Noncancer End Points
This chapter reviews the U.S. Environmental Protection Agency (EPA) assessment of the noncancer end points, including immune function, reproduction and development, diabetes, thyroid function, lipid levels, and other effects related to exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD, also referred to as dioxin), other dioxins, and dioxin-like compounds (DLCs) in animals and humans. The purpose of this chapter is to critically assess, to the extent possible, whether EPA has met the criteria set forth in the “Statement of Task” with respect to the noncancer effects of TCDD, other dioxins, and DLCs. Toward this end, this chapter focuses on the uncertainties and assumptions made by EPA in determining whether TCDD, other dioxins, and DLCs have noncancer effects in humans; determining the methods and models used for assessing these effects; determining the breadth and robustness of the studies used and the balance with which the studies are presented in the Reassessment,1 and finally, determining whether the conclusions reached by EPA are consistent with the current scientific peer-reviewed literature.
EPA uses a sizeable immunotoxicology database derived largely from laboratory animal studies and a smaller number of epidemiological studies in
its assessment of immunotoxicity produced by TCDD, other dioxins, and DLCs.
The assessment of changes in immune competence, regardless of the cause, is complex, as the immune system is composed of a large and diverse group of cellular and soluble components. In addition, compensatory and overlapping mechanisms in host immunity can make it difficult to identify subtle or modest changes within the immune system.
Because of the many different cell types and soluble factors, which alone or cooperatively mediate a wide variety of evocable immunological responses, there is no one test or assay that can measure all the different elements. Therefore, the approaches used to identify changes in immune status are multifaceted, including pathological examination of lymphoid organs, enumeration of leukocyte subpopulations, quantification of soluble mediators, and measurement of immune function responses, such as susceptibility to infection or reduced immunological responses to vaccines. Standard assays are available for all the above determinations; however, because of the sheer enormity of the task to measure them all, immune competence is typically assessed by using a small number of immunological end points, often quantifying various aspects of innate, humoral, and cell-mediated immunity. Therefore, the EPA review draws on a large but diverse database of studies that are often difficult to compare because different assays, model systems, responses, and animal species were used.
EPA draws several important conclusions about the immunotoxicity of TCDD, other dioxins, and DLCs that are summarized in the Reassessment, Part III, Integrated Summary and Risk Characterization. The first is that “there appears to be too little information to suggest definitively that 2,3,7,8-TCDD at levels observed (in the reported studies) causes long-term adverse effects on the immune system in adult humans” (p. 2-34; lines 19 to 21). The second is that “cumulative evidence from a number of studies indicates that the immune system of various animal species is a target for toxicity of TCDD and structurally-related compounds, including PCDDs, PCDFs [polychlorinated dibenzofurans] and PCBs [polychlorinated biphenyls]” (p. 2-34; lines 24 to 26). Third, animal studies show that TCDD suppresses both “cell-mediated and humoral immune responses, suggesting that there are multiple cellular targets within the immune system that are altered by TCDD” (p. 2-34; lines 26 to 28). EPA goes on to state that “it can be inferred from the available data that dioxin-like congeners are immunosuppressive” in animals. Fourth, the weight of evidence from animal studies in vivo and in vitro “supports a role for Ah-mediated immune suppression” by DLCs (p. 2-35, lines 27 to 28); “other in vivo and in vitro data, however, suggest that non-AHR (aromatic hydrocarbon receptor)-mediated mechanisms may also play
some role in immunotoxicity” (p. 2-35, lines 28 to 30). Finally, EPA concludes that “there are insufficient clinical data from these studies to fully assess human sensitivity to TCDD exposure. Nevertheless, because of extensive animal work, the database is sufficient to indicate that immune effects could occur in the human population from exposure to TCDD and related compounds at some dose level. At present, it is EPA’s scientific judgment that TCDD and related compounds should be regarded as nonspecific immunosuppressants and immunotoxicants until better data to inform judgment are available” (p. 2-37, lines 4 to 10). The strengths and weaknesses of those conclusions are discussed below.
Uncertainties and Assumptions in Determining Whether TCDD Is Immunotoxic in Humans
Is the Assumption Correct That the Immune System in Humans and in Animal Models, Primarily Mice, Are Similar?
Historically, the mouse has been the animal model of choice for immunologists; it has also been widely embraced as the model of choice for immunotoxicological studies. Hence, most immunotoxicological studies used by EPA in preparing its report are based on studies in mouse models. The human and mouse immune systems are similar in composition and function; therefore, from the standpoint of comparisons based on the composition of this target organ, it is reasonable to assume that studies in mice provide important qualitative insights into the mechanism of action of TCDD, other dioxins, and DLCs on the human immune system. Providing, insights into the mechanism of action is one of the primary strengths of the mouse model for which there are genetically defined AHR high- and low-responding mouse strains, congenic mice at the Ahr locus, and AHR null (Ahr-/-) mice. More reagents, assays, and biological probes are available for the mouse immune system than for any other species except the human immune system. However, as is the case for other toxicological end points, information-derived from animal studies is qualitative in that the pharmacodynamics, pharmacokinetics, half-lives of the compounds, affinity of AHR, linkage of the receptor to signal transduction pathways, and numerous other factors can be significantly different, at least quantitatively, between humans and other animal species. Therefore, direct quantitative extrapolations from animal models to humans can result in a significant underestimation or overestimation of risk. When strong scientific evidence exists concerning specific species differences, these factors should be incorporated into the risk characterization.
Is the Toxic Equivalency Factor/Toxic Equivalent Quotient Approach for Estimating the Immunotoxicity of Mixtures Scientifically Justified?
The toxic equivalency factor/toxic equivalent quotient (TEF/TEQ) approach is based on a well-defined structure-activity relationship for persistent dioxins, other than TCDD, and DLCs for which there is a positive correlation between AHR affinity and toxic potency. There is good general agreement in comparisons of results across studies, primarily in mice, where the acute immunotoxic effects for individual dioxinlike congeners were examined, suggesting that the immunotoxic potency for various congeners and AHR activation exhibit the same qualitative rank order. Few immunotoxicological studies investigated structure-activity relationships for immunotoxic potency. Of those studies, with only a few exceptions, the rank order immunotoxic potency correlated positively with AHR activation. Several exceptions to this relationship are found in the current literature, and some are discussed in the Reassessment, Chapter 4, Part II. For example, 2,7-dichlorodibenzo-p-dioxin, a congener that would be predicted to exhibit low binding affinity for AHR, was found to be equipotent in suppressing the anti-sheep red-blood-cell (anti-SRBC IgM) antibody-forming response to TCDD. For this example, the TEF/TEQ approach would not provide a reasonable estimate of immunotoxicity. Because in vivo immunotoxicity data for 2,7-dibenzo-p-dioxin are available only in the mouse, it is unclear whether the unexpected potency of this congener occurs in other animal species and humans. A second example pertains to certain halogenated aromatic hydrocarbons (HAHs), including several diortho PCB congeners, which exhibit antagonist activity when administered in a mixture with AHR agonists. The latter point may not be trivial, as several of the diortho PCB congeners (e.g., PCB153) are abundant environmental cocontaminants with dioxins, other than TCDD, and DLCs. The aforementioned examples of exceptions for which the TEF/TEQ approach might not predict toxic potency accurately are presented and discussed in a balanced manner (Part II, pp. 4-6 to 4-7, lines 19 and 20).
In spite of the aforementioned caveats, based on what is known about the cell biology of AHR and the mechanisms of immunotoxicity for TCDD and related compouds, the TEF/TEQ approach for assessing the immunotoxic potency of mixtures of persistent dioxins, other than TCDD, and DLCs is scientifically justified. Having said that, the effective application of the TEF/TEQ approach for assessing immunotoxic potential ultimately would critically depend on the TEF values assigned to individual congeners for a given immunological response. What is unclear is which immune responses should be used in risk management, as all are not equally sensitive to modulation by HAHs.
What Is the Profile of Immune Toxicity in Humans Exposed to DLCs?
Well-documented human exposure to TCDD, other dioxins, and DLCs has occurred in the occupational setting and within the general population from industrial accidents and through consumption of contaminated food. Even though immune function and status have been examined in exposed individuals, only a small number of studies have been conducted with appropriate controls and accurate measures of exposure. Likewise, a small number of epidemiological investigations evaluating immunologically related outcomes from chronic exposures have been reported. The results from these human studies for the most part have yielded inconclusive results. In several studies, modest changes in immune status were observed; yet in other studies, the findings were not reproduced or were even contradicted. In animal studies, it is clear that TCDD, other dioxins, and DLCs markedly suppress both humoral and cell-mediated immune responses. This profile of activity has not been unequivocally demonstrated in humans. The most obvious reason for the inconclusive findings in human studies is that, in many cases, a very small number of subjects were evaluated, their primary exposures were often long before measurement of immune end points and concentrations of TCDD and related compounds, and the nature of those exposures were often unclear. Furthermore, obtaining an accurate estimate of the level of exposure through back-extrapolation from current body levels may be more complex than simply using a single half-life throughout.
