates for 2-year-olds and 1% of the intake estimates for 6-year-olds exceed the tolerable daily intake by about 10%.” When these media concentration measurements below the LOD are set to zero (when only concentration values actually measured are used), “only 1% of intake estimates exceed the tolerable daily intake for 2-year-olds.” The committee notes that this reveals the problem of interpreting a “mean” concentration. The arithmetic mean among individuals in these cases is quite sensitive to the treatment of samples below the LOD. One alternative is to avoid the use of sample means and instead consider comparisons based only on percentile concentrations (e.g., median values and 90th percentile individual values). These percentile values only require information about the rank of a sample and thus avoid the impact on central value estimates introduced by LOD assumptions.
In both American (Charnley and Doull 2005) and Dutch (Baars et al. 2004) populations, meat and dairy products account for approximately 50% of the TCDD TEQ consumed in food, but the Dutch consume more TCDD TEQ in fish than do Americans—16% and 5.8%, respectively. Additional data from the U.S. Department of Agriculture Food Safety and Inspection Service confirmed that the contents of TCDD, other dioxins, and DLCs measured in 2002 and 2003 in meat products sold in the United States, including hogs, steer, heifers, young chickens, and young turkeys, have declined significantly from the contents measured in 1994 through 1996, although methodological differences preclude a precise calculation of the decrement (FSIS 2005).
Recognizing that some data gaps will remain in the source inventory, in the environmental media concentrations describing the distribution and environmental fate of these chemicals, and in various parts of the food
chain and human tissue concentrations (e.g., breast-milk and serum concentrations), the committee notes that it would be helpful if EPA could set up a congener-specific database of typical concentrations in foods for the whole range of PCDDs, PCDFs, and dioxin-like PCBs (those included in the WHO TEF list). Such a database would need to fulfill clear requirements of data quality and traceability (e.g., chemical analysis, representative and targeted sampling, data representative consumer exposure, presentation of data, and handling and presentation of values below the LOD). Making such a database available could improve the transparency of how EPA came to some of the conclusions in the Reassessment. Moreover, if TEF values change, TEQ values can be easily recalculated. Such a database could be updated on a regular basis to evaluate temporal trends. Here, it is important to consider methodological aspects (e.g., reproducibility, sensitivity, specificity of the analytical determinations, inclusion of reference samples, and comparable sampling strategy) to ensure that such a time-trend analysis is useful.
The section of the Reassessment that addresses background exposures provides a summary of information on human tissue levels, intake estimates, and variability in intake levels.
The section of the Reassessment addressing tissue levels evaluates data on concentrations of TCDD, other dioxins, and DLCs in human tissues expressed per gram of lipid and the changes in these concentrations that have occurred in recent decades. The Reassessment acknowledges the difficulty of comparing different data sets because some do not include coplanar PCBs in the estimation of TEQ values. It is clear from the data in Part III, Table 4-5, that TCDD per se is not the main source of TEQs in human lipid. The Reassessment uses the calculation of body burden at steady state along with EPA’s associated assumptions given in section 1.3 to calculate the TEQ concentration in human lipid based on the best estimate of current adult intake and the assumption of 25% body fat. The result is about one-half of that actually measured in human lipid. EPA assumes that the discrepancy arises from the presence of an historical body burden and lipid concentration. Given the various assumptions in the estimation of body burden at steady state, especially in relation to application of the TCDD model to congener TEFs, it is reassuring that the TEQ in human lipid predicted by the model are somewhat consistent with the estimated values.
Intake Estimates and Variability in Intake Levels
These sections describe intake estimates and the variability and age-related changes in intake—in particular, by nursing infants.
Potentially Highly Exposed Populations or Developmental Stages
In compiling and evaluating available data on highly exposed populations, EPA considered contamination of food, exposures to workers, and exposures to nursing infants.
In the Reassessment, EPA assumes that contamination incidents in food probably have not and will not lead to disproportionate exposures to populations living near where they occurred. The basis for this assumption is that meat and dairy products in the United States are widely distributed on a national scale. As a result of this assumption, the Reassessment does not comment on any disproportionate exposures due to interaction with contaminated sites.
