CALL (812) 333-8888 FOR MORE INFORMATION OR E-MAIL info@copa.org
| |||||||||||||||
CALL (812) 333-8888 FOR MORE INFORMATION OR E-MAIL info@copa.org | |||||||||||||||
 |
|||||||||||||||
Reports: EPA | |||||||||||||||
Dioxin Reassesment ReviewU.S. Environmental AgencyScience Advisory Board Section 4 - Part A 4.6 Developmental Toxicity and Animal NOAELs (Charge Question 14) The specific issue of animal No Observed Adverse Effect Levels (NOAEL), as it is framed in the Charge, is inconsistent with the question posed concerning the use of the Reference Dose (RfD) in evaluating incremental exposures (see health question 18), and on pages 9-69 ff of the reassessment document. The latter question derives from the argument that RfDs (which are based on NOAELs), are inappropriate for the current assessment because background levels may be significant. If that argument is sound, why should it not apply to animal models? Furthermore, even if the effect level procedure could be defended for animals, is it the optimal metric? The basis for Charge question 18 is the reasoning that the proper evaluation of risk in this context is the increment per arbitrary unit of exposure, expressed as some measure of body burden. This is one instance in which dose-response modeling, as by Benchmark doses, may confer a substantial advantage. What is gained by using NOAELs and RfDs? In a somewhat analogous situation--the relationship between lead exposure (defined as blood lead level) and IQ--the risk is better expressed as 0.25 IQ points for each 1 g/dL than as an RfD. This would be a more appropriate model for TCDD because, as with lead, the location of a threshold for developmental outcomes is rather uncertain. If, for consistency with traditional EPA practices, a NOAEL were to be extracted from the animal data, 1 ng/kg day-1 appears to lie in an ambiguous zone. Murray et al. (1979), in a multi-generation study, found that 10 ng per kg-1 per day-1 of TCDD lowered the body weights and food consumption of f1 and f2 rats and also affected postnatal survival. At 1 ng per kg-1 per day-1 both increases and decreases were noted among the survival indices of f1 litters. A NOAEL of 1 ng per kg-1 per day-1 would be a reasonable figure if based on these data. On the basis of other endpoints, however, it may not be as reasonable. Monkeys whose mothers were undergoing exposure to TCDD in food (Schantz et al., 1989) were impaired, relative to controls, in a behavioral task called reversal learning. They also displayed differences from controls in social behaviors (Schantz and Bowman, 1992). The mothers had been fed a diet containing 5 ppt of TCDD. The offspring tested for learning, from two cohorts, were born a mean of 16 months or 36 months, respectively, from the initiation of exposure. Social behavior was assayed in the first cohort. On the basis of total TCDD consumed by the time of birth, the mothers had been exposed to about 0.125 ng/kg daily. Copulatory behavior in male rats and measures of morphological and endocrine development showed adverse effects of prenatal exposure to TCDD at a dose of 64 ng/kg to the mother on gestation day (GD) 15 (Mably et al., 1992). If a half-life in rats of 24 days was assumed for 1 ng/kg day-1, steady state would be reached in about 120 days, and total body burden would be about 34 ng/kg. For this reason, because 64 ng/kg is actually a frank effects level (FEL), 1 ng/kg/day would be a dubious NOAEL. The problem lies in distin guishing between the acute effects of a dose delivered on GD 15 and an equivalent body burden stored largely in fat tissue. TCDD stored in this way might be viewed as function ally dormant, although the Schantz and Bowman data cited above argue against such an interpretation. Comparisons between acute and steady-state exposures for endpoints such as those above have not been carried out despite the significance of this issue. Such studies should be relatively straightforward, though tedious, to conduct. In humans, TCDD and other lipophilic agents could enter the blood if fat stores were to be drawn upon during periods of caloric deficit. Weight-loss diet books consume enormous shelf space in bookstores and weight-loss regimens are not a guarantee against pregnancy. Although the increment in blood level of lipophilic toxicants during such a regimen may be small, it may not be insignificant, and fetal tissue might accumulate these increments. Release of such agents during gestation, in the small proportion of cases in which the mother fails to consume adequate nutrition, might lead to their accumulation in fetal tissue. Moreover, the fetus will deplete the mother of nutrients even at the cost of her own nutritional status. Lactation is another situation in which lipophilic agents are released and consumed, as discussed in the document. [A study just published, and not available to the Committee at the time of the review, reports that higher levels of PCBs, PCDDs, and PCDFs in breast milk were related to reduced neurological optimality, and higher levels of planar PCBs were associated with a higher incidence of hypotonia (Huisman et al., 1995).] The lead parallel should again be considered. During preg nancy, lead