Another contributing factor to the inconsistent results is that immune responses to defined stimuli are highly variable among humans. This variability can be attributed to genetic variability, age, environmental history, and other still undefined causes. This variability in individuals substantially limits the ability to identify subtle and even moderate alterations of immune function after exposure to agents, especially in human populations. Therefore, in the absence of more comprehensive immunotoxicological human data, it is reasonable to assume that TCDD, other dioxins, and DLCs will exert a profile of immunotoxicity comparable to that observed in animal models, such as mice. For risk analysis, the more critical issue is whether the immunotoxic potency observed in certain animal models is significantly greater (by orders of magnitude) than in humans.
What Is the Sensitivity of the Human Immune System to DLCs?
For many of the same reasons as discussed in the previous section, the sensitivity of humans to immune suppression by TCDD, other dioxins, and DLCs is also currently unclear. There are four separate reports of a longitudinal study in a cohort of Dutch children suggesting that the developing
human immune system may be susceptible to immunotoxic alterations from exposure to Western European environmental levels of TCDD, other dioxins, and DLCs (Weisglas-Kuperus et al. 1995, 2000, 2004; ten Tusscher et al. 2003). Exposures before 1990 in the Netherlands resulted in breast milk concentrations of TEQ (at the start of the study) at 30 to 60 parts per trillion (ppt) (lipid), whereas concentrations in the United States at approximately the same time period were in the 15- to 24-ppt range. Three industrial areas were compared with a rural area with about 20% less PCBs in maternal plasma (Koopman-Esseboom et al. 1995). The Dutch study measured the major PCB congeners in plasma in the mother and the newborn and all TCDD and related compounds in maternal breast milk 10 days and 3 months postpartum (Feeley 1995). About half the study subjects were fed a formula that had TCDD and related compounds at less than 2 ppt, and the other half of the test subjects were breast-fed and could be divided into groups breast-fed less than 4 months and those breast-fed more than 4 months. Therefore, individual calculations of total exposure could be correlated with plasma PCB concentrations in children at 3 months, 18 months, 42 months, and 9 years. Three more recent studies (Weisglas-Kuperus et al. 2000, 2004; ten Tusscher et al. 2003) are important because some of the same findings observed in these later studies were also reported in the 1995 studies, indicating persistence of alterations. Weisglas-Kuperus et al. (2000) reported on 207 healthy mother-infant pairs with increased prenatal exposure to TCDD, other dioxins, and PCBs; the results showed an association between exposure and immunological changes, which included an increase in number of lymphocytes, γ-δ T cells, CD3+HLA-DR+ (activated) T cells, CD8+ cells, CD4+CD45RO+ (memory T cells), and lower antibody levels after mumps and measles vaccination at preschool age. In addition, an association was found between prenatal exposure and decreased shortness of breath with wheeze, and current PCB burden was associated with a higher prevalence of recurrent middle-ear infections and chicken pox and a lower prevalence of allergic reactions. Although an association between TCDD and PCB exposure and changes in immune status was observed, all infants were found to be in the normal range. In a second study, ten Tusscher et al. (2003) reported modest but persistent changes in immune status in children with perinatal exposure to dioxin as evidenced by a decrease in allergy, persistently decreased thrombocytes, increased thrombopoietin, increased CD4+ T cells, and increased CD45RA+ cell counts in a longitudinal subcohort of 27 healthy 8-year-old children with documented perinatal exposure to TCDD. The original cohort at 42 months demonstrated an association between reduced vaccine titers, increased incidence of chicken pox, and increased incidence of otitis media with higher TEQ. However, by 8 years of age, the more frequent recurrent ear infections were still apparent
(overall), although the chicken pox frequency showed an inverse correlation with PCB and TCDD concentrations. The subcohort in the ten Tusscher study was small, and therefore the results are not as robust as those in the Weisglas-Kuperus et al. (2004) study, which included almost 91% of the original 2000 study cohort.
Animal studies also suggest that the developing immune system is sensitive to persistent changes in immune function or status, especially when exposure occurs in the perinatal and neonatal stage and especially in T-cell-mediated immunity. Less compelling studies exist from which to estimate the sensitivity of the adult human immune system.
EPA concludes that
there is insufficient clinical data from these studies to fully assess human sensitivity to TCDD exposure. Nevertheless, based on the results of the extensive animal work, the database is sufficient to indicate that the immune effects could occur in the human population from exposure to TCDD and related compounds at some dose level. At present, it is EPA’s scientific judgment that TCDD and related compounds should be regarded as nonspecific immunosuppressants and immunotoxicants until better data to inform judgment are available. (Reassessment, Part III, p. 2-37, lines 4 to 10)
Indeed, based on the extensive animal data, it is reasonable and prudent for EPA to regard TCDD as an immunotoxicant. Furthermore, the Dutch study provides some suggestive evidence for this conclusion. However, it is unclear in EPA’s conclusion what is meant by TCDD and related compounds being regarded as “nonspecific immunosuppressants and immunotoxicants” and this should be clarified.
How Persistent Are the Immunotoxic Effects of TCDD on the Human Immune System (Reversible Versus Irreversible Changes)?
Because it has not yet been unequivocally established that TCDD induces immune suppression in humans, it is not possible at this time to delineate the persistence of TCDD-mediated immunotoxicity in humans. The lowered total white-blood-cell numbers reported in the studies of Dutch children (Ilsen et al. 1996) were no longer evident 2 years after birth (Weisglas-Kuperus et al. 2000). The elevated T-cell subpopulations in Dutch children at 42 months of age did not appear to persist at 8 to 10 years of age (ten Tusscher et al. 2003). However, Weisglas-Kuperus et al. (2004) also reported a positive association in children (3 to 7 years of age) between increased postnatal PCB exposure and increased prevalence in recurrent middle ear infections. In addition, there was a positive association between increased prenatal PCB exposure and decreased chicken pox frequency as well as allergy and asthma. A second recent report (Van den Heuvel et al.
2002) suggested that TCDD, other dioxins, and DLCs may produce subtle but persistent changes in immune status as evidenced by a reduction in allergy and asthma. In this study of 200 Flemish adolescents, a negative correlation was observed between exposure to TCDD, other dioxins, and DLCs with TCDD TEQ measured and allergic responses in airways. In addition, serum immunoglobulin G levels were also negatively correlated with PCB exposure.
Relevance of Rodent Models to Human Quantitative Risk Assessment for Immunotoxicity
In the Reassessment, Part III, Appendix A, Table A-1, EPA centers its risk characterization for adult immunological effects on four studies conducted in mice (Vecchi et al. 1983; Narasimhan et al. 1994; Smialowicz et al. 1994; Burleson et al. 1996) and its developmental immunotoxicological risk characterization on a single rat study (Gehrs and Smialowicz 1999) (see Appendix A, Table A-1; for immune end points; see Figures 12 to 15, pp. A-7 to A-9; same studies identified in Table 5-6, p. 5-39, in the Reassessment, Part III). Based on these studies, lowest-observed-adverse-effect levels (LOAELs) and no-observed-adverse-effect levels (NOAELs) are used to derive ED01 (1% effective dose) and human equivalent intake values. Because of the importance that these studies have to the Reassessment and the potential importance that the derived values may have for risk management, some additional comments are provided here.
The study of Burleson et al. (1996) showed the lowest LOAEL (6 ng/ kg), and NOAEL (3 ng/kg) values, which result in calculated “human equivalent intakes” of 1 pg/kg/day and 2 pg/kg/day, respectively. The study showed that the sensitivity to alteration by TCDD of host resistance of mice to H3N2 influenza A (Hong Kong/8/68) virus is strikingly sensitive compared with other LOAELs for immunotoxicological end points in adult rodents given in Table A-1, the LOAEL ranging from 100 to 1,200 ng/kg. In fact, the sensitivity to TCDD in the Burleson et al. study is striking even when compared with similar and more recently published host-resistance studies using influenza virus. For example, Nohara et al. (2002) showed that TCDD doses up to 500 ng/kg did not increase mortality in a number of different strains of mice, including B6C3F1, C57Bl/6, Balb/c, and DBA/2 mice infected with influenza A virus (A/PR/34/8, H1N1). More mice per group were used in the Nohara study than in the Burleson study, thus providing even greater statistical power. In a study using influenza A/ HKx31, Warren et al. (2000) reported that in certain experiments, TCDD treatment (1 to 10 µg/kg) increased mortality, whereas in other experiments no mortality was observed. Furthermore, Warren and coworkers stated that in some experiments TCDD doses as high as 10 µg/kg produced no
mortality, whereas 80% mortality was observed at the same dose in other experiments. Such data emphasize the variability typically observed in host-resistance studies. The reason for the significantly greater sensitivity to TCDD in the Burleson et al. (1996) study is unclear but strongly suggests that further studies are needed before using results from this study for risk characterization.
The LOAEL, NOAEL, ED01, and human equivalent intake values for immunotoxicity are based on suppression by TCDD of the anti-SRBC IgM antibody-forming cell response in studies by Vecchi et al. (1983), Narasimhan et al. (1994), and Smialowicz et al. (1994) (the other three studies identified in Table A-1 and used for risk characterization of the adult immune system). Numerous laboratories have demonstrated suppression of the antibody-forming cell response by TCDD, and in general, there is good concordance in the ED50 doses (600 to 770 ng/kg) derived from these studies (see Table 4-1, p. 4-38) (Vecchi et al. 1980; Davis and Safe 1988; Kerkvliet and Brauner 1990; Kerkvliet et al. 1990). In contrast, some variability in the LOAEL values identified in Table A-1 were observed in three other studies: 100 ng/kg (Narasimhan et al. 1994), 300 ng/kg (Smialowicz et al. 1994), and 1,200 ng/kg (Vecchi et al. 1983). The variation is due primarily to dose selection in each of the studies. It is clear how the LOAEL and, in the case of the Narasimhan study, the NOAEL were identified from the data presented in each of the published reports. It is not clear, however, how the ED01 values were calculated.
