Comparisons of estimated human body burdens of dioxinlike chemicals and TCDD body burdens in experimentally exposed animals.

Humans are exposed to mixtures of polyhalogenated aromatic hydrocarbons, and the potential health effects of these exposures are uncertain. A subset of this class of compounds produce similar spectra of toxicity in experimental animals as does 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), and these chemicals have been classified as "dioxins." In this study, we compared the body burdens of dioxins that produce effects in experimental animals to body burdens associated with these effects in humans. Human body burdens were estimated from lipid-adjusted serum concentrations of dioxins, assuming dioxins are equally distributed in body fat and an adult has 22% body fat. The toxic equivalency factor (TEF) method was used to calculate body burdens of dioxins in humans. These calculations included dibenzo-p-dioxins, dibenzofurans, and polychlorinated biphenyls. In the general population, average background concentrations were estimated at 58 ng TCDD equivalents (TEQ)/kg serum lipid, corresponding to a body burden of 13 ng TEQ/kg body weight. Populations with known exposure to dioxins have body burdens of 96-7,000 ng TEQ/kg body weight. For effects that have been clearly associated with dioxins, such as chloracne and induction of CYP1A1, humans and animals respond at similar body burdens. Induction of cancer in animals occurs at body burdens of 944-137,000 ng TCDD/kg body weight, while noncancer effects in animals occur at body burdens of 10-12,500 ng/kg. Available human data suggest that some individuals may respond to dioxin exposures with cancer and noncancer effects at body burdens within one to two orders of magnitude of those in the general population.

Over the last 30 years, an abundance of studies have clearly demonstrated that 2,3,7, 8-tetrachlorodibenzo-p-dioxin (TCDD) is extremely toxic to experimental animals (1)(2)(3). Fish and wildlife are also sensitive to the toxic effects of this chemical (4). TCDD is carcinogenic in male and female rats and mice, male hamsters, and male and female fish (5,6). Reproductive and developmental toxicity has been observed in all experimental animals tested. Immunotoxic effects occur in mice, rats, and nonhuman primates exposed to low doses of TCDD (7). Evidence to date indicates that the actions of TCDD are mediated by the Ah receptor (8,9) which functions as a signal transducer and transcription factor. In many ways the actions of the Ah receptor are similar to those of the steroid hormone receptors (10,11), although the Ah receptor is not a member of this superfamily of proteins (12,13). Other halogenated dibenzo-p-dioxins and dibenzofurans substituted in all four lateral positions also have high binding affinity to the Ah receptor and induce the same spectrum of toxicity as TCDD (14). In addition, certain polyhalogenated biphenyls, naphthalenes, and diphenyl ethers are Ah receptor agonists. Humans are exposed to complex mixtures of these chemicals; estimates of daily exposure to TCDD or "dioxinlike" (all 2,3,7,8-halogenated dibenzo-p-dioxins and dibenzofurans as well as the dioxinlike polychlorinated biphenyls) chemicals is 3-6 pg TCDD equivalents/kg/day in the United States (15,16). The subclass of the polyhalogenated aromatic hydrocarbons with dioxinlike activity are referred to as dioxins in this article.
Although the toxic effects of dioxins in experimental animals are unequivocal, their toxic effects in humans are less certain. Chloracne is the only toxic effect induced by dioxins for which there is unequivocal evidence linking exposure to effect in humans (17). The uncertainty of other toxic effects of dioxins in humans is due to the scarcity of human populations with high dose exposures, limited data on the body burdens of dioxins present in these populations, the difficulty in assessing sensitive toxic endpoints in humans, and the lack of knowledge about likely, but unknown, genetic factors that may influence the relative susceptibility of individuals. Dioxins produce some of the same biochemical alterations in humans and experimental animals (18). Several recent epidemiological studies suggest an association between dioxin exposure and increased incidence of cancer (19)(20)(21)(22)(23) and increased incidence of altered glucose tolerance in exposed populations (24,25). One way to determine the strength of an association between dioxin exposure and a toxic effect in humans would be to compare the dose of dioxin that is required to produce an effect in animals to the dose of dioxin in humans that is associated with a similar toxic effect. While it is clear that for some toxic effects, such as lethality and body weight loss, there are marked species differences in susceptibility to dioxins, many recent studies have also noted that for other endpoints, such as reproductive and developmental effects, most animal species respond at similar doses (9,26). Thus, the dose of dioxin that produces a particular effect in experimental animals might be expected to be similar to the dose of dioxin associated with that same effect in humans.
Although the hypothesis that toxic doses of dioxins in animals and humans are similar for most responses is theoretically testable using data from accidentally exposed human populations, there are some difficulties. In particular, it is often difficult to determine the human dosage at the time of exposure. In experimental studies, animals are administered a known amount of dioxin and evaluated at a specific time after the treatment. In humans the actual exposure is unknown and often difficult to estimate. Several epidemiological studies determined serum concentration of dioxins in exposed and control populations (19)(20)(21)(22)(23)(24)(25). Although the dose to the individuals in these studies is uncertain, the body burdens of dioxins in these populations can be estimated at a specific point in time. In addition, serum and tissue dioxin concentrations from populations in the United States with-out any unusually high exposures have been reported from several different laboratories (27)(28)(29)(30). All humans in industrialized countries are presumed to carry a body burden of dioxins based primarily on consumption of minute quantities of dioxin in the food supply. Here we compare the body burdens of dioxins that produce effects in experimental animals to the body burdens associated with effects in humans, based on the clinical findings observed during epidemiological studies. A comparison of the in vitro effects of dioxins on human and animal tissues and cell cultures is also presented. This analysis suggests that some of the effects observed in experimental animals also occur in humans and that the body burdens of dioxins associated with these effects (adaptive and/or toxic) are similar between animals and humans.