is released from bone as a consequence of to hormonal changes. The same mechanism would not pertain to TCDD, but the details of gestational pharmacokinetics is a question that deserves to be further pursued. Endpoints other than those discussed above (sexual function in rats and learning in monkeys) should also be considered. The role that developmental toxicity has assumed in risk assessment stems from an extensive body of literature indicating the exquisite vulnera bility of the fetus. The developing brain is especially sensitive. But the rodent brain may not reflect all aspects of this sensitivity. Recall that rats are born, from the standpoint of brain development, at the end of the second human trimester. Processes that occur in humans during the third trimester, such as synaptogenesis and the maturation of neurotransmitter systems, occur postnatally in rats. For this reason, confining developmental treatments in rats to the prenatal period may underestimate the full impact in humans of gestational exposure. The behavioral pharmacology literature offers many examples of the sensitivity of the neonatal period in rodents and Bjerke et al. (1994) and Bjerke and Peterson (1994) reported that lactational as well as prenatal exposure proved necessary to induce feminiza tion of male rat copulatory behavior. Further, although sexual development has become a focus of TCDD toxicity research, it is crucial to recognize that the complex unfolding of brain developmental processes in the presence of certain levels of TCDD could also exert an impact on other indices of brain function and structure. The substantial levels of the Ah receptor in developing brain and its virtual disappearance later in life argue that sexual development is unlikely to be its only role. The report by Schantz and Bowman (1989) hints at additional outcomes. Note, further, that copulatory behavior and genital structure reflect only limited facets of sexual maturation. Copulatory behavior and reproductive morphology may not be the best endpoints to examine for prenatally-exposed females. Other sexually dimorphic behaviors, based on cognitive function, for example, might also reveal consequences of TCDD exposure if examined. And, for males, sexual motivation is an arena independent of copulation itself. The preceding comments presume that the total exposure is to TCDD. They make no assumptions about the validity of TEQs, which is a separate, but related issue. Chapter 5 of the reassessment document neither cited nor discussed the findings of several researchers on the effects of PCBs (e.g., Jacobson and Jacobson, 1994 and Gladen, et al., 1988). Jacobson and Jacobson (1994) reported that higher gestational exposure to PCBs, as reflected by cord serum levels and maternal consumption of Lake Michigan fish, was correlated with lower scores on tests of psychological development. A large cohort in North Carolina was reported by Gladen et al. to show also evidence of poorer performance associated with higher prenatal exposure to PCBs. Jacobson and Jacobson suggested that the PCBs themselves, particularly the coplanar PCB congeners, could be more responsible for such effects than the much lower concentrations of dioxins or dibenzofurans in these mixtures, but the relative potency issue would have to be clarified. PCDFs in the Yusho and YuCheng exposures are currently held to be primarily responsible for the observed toxicities. The North Carolina and Lake Michigan studies, however, are significant because they indicate adverse neurobehavioral development associated with current environmental levels of this class of compounds. [These findings were not discussed at the Committee s public meeting, but came to light during the preparation of this report.] Another reason for reevaluating the current NOAEL are data strongly suggesting that TCDD is a more potent teratogen in mice than previously supposed, and that this effect occurs at doses considerably lower than those causing liver enzyme induction. (Bjerke et al., 1994) These data are coupled with findings from various laboratories showing changes in hormone levels controlling reproductive and developmental processes and reproductive success in offspring of treated animals. In summary, the current NOAEL of 1 ng/kg day-1 rests on a debatable foundation, and it would be appropriate to reevaluate it. [Several Members of the Committee suggest that the immediate inference of this statement is that the NOAEL should be lowered.] One reason for a fuller examination of these issues is that such a NOAEL is likely to be used in evaluating the risks of human exposures, given that crude clinical evidence for developmental and reproductive effects attributable to TCDD is necessarily quite limited. The neurobehavioral studies on PCBs offer directions for additional research, but forcing such studies into an effects level mold, as noted earlier, is a premature use of the data. 4.7 Human / Animal Databases: Potential for Immunotoxicity (Charge Question 15) There was a consensus among most Committee Members that, overall, Chapter 4 of the reassessment document provided an accurate, current summary of the immunotoxicology associated with TCDD