Several concerns also exist about using the Gehrs and Smialowicz (1999) study for characterizing developmental immunotoxicological risk by TCDD. The Gehrs and Smialowicz study gives no indication as to the number of rat offspring studied; therefore, it is unclear whether the results are robust. In addition, both males and females were found to be more sensitive to immune suppression by TCDD after 14 months of age than at 4 months of age, as measured by a delayed type hypersensitivity response, which is somewhat puzzling. Although hypotheses could be advanced to explain these unexpected findings, it would be valuable and prudent to repeat that study before using those results for characterizing developmental immunotoxicological risk by TCDD.
Congruence with Full Document
In Part II, Chapter 4 of the Reassessment, a comprehensive and balanced synthesis of the immunotoxicology literature on TCDD, other dioxins, and DLCs is presented. Results from more than 200 published studies are discussed in an organized and logical manner. Moreover, there is good congruence between Chapter 4 and section 2.2.3 on immunotoxicity in the Executive Summary.
CONCLUSIONS AND RECOMMENDATIONS ON THE IMMUNOTOXICITY OF TCDD, OTHER DIOXINS, AND DLCS
Present clinical findings are inconclusive about whether or in what way TCDD, other dioxins, and DLCs are immunotoxic in humans, a conclusion that EPA acknowledges, and human data are also sparse. Perhaps the most compelling data that these compounds are human immunotoxicants, at possibly relevant environmental levels, come from the studies of the Dutch children’s cohort. These studies show an association between prenatal exposure and changes in immune status. However, the effects are modest and do not lie significantly outside the full range of normal. The correlation of increased otitis media in the very young with perinatal TEQ is the only statistically significant immunological clinical finding. Some of the same findings were made in acutely exposed Taiwanese and Japanese cohorts (Yu et al. 1995). Concordant with findings in Dutch children are a number of animal studies that also suggest that the developing immune system is especially sensitive to modulation by TCDD, other dioxins, and DLCs. Collectively, in light of the large database showing that these compounds are immunotoxic in laboratory animal studies together with sparse human data, EPA is being prudent in judging TCDD, other dioxins, and DLCs to be potential human immunotoxicants in the absence of more definitive human data.
EPA’s conclusion that TCDD, other dioxins, and DLCs are immunotoxic at “some dose level” by itself is inadequate. At a minimum, a section or paragraph should be added that discusses the immunotoxicology of these compounds in the context of current AHR biology. Specifically, there is evidence showing that the affinity of TCDD for the human AHR is at least an order of magnitude lower than that in high-responding Ahrb-1 mouse strains (Ramadoss and Perdew 2004), which has been the most commonly used animal model for investigations of immunotoxicity of TCDD, other dioxins, and DLCs. Other properties of AHR, in addition to binding affinity, such as specificity for target genes and transactivation potential, will contribute to the toxicity produced by AHR ligands. Nevertheless, EPA supports a TEF/TEQ approach for estimating the immunotoxic potency of mixtures of dioxins, other than TCDD, and DLCs. The Reassessment assumes that immunotoxicity is therefore primarily mediated through an AHR-dependent mechanism, so some discussion should be included acknowledging the possibility that rodents, especially certain mouse strains expressing Ahrb-1 might be significantly more sensitive to the immunotoxic effects of TCDD, other dioxins, and DLCs than humans. Some discussion should also be included on the strengths and weaknesses of using genetically homogeneous inbred mice to characterize immunotoxicological risk in the genetically variable human population. Expanding the
discussion to include the above crucial points would provide additional balance to Part III of the Reassessment.
EPA centers its risk characterization for adult immunological effects on four studies conducted in mice (Burleson et al. 1996; Smialowicz et al. 1994; Narasimhan et al. 1994; Vecchi et al. 1983), and its developmental immunotoxicological risk characterization on a single rat study (Gehrs and Smialowicz 1999) (see Part III, Appendix A, Table A-1; for immune end points; see Figures 12 to 15, pp. A-7 to A-9). Concerns about Table A-1 are the following:
The calculations of ED01 values and the scientific assumptions made in deriving those values need further clarification. Likewise, EPA should provide a clear scientific rationale for selecting ED01 as a benchmark dose.
Considerations of the Burleson et al. (1996) study with no consideration of two similar studies—Nohara et al. (2002) and Warren et al. (2000)—that yield very different results requires justification.
On the basis of concerns discussed earlier, it would be prudent to replicate the Gehrs and Smialowicz (1999) study before using its results for characterizing developmental immunotoxicological risk of TCDD.
An important animal study by Oughton et al. (1995) was not included in either Part II or the tables in the Executive Summary of the Reassessment. The importance of the study is that it is the only low-level chronic exposure investigation published (TCDD at 200 ng/kg/week once a week to Bb mice at 2 to 16 months of age) in which immunotoxicological parameters have been assessed—specifically, a phenotypic analysis by flow cytometry of major cell subpopulations in the mouse spleen, thymus, and peripheral blood. The study showed only subtle alterations in the immune system as demonstrated by a modest increase in γ-δ T cells, which the authors considered “questionable biological relevance,” and a small decrease in the frequency of memory CD4 cells (by phenotype). However, these changes, although of questionable biological relevance, have also been observed in humans and in high-exposure animal studies.
REPRODUCTION AND DEVELOPMENT
EPA provided an overview of the effects of TCDD, other dioxins, and DLCs on development and reproduction based on published animal studies and accidental human exposures. Determination of the alterations in development and reproduction is a highly complex process because hormonal as well as intracellular processes and compensatory mechanisms, including hormonal feedback mechanisms, are affected. The Reassessment compre-
hensively covers developmental aspects in a wide variety of models. Two major rodent models have been used to study the effects of TCDD on reproduction and development. In the first model, TCDD is given during pregnancy (an in utero and lactational exposure model). This model tested the ability of TCDD to disrupt development of the pups and assessed the effects on reproduction and reproductive behavior later in life. A comprehensive overview of the in utero and lactational exposure model is presented, but the doses used and how the model relates to the human reproductive and developmental toxicity are not emphasized. For example, maternal concentrations of TCDD in plasma are needed at designated times during pregnancy and lactation in the rat dam; those data would allow comparison to human data and determination of whether concentrations in rodents are higher or lower than those in humans accidentally exposed to TCDD. In the second model, adult rats and an immature gonadotropinprimed model are used to assess the effects of TCDD on ovulation. These models were not adequately discussed. In addition, there is uncertainty in the risk assessment based on differences in TEF reported by the World Health Organization (WHO) (Van den Berg et al. 1998) when compared with published data. For example, the WHO 1998 TEQs appear to be 2.5-fold higher (Van den Berg et al. 1998) than actual potency data determined with these models (Safe 1990; Gao et al. 1999, 2000a,b). Studies have been conducted using the 1998 TEFs (e.g., Hamm et al. 2003), and the conclusions seemed to indicate that mixture doses two to three times higher than the calculated TEQ appeared to be required to elicit the same alterations. The study by Hamm et al. is comprehensive and revealed numerous adverse effects on male and female reproduction, such as prolonged time to puberty, decreased seminal vesicle and ventral prostate weights, and, in the female, increased the incidence of vaginal threads. The lowered responses to the mixture of TCDD, other dioxins, furans, and coplanar PCBs were attributed to decreased transfer of mixture components to the offspring, whereas a miscalculation of the TEQ might have also contributed to the lowered response (Hamm et al. 2003). In addition, if the mixture was altered to favor what might have been present in the diet in nature, then the true TEQ might not have been accurately represented in the Hamm et al. study. However, given that the WHO TEFs are order-of-magnitude estimates of the relative potency of a chemical and derived from all toxicological outcomes in a variety of species, it is not surprising that there is a lack of absolute concordance between a calculated TEQ and an actual TEQ, measured in one species, for male and female reproduction end points. The generation of end-point-specific TEFs would probably resolve the observed differences between calculated and measured TEQs on these end points.
Examples of complete dose responses on various reproductive parameters in females using various polychlorinated dibenzo-p-dioxins (PCDDs),
polychlorinated dibenzofurans (PCDFs), and PCBs are given below. Trace levels of PCDDs have been detected in fish, wildlife, and humans (Van den Berg et al. 1985; Tiernan et al. 1985), and PCDDs have toxicological effects on the reproduction and development in vertebrates, including rodents and nonhuman primates (Van den Berg et al. 1994). Few studies have evaluated the effects of complex mixtures of PCDDs and DLCs on the female reproductive system. Previous studies have validated the TEQ concept for several PCDDs in acute and subchronic/chronic experiments using several biological end points (Stahl et al. 1992, Weber et al. 1992, 1994; Rozman et al. 1993, 1995; Viluksela et al. 1997a,b and 1998a,b).