In considering the distribution of exposures to TCDD, other dioxins, and DLCs in the U.S. population, EPA suggested that variability in exposure probably regresses toward the mean because Americans consume varied diets from multiple sources, meaning that EPA assumed that variations in diet would prevent either very high or very low extremes of exposure. EPA reported that this pooling of the food supply reduces the potential high exposure that could result from high consumption of certain food products. This assumption may be valid, but EPA should provide additional analyses to support it and should also explicitly consider the possibility of populations who violate the assumptions with respect to varied diets and multiple sources (e.g., those who rely on home-produced foods or sustenance fishing). It is of interest that only in the last paragraph of this section is there discussion of measurements reflecting potentially highly exposed groups. Here it is mentioned without further discussion that several European studies showed increased TCDD, other dioxins, and DLC levels in milk and other animal products near combustion sources. EPA did not consider the implications of this finding for the U.S. population. It thus seems that EPA is implicitly assuming that this problem does not exist in the United States.
The Reassessment suggests that no clear evidence demonstrates that increased exposures to TCDD, other dioxins, and DLCs are currently occurring among U.S. workers, but the Reassessment does not document the level of ongoing monitoring and assessment to support this conclusion. Low levels of occupational exposure are not congruent with their reported inventory of sources.
To evaluate the impact of nursing on infants, EPA estimated changes in body burden with a model developed by Lorber and Phillips (2002). This
model includes a number of assumptions, including that the fraction of TCDD, other dioxins, and DLCs absorbed by an infant after ingestion is 0.80 and that the dissipation rate of the ingested TEQ is rapid. The developers evaluated the model with the data (from Germany) of Abraham et al. (1998). The EPA evaluation does not necessarily confirm the numerous assumptions (e.g., half-life and uptake). Moreover, the evaluation does not capture variability or uncertainty in the model because of the assumptions. The conclusions at the end of the Reassessment, Part III (p. 4-23, lines 7 to 15), include the presentation of model predictions that are implied to be very precise. Yet, in view of the various assumptions, these results might or might not reflect reality. In light of the amount of supporting information available from other sources, it is unclear why EPA relied primarily on a relatively detailed model with all its inherent uncertainties to report that the annual infant TCDD-TEQ intake from nursing significantly exceeds the currently estimated adult intake of 1 pg TEQ/kg/day. This observation can be easily demonstrated from qualitative findings and simple assessments based on TCDD half-life and lipophilicity, infant body size, breast-milk composition, and breast-milk intake. The committee recommends that EPA consider the value and availability of any data to confirm this modeling result.
Is EPA’s Exposure Assessment Scientifically Robust?
In preparing its findings, the committee notes that those who will make use of the Reassessment are likely to be interested in issues beyond risk characterization and risk assessment methodology. For example, some users will want to use the Reassessment to decide whether U.S. exposures to TCDD, other dioxins, and DLCs pose an undue health risk, whereas others will want to use the Reassessment to consider alternatives for reducing exposures to these compounds and identifying strategies for achieving reductions of TCDD-TEQ burdens in the U.S. population. In preparing its findings, the committee considered a range of potential uses for the Reassessment—including the following alternatives.
Clearly, an important opportunity that EPA overlooks is checking the observed decline in overall environmental concentrations against body burden changes over time. For example, the emissions estimates for PCBs and mass-balance evaluation provided recently by Breivik et al. (2002a,b) provide a better opportunity to consider global-scale chemical PCB fate by
comparing model results with measured concentrations of PCBs at monitoring stations located in regions of the Northern Hemisphere over the 70-year period from 1930 to 2000. Calculations based on this 70-year estimate of emissions will introduce uncertainties, but such an analysis could build confidence about trends and better inform future investigation.
EPA did not fully address the issue of reservoir sources or explore their potential impacts on the long-term distribution of TCDD, other dioxins, and DLCs as well as the distribution of TCDD TEQ. It also did not fully consider how reservoir effects vary among different congeners and thus cause the TEQ from reservoir sources in soil and sediments to evolve and change in time. Finally, EPA did not address the issue of when reservoir sources are likely to become dominant relative to anthropogenic sources. For example, some studies provide experimental evidence for how TCDD, other dioxins, and DLCs are incorporated in soil and then reemitted (Brzuzy and Hites 1995, 1996; Cousins et al. 1999a,b; Cousins and Mackay 2001; McKone and Bennett 2003).
One of the most important aspects of the analysis emerges in the discussion of the trends over time. Given the importance of properly estimating TEQ and the need for risk analysts to consider the impacts of exposure timing for some potential dose metrics, the EPA inventory should yield estimated TEQs associated with each identified source more transparently. Part III of the Reassessment and the background documents do not provide sufficient information for the committee to review the emissions inventory table inputs, either the qualitative assessments or the quantitative estimates. Although inventories shifted over time with the identification of new sources, EPA did not examine the extent of that shift.