and related compounds in humans and experimental animals. However, the Committee has some concerns (and noted some omissions) pertaining to immunology and the interpretation of certain experimental results, in Chapter 4, and particularly, Chapter 9. Although the overall data suggest that dioxin and related compounds can produce immune effects, there are insufficient supporting data to establish fully whether these effects can occur at or near two orders-of-magnitude above background levels. The document hedges on whether there was sufficient evidence to state that this class of compounds can cause immune effects in humans. In some instances there was a YES and in other sections a statement of UNSURE. Based upon the extensive experimental animal, and the very limited human, database the majority of the Committee agreed that sufficient data exist to indicate immune effects could occur in the human population from exposure to dioxin or dioxin-like agents at some dose levels. However, the large variability in the immune response in humans, the limited numbers of tests conducted, and the poor exposure characterization of the popula tions that have been studied prevent definitive conclusions as to sensitivity. This is not to say that humans are more or less sensitive than other species, only there are not sufficient clinical data to assess human sensitivity. The most notable documented immune effects in humans occurred in the Taiwan population exposed to contaminated rice oil where both immunosuppression and increased infections were observed, presumably resulting from exposure to furans and PCBs (Wu et al., 1984; Chang et al., 1982a; Chang et al., 1982b: Chang et al., 1981). Some studies have reported changes in immunoglobin (Ig) levels (Jennings et al., 1988) and NK cell activity (Jennings et al.,1988; Svensson et al., 1991). Other studies have reported no effects (such as the Ranch Hand (USAF, 1991) and Seveso studies (Mocarelli et al., 1986; Mocarelli et al., 1991)). but, as with the positive studies, the actual exposure levels and study design, were not adequately addressed in the document nor were they critically reviewed. Although not well-established, several studies of humans (Chang, et al., 1981; Bekesi et al.,1985) exposed to halogenated aromatic hydrocarbons (HAH), as well as of monkeys (Hong et al., 1989; Tryphonas et al., 1989), reported that a slight reduction in CD4+ cells occurred. Although this may or may not translate to a significant health effect, these cells are involved in regulating immune responses and reduced CD4/CD8 ratios are a hallmark of immunosuppression. It may be argued that any reduction in CD4 cells could lead to such potential health effects as increases in infectious disease, given that a large population is affected. Of interest to this discussion is a recent study by Oughton et al. (1995) which found no decrease in total CD4+ cells in TCDD-treated mice following low-level chronic exposures. However, within the CD4+ subset, a modest decrease was observed in the proportion of CD4+ memory cells as defined by concomitant expression ofPgp-1 CD45RB. The clinical significance of this change on immunocompetence is presently unclear. Similar changes in immune system functioning have, however, been suggested by some investigators to be significant to HIV pathogenesis (Janossy et al., 1993; Lim et al., 1993; Cameron et al., 1994; Jaleco et al., 1994), indicating that the decrease in these cells may be associated with a decrease in immunocompetence. Although the immune system is a sensitive target to HAHs in experimental animal species, as presented, the EPA document does not provide convincing evidence to indicate that background or near background exposure levels to dioxin-like compounds in industrial countries are sufficient to affect the immune system. Given the current methods available for testing, it would be unlikely that this could even be determined in humans, and one would have to rely on experimental animal data or highly exposed populations to determine effects at the low-end of the dose-response curve, or in vitro approaches using primary isolated human lymphocytes and human lymphoid cell lines. Changes reported at very low levels of exposure in two or three of the experimental animal studies are certainly of concern, but need to be confirmed and reproduced and, until then, cannot yet be used to support a back ground level effect. However, the No Observed Eeffects Level (NOEL) and ED50 (dose effective for 50% of the recipients) for suppression of the T-dependent antibody response, in sensitive mouse strains, has been reproduced in many labs using different experimental designs and can be used to help support or refute that background or one to two orders above background is significant. In this respect, the recent paper by McGrath et al.