Methods
Comparisons of animal and human tissues or cell lines studied under in vitro conditions are shown in Table 1. This list is not meant to be exhaustive. The data presented are from peer-reviewed literature and include only those papers that compared animal and human tissues in the same study or laboratory.
We estimated human body burdens based on analyses of dioxin in serum or tissue in the cited literature. Several assumptions were used to derive body burdens from these values. Dioxins are assumed to be equally distributed in the body lipid with all tissues having the same concentration of TCDD when expressed on a lipid-adjusted basis (31)(32)(33). Thus, serum levels presented as lipid-adjusted are assumed to be equivalent to adipose tissue levels expressed as lipid-adjusted values. In addition, we assumed that for the average person, 22% of the body weight is lipid or fat (34). To estimate body burdens in humans, lipidadjusted serum or adipose tissue concentrations (expressed as ng TCDD/kg or TEQ/kg) were multiplied by 0.22 (34), the fraction of body weight that is fat. Some of the body burden estimates in humans presented here are based on tissue concentrations of 2,3,7,8-tetrachlorodibenzo-p-dioxin alone. In all cases, humans were likely to have been exposed to many dioxinlike chemicals that bind to the Ah receptor and produce the same spectrum of toxic effects in experimental animals as TCDD (2,14,26). To account for exposure to additional dioxins, the toxic equivalency factor method (TEF) was used (14,(35)(36)(37)(38). TEFs are relative potency factors used to convert the amount of dioxins in a sample to TCDD equivalents or TEQs (14,(35)(36)(37)(38). TEFs were assigned only to 2,3,7,8-chlorine substituted dibenzo-p-dioxins and dibenzofurans, the coplanar polychlorinated biphenyl(s) (PCBs) (IUPAC nos. 77, 81, 126, and 169) and the mono-ortho-substituted PCBs (IUPAC nos. 105, 114, 118, 156, 157, 167, and 189). The TEF values used for the dibenzo-p-dioxins and dibenzofurans were the U.S. EPA interim TEF values, which represent an internationally accepted convention for assessment of dioxins (37,38). The TEF values used for the dioxinlike PCBs were the World Health Organization values, which resulted from a recent international meeting of dioxin and PCB experts (38). Hence, body burdens for this complex mixture of related chemicals are expressed in terms of TEQs.
Body burden estimates in populations exposed to background levels of dioxins were based on published studies that measured serum concentrations of 2,3,7,8chlorine substituted dibenzo-p-dioxins (CDDs) and dibenzofurans (CDFs) and dioxinlike PCBs in populations with no unusually high exposure to dioxins (27)(28)(29)(30)39). Serum concentrations of CDDs and CDFs have been measured in a number of different populations from several studies. Schecter (27) presented data indicating that the average whole-blood CDD/CDF concentration in U.S. (n = 100) and German (n = 85) populations were similar when presented on a TEQ basis (41 and 42 ng TEQ/kg whole blood, lipid adjusted). More extensive studies of U.S. populations indicate that the national average for serum CDD/CDF concentrations is 28 ng TEQ/kg serum lipid (39). Much smaller studies of congener-specific PCB serum or adipose tissue concentrations have been published that indicate that average dioxinlike PCB concentrations range from 8 to 17 ng TEQ/kg tissue lipid in U.S. populations (28,30). The range of average tissue TEQ concentrations for CDDs/CDFs is 28-41 ng TEQ/kg lipid and for the PCBs the range is 8-17 ng TEQ/kg lipid. Based on these studies, average background dioxin tissue concentrations range from 36-58 ng TEQ/kg lipid. In these populations, TCDD contributes approximately 15% of the total TEQ. Body burden estimates in exposed populations were based on the published literature. These populations were assumed to have background exposures, in addition to the specific exposures determined in the study. The level of dioxins in exposed populations were often determined years after the initial exposure. Body burdens were estimated at the time of maximal exposure assuming the rate of total body elimination of dioxins is linear with respect to time and dose and a assuming 7.1-year half-life (40).
Determination of maximum body burdens in experimental animals was based on the administered dose and the rate of elimination of dioxin from the animal. Total body half-life of TCDD in experimental animals was assumed to be first order with respect to time and dose. In several cases, body burdens in animals were based on tissue levels determined in the study.
Effects seen in epidemiological studies have been divided into two categories. The first category (Table 2) is for effects that have been causally associated with exposure to dioxins. These are effects for which there is strong evidence that the responses observed are due to exposure to dioxins and/or related compounds. Typically, adverse effects with demonstrated causality Abbreviations: EGFR, epidermal growth factor receptor; LOEL, lowest observed effect level; TCD tetrachlorodibenzo-p-dioxin; PCDD, polychlorinated dibenzodioxin; PCDF, polychlorinated dibe PCB, polychlorinated biphenyl. "The data and methodology used to determine each value are presented in the appendix under indicated. are associated with high-level exposure and are severe enough to dearly indicate a relationship with such exposure. Chloracne is an example of such an adverse effect. In addition, biochemical changes such as induction of CYPlAI and decreases in EGF receptor autophosphorylation are included in this category because there is significant experimental evidence that these effects occur through activation of the Ah receptor and are therefore causally related to exposure to dioxinlike chemicals. A second category (Table 3) was assigned for effects associated with dioxin exposure for which a causal link has not been definitively proven. Effects included in this category are decreased birth weight, decreased growth, delayed developmental milestones, cancer, decreased testosterone levels, and increased risk of diabetes. In both Tables 2 and 3, body burdens in experimental animals are presented for comparable toxic effects to those seen in the epidemiological studies. Table 4 presents body burdens in experimental animals that produce an effect for which no comparable human epidemiological data are yet available. The current epidemiological database consists primarily of studies on adult male populations; few studies of women or children are available. Only effects seen at low body burdens in experimental anin chosen for this table to estimate end of the animal effect range; effi as thymic atrophy, the wasting sy or death are not included. The assumptions and data used to der value presented in Tables 1-4 are p in the appendix.