Environmental Fate Assessment
EPA’s finding regarding the wide distribution of TCDD, other dioxins, and DLCs is supported by environmental sampling. There have been sufficient measurements to conclude that, as a chemical class, these compounds are widely dispersed in the environment. With regard to individual congeners of PCDDs, PCDFs, and dioxin-like PCBs, a sufficient number of samples are not available to conclude that each individual congener is widely dispersed in the environment.
Although consideration of individual PCDD, PCDF, and PCB congeners would be informative and useful, doing that for more than 200 congeners would be excessive; summing up mass quantities instead of TEQ contributions would be equally bad, and most other inventories (e.g., in Europe and Japan) were also done in TEQs. According to the Reassessment (Part III, p. 1-8), five congeners contribute approximately 80% of the total TEQ in humans: 2,3,7,8-TCDD, 1,2,3,7,8-PCDD, 1,2,3,6,7,8-hexachlorodibenzo-p-dioxin, 2,3,4,7,8-
PCDF, and PCB126. Thus, it would be informative to provide congener-specific emissions estimates for these congeners in place of the TEQ estimates.
Environmental Media and Food Concentrations
In reviewing the EPA assessment of environmental media and food concentrations, the committee had the following concerns:
In using food concentration data to estimate intake, the choice of LOD has significant impact on calculated mean values. EPA was not clear about (1) how it made use of values below the LOD in making intake estimates based on food concentration data, and (2) how its treatment of the LOD had an impact on results. Because the committee found no basis for making recommendations on other aspects of the food intake calculation and because food supply issues are covered in the IOM (2003) report, the committee elected to focus on the LOD issue.
EPA did not make clear its criteria for distinguishing background from non-background concentrations.
Relative to dioxin and furan congeners, data on environmental media and food concentrations of dioxin-like PCBs were generally lacking.
TCDD-TEQ intake estimates from fish consumption did not include direct consumption of fish oils.
The committee finds value in EPA’s establishing a congener-specific database of typical concentrations for the range of PCDDs, PCDFs, and dioxin-like PCBs (those included in the WHO TEF list). The details of such a database are described above in the overview and commentary section in the subsection on environmental media and food concentrations.
Estimates of Background Exposures
The committee found the text in this section noncontroversial and the conclusions valid. The committee did not find any important errors in this text, but issues arose concerning the interpretation of the background exposure data. It is not clear that the existing database as used by EPA covers all foods consumed by the U.S. population (e.g., data were missing on fish oils). It would be helpful to include or reference in the exposure estimates the most recent data on food intake as produced by FDA. The committee believes that EPA can make more efficient use of the existing data sets on occurrence in foods and on food consumption to assess the distribution of intakes of TEQ for the general U.S. population (different age groups, expressed in picogram per kilogram of body weight per day) as well as intra-and inter-person variability.
Exposures in Highly Exposed Populations or at Key Developmental Stages
In compiling and evaluating available data on highly exposed populations, EPA failed to draw informative conclusions from the numerous studies described in the full document (Part I). Part III of the Reassessment is missing a summary that integrates the information compiled in Part I.
The Reassessment makes statements discounting the potential for having highly exposed groups without clearly documenting the basis for these statements. First, it suggests that, with regard to the commercial food supply, the incidents of contamination by TCDD, other dioxins, and DLCs are likely to be low. Yet the Reassessment provides no formal assessment and almost no data to support that determination. It also states that there is no clear evidence that increased exposures are occurring among U.S. workers. Finally, it reports that no or few studies show evidence of groups in the United States being exposed to highly increased levels of TCDD, other dioxins, and DLCs in situations in which people consume large quantities of foods with high levels of these compounds. In spite of giving substantial attention to nursing infants as a highly exposed group, EPA provides no comment on the potential level of increased exposure that may have arisen during recent contamination episodes involving the commercial food supply (e.g., the ball clay incident and high levels in beef and dairy animals due to PCP-treated wood).
Is There a Clear Delineation of All Substantial Uncertainties and Variabilities?
Overall, the committee finds that EPA has qualitatively identified a number of important uncertainties and variabilities. However, there are some areas for which even the qualitative information provided by EPA was unclear or incomplete. What is more important is that the Reassessment does not quantitatively characterize either variability or uncertainty in exposure except in the limited sense of demonstrating increased average daily dose estimates for children (on a body-weight basis) and analyzing potentially increased exposures for nursing infants during their first few years of childhood.