(1994) is relevant to this issue. Human studies undertaken to study the immune system of exposed populations have not used the appropriate test battery for this class of chemicals. The gold-standard test (i.e., suppression of the primary antibody response following immunization) was not employed in any of the human test panels, although this is a hallmark in experimental animals. The exception to this is the as yet, unpublished study on the Inuit population (Dewailly, in press). Chapter Four discussions pertaining to in vitro effects, although complete, concluded these tests have limited relevance as culture conditions may play a significant role (i.e., serum effects). The Committee felt that this was not a legitimate argument as numerous investiga tors have successfully reproduced in vivo observations using well-defined in vitro culture conditions. The in vitro studies have provided considerable understanding of the cellular and molecular mechanisms of TCDD and should not be understated. Although data exist suggesting that non-Ah receptor mechanisms may play some role in immunotoxicity, definitive evidence for this is lacking and will require using novel approaches such as receptor knockout mice or pure binding antagonists. The Committee agreed that the majority of evidence indicates that imununotoxicity (particularly suppression of the antibody response) by dioxins is presumably Ah-receptor dependent. The disappear ance of quantitative differences in immunosuppression between Ah-low and Ah-high responsive mice after sub-chronic exposure suggests that chronicity can override acute exposure resistance and may suggest an even greater hazard. Based upon existing evidence, the involvement of an endocrine-related non-Ah receptor mechanism impacting on immuno competence may be overstated in the document. Numerous studies suggest that the immune system is a sensitive target for dioxin-like compounds in experimental animals. There are species/strain differences in the sensitivity, but the effects tend to be similar with the most sensitive indicator (at least in adult animals) being changes in the primary antibody responses; similar effects occur in many test species (guinea pig, rabbit, monkey, etc.). As such, it would be more appropriate to indicate that differences in animal sensitivity exists rather than variability in response as this suggests a different meaning. One might expect that similar variability would also exist in the human population, but this has not been examined when the limited clinical studies have been undertaken. Multiple cellular targets exist for immunotoxicity by dioxin-like compounds including both T and B lymphocytes as well as lymphoid-associated tissue (e.g., thymus epithelium) and marrow stromal elements. Debate exists as to the most sensitive or most proximate target and not which cell is the target. The reassessment document should be more clear on this point. (Table 9-2 is incorrect as the immune system of rabbits and fish are also affected). Studies conducted in a number of experimental species, including mice and monkeys, indicate that the antibody response to a T-dependent antigen is the most sensitive and reproducible indicator for immunotoxicity in adult animals. The ED50 in sensitive strains of mice is approx. 0.7 g/kg. Several other responses have been shown to be more sensitive but have not been confirmed or reproduced by other research groups. Chapter 9 gives undue weight to these unconfirmed or limited studies (see Table 9-5), and fails to discuss the highly reproducible and widely used primary antibody suppression studies. Results for the host resistance tests are, for the most part, consistent with the immune affects. One obvious exception to this are the influenza challenge studies conducted in mice (G. Burelson, in press), where disease occurs at much lower dose levels than do immune changes. As this directly relates to human health, the mechanism and relevance of these observations, which of course were not available when the reassessment was developed, need to be addressed in the future revision. The recent observations reported by the same author in rats (G. Burleson, in press) should also be discussed. [ The two in-press studies noted here were not available to most Members of the Committee during the review process; several Members therefore cannot endorse these specific findings at this time.] This new study helps increase the validity of the preceding observation, although additional studies are warranted to help elucidate the mechanism. It would argue that a very specific component (perhaps immune/perhaps not) is altered, and at extremely low concentrations. Chapter 9 should also include a table of confirmed laboratory results (i.e., the PFC primary response) which provide ED50, ED1, and/or NOELs. A separate table for suggestive results (not yet extended at low doses) can then be included and identified as such. The text of these tables should include a critical review of the data, the most reproducible and sensitive indicators, and a clear and logical presentation of how these data were used to determine that exposures at 1 or 2 orders-of-magnitude above background levels have potential human health effects. 4.8 Other Effects (Charge Question 16) The Committee had no specific concerns with the manner in which the topic of other effects is covered in the reassessment document. There has been considerable expansion of the knowledge base since the document was issued, however, and these gains should be addressed in any revision. Specifically, major growth has taken place in our understanding of the biological and biochemical effects of TCDD and related members of this class, and in the whole area of receptor biology and signaling biology, and these gains should be factored into the revision. 