Results
Comparisons of the in vitro ef TCDD on animal and human ti cell lines are shown in Table 1. A of investigations have found the A tor present in humans to have a sir slightly lower binding affinity for than the Ah of many other species The concentration of TCDD req produce equivalent effects in ani human tissues is not significantly for responses as varied as indu CYPlAI in lymphocytes and th proliferation ( Table 1). For responses, the effective concentr TCDD differs in animal and hum by an order of magnitude or Cytotoxic effects induced by T organ cultures of developing palate concentrations 1000 times lower i tissue than in either human or r (48). Cultures of embryonic human and rat acts in palatal shelves respond at the same concentrations (48). Inhibition of lymphocyte ipendix proliferation and secretion of IgM in iotea mouse splenic lymphocytes requires 10 f times the concentration of TCDD comg pared to human tonsilar lymphocytes (50). h Comparisons of body burdens associated with in vivo effects demonstrate similar correlations between animals and humans. Body burden estimates in individuals with k chloracne vary by almost two orders of magnitude (Table 2). In subjects with chloracne, exposures resulted from either industrial or accidental poisonings. In k experimental animals, species differences in body burdens of TCDD that induce m chloracne vary by almost three orders of n magnitude, with the rabbit the most sensitive and the hairless mouse the least sensi- Xnzofuran; Body burdens in the general population were determined based on TCDD alone, the letter total PCDDs/PCDFs, and total PCDDs/PCDFs/PCBs ( Table 2). The average body burden of TCDD in the gendoses or eral population is approximately 1.1 ng ials were TCDD/kg body weight. The average body the low burden in the general population for total ects such PCDDs/PCDFs is 9 ng TEQ/kg body ndrome, weight and for total PCDDs/PCDFs/PCBs specific is 13 ng TEQ/kg body weight. rive each Rice oil contaminated with PCDFs and resented PCBs, among other contaminants, was ingested by men and women from Taiwan (Yu-Cheng incident); these individuals have been carefully studied since the poisoning ffects of incident (18,56,(63)(64)(65). Biochemical issues or changes in placentas from the women .number exposed during the Yu-Cheng incident are Lh recep-similar to the biochemical changes in rodent nilar but liver from animals exposed to TCDD. Near TCDD maximal downregulation of human placen- (42)(43)(44)(45).
tal epidermal growth factor receptor [uired to autophosphorylation occurs at similar body imal and burdens, as do comparable decreases in different hepatic epidermal growth factor receptor in ction of rats and mice ( u aThe data and methodology used to determine each value are presented in the appendix under the letter indicated. mice (61), respectively, which is within the range of background human body burdens of 13 ng TEQ/kg body weight. Disposition of dioxins is dose dependent in animals and humans (62). The body burden necessary for hepatic sequestration is similar for rats and humans (624. In animals, the body burden of TCDD that produces a carcinogenic effect ranges from 944 ng TCDD/kg body weight in mice (70) to 137,000 ng TCDD/kg in hamsters (68) ( Table 3). Body burdens in animals exposed to carcinogenic doses of TCDD are 73to 10,500-fold greater than background human TEQ body burdens. In epidemiological studies that indicate an association between TCDD exposure and increased incidence of cancer, body burdens were estimated between 109 and 7,000 ng TCDD/kg at the time of highest human exposure. Background human TEQ body burdens are approximately 8-540 times less than human TEQ body burdens estimated from the studies that associated dioxin exposure with increased cancer incidence.
Decreased birth weights were reported in children born to women exposed during the Yu-Cheng incident (18,56). These women were highly exposed and had an average body burden of approximately 2,130 ng TEQ/kg body weight. Body burdens of dioxins in experimental animals that decrease birth weight range from 400 to 2,000 ng TCDD/kg body weight in rats and hamsters (73,74) (Table 3).
Children of the Yu-Cheng mothers are not only smaller at birth but remain smaller throughout childhood compared to children of unexposed women (63). In rats, pups of dams exposed to 1,000 ng TCDD/kg body weight not only have decreased birth weights but consistently weigh less than controls up to 63 days of age, though they do recover upon reaching sexual maturity (75).
Behavioral effects after perinatal TCDD exposure have been observed in rhesus monkeys born to mothers exposed to approximately 5 ppt TCDD in the diet (66). Body burdens in the rhesus mothers were 42 ng TCDD/kg body weight, which is approximately 51 times less than the TEQ body burden in the Yu-Cheng women, but only 3.2 times higher than average TEQ body burden in the general population. Although some of the responses seen in experimental animals appear to occur in humans at similar body burdens, there are significant differences in the body burden estimates for decreased testosterone levels (41) and between human and animals. Based on these limited data, if decreased testosterone in humans is due to dioxin toxicity, then some humans may be approximately 280 times more sensitive than are rats for dioxin-induced decreases in testosterone.
Increased incidence of diabetes in populations exposed to dioxins has been reported in two studies with body burdens ranging from 99 to 140 ng TEQ/kg. While TCDDinduced diabetes has not been studied in experimental animals, there are reports of altered glucose homeostasis. Alterations in glucose uptake in adipocytes isolated from guinea pigs treated with TCDD occurs at body burdens 3-4 times lower than human populations with increased incidence of diabetes and altered glucose tolerance (24,25). Decreased serum glucose in rats occurs at body burdens 14-20 times higher than the increased incidence of diabetes and altered glucose tolerance in humans.