The magnitude, type, geographic distribution, and time history of TCDD, other dioxins, and DLC sources are essential components for risk characterization. The interpretation of these factors is an important input to decisions about managing both new and historical (reservoir) sources.
Any errors in interpretation could lead to policies and regulatory actions that are inefficient or ineffective in reducing human exposures to TCDD, other dioxins, and DLCs. EPA exposure characterization excludes basic data quality checks that could provide an opportunity to evaluate key assumptions. The committee notes that EPA did not explore an alternative top-down approach in an effort to evaluate the results from its bottom-up approach for source-to-intake characterization. The Reassessment clearly notes significant uncertainties in estimates of emissions and communicates these uncertainties using qualitative confidence scores (A, B, and C). However, given this clear acknowledgment of significant uncertainties in the emission estimates, the committee questions the reliability of the Reassessment’s trend analysis of emissions from 1987 to 2000. EPA does not communicate these uncertainties in the Reassessment’s summary and other sections where the trend analysis of emissions is discussed
Environmental Fate Assessment
EPA’s finding that atmospheric transport and deposition of TCDD, other dioxins, and DLCs are a primary means of their dispersal throughout the environment is strongly supported by theoretical models in combination with observations of these compounds globally that are more uniform than emissions sources and far from regions of release. However, there is considerable uncertainty about the nature and magnitude of the re-emission process that takes place after deposition.
The EPA finding that the two primary pathways for TCDD, other dioxins, and DLCs to enter the food chain and human diet are air to plant to animal and water and sediment to fish is supported by environmental sampling, but significant uncertainty remains about mechanisms and rates of transfer through food webs. There have been sufficient measurements to conclude that, as a chemical class, TCDD, other dioxins, and DLCs, particularly the more persistent ones, enter humans primarily through animal products and fish. With regard to individual congeners of PCDDs, PCDFs, and dioxin-like PCBs, samples are insufficient to conclude that each individual congener enters humans by these two primary pathways.
The committee concurs with EPA that, although it is appropriate to use TEQ as a metric of release, it must clearly emphasize the uncertainty and limitation of using the TEQ approach.
Environmental Media and Food Concentrations
It is uncertain whether the existing information on background levels in environmental media and food adequately captures the full national variability.
The effect of cooking and processing on concentrations of TCDD, other dioxins, and DLCs in foods was considered too limited to draw firm conclusions.
EPA had only limited access to data that support any conclusions about temporal trends in the occurrence of TCDD, other dioxins, and DLCs in environmental media and foods.
Estimates of Background Exposures
When EPA assumes that less-than-LOD samples equal zero, there are often significant differences between values of TEQ-based estimates of background intake compared with estimates obtained when EPA assumes that less-than-LOD samples equal half or the whole detection limit. This illustrates the importance of analytical method sensitivity in limiting the ability to determine the full range of population variation of PCDDs, PCDFs, and dioxin-like PCBs in human and other tissues. EPA missed the opportunity to quantify the effect of these differences in an uncertainty analysis of the current exposure estimates.
Exposures in Highly Exposed Populations or at Key Developmental Stages
As noted above, it is not clear to the committee why EPA relied entirely on a model with all its inherent uncertainties to conclude that the annual infant intake of TCDD, other dioxins, and DLCs from nursing significantly exceeds the currently estimated adult intake of 1 pg TEQ/kg/day. EPA failed to provide any measurements or environmental samples to support the conclusions drawn from the model. Providing this information would increase the confidence in its conclusions on this issue.
With regard to sources and emissions, the most appropriate way to characterize historical sources of TCDD, other dioxins, and DLCs is to compile a list of all known sources, make emissions estimates for each class from the available literature, and then combine these emissions to establish historical trends. The committee finds this assumption reasonable and sufficiently documented but finds that it would be valuable for EPA to consider alternative approaches (e.g., the top-down approach) for confirming or revising this approach.
In its consideration of highly exposed subpopulations, EPA found information indicating that breast-feeding might result in higher TCDD-TEQ body burdens of the nursing infant compared with those of non-nursing
infants. The issue that exposure of the developing infant is already starting during pregnancy (in utero exposure) is not addressed in this section or not clearly mentioned in the full Reassessment. EPA did not consider this information in their overall conclusions about exposure. Moreover, because of the potential for causing anxiety among nursing mothers, EPA should expand its discussion about the multiple known benefits of breast-feeding as a footnote to the section describing exposures to nursing infants.