Female reproductive toxicity of TCDD is evidenced by reduced ovulation (Li et al. 1995a,b) and developmental defects (Heimler et al. 1998a), which were orders of magnitude less than the cancer response (Kociba et al. 1978; 1979; Rozman et al. 1993), indicating that ovulation and development are more sensitive end points because lower doses are needed to disrupt reproductive processes than to increase the incidence of cancer. Studies using gonadotropin-treated immature rats revealed that complex mixtures of PCDDs, such as TCDD, pentachlorodibenzo-p-dioxin (PeCDD), and hexachlorodibenzo-p-dioxin (HxCDD), as well as each congener alone produced dose responses that lowered ovarian weights and the number of ova shed (Gao et al. 1999). In addition, the effects of PCDDs were additive when an equipotent mixture of the PCDDs was given. The slopes of the dose-response curves were not statistically different among the various congeners. Thus, the additive effect and parallel dose-response curves indicated a similar mechanism of action. The PCDFs and PCBs, with TCDD-like actions, also have a similar inhibitory effect on ovulation. The studies of Gao et al. (1999) and others (Krishnan and Safe 1993) are in close agreement and indicate a TEF of 0.12 to 0.2 for PeCDD, which differs from the TEF of 0.5 proposed by WHO (Van den Berg et al. 1998). The doses required in the ovulation study for PCDDs (Gao et al. 1999) to produce the same effect increased approximately 5-fold for each chlorine added to TCDD. Those observations are consistent with prior studies (Stahl et al. 1992) and imply a TEF based on female reproductive effects of 0.2 for PeCDD and 0.04 for HxCDD, which differ from the WHO report in which a TEF of 1 was given for PeCDD (Van den Berg et al. 1998) and used by Hamm et al. (2003) for numerous reproductive studies in males and females.
TEFs for the pentachloro-isomers of PCDPs and PCBs are in the same range as those previously reported for other end points (Safe 1990; Van Birgelen et al. 1994a,b, 1996). However, the WHO conference (Van den Berg et al. 1998) reported values of 0.5 and 0.1 for 2,3,4,7,8-pentachlorodibenzofuran (PeCDF) and 3,3′,4,4′,5-PCB, respectively, which are twice as high as most studies report. The doses of the pentachloroisomers of PCDs and PCBs studied by Gao et al. (2000b) had 10-fold lower
potency than the ED50 for TCDD in blocking ovulation (8 µg/kg) (Gao et al. 2000b). Generally, TEF values are combined from all end points, and development of end-point-specific TEFs might ultimately be useful. However, it must be noted that TEFs are not expected to be exact, and the values determine in the reproductive studies are well within an order of magnitude.
The Reassessment was mainly directed at understanding the adverse effects of TCDD administered during pregnancy on development of the pups; that section was superbly written and covered numerous aspects of exposure to TCDD in utero and during lactation. However, little risk estimate information is given. Also, an important part of the literature on the adult female reproductive system was not addressed in the Reassessment. This included mechanisms of ovulatory blockage at the level of the hypothalamic-pituitary axis and the ovary and endocrine disruption of reproductive processes by TCDD in adult rodents (Goldman et al. 2000; Petroff et al. 2001; Valdez and Petroff 2004). The committee summarizes some of those studies in the following section.
Effects of PCDDs on the Ovary
Studies have shown that PCDDs adversely affect ovarian function by direct actions on the ovary and the hypothalamic-pituitary axis (Petroff et al. 2000; Valdez and Petroff 2004). Human ovarian follicular fluid has been found to contain PCDDs (Tsutsumi et al. 1998), implicating PCDDs in possible adverse ovarian effects. Exposure of adult female rats to PCDDs disrupted estrous cycles, delayed ovulation, and lowered ovarian weights (Li et al. 1995a; Cummings et al. 1996). Irregular menstrual cycles were observed in female rhesus monkeys fed TCDD in the diet (Allen et al. 1977; Barsotti et al. 1979). Mice are less prone to the adverse ovarian effects of PCDDs in some studies (Cummings et al. 1996), although TCDD caused the formation of ovarian cysts in CD-1 mice (Gallo et al. 1986). In rats, administration of TCDD before mating interrupts fertility by affecting both ovulation and implantation (Giavini et al. 1983). In the immature gonadotropin-primed rat, the adverse effects of PCDDs on the ovary were characterized by small ovaries, the absence of corpora lutea, and numerous unruptured preovulatory follicles (Gao et al. 1999, 2000b; Petroff et al. 2001). In the immature rat primed with gonadotropin, the number of ova shed in response to PCDDs was dose-dependently inhibited with an ED50 of TCDD at 8 µg/kg of body weight. This supported the TEQ for several other AHR agonists, including PeCDD, HxCDD, PeCDF, and pentachlorobiphenyl (PeCB) (Li et al. 1995a,b; Gao et al. 1999, 2000b; Son et al. 1999; Petroff et al. 2000, 2001). TCDD suppressed follicular development as determined by a reduction in the number of antral and preantral follicles in
the pups of pregnant rats exposed to TCDD in utero and during lactation (Heimler et al. 1998a). The anovulatory effect of PCDDs was observed in gonadotropin-primed hypophysectomized rats (Gao et al. 1999; Petroff et al. 2000; Roby 2001), and direct application of TCDD to the ovary blocked ovulation as well (Petroff et al. 2000). Thus, PCDDs have direct effects on the ovulatory follicle that are sufficient to block ovulation.
The rat ovary expressed AHR mRNA (Son et al. 1999), and macaque granulosa cells also expressed AHR mRNA which was increased by human chorionic gonadotropin (Chaffin et al. 1999). β-Naphthoflavone (Bhattacharyya et al. 1995) and TCDD (Son et al. 1999) also increased ovarian Cytochrome P4501A1 protein (CYP1A1) mRNA in rats. The direct effects of PCDDs on ovarian steroid production are less clear, despite consistent blockade of ovulation after systemic and local ovarian exposure to PCDDs. In immature gonadotropin-primed female rats, pretreatment with PCDDs increased serum estradiol during the preovulatory period and reduced serum progesterone concentrations consistent with blockage of ovulation and reduced luteinization (Gao et al. 1999, 2000b). In addition, in the immature rat model, PCDDs blocked the surges of follicle-stimulating hormone (FSH) and lutenizing hormone (LH) in sera on expected proestrus (Li et al. 1995b; Gao et al. 1999, 2000b). Collectively, these results indicate that the adverse effects of PCDDs may be due to effects on gonadotropin release as well as to direct effects on the ovary (Son et al. 1999; Petroff et al. 2000; Roby 2001). In CD-1 mice and avian species, PCDDs did not alter serum concentrations of estradiol (DeVito et al. 1992; Janz and Bellward, 1996).
In vitro models have been used to assess the effects of PCDDs on ovarian steroidogenesis. PCDDs decreased cellular uptake of glucose and reduced protein kinase A activity and secretion of progesterone and estradiol in human granulosa cells (Enan et al. 1996a,b). However, another study reported an initial inhibition of estradiol in human luteinized granulosa cells that was followed by increased estradiol accumulation at 36 and 48 hours (Heimler et al. 1998b). A decrease in aromatase activity and reduced messenger ribonucleic acids (mRNAs) for P450ssc and P450arom in FSH-stimulated rat granulosa cells exposed to PCDDs has been reported (Dasmahapatra et al. 2000). In contrast, PCDDs failed to alter progesterone, androstenedione, or estradiol secretion in in vitro cultures of whole ovarian dispersates, granulosa cells, or thecal-interstitial cells derived from immature rats (Son et al. 1999) Although, this lack of in vitro action is also seen in immune cells in vitro.
One target of PCDDs may be alterations in follicular proteolysis and tissue remodeling during the periovulatory period as ovulation is blocked after acute exposure to TCDD, other dioxins, and DLCs (Gao et al. 1999, 2000b; Petroff et al. 2001). Potential mechanisms blocking degradation of the follicular wall may involve modulation of steroid action since decreased
expression of ovarian cyclooxygenase-2 (COX-2) and AHR coincide with increased plasminogen activator inhibitor-1 (PAI-1) and tissue plasminogen activator (PA) (Mizuyachi et al. 2002). Because PA participates in ovulation in the rat (Tsafriri 1995), TCDDs may increase PAI-1, reduce overall PA activity, and block ovulation. It is well known that PA activity increases after the ovulatory surges of LH and FSH as a result of increased granulosal cell prostaglandin secretion, a process dependent on COX-2 (Richards et al. 1987). COX-2 has TCDD response elements. Thus, TCDD, other dioxins, and DLCs may block ovulation by inhibiting granulosal prostaglandin secretion, reducing COX-2 in the preovulatory follicle, before reducing PA activity after an increase in ovarian PAI-1.
TCDD reduces expression of the progesterone receptor (PR), and PR null mice do not ovulate (Lyndon et al. 1996). TCDD is well known to inhibit estradiol-induced PR in the breast cancer cell line MCF-7 through an AHR-mediated mechanism (Harper et al. 1994). However, within 24 hours after administration of TCDD to immature rats, estrogen receptor (ER)α, ERβ, and PR were unaffected in the ovaries (Mizuyachi et al. 2002). Thus, the role of the PR in the anovulatory effects of PCDDs is unresolved.