Environmental Health Perspectives * Volume 103, Number 9, September 1995 Table 4 presents estimated body burdens of TCDD in experimental animals from studies that report low-dose effects for which no comparable human studies are available. LOELs for decreased offspring viability/fetal viability vary from 345 ng/kg in monkeys to 18,000 ng/kg in hamsters. Alterations in lymphocyte subsets in juvenile marmosets is 10 ng TCDD/kg body weight (82,83). Enhanced viral susceptibility, as measured by increased mortality, occurs in mice at body burdens of approximately 10 ng TCDD/kg (84), which is equivalent to the body burden seen in unexposed humans and approximately twice the level in untreated mice. Effects such as increased incidence of endometriosis in rhesus monkeys (85) and decreased sperm count in offspring of rats treated with TCDD (74,86) occur at body burdens approximately five times that of unexposed human populations.

Discussion
A number of investigators have found the Ah receptor present in human tissues to have a similar, but slightly lower, affinity for TCDD than those receptors present in many other species (42)(43)(44)(45). For example, a recent study determined that the apparent binding affinity of TCDD to the Ah receptor ranged from 0.4 to 15 nM in 115 human placentas and from 1 nM in the TCDD responsive C57BI/6J mouse to 16 nM in the TCDD nonresponsive DBA/2 mouse. The binding affinity of TCDD to the Ah receptor is similar in mice, rats, hamsters, guinea pigs, and monkeys (87), and there is no obvious correlation between TCDD binding affinity to the Ah receptor and species sensitivity to the lethal or toxic effects of TCDD (87). Thus, our knowledge of the quantitative relationship between binding affinity and interspecies responsiveness does not provide adequate information to determine whether humans are more or less responsive than other species based solely on the binding affinity of TCDD to the Ah receptor.
Comparisons of human tissues or cell lines with similar animal tissues or cell lines demonstrate that from relatively simple responses, such as enzyme induction to more complex phenomena, such as cytotoxicity and proliferation, human tissue responds in the same manner as animal tissue and at similar concentrations (Table 1). These in vitro studies suggest that humans will respond to dioxin and that some of these responses may be adverse.
The doses of dioxins that produce lethality in experimental animals can vary by more than three orders of magnitude; guinea pigs are the most sensitive and ham-sters are the least sensitive (1-3). Because of this large variability in lethal effects, there has been an expectation that large species differences exist for all other effects. The data presented in the tables indicate that for a particular effect, some species may be extremely sensitive and some may be resistant, but many species respond at similar doses (i.e., within an order of magnitude). All experimental mammalian species examined respond to most of the adverse effects of dioxins at some dose. It is possible that humans may be resistant to some of the toxic effects of dioxins, but it seems highly unlikely, given the data currently available, that humans are refractory to all of the toxic effects of these chemicals.
Dioxins are unequivocally potent toxicants in experimental animals, yet the human health effects of exposure to these chemicals remain controversial. Comparisons of human and animal body burdens alone cannot prove a cause-and-effect relationship between toxicity and exposure in humans observed in an epidemiological study. However, this information can be used to increase or decrease our confidence that a particular adverse health effect observed in an epidemiological study was associated with the exposure to dioxins.
In addition, the present analysis required several assumptions in estimating both animal and human body burdens. These assumptions were required due to the lack of complete data on pharmacokinetics, toxic equivalency factors, species extrapolation, and, for humans, lack of information on daily dose or exposures. Hence, the information presented here can be used to direct research efforts to provide more accurate information on these topics.
There are some uncertainties associated with the assumptions used to estimate body burdens of dioxins in animals and humans. Unlike the experimental animal toxicology studies examined, humans are exposed to multiple chemicals. However, in the epidemiological studies, many of these chemicals interact with the Ah receptor as either agonists, partial agonists, or possibly antagonists. Assumptions of the relative potency of the chemicals and their distribution in the humans will result in uncertainties that are difficult to quantify given the present database. However, these uncertainties are likely well within an order of magnitude because body burdens of TCDD alone represent 10% of the total TEQ body burden due to all the PCDDs, PCDFs, and PCBs (Table 2).
Human body burdens are estimated using the TEF methodology. The TEF values derived by the U.S. EPA and the World Health Organization were based on scientific judgment as well as experimental data (37,38). In setting a TEF value, more weight was given to long-term, in vivo studies than to in vitro or acute in vivo studies (14,(36)(37)(38). In fact, although wide ranges of TEF values have been reported for specific congeners, the variability is within a factor of 10 when the in vivo data are used to set the TEF value (14,37,38).
The TEF methodology assumes additivity of toxic potential. The use of the TEF methodology has been validated for complex mixtures of chlorinated dibenzop-dioxins for effects such as enzyme induction and tumor promotion (88). The interaction of mixtures containing both dioxinlike and non-dioxinlike chemicals has not been studied as thoroughly. There are reports of antagonistic (89)(90)(91) and synergistic (92,93) interactions of dioxins and non-dioxinlike PCBs. The demonstration of nonadditive interactions increases the uncertainty of these values. Finally, the TEF scheme includes only full agonists of the Ah receptor. The use of TEFs and the assumption of additivity have been approved by both the World Health Organization and the U.S. EPA as a default, but interim, approach given the enormity of the task to test for all possible interactions of complex mixtures and in the relative absence of consistent data to the contrary (94). Clearly, the TEF values and assumptions regarding additivity need to be updated as more data become available.
Estimates of body burdens in animals and humans assume that the half-life of elimination of dioxins is a first-order process which is independent of the body burden or dose. There is significant evidence that disposition of TCDD is dose dependent (95)(96)(97). Induction of a binding protein in the liver has been proposed by Andersen et al. (98) to explain the dosedependent disposition of TCDD seen in experimental animals. Similar dose-dependent hepatic sequestration has been proposed in humans (62). These data suggest that elimination of these chemicals may not be a first-order process and the use of a single one-component half-life to estimate body burdens may not adequately predict these values.