In characterizing exposures, EPA relied primarily on measurements combined with assumptions for emissions and relied almost completely on measurements of environmental and tissue levels for estimating exposure and body burdens. With the exception of their toxicokinetic model for nursing mothers, they did not rely on models for assessing transport and distribution from sources to environmental (such as air, water, and soil) and exposure (food products) media.
EPA’s finding that, for purposes of environmental fate modeling, it is important to use the individual PCDD, PCDF, and PCB congener values rather than TEQ is self-evident and robust. The committee concurs with the EPA finding that TEQ should not be used in place of individual congener concentration as the variable in fate models for TCDD, other dioxins, and DLCs.
Were the Most Appropriate Studies Relied Upon?
For characterizing emissions, EPA developed a comprehensive inventory of all known emissions of TCDD, other dioxins, and DLCs but did not fully characterize the work of those researchers who looked at a top-down approach for characterizing historical emissions of PCDD and PCDF compounds. Rappe (1991), Harrad and Jones (1992), Brzuzy and Hites (1995), and Eisenberg et al. (1998) used fate modeling and mass-balance analyses applied to environmental samples and a number of assumptions to determine the extent to which deposition matches emissions.
CONCLUSIONS AND RECOMMENDATIONS
To assess the total magnitude of emissions of TCDD, other dioxins, and DLCs, EPA used a bottom-up approach in which they attempt to identify all source categories and estimate the magnitude of emissions for that category. EPA also should use a top-down approach that attempts to account for observed levels and consider what emissions would be required to account for these levels. These alternative approaches give rise to significantly different estimates of the historical levels of emissions. Both ap-
proaches come with uncertainties. Thus, the readers of the Reassessment could benefit substantially from EPA using both approaches simultaneously to set plausible bounds on the historical and current trends in emissions.
EPA needs to be explicit about how they dealt with measurements below the LOD in environmental and exposure media samples. Whether the less-than-LOD samples are assumed to be zero, assumed to be one-half LOD, imputed by fitting a censored regression model, or dealt with by using some other assumption could have significant impacts on estimates of TCDD, other dioxins, and DLC intakes. EPA should explicitly address how its assumption affects the magnitude and range of estimated intakes relative to alternative approaches. Moreover, EPA should describe how the changing LOD affects its estimate of the time trend of TEQ intake.
Because many users of the Reassessment will be interested in reducing exposures to TCDD TEQ and identifying strategies for achieving reductions in body burden, EPA should add some discussion in the exposure chapter about what factors (such as diet, activities, and location) tend to increase or decrease TEQ intake.
EPA should construct their reports so that information in the summary emissions inventory table of Part III can be more clearly and more easily traced back to the source chapters that provide background information.
EPA should evaluate the impact on early emission-inventory estimates (1987, 1995) of sources added in more recent assessments (2005) so that the overall percentage declines reflect all sources. Such an evaluation would help to confirm dramatic decreases in TEQs that appear to have occurred over time.
EPA should define a strategy for collection of samples and reanalysis of archived samples to answer a number of remaining questions about exposure trends and to fill in some important data gaps. (The committee does not consider it particularly useful or cost-effective for EPA to obtain and analyze more environmental media samples for the full range of TCDD, other dioxins, and DLCs.)
EPA should create a congener-specific and active database of typical concentrations for the whole range of PCDDs, PCDFs, and dioxin-like PCBs (included in the WHO TEF list). This recommendation applies to work separate from the Reassessment. The database should be based on a compendium of all available data and be updated on a regular basis with new data as they are published in the peer-reviewed literature. Maintaining the database would not require EPA to conduct its own sampling program. Such a database would need to fulfill clear requirements of data quality and traceability, including chemical analysis, representative and targeted sampling, data representative of consumer exposure, presentation of data, handling, and presentation of less-than-LOD samples.
In view of the number of sites with increased levels of PCBs in the
environment and anticipating that those levels could result in higher contributions of the dioxin-like PCB fraction to total TEQ exposure (e.g., through local fish consumption), EPA should explicitly characterize the variability of population exposures to PCBs. EPA should estimate the magnitude of the ratio of high-end to median and mean exposure, the factors (e.g., proximity to sources, geographic region, and eating habits) that give rise to high-end exposure, and the relative uncertainty with which high-end exposures can be estimated.