4.9 Dose-Response 4.9.1 Approaches to Dose-response Determination for Cancer (Charge Question 17) There was public expectation that the reevaluation of the carcinogenic potency of dioxin would be comprehensive and would incorporate extensive new data generated for that purpose. The Committee was disappointed to see that, in addition to data from the Kociba et al. (1978) bioassay (which formed the basis for EPA's earlier estimate of cancer potency for TCDD), the only additional data used in EPA's quantitative analysis was from the Maronpot et al. (1993) gavage study. The only use that EPA made of pharmacokinetic modeling in its quantitative analysis was in making correlations between estimates of two-stage model parameters and outputs of pharmacokinetic models (Appendix D of the reassessment document). This is a very limited use of the physiological and pharmacokinetic information (although the Committee is aware of the problems involved in using the extant PBPK models--see the discussion in sections 4.1.1 and 4.1.2); moreover, the description provided in Appendix D is not sufficient to enable the reader to understand what was done. This lack of clear exposition needs to be corrected in the revised document. Not only is EPA's risk assessment of dioxin extremely important in its own right, it represents EPA's first application of so-called biologically based models. Consequently, it may set a precedent for future analyses. It is thus important that the analysis be clearly presented so that its scientific justification and usefulness can be assessed. EPA should describe its analysis in sufficient detail that it can be fully understood by the reader, to the point of reproducing the analysis if desirable. The reasoning and analysis that led EPA to propose its preferred model must be clearly explained. The sensitivity of the results to alternative models or assumptions must also be presented. Unfortunately, EPA's description of its analysis is lacking in clarity, in details, and in supporting documentation. The description of the analysis relied upon by EPA consists of a single paragraph at the bottom of page 8-45. The resulting model is described in broad terms as the most parsimonious two-stage model that agrees with the tumor incidence data and focal lesion data. EPA must describe its analysis in sufficient detail to allow the reader to understand how EPA arrived at its preferred model and how robust those results are -- i.e., to what extent would other assumptions be reasonable, adequately fit the data, and lead to different levels of risk. For example, EPA's preferred analysis uses data from a feeding study and a gavage study and thus had to make assumptions in order to combine data from two different types of studies. EPA does not describe how this issue was handled, nor detail the necessary assumptions. Also, EPA does not describe clearly what data were used in its preferred analysis. It states only that the tumor incidence data from the Kociba et al. (1978) study and the focal lesion data from Maronpot et al. (1993) were used. The precise form of the tumor data (whether individual animal time-to-tumor data or summary data) is not stated. Similarly, the reader is not informed as to whether the data from Maronpot. et al. involving initiation with DEN or the data not involving such initiation were used in EPA's preferred analysis. The focal lesion data used in developing EPA s cancer risk model are not from the published Maronpot et al. (1993) paper but are unpublished results from that study. That fact should be clearly indicated. More importantly, the actual data used by EPA should be made accessible to the reader, by including them in an appendix, if necessary. If time-to-tumor data were used, an assumption was required as to the context in which the hepatocellular tumors from the Kociba et al. (1978; 1979) studies were observed (whether incidental, fatal or some combination). This assumption is not stated, and it may have an important effect upon EPA's conclusions regarding the fits to the data of various models. Appendix C contains additional details of some analyses, but not the ones used to derive EPA's preferred cancer model. A considerable portion of Appendix C is devoted to discussing the non-identifiability of some parameters in certain applications of the two-stage model. In some of these instances, EPA is attempting to estimate up to 16 parameters from what are essentially four data points. It is not surprising that some of the parameters are not identifiable. The last paragraph in Appendix C is particularly disturbing. It starts with the assertion that the calculations have hidden assumptions that can have a bearing on the interpretation of the results. It then goes on to list some of these hidden assumptions and ends with the statement that the analysis is very sensitive to the choice of values for the radius of a cell and to the (assumed) minimum size of a