Two different methods were used to estimate body burdens in experimental animals. One method involved classical pharmacokinetic calculations, and the second method used tissue concentration data presented in the papers. These methods resulted in similar body burden estimates for some cases where the appropriate data were available. For example, in mice receiving 1.5 ng TCDD/kg/day, estimated body burdens using classical pharmacokinetic Volume 103, Number 9, September 1995 * Environmental Health Perspectives Reviews * Body burdens of dioxins in humans and animals calculations were 14 ng TCDD/kg body weight and 23 ng TCDD/kg body weight using TCDD tissue concentrations. Body burden estimates from a tumor promotion study with rats receiving 125 ng TCDD/ kg/day produces estimates of 3615 ng TCDD/kg body weight using pharmacokinetic calculations and 2582 ng TCDD/kg body weight using TCDD tissue concentrations. These results suggest that the use of either method to derive body burdens will result in reasonably accurate estimates.
In estimating human body burdens, we assumed that dioxins distribute solely to the lipid portion of the body and that the concentration of dioxins in serum lipid is directly correlated to the concentration of dioxins in total body lipid. Several studies have demonstrated direct correlation between lipid-adjusted serum and adipose tissue concentrations of dioxins from human biopsy samples for the lower chlorinated dibenzo-pdioxins and dibenzofurans (30)(31)(32)(33). This relationship is not as certain for the higher (six or more chlorine substitutions) chlorinated analogs. Furthermore, in humans exposed to background levels of dioxins, the absolute or lipid-adjusted concentrations of CDDs and CDFs in adipose tissue and liver are not directly related and liver/fat ratios vary between 1.22 and 15.42 depending on the congener and possibly on dose (99). The highly chlorinated dibenzo-p-dioxins and dibenzofurans are found in greater concentration in the liver compared to the fat (liver/fat ratio 7.4-15.42). In the same samples, TCDD had a liver/fat ratio of approximately 2 (96). The human liver appears to accumulate these chemicals in greater proportion than adipose tissue, similar to what has been observed in experimental animals. In experimental animals, liver/fat concentration ratios are not only different for different compounds, but they are dose dependent. As the dose of dioxins are increased, so is the liver/fat ratio (95)(96)(97)(98).
Using the assumption that dioxins are equally distributed in the body lipid may underestimate the body burden of these chemicals due to chemical and dose-dependent sequestration in the liver. The magnitude of underestimation can be determined if several assumptions are used: that the liver/fat ratio for all dioxins is 15 and that liver is 10% of the body weight and is 10% lipid by weight. A liver/fat ratio of 15, as determined for the hexachlorodibenzofurans in humans, is used as a worst-case scenario for hepatic sequestration. Using these assumptions, the present estimate of dioxin TEQ body burdens in background populations will change from 13 to 21 ng TEQ/kg body weight. Hence, the assumption that dioxins are equally distributed in body lipid may slightly underestimate the body burdens of these chemicals, but the magnitude of error will be less than a factor of two. A better understanding of the pharmacokinetic properties for this class of compounds in humans is dearly indicated.
Chloracne has been described as the hallmark of dioxin toxicity in humans (17). Dioxin exposure in several animal species results in a chloracnegenic response and the body burdens which produce this response in animals are similar to the body burdens of dioxins in humans with chloracne. The chloracnegenic response has been thought to be a relatively high-dose phenomenon; however, the variation in human sensitivity to the chloracnegenic effects of TCDD is almost two orders of magnitude. For example, there are individuals who developed chloracne at body burdens approximately three times background (51). In contrast, there are subjects with body burdens of 1450 ng TEQ/kg body weight who have not developed chloracne (51). These data suggest that humans differ widely in sensitivity to the chloracnegenic actions of dioxins.
There are two points of caution when interpreting the chloracne data. First, human body burdens may not be an accurate measure of chloracnegenic potential if point-of-contact concentrations are important. For example, if dermal exposure results in a localized chloracnegenic response, body burdens estimated from serum or adipose tissue levels may not accurately reflect the concentration of dioxins at the site of effect. Also, the lack of chloracne in highly exposed patients does not necessarily indicate that these individuals are resistant to all the effects of dioxins. In mice, gene products, in addition to the Ah receptor, regulate the chloracnegenic response (100). It seems likely that multiple genetic factors may influence the relative susceptibility of individuals in a response-specific fashion.
Human responses to dioxins other than chloracne are not as obvious. In the Yu-Cheng poisoning incident, increased rates of toxic effects such as miscarriages, stillbirths, low birth weight infants, and developmental delays have been observed in offspring of women exposed to high levels of PCDFs and PCBs. However, it has been difficult to determine if the effects are due to the dioxins in the mixture, the nondioxinlike PCBs, or to the combination of these chemicals. Researchers have tried to correlate effects with serum concentrations of either the PCDFs or PCBs (56). Birth weights were negatively correlated with PCDF levels in these individuals (56). Other effects such as induction of arylhydrocarbon hydroxylase activity, a marker for CYPlAI, were not correlated with either the polychlorinated dibenzofurans or the PCB concentrations, but decreased placental EGF receptor autophosphorylation was correlated with total PCB concentrations (56). However, due to the nature of the exposure, patients with high levels of dibenzofurans will likely have high levels of PCBs, making such correlations difficult to interpret. Also, the presence of dioxinlike and non-dioxinlike PCBs adds to the complexity of these correlations.
We compared the body burdens of dioxins in the Yu-Cheng population to body burdens in experimental animals to determine the role of dioxins in the toxic effects seen in these individuals. Women who were pregnant at the time of exposure or became pregnant thereafter had children with lower birth weights compared to unexposed women, and the decrease in size persisted years after birth (63). Body burdens in the Yu-Cheng mothers were estimated at 2130 ng TEQ/kg. In experimental animals the body burdens that result in decreased birth weights range from 400 to 2000 ng TCDD/kg, while decreased growth occurs in rats at 1,000 ng TCDD/kg. The similarities between the body burdens in animals and humans suggests that dioxins may play a role in the decreased birth weights.
The behavioral effects of dioxins have not been thoroughly studied in experimental animals. One study reported deficiencies in object learning in rhesus monkeys prenatally exposed to TCDD. Delayed developmental milestones were seen in children born to Yu-Cheng mothers, but the body burdens are approximately 51 times higher in humans than in the monkeys. There is recent evidence that some of the non-dioxinlike PCBs may have neurotoxic actions (101). The absence of studies in experimental animals examining the developmental behavioral toxicity of dioxins makes it difficult to assess the role of either the dioxins or the non-dioxinlike PCBs in the developmental effects of the children of the Yu-Cheng patients.
In experimental animals, some biochemical changes produced by dioxins occur at lower body burdens than do the toxic effects (57-61,71). Induction of CYPlAl and decreased hepatic EGF receptor are two well-characterized biochemical responses to TCDD. Earlier studies comparing the induction of CYPlAl and decreased EGF receptor in human placenta and rat liver suggested that humans may be more sensitive when compared on a tissuedose basis (18). However, it is possible that the difference in sensitivity is not entirely due to species differences but due to altered tissue sensitivity. For example, induction of GCYPlA1 is similar in lung, liver, and skin of mice based on administered dose (102). In contrast, when the sensitivity of these tissues is compared on a tissue-dose basis, the lung is much more sensitive than the liver or skin (102). The present study indicates that humans and rats are equally sensitive to TCDD-induced biochemical changes when compared on a total body burden. Thus, when comparing the relative sensitivity of human or animal tissues to TCDD-induced biochemical changes, it may be more appropriate to compare body burdens than tissue concentrations. In addition, these data provide support for our approach.
TCDD is clearly carcinogenic in experimental animals. All species and both sexes of experimental animals that have been chronically exposed to TCDD exhibit a dose-dependent increased incidence of tumors (5). Several recent epidemiological studies have indicated an association between TCDD serum concentrations and increased incidence of tumors (19)(20)(21)(22)(23). Body burdens in rats and mice with increased tumors are comparable to the body burdens in the human cohorts that have increased incidence of tumors thought to be associated with dioxin exposure. Although these data are not conclusive, they are consistent with the hypothesis that exposure to TCDD was an important factor in the increased incidence of tumors in these cohorts. It is interesting to note that based on body burdens, mice are more sensitive to the carcinogenic effects of TCDD than are rats.
Carcinogenic responses are seen in hamsters, but the carcinogenic doses produce body burdens 46-1,300 times that seen either in humans, rats, or mice. Hamsters are insensitive to the lethal effects of dioxins, and they may also be less sensitive to the carcinogenic response. However, responses such as cancer are dose dependent as well as time dependent. Thus, the apparent differential sensitivity of the hamster may be due to differences in the dose-time regimens used in the hamster compared to the rat and mouse studies. It would be useful to compare these species under similar exposure protocols.
Decreases in serum testosterone have been reported in a National Institute of Occupational Safety and Health (NIOSH) cohort (41). There was a decrease in testosterone concentrations in individuals with serum concentrations of TCDD as low as 20 ppt at the time of tissue sampling, which is 3-4 times background TCDD levels and only a 33% increase over total average body burdens. Although the decrease in testosterone concentrations was statistically significant, the decrease was minor, and average levels were still within the normal range. In addition, a clear association between serum TCDD concentrations and effect was not readily apparent in the data (41). If differences in exposure patterns in the individuals are taken into account by back-calculating serum TCDD concentrations to the time of exposure, there is a clearer association between serum TCDD concentrations and lower testosterone concentrations. Here the lowest serum TCDD concentration associated with decreased testosterone concentration is 140 ppt (200 ppt TEQ). In experimental animals, high doses of TCDD decrease testosterone concentrations in rats at a body burden of 12,500 ng TCDD/kg body weight (73). These data suggest that some humans may be approximately 280 times more sensitive to the testosteronedecreasing effects of dioxins compared to rats. Alternatively, the decreased testosterone levels in the NIOSH cohort could be related to the concomitant exposure to other chemicals involved in the manufacturing process. Future studies examining the sensitivity of other species to the testosterone-decreasing effects of dioxins and epidemiological studies of other populations may provide additional information to adequately assess the association between dioxin exposure and decreased testosterone concentrations in some human populations.
Many of the effects of TCDD have been studied following an acute exposure in experimental animals. In contrast, humans receive low daily doses of these chemicals. One of the assumptions in extrapolating these effects to humans is that the effects are solely related to body burdens. For some of these endpoints, such as decreased testosterone, this assumption has not been adequately tested. Effects such as cancer are clearly related to both dose and time. It is possible that, in addition to dose and body burden, length of exposure may also have a significant effect on toxicity. Analysis of the area under the total body concentration-time curve may be a more appropriate marker for dose, and analysis of these data sets is ongoing.
The clinical significance of some of the endpoints studied is uncertain. Induction of CYPlAI and CYP1A2 by TCDD are some of the most sensitive markers of dioxin exposure, yet their relevance to toxicity is unclear. Recent studies have suggested an association between PAH exposure and CYPlA1/lA2 induction for lung and colorectal cancer and athelerosclerosis (103)(104)(105). However, these associations are speculative and not proven. At present, one could conclude that low doses of dioxins produce effects such as enzyme induction in experimental animals and that humans are exposed to levels of dioxins that induce CYPlA1/lA2 in experimental animals, but the relationship between these effects and disease are uncertain.
One of the most sensitive targets for TCDD toxicity in experimental animals is the immune system. Immune alterations, including increased viral sensitivity in mice and altered lymphocyte subsets in marmosets, have been reported at body burdens equivalent to human background exposures. However, the evidence for immunotoxicity of dioxins in humans is inconclusive. There are reports of subtle immune alterations in populations heavily exposed to dioxins. The incidence of intestinal and upper respiratory tract infections correlated with chloracne state and increased with increasing serum TCDD concentrations (106). One year after the Yu-Cheng poisoning episode, patients exhibited decreases in percentage of total T-cells, active T-cells, and T-helper cells, which recovered by the 3-year follow up study (107). Recent studies of occupationally exposed individuals with slightly elevated body burdens of approximately 72 ng TEQ/kg showed no alterations in lymphocyte subsets (108). However, in mice, a dose of TCDD that suppresses the antibody response to sheep red blood cells is not associated with alterations in lymphocyte subsets (109). Thus, immune function may be altered without altering lymphocyte subsets. Although some of these data suggest that the human immune system may be sensitive to the effects of dioxins, our present understanding of immunology does not support a conclusion that these alterations are or are not clinically significant.
The present study indicates that in vitro similar responses are seen in human and animal tissues after similar dioxin exposure. Human populations exposed to high concentrations of dioxins exhibit symptoms that are similar to the signs of toxicity seen in some experimental animals exposed to dioxins. These effects are seen at equivalent body burdens, strongly indicating that dioxins are responsible for some of these toxic effects in humans. For most of the toxic effects of dioxins, background exposure is well below those associated with overt toxicities. However, the background level used in this evaluation (13 ng TEQ/kg body weight) is an average background. Body burdens of dioxins appear to be log-normally distributed in humans (110), thus it would not be unusual to see populations with body burdens three to four standard deviations beyond the mean body burden. Recent studies in the Netherlands indicate that plasma TEQ concentrations in the 95th percentile of the Volume 103, Number 9, September 1995 * Environmental Health Perspectives population are twice that of the mean (113), suggesting that at least 5% of the population has two times the mean body burden. In addition, there are subpopulations such as subsistence fishermen who are likely to have much greater body burdens. There are also some toxic effects, such as endometriosis and increased viral sensitivity, which occur in experimental animals at body burdens less than 10 times the average background exposures to humans. Finally, human exposures that result in adverse health effects, such as chloracne, decreased birth weights, developmental delays, and cancer are 3-540 times the present average background exposure to these chemicals. Nevertheless, the available data indicate that high-level human exposure to dioxins produce adverse health effects and that humans are a sensitive species to the toxic effects of dioxins. Whether these lowdose effects are occurring in the general population or the more highly exposed subpopulations remains to be determined.

Appendix. Table Notes
(Some notes appear in more than one table.) Table 1 a) Apparent equilibrium binding dissociation constants are presented (42). Under conditions of infinite dilution, an apparent Kd of 9 pM has been determined for the Al) allele in the C57BI/6 mice; this value is close to the estimated true Kd (43). b) Splenic lymphocytes from C57BI/6 mice and peripheral blood lymphocytes were isolated, cultured, and exposed to TCDD. Ethoxyresorufin-O-deethylase (EROD) activity, a marker for CYPlAl, was determined following TCDD exposure (43). c) The authors (47) compared the cytotoxic effects of TCDD on organ culture of human, mouse and rat embryonic palatal shelves. Embryonic palates from human, mouse and rat were grown in the same organ culture system and exposed to TCDD. Cytotoxicity was detected using transmission electron microscopy. d) Thymocytes were isolated from either murine or human sources and cocultured with either murine (48) or human (49) thymic epithelium culture. The incorporation of tritiated thymidine into DNA was determined in cells treated with TCDD following antigen stimulation. e) Human tonsilar lymphocytes and murine splenic lymphocytes were used as a source of B-cells. Human and murine B-cells were grown under identical conditions and exposed to TCDD. Proliferation and IgM secretion were determined in response to different concentrations of TCDD ranging from 0.3 to 30 nM (50). Table 2 f) The lower value, 96 ng TEQ/kg body weight, is the body burden estimate of a patient with the lowest reported adipose dioxin concentration for any patient with chloracne (51). This individual was exposed to a mixture of CDDs and CDFs in 1969 and developed chloracne. At the time of exposure this individual had adipose tissue CDD/CDF concentrations of 419 ng TEQ/kg adipose tissue (51). An additional 17 ng TEQ was added to this value to include the PCBs. The values of dioxins at the time of exposure were estimated by the authors (51). The higher of the two values represents the average body burden of dioxins (TEQs) in individuals from Yusho with chloracne (52). Estimates of body burdens from these individuals were determined by Ryan et al. (52). g) Rhesus monkeys were administered 1 pig/kg TCDD, and it is assumed that essentially no TCDD was eliminated when the animal developed a chloracnegenic response. This is a LOEL dose; no lower doses were tested (53). h) Assumes the rabbit and the rat have the same rate of elimination, a half-life of 23.7 days (88) and that the rabbits weighed 2.5 kg throughout the experiment. This is a LOEL dose; no lower doses were tested (52).
i) Assumes the half-life of TCDD in mice is 11 days and that the mice weigh 25 g. This is a LOEL dose; no lower doses were administered (5). j) In highly exposed patients from the Yu-Cheng incident, there is a decrease in birth weights of children born from these patients compared to unexposed control populations (18,56). In addition, the Yu-Cheng mothers have altered levels of placental epidermal growth factor receptor (EGFR) and CYPlAI. The data indicate that the changes in placental EGFR and CYPlAl in these patients were maximal. Body burdens determined based on levels of 2,3,4,7,8-pentachloro-dibenzofuran (TEF = 0.5) and 1,2,3,4,7,8-hexachlorodibenzofuran (TEF = 0.1) in placenta tissue. Lipid content of the placenta is estimated at 1% (112) and the average percent body fat of a women is assumed to be 22%. These body burden estimates were also used as body burdens of Yu-Cheng mothers whose children demonstrate decreased growth (63) and delayed developmental milestones (64,65). k) In a rat liver tumor promotion study, rats initiated with diethylnitrosamine were exposed to doses of TCDD from 3.5 to 125 ng/kg/day. Statistically significant increases in numbers of altered hepatic foci were observed in rats treated with 125 ng TCDD/kg/day (62). At the end of the study, liver concentrations of TCDD were approximately 20 ppb (60); assumes 20% body weight is adipose tissue and that at this dose, the liver has three times the concentration of TCDD than adipose tissue. Body and liver weights were reported (67) for these animals. The body burden calculation assumes that liver and fat account for 85% of the body burden in these animals. For tumor promotion, 125 ng TCDD/kg/day is the LOEL and 35 ng TCDD/kg/day is the NOEL for tumor promotion (67). For induction of CYPlAl (60) and downregulation of EGFR (59), 125 ng TCDD/kg/day was assumed to produce a maximal response. 1) Mice were administered 10 pg TCDD/ kg and sacrificed 7 days after treatment. EGFR binding was determined in hepatic plasma membrane (58 Table 3 s) Estimated highest body burden at time of last exposure. Calculations based on measured TCDD levels in serum (lipid adjusted) and assuming a first-order elimination kinetics and a half-life for elimination of 7.1 years. Also assumes a body weight of 70 kg and 22% body fat. Calculations for estimated serum concentrations at last time of exposure performed by authors (18,23). t) Animals administered 100 pg TCDD/kg 6 times every 4 weeks over a 24-week period; assumes a half-life of 14.9 days (111). Body burdens are estimated immediately after the last treatment with TCDD. The administration of 50 pg TCDD/kg 6 times every 4 weeks over a 24-week period did not increase the incidence of any types of tumors in 10 hamsters (68). u) Assumes a single first-order elimination rate constant and a half-life for the whole body elimination of 23.7 days (85) and a gastrointestinal tract absorption of 86% (85). Increased incidence of hepatocellular carcinomas were observed at 100 ng/kg/day and 10 ng/kg/day is the NOEL (69). Decreased testis weight and testosterone concentrations were observed after 12.5 ng TCDD/kg 7 days later (76). Decreased serum glucose levels were observed in rats treated with 100 ng/kg/day for 30 days (79).
v) Assumes an apparent half-life of 11 days and a body weight of 20 g. Mice receiving 71.4 ng/kg/day for 2 years had a statistically significant increase in hepatocellular carcinomas (7().
w) Assumes neonatal rats and hamsters are exposed to an equal dose of TCDD as are the dams on a weight basis and assumes all alterations are due to the neonatal exposure. For decreased body weight in pups 400 ng/kg is the LOEL; a dose of 64 ng/kg to the dam was the NOEL for this response (73). For decreased sperm count the LOEL is 64 ng/kg and no lower doses were tested (86). In hamsters only one dose was tested (2000 ng/kg) for decreased sperm counts (74). Decreased growth in rats is indicated by decreased body weights up to postnatal day 63 (75). The incidence of fetal mortality was increased in hamsters at a dose of 18 pg/kg but not at a dose of 6 pg/kg (81). x) Assumes a single first-order elimination rate constant and a half-life for the whole-body elimination of 400 days (81) and a gastrointestinal absorption of 86% (88). This is the LOEL from this study; no lower doses tested. Monkeys exposed to a diet of approximately 5 ppt had a daily intake of 0.151 ng/kg/day. Monkeys exposed to approximately 25 ppt in the diet had a daily intake of approximately 0.76 ng/kg/day. For animals with decreased object learning, the TCDD-exposed offspring were born after 16.2 months of maternal TCDD exposure of a diet of 0.151 ng TCDD/ kg/day. Animals with increased incidence and severity of endometriosis had a daily intake of 0.151 ng/kg/day for 4 years, and body burdens were determined at the end of the exposure period. Monkeys exposed to 0.76 ng TCDD/ kg/day for 16.2 months had significant decreases in offspring viability. y) The authors extrapolated serum concentrations of TCDD at the time of sampling to initial exposures (41). Workers with serum TCDD concentrations of 140-496 ng/kg (lipid adjusted) have a greater incidence of low testosterone concentrations (41). Extrapolation assumed a half-life for TCDD of 7.1 years. To estimate body burdens in these workers, it was assumed that the background TEQ was 60 ng/kg, thus the total serum TEQ was 140 ng TCDD/kg + 60 ng TEQ/kg = 200 ng TEQ/kg (lipid adjusted). z) Assumes that high-exposed group (>33 ng/kg) had a background of 60 TEQ ng/kg. This group had at least 93 TEQ ng/kg. Assumes average subject was male, weighing 70 kg with 22% body fat. aa)Workers with increased glucose tolerance and diabetes have serum levels of 640 ppt TEQ (24). bb) Guinea pigs received 30 ng TCDD/kg intraperitoneally and sacrificed 24 hr after dose. Assumes that no TCDD was eliminated at this time. This is a LOEL, no other doses tested (78). Table 4 cc) Assuming a single first-order elimination rate constant and a half-life of 6-8 weeks. Body burdens calculated by authors (82). Animals treated with a single dose of TCDD were tested 2 weeks after treatment (83). dd) Mice were treated with TCDD and challenged with influenza virus 7 days later (84). ee) Mice were administered 100 rig/kg and examined 30 days after receiving the treatment (77).
GW. Decreased human birth weights after in utero exposure to PCBs and PCDFs are associated with decreased placental EGF-stimulated receptor autophosphorylation capacity. Mol Pharmacol 32:572-578 (1987). 57. Sewall CH, Lucier GW, Tritscher AM, Clark GC. TCDD-mediated changes in hepatic epidermal growth factor receptor may be a critical event in the hepatocarcinogenic action of