Effects of PCDDs on the Hypothalamus and Pituitary Gland
TCDD, other dioxins, and DLCs reduce pituitary secretion of LH and FSH at the time of the LH and FSH surges but premature surges of LH and FSH have been reported in immature rats (Gao et al. 1999; 2000a,b). In the Han Wistar rat that is resistant to TCDD, (50 µg/kg) caused atrophy of the pituitary with little to no loss of weight and no mortality (Pohjanvirta et al. 1993). However, exposure of fetal (in utero) or neonatal (via mother’s milk) mice to TCDD reduced pituitary weights of male offspring (Theobald and Peterson 1997).
LH synthesis in the pituitary is controlled by gonadotropin-releasing hormone (GnRH) and gonadal steroids feed back negatively to reduce secretion. LH and FSH secretion were altered in gonadotropin-primed female rats pretreated with TCDD, other dioxins, and DLCs (Li et al. 1995b; Gao et al. 1999, 2000a,b). TCDD-treated animals had reduced gonadotropin secretion during the preovulatory period compared with controls. Culture of pituitary halves with TCDD dose-dependently reduced LH secretion, but no effect of TCDD was observed in primary pituitary cell cultures (Li et al. 1997).
Preovulatory increases in estradiol are required through a positive-feedback mechanism for induction of the LH and FSH surges on proestrus. TCDD has antiestrogenic effects and inhibits ovulation through blockage of the LH and FSH surges. However, serum concentrations of estradiol in intact control and TCDD-treated rats are similar during the preovulatory
period, indicating the possibility that the lack of estradiol action was causal in blocking the surges (Li et al. 1995b; Gao et al. 1999, 2000a,b). This appeared to be the case, as a long-acting exogenous estradiol administered during TCDD treatment overcame the blockade on ovulation and restored the LH and FSH surges (Gao et al. 2001).
Exogenous GnRH also overcame the inhibitory effects of TCDD on ovulation by restoring the LH and FSH surges in the immature gonadotropin rat model (Gao et al. 2000a). Controls exhibited normal LH and FSH surges, whereas such surges diminished in rats treated with TCDD. GnRH treatment increased secretion of LH and FSH to surge levels in TCDD-treated rats and partially restored ovulation. Those data indicate that GnRH secretion may have been reduced by TCDD. The failure of the gonadotropin surges to completely restore ovulation in rats receiving TCDD and GnRH indicates the possibility that adverse direct effects of TCDD on the ovaries may have reduced the number of ovulations.
Effects of TCDD on the Cardiovascular and Pulmonary Systems
Since the publication of EPA’s draft Reassessment, a substantial body of literature has emerged concerning the effects of TCDD on heart and vascular development. The developing vascular system appears to be a target very sensitive to TCDD in vertebrate embryos. Much of the work in this area has been performed in zebrafish (Danio rerio) embryos, a model that has the advantage over mammalian and avian models of allowing for direct visual observation of many developing organ systems, including the heart and associated vasculature. Studies have also been performed in avian and rodent models.
Several studies have indicated a fundamental role for the AHR system in vascular development and hence a theoretical basis for the TCDD sensitivity of vascular development. Lahvis et al. (2000) generated Ahr-/- mice that displayed reduced liver size. Developing mice exhibited altered vascular architecture, including massive portosystemic shunting due to a patent ductus venosus, resulting in reduced blood flow to the liver and hence reduced hepatocyte size and liver mass. This failure of the ductus venosus to close in Ahr-/- mice was subsequently associated with major hepatic veins failing to decrease in size, as observed in wild-type mice, which may result in increased blood pressure or a failure in vasoconstriction (Lahvis et al. 2005). Walisser et al. (2004b) observed that mice engineered to contain a hypomorphic Arnt allele (underexpressing ARNT, the AHR nuclear translocator protein) demonstrated the same vascular phenotype and were resistant to TCDD toxicity versus wild-type mice. Together with the AHR studies, this indicated essential roles for ARNT and for AHR-ARNT dimerization for both the purported developmental and TCDD toxicity roles of
the AHR pathway. TCDD exposure during a specific time frame of embryonic development rescued vascular development in both Ahr and Arnt hypomorphs, indicating the requirement for activation of the AHR-ARNT heterodimer for normal vascular development (Walisser et al. 2004a).
Studies with fish models, particularly zebrafish, have demonstrated the sensitivity of the cardiovascular system, including cardiomyocytes, to TCDD during embryonic development (Antkiewicz et al. 2005). Studies with morpho-lino antisense oligonucleotides to knock down expression of specific genes in the zebrafish embryo have supported the key role of AHR in the developmental effects of TCDD. Knockdowns of AHR2 prevented TCDD-induced pericardial edema, trunk circulation failure, and anemia in developing zebrafish (Prasch et al. 2003; Dong et al. 2004). (Due to gene duplication, zebrafish have two AHRs, AHR1 and AHR2; TCDD-mediated effects are associated with binding to AHR2, and not to AHR1.) In these studies, the AHR2 morpholinos were highly effective at blocking TCDD-induced cytochrome P4501A protein (CYP1A) expression in the vascular endothelium. Carney et al. (2004) showed that, whereas an AHR2 morpholino protected zebrafish from TCDD-mediated effects of reduced blood flow to trunk segments and pericardial edema, the CYP1A morphlino did not provide protection against TCDD toxicity in contrast to the findings of Teraoka et al. (2003). Collectively, these studies demonstrate that these developmental effects of TCDD are AHR2 mediated in zebrafish, but the role of CYP1A remains unresolved. TCDD has also been demonstrated to perturb cardiovascular development in the chicken embryo (Sommer et al. 2005) and in maternally exposed fetal mice (Thackaberry et al. 2005a,b). In addition, cardiovascular function is compromised in Ahr-/- mice (Lund et al. 2005; Vasquez et al. 2003).
These studies addressing the effects of TCDD on cardiovascular development were not performed with the objectives of quantitative risk assessment in mind. However, given the sensitivity of this end point at a very sensitive lifestage, EPA is encouraged to consider these and related studies identifying adverse effects of TCDD on cardiovascular development and function in its risk assessment for noncancer end points.
The Reassessment extensively documents the known reproductive, developmental, and ectodermal consequences of TCDD exposure in a variety of animal species (Part II, Chapter 5) and describes to a lesser extent various other noncancer consequences, including hepatic, thyroid, and cardiovascular effects observed in animals other than humans (Part II, Chapter 7, part B). In assessing the potential for related risks in humans, EPA makes several critical assumptions.
Assumption: Because dioxins are proven causes of reproductive, developmental, and other abnormalities in various animal species, they may, therefore, cause similar effects in humans. (Part III, p. 2-33, lines 3 to 5; p. 6-1, lines 21 to 22; p. 6-3, lines 14 to 16).
For reproductive, developmental, and ectodermal effects, this assumption is readily justified given the nature and extent of the animal data. Further, the profiles of reported human reproductive, developmental, and ectodermal effects after exposures to TCDD, other dioxins, and DLCs are similar to the effects found in animals, thus lending overall general support to the assumption. Similarities in developmental effects are most compelling at the highest levels of exposure such as those reported in the Yusho and Yu-Cheng poisonings (Part II, pp. 5-15 to 5-16) because “all four manifestations of developmental toxicity (reduced viability, structural alterations, growth retardation, and functional alterations) have been observed to some degree” (Part II, p. 5-97, lines 1 to 3).
Even so, the developmental effects are not entirely consistent and the Reassessment appropriately notes that other than the mouse “no other species develops cleft palate except at maternal doses that are fetotoxic and maternally toxic” (Part II, p. 5-19, lines 10 to 11) and that “studies in humans have not clearly identified an association between TCDD exposure and structural malformations” (Part II, p. 5-19, lines 15 to 17). As discussed below, the effects of low-level TCDD exposure on reported human developmental effects are less compelling. Although the spectrum of reported human reproductive and hormonal abnormalities following TCDD exposure is generally similar to that found in animals, the strengths of the individual associations in studies thus far, are weak, and confidence in the causal nature of these associations while suggestive is not yet compelling.
In reference to other noncancer consequences of TCDD exposure, the assumption remains equally valid, although the animal evidence for other noncancer end points, such as adverse effects on hepatic enzymes (Part II, section 220.127.116.11.3), pancreatic islet function (Part II, section 18.104.22.168.2), thyroid hormone dysregulation (Part II, section 22.214.171.124; Part III, section 126.96.36.199), lipid abnormalities (Part III, section 188.8.131.52), and cardiopulmonary or circulatory disturbances (Part II, section 184.108.40.206; Part III, section 220.127.116.11), is often more limited in scope.
Assumption: Humans are neither more nor less sensitive than animals as far as the adverse effects of dioxins are concerned (Part II, p. 8-4, lines 6 to 27; Part III, p. 2-3, lines 28 and 29; p. 2-32, lines 14 to 16]. Given the paucity of systematic in vivo human data, this assumption is the parsimonious choice and also the most defensible based on in vitro data (Part II, pp. 8-4 to 8-5). Nevertheless, EPA acknowledges the uncertainty and imprecision of this assumption noting that (1) “for most toxic effects produced by dioxin, there is marked species variation” (Part II, p. 8-5, line 31); (2)
human epidemiological studies are confounded by the fact that the unexposed “cohorts contain measurable amounts of background exposure to PCDDs, PCDFs, and dioxin-like PCBs” (Part II, p. 8-5, line 35; p. 8-6, line 1); (3) “many epidemiological studies are hampered by small sample size, and in many cases the actual amounts of TCDD and related compounds in human tissues were not examined” (Part II, p. 8-6, lines 2 and 3); (4) “it is often difficult, if not impossible, to assess in humans the same endpoints that might be determined in experimental animals” (Part II, p. 8-6, lines 4 and 5); and (5) “it is essentially impossible to determine the contribution of TCCD-like versus non-TCDD-like congeners to fetal/neonatal toxicity” (Part II, p. 5-15, lines 14 and 15) in the poisoning episodes where complex mixtures containing a variety of toxicants were ingested accidentally (Part III, p. 2-23, lines 32 to 35).
Assumption: Noncancer effects can occur at body burden levels in animals equal to or less than body burdens calculated for tumor induction in animals (Part III, p. 5-25, lines 28 and 29). Although critical to the discussion of noncancer end points in humans, the strength of this assumption is unknown and the uncertainty is possibly large. The propagated uncertainties leading to this assumption are highly dependent on the inherent uncertainties in the use of TEQs, the calculation of the historical body burden, and the modeling of dose-response effects, as discussed in detail in Chapters 2 and 3. Because of limited epidemiological evidence, further uncertainty is introduced by the inability to demonstrate convincing associations and dose-response relationships between TCDD exposure and noncancer end points in humans (Part III, p. 2-23, lines 20 to 22), as discussed below.
Assumption: ED01is an acceptable departure point for calculating the risks of noncancer end points. As noted above, the limitations of this assumption are highly dependent on the inherent uncertainties in the use of TEQs, the calculation of body burden, and the modeling of dose-response effects, as discussed in detail in Chapters 3 and 5.
The EPA Reassessment does not adequately discuss the level of confidence that should be accorded results whose statistical significance is associated with wide uncertainty limits. Attention should also be directed to addressing the potential biological significance of very small statistically significant physiological or biochemical changes that remain well within the normal range of variation and adaptation.
Furthermore, the EPA Reassessment continues to rely on the approach that diverse human data collected across disparate studies of different types and inherent strengths can be interpreted with confidence without applying the more formalized tools of evidence-based medicine. Thus, the EPA Reassessment (as well as Institute of Medicine [IOM] committee report) relies
largely on committee-based, consensus evaluation of the available data rather than on specifically commissioned, rigorous analyses constructed according to established criteria that both formally evaluate the strengths of the available evidence and integrate, by quantitative systematic review, the data across available studies (Sackett et al. 2000; NCI 2002; CEBM 2005; Guzelian et al. 2005).
On the whole, the potential for increased risk of noncancer end points after exposure to TCDD at or near background levels is cautiously presented in the Reassessment. However, the Reassessment explicitly characterizes TCDD as “developmental, reproductive, immunological, endocrinological, and carcinogenic hazards” (Part III, p. 6-3, lines 10 and 11), although the formal criteria for defining human hazard in the context of these noncancer end points are not defined precisely in the Reassessment. Further, although the Reassessment acknowledges that “some have argued that in the absence of better human data, deducing that a spectrum of noncancer effects will occur in humans overstates the science” (Part III, p. 6-3, lines 33 and 34), the EPA position is that an inference of human effects “is reasonable given the weight of evidence from available data” (Part III, p. 6-4, lines 1 and 2). Nonetheless, as EPA concedes, available human data currently do not permit resolution of these divergent evaluations.
Human Reproductive and Developmental Outcomes
The available human reproductive and developmental studies available at the time of the Reassessment draft are presented in detail, although a number of the more recent follow-up studies are obviously not reported, as mentioned below. EPA provides an overall conclusion that “subtle effects, such as the impacts on … developmental outcomes … or the changes in circulating reproductive hormones in men exposed to TCDD, illustrate the types of responses that support the finding of subtle yet arguably adverse effects at or near background body burdens” (Part III, p. 6-2, lines 6 to 11). The committee agrees that the results are subtle but disagrees that the reported effects are truly clinically adverse, especially when confidence in the observations is low and the reported changes could be non-significant at the biological level and clinical outcome. In this context, the Reassessment also notes that “there is no reason to expect, in general, that humans would not be similarly affected [as animals] at some dose, and a growing body of data supports this assumption. On the basis of the animal data, current margins of exposure are lower than generally considered acceptable, especially for more highly exposed human populations. The human database supporting this concern for potential effects near background body burdens is less certain” (Part III, p. 6-32, lines 19 to 23).
Male Reproductive Hormones
The Reassessment’s description of the National Institute of Occupational Safety and Health (NIOSH) study report (Egeland et al. 1994) showing a significant positive correlation of serum LH and FSH levels with serum TCDD does not discuss the weak nature of this correlation, the wide confidence intervals (CIs) around the regression, or the hormone values within the normal range (Part II, section 18.104.22.168). Similarly, the text further describes a two to four times higher prevalence of low testosterone levels among workers exposed to TCDD but does not report that the CIs around the risk ratios at the higher serum TCDD levels not only are very broad but also cross 1.0, indicating limited confidence in the significance of the relationships (Part II, section 22.214.171.124). Nor does the EPA Reassessment report that no dose-response effect was observed (odds ratio = 3.9 at lowest range of TCCD levels and 2.1 at highest levels), although the 95% CIs of the odds ratios themselves are so broad as to raise significant uncertainty about whether there is indeed a dose response relationship indicated by these studies (Part II, section 126.96.36.199).
Similarly, the Reassessment states that the Ranch Hand study (Roegner et al. 1991) (Part II, section 188.8.131.52) reported lower serum testosterone levels in Ranch Hand veterans with current serum TCDD levels exceeding 33.3 pg/g, although the reported difference (10.2 ng/dL) was “statistically nonsignificant” and unlikely to have a measurable physiological effect. The EPA Reassessment also describes three additional negative studies (CDC 1988; Grubbs et al. 1995; Henriksen et al. 1997), concluding that “the human data offer some evidence of alterations in male reproductive hormone levels associated with substantial occupational exposure to 2,3,7,8-TCDD” (Part II, p. 7B-38). Thus, although “some evidence” has been reported, the bulk of the reported evidence is either negative or uncertain to a degree.
The Department of Defense (DOD) released the latest report of the Ranch Hand study in 2005. The committee did not have the opportunity to review the report in detail because its release coincided with the end of the committee’s deliberations. However, the document reports that “the difference in adjusted free testosterone means in Ranch Hand versus Comparisons was 10.95 versus 10.47, respectively. The LH means for Ranch Hand and Comparison officers were 4.49 mIU/mL versus 4.09 mIU/mL, respectively. Both were well within one standard deviation of normal-age matched populations. No evidence of a dose-response effect was seen based on categorized dioxin or 1987 dioxin levels” (DOD 2005, p. 18-156). The report concludes that “the association of dioxin with … gonadal abnormalities appeared weak at best and unlikely to be clinically significant” (DOD 2005, p. 18-156) and “associations between dioxin level and …
gonadal hormone abnormalities were unlikely to be clinically important” at these levels (DOD 2005, p. 21-8).
Female Reproductive System
In the Reassessment’s discussion of potential effects of TCDD exposure on endometriosis, the more recent Seveso data (Eskenazi et al. 2002a) are not included. Compared with women with TCDD concentrations of ≤20 ppt, the relative risk of endometriosis is 2.1 in women with TCDD concentrations >100 ppt, but the 90% CI ranges from 0.5 to 8.0, indicating little confidence in the true magnitude of the rate ratio or the significance of the reported average relative risk of 2.1. One conclusion from these data might be that women whose serum TCDD levels were >20 ppt had no more endometriosis than those whose serum TCDD concentrations were ≤20 ppt. Another defensible conclusion might be that the study did not have the power to come to any convincing conclusion on this issue. The authors of the study, Eskenazi et al. (2002a), chose to describe their findings as a “doubled, nonsignificant risk for endometriosis among women with serum TCDD levels of 100 ppt or higher, but no clear dose response.” Finally, in a recent review of the nonhuman primate and the human data assessing the relationship between TCDD exposure and endometriosis Guo concluded that “there are no solid, credible data available at this moment to support the hypothesis that dioxin exposure may lead to the development of endometriosis” (Guo 2004).
Data published within the past 2 years on effects of exposure to TCDD, other dioxins, and DLCs on the menstrual cycle in women are obviously not referenced in the 2000 Reassessment. Thus, data from the Seveso incident surveying women who were exposed to TCDD postnatally, but while they were prepubertal, found “no change in the risk of onset of menarche with a 10-fold increase in TCDD,” and there was “no evidence of a dose-response trend” (Warner et al. 2004). Likewise, postmenarchal women exposed in Seveso showed no association of TCDD exposure with menstrual cycle length, but, in women exposed before menarche who had a 10-fold increase in serum TCDD concentrations, the menstrual cycle was lengthened by 0.93 day, although the 95% CI ranged from −0.01 to 1.86, and the strength of the relationship between menstrual cycle length and serum TCDD concentration shown in Figure 1A of the report is not convincing, with widely scattered data points (Eskenazi et al. 2002b). An observational study of wives and sisters of Swedish fishermen found a 0.49-day shorter menstrual cycle (95% CI 0.03 to 0.89) in those with a high dietary exposure to polychlorinated organochlorine compounds, including TCDD, but found no association with early life exposure (Axmon et al. 2004).
The discussion on spontaneous abortions briefly mentions the study on the NIOSH cohort as “in press” (Part II, section 184.108.40.206.5). This study has now appeared (Schnorr et al. 2001) and found no effect on the incidence of spontaneous abortion or on the sex ratio of offspring. The authors concluded that the study provided “additional evidence that paternal TCDD exposure does not increase the risk of spontaneous abortions at levels above those observed in the general population.” Likewise, recent data from the Seveso cohort (Eskenazi et al. 2003) showed no association of TCDD with spontaneous abortions.
Other recent relevant studies include birth-weight results reported for the NIOSH cohort (Lawson et al. 2004) and the Seveso cohort (Eskenazi et. al. 2003). The recent NIOSH report (Lawson et al. 2004) found that paternal TCDD exposure had no effect on birth weight for term infants, and a “somewhat protective” association of preterm delivery with paternal TCDD (odds ratio = 0.8), although the 95% CI ranged from 0.6 to 1.1. There was no obvious increase in birth defects, although the results were descriptive only. The authors concluded that “because the estimated TCDD concentrations in this population were much higher than in other studies, the results indicate that TCDD is unlikely to increase the risk of low birth weight or preterm delivery through a paternal mechanism.” The recent Seveso follow-up (Eskenazi et al. 2003) also showed no association of TCDD concentration with offspring birth weight or with the birth of infants small for gestational age. Finally, the Reassessment does not discuss the female Vietnam veterans study reported by Kang et al. (2000), which also reported no increase in spontaneous abortions, stillbirths, low-birth-weight infants, or infant deaths among women veterans who had served in Vietnam (and possibly exposed) compared with those who had served in the United States, although there are no body burden measurements made.
The data on birth-weight effects were described adequately (Part II, section 220.127.116.11), and the summary comment (Part II, section 18.104.22.168) reflected appropriately the uncertainty of whether there were any birth-weight effects of exposure to TCDD at the time of the Reassessment. However, the likelihood of TCDD exposure having a measurable effect on birth weight has been substantially reduced by the recently reported studies of Kang et al. (2000), Eskenazi et al. (2003), and Lawson et al. (2004). To reflect and appropriately weigh this new information, EPA should correspondingly modify the summary comment (Part II, section 22.214.171.124).
For the state of available information in 2000, the Reassessment describes adequately the observed effects of TCDD exposure on offspring sex ratio (Part II, sections 126.96.36.199, 188.8.131.52.8). As noted in the report, increased female births were observed after the Seveso accident (Mocarelli et al. 1996, 2000). They were also observed in a study of offspring of Russian pesticide producers exposed to TCDD (Ryan et al. 2002). However, the
Schnorr study mentioned in the Reassessment as “in press” has now been published (Schnorr et al. 2001) and found no effect of TCDD exposure on sex ratio of offspring in the NIOSH cohort.
In the United States, both exposure to TCDD and the male-to-female sex ratio at birth have declined since the early 1970s (Matthews and Hamilton 2005). This parallel decline is opposite that postulated from TCDD poisoning incidents. Thus, because sex ratios at birth not only undergo temporal trends but also show racial and nationality differences and are affected by both maternal age and infant birth order (Matthews and Hamilton 2005), EPA should also acknowledge the uncertainties inherent in evaluating sex ratios at birth without properly controlling for the aforementioned variables. The committee recognizes, however, that the TCDD exposure studies that showed altered gender ratios at birth have reported ratio values that were greater than the changes that might normally be expected to be caused by the variables mentioned above.
Childhood Growth and Postnatal Development
The Reassessment text describes appropriately the cited growth data and conveys adequately the uncertainty of whether TCDD exposure has effects on postnatal growth in humans (Part III, section 184.108.40.206). The issue would not be further clarified by including the omitted Swedish fish exposure study (Rylander et al. 1995) that reported diminished height, but not weight, at age 18, because this report includes neither TCDD nor TEQ data. Two of the longest-term studies of chlorinated toxicant effects on growth published subsequently (Blanck et al. 2002; Gladen et al. 2000) deal with PCB exposure and thus do not contribute to resolving the debate about the effects of TCDD on childhood growth.
Similarly, the longest neurodevelopmental follow-up studies (Jacobson and Jacobson 1996; Gray et al. 2005) are reports on PCB exposure and do not directly contribute to the current TCDD issues since no TEQ is derived. However, the ongoing Dutch follow-up study referenced repeatedly in the Reassessment has now published its findings in 6.5-year-old children (Vreugdenhil et al. 2002a). At that age, there were no cognitive or motor differences between breast-fed infants (primarily postnatally exposed) and formula-fed infants (primarily exposed in utero with background postnatal exposure), including no overall differences in global cognitive index, memory, or motor performance, except when children from “less optimal homes” were analyzed separately. This observation suggested to the authors that less optimal home environment may allow the effects of TCDD on neurodevelopment to become manifest more readily while, in more optimal home environments, the additional beneficial environmental influences overcome the detrimental effects of exposure to TCDD. This is clearly
a hypothesis at this stage, given the availability of only this single study that has addressed the issues. In an additional report, Vreugdenhill et al. (2002b) also described decreased masculinized play in boys and increased masculinized play in girls at age 7.5 years. Although statistically significant, the biological relevance of these conclusions remains uncertain given the wide scatter of the data and the regression coefficients reported.
Cardiovascular and Pulmonary Systems
The Reassessment discusses in detail the available data on potential human cardiopulmonary consequences of TCDD exposure highlighting the difficulty of supporting firm conclusions about the presence of a relationship (Part II, sections 7.13.9, 220.127.116.11, 7.13.10). Recently, from the latest data on Ranch Hands (DOD 2005), DOD concluded that “no consistent evidence suggested that herbicides or dioxin were associated with ill effects on respiratory health” (p. 21-9). On the other hand, “the presence of heart disease was found to be higher among Ranch Hands than Comparisons in enlisted flyers” (p. 21-6), and “an increased percentage of Ranch Hands in the high dioxin category were found to have abnormally high diastolic blood pressure. Ranch Hands in both the low dioxin category and the low and high dioxin categories combined were found to have a lower mean systolic blood pressure. Similarly, a smaller percentage of Ranch Hands in both the low dioxin category and the low and high dioxin categories combined had an abnormally high systolic blood pressure” (p. 21-6). However, the report notes that “the prevalence of cardiovascular disease was not increased in the Ranch Hand cohort. In only one analysis, that of diastolic blood pressure noted above, was there any evidence of an increased risk with increased body burden of dioxin” (p. 21-7).
OTHER NONCANCER END POINTS
The Reassessment (Part II, sections 7.13.6, 18.104.22.168, 22.214.171.124.2; Part III, section 2.2.5) presents in detail the then available data on the relationship between TCDD exposure and the development of Type 2 diabetes. This relationship was evaluated in greater depth by an IOM committee (IOM 2000), which concluded that “there is limited/suggestive evidence of an association between exposure to the herbicides used in Vietnam or the contaminant dioxin and Type 2 diabetes.” This is an adequate statement of the state of the science concerning this noncancer end point, and the committee recommends that EPA revise the Reassessment to include the analysis provided by the IOM committee. The Reassessment should also incor-
porate the data of the recent Ranch Hand study report (DOD 2005) showing that “mean fasting insulin and the risk of diabetes requiring insulin control increased with initial dioxin. C-peptide and time to diabetes onset decreased as initial dioxin increased.” The risk of diabetes requiring insulin control was increased in the Ranch Hand high-dioxin category. An increase in the risk of diabetes requiring oral hypoglycemic or insulin control was observed as 1987 dioxin levels increased. Time to diabetes onset decreased as 1987 dioxin levels increased. The risk of an abnormally high hemoglobin A1c increased with 1987 dioxin levels. Some findings in the DOD (2005) report appeared inconsistent with the results presented above, such as a decrease in the risk of 2-hour postprandial urinary glucose abnormalities with 1987 dioxin levels. The findings appear consistent with the previously noted association between Type 2 diabetes and dioxin in Ranch Hand veterans. Increased risks of diabetes requiring insulin control were found with initial dioxin, in the high-dioxin category, and with 1987 dioxin levels. In contrast, “associations between dioxin level and thyroid and gonadal hormone abnormalities were unlikely to be clinically important” (DOD 2005, p. 21-8).
These data led to the conclusion that “the association noted at previous Air Force Health Study examinations between Type 2 diabetes mellitus and dioxin persisted. A higher prevalence of diabetes, as well as severity, as dioxin increased was evident, even after adjustment for such factors as age and body mass index” (DOD 2005, p. 18-156).
The Reassessment acknowledges that despite the fact that “many effects of TCDD exposure in animals resemble signs of thyroid dysfunction or significant alterations of thyroid-related hormones” (Part III, p. 2-41, lines 32 and 33), the results of human studies “are mostly equivocal” (Part III, p. 2-42, line 1). The Reassessment reports that Pavuk et al. (2003) showed “elevated TSH [thyroid stimulating hormone] means among the high TCDD exposure group in the 1985 and 1987 follow-ups, with an increasing trend across the decade 1982-1992, but no association with the occurrence of thyroid disease” (Part III, p. 2-42, lines 2 to 4). The discussion does not address the fact that the TSH differences, although statistically significant, are quantitatively extremely small and well within the normal range of circulating TSH levels. Further, limitations of the study are not described, including 86 exclusions “from the longitudinal analyses because they had undergone thyroidectomy, or had endocrine cancer, or were on thyroid medication” (Pavuk et al. 2003), the exclusion of one Ranch Hand because of an extremely low TSH value that might have indicated the earliest sign of hyperthyroidism and the fact that “over the five examina-
tions, three different radioimmunoassays were used to measure TSH” (Pavuk et al. 2003). In addition the Reassessment does not report that no significant differences “with regard to mean levels of total thyroxine (T4), triiodothyronine (T3)% uptake, or free thyroxine index were observed at any examination” (Pavuk et al. 2003). Recently, DOD (2005) reaffirmed that “as for a dioxin effect related to thyroid disease, the 2002 examination data did not support such a relation.”
Further, while the most recently published Ranch Hand follow-up report (DOD 2005) found a difference in adjusted mean TSH levels (1.653 microunits/mL versus 1.557 microunits/mL) in Ranch Hands and Comparisons, respectively, this difference was “not considered clinically significant because a 1% difference is difficult to measure. The same was true from the free T4 values in enlisted flyers (mean of 1.115 ng/dL in Ranch Hands versus 1.054 ng/dL in the comparison groups). If a primary thyroid effect were present, one would expect the TSH to move in the opposite direction of the free T4, which was not seen in these data” (DOD 2005, p. 18-156).
The draft Reassessment also highlights the higher TSH values reported in human infants by Pluim et al. (1993) and by Koopman-Esseboom et al. (1994) (Part III, 126.96.36.199 and Part II,188.8.131.52.2), but does not discuss the fact that the TSH changes were very small and possibly not of physiological or clinical significance. Follow-up of the Dutch children’s cohort has now been carried out for more than a decade, and changes in thyroid status in this cohort have not been reported, although it is unclear whether they were in fact thoroughly assessed. The text in the Reassessment does include lengthy hypothetical discussions (Part III, 184.108.40.206 and Part III, 220.127.116.11) of plausible mechanisms for perturbations of human thyroid function, although there is limited human data to support or refute such mechanisms.
Because ectodermal abnormalities are common findings in animal studies, including nonhuman primates, as well as in the human Yusho and Yu-Cheng exposures, the enamel hypomineralization found in 6- to 7-year old Finnish children (Alaluusua et al. 1996, 1999) is highlighted twice (Part II, section 18.104.22.168.1; Part III, section 22.214.171.124). Because the enamel mineralization scores were largely subjective, it is imperative that the observers were blinded to the prior breast-feeding status of the children, but this issue is not specifically mentioned in the publications. Additionally, the Reassessment highlights several other limitations of this study and, in the related commentary (Part II, section 126.96.36.199.2) EPA acknowledges that “the presentation of the results is incomplete.”
The Reassessment adequately summarizes the uniformly agreed upon and well-documented association between TCDD exposure and the development of chloracne.
Elevated γ-Glutamyl Transferase
The Reassessment adequately summarizes the data showing “a consistent pattern of increased γ-glutamyl transferase (GGT) levels among individuals exposed to TCDD-contaminated chemical.” (Part III, p. 2-40, lines 23 and 24) and notes that “long-term pathological consequences of elevated GGT have not been illustrated by excess mortality from liver disorders or cancer; or in excess morbidity in the available cross-sectional studies” (Part III, p. 2-41, lines 23 and 24). The report also acknowledges that “the consistency of the findings in a number of studies suggests that the finding may reflect a true effect of exposure but for which the clinical significance is unclear” (Part II, p. 7B-116, lines 20 to 22).
The most recent report of the Ranch Hand study found no relationship between TCDD exposure and GGT (DOD 2005, pp. 13-51 to 13-57).
The Reassessment accurately notes that “neither adults nor children from Seveso had lipid levels above the referent level” (Part II, p. 7B-118) despite very high exposures and that “the most recent data suggest that high exposure to 2,3,7,8-TCDD contaminated substances are not related significantly to increased lipid concentrations, specifically total cholesterol and triglycerides” (Part II, p. 7B-118). The Reassessment adds that “slight but chronic elevations in serum lipids may put an individual at increased risk for disorders such as atherosclerosis and other conditions affecting the vascular system” (Part II, p. 7B-118) and that “risk factors such as dietary fat intake, familial hypercholesterolemia, alcohol consumption, and exercise” (Part II, p. 7B-118) were not considered in the Seveso study, even though no effects of TCDD exposure were found.
The most recent Ranch Hand follow-up showed no relationship between TCDD exposures and total high-density-lipoprotein cholesterol, although ranch hands had an increased percentage of individuals with increased triglyceride values (DOD 2005, pp. 13-78 to 13-103). This report concluded that “based on the analysis of triglycerides, a subtle relation between dioxin and lipid metabolism cannot be excluded” (DOD 2005, p. 13-218).
The various remaining noncancer end points, for which there are fewer data and even less suggestive evidence than the associations discussed above, are adequately summarized throughout the Reassessment.
CONCLUSIONS AND RECOMMENDATIONS ON THE REPRODUCTIVE, DEVELOPMENTAL, AND OTHER NONCANCER END POINTS OF TCDD, OTHER DIOXINS, AND DLCS
Embryonic and fetal development and female and male reproduction are sensitive end points of toxicity from TCDD, other dioxins, and DLCs in rodents because, as discussed earlier, responses occur at lower administered doses than other end points. However, the sensitivity of these end points in humans is less apparent.
The fetal rodent is more sensitive than the adult rodent to adverse effects of TCDD.
In humans, there is a clear association of TCDD exposure with chloracne and available studies have shown suggestive associations of TCDD exposure with Type 2 diabetes, but the latter data are not yet robust.
In humans, the association of TCDD exposure with other reported, detrimental noncancer effects has not been convincingly demonstrated. The available studies have not yet shown clear associations among TCDD exposures and the risks of individual, clinically significant, noncancer end points.
In reference to human disease risks, the overall conclusions about noncancer risks due to TCDD exposure are, in general, cautiously stated, and the uncertainty of suspected relationships is acknowledged. Nonetheless, the limitations of specific human studies are not uniformly addressed, and the broad 95% CIs accompanying some reported statistically significant effects are not discussed in the context of the uncertainty that these broad confidence limits imply. Conversely, statistically insignificant effects are sometimes highlighted.
The divergent data across the diverse studies assessing human noncancer end points have not been subjected to systematic review according to currently accepted approaches, including meta-analysis when appropriate, nor has there been formal grading of the quality of the evidence according to accepted principles currently applied in other areas of clinical pathophysiology, including one report of the relationship of TCDD exposure to cancer end points (Crump et al. 2003).
EPA should discuss how the ED used in the in utero and lactational exposure rat models relates to human reproductive and developmental toxicity and risk information, including TEFs and TEQs.
EPA should more clearly describe how and why ED01 values were determined in animals and transferred to human equivalents for the various noncancer end points and address risk estimate calculations using alternative assumptions (e.g., ED05). Whereas ED01 is conceptually a viable POD, the committee has concerns about how the ED01 is computed and whether there are adequate data at the ED01 level to ensure an acceptable level of confidence in the conclusions derived from using the ED01. The dynamic range approach EPA used to compute ED01 for continuous response is flawed in that the change of 1% total range may not identify any meaningful toxic effects, that 1% change may be well within random variation in the absence of exposure, and that the use of total range is less sensitive than use of a control range because total range can be much wider.
EPA should provide a discussion of the dose-response effects of TCDD, other dioxins, and DLCs on the adult female reproductive system that result in endocrine disruption in animals. The impact of the dose-response data provided in these studies on human risk assessment should be presented.
With respect to human noncancer end points, the Reassessment text should be revised to include the relevant, more recent data and, when appropriate as discussed above, to reflect study quality and data uncertainty of the studies referenced.
For available human, clinical, noncancer end point data, EPA should establish formal principles of and a formal mechanism for evidence-based classification and systematic statistical review, including meta-analysis when possible. The application of systematic review, followed by evidence-based classification, leads to a more explicit statement of, and concrete appreciation of, the level of certainty (and, correspondingly, of uncertainty) that can be accorded the answers to specific questions in a particular field.
When the mechanism is established, currently available and newly available human clinical studies should be subject to such systematic review and formal evidence-based assessment. The quality of the available evidence should be reported, and the strength or weakness of a presumptive association should be classified according to currently accepted criteria for levels of evidence. Animal studies have shown that TCDD can cause a variety of noncancer effects. These studies support both the EPA position of the plausibility of corresponding human effects and the need to devise adequately designed investigations that will answer the questions in man.
In making its final recommendations, EPA should incorporate and integrate the relevant data from both human and animal studies, as appropriate, according to the levels-of-evidence hierarchy devised.
EPA is encouraged to review newly available studies on the effects of TCDD on cardiovascular development in its risk assessment for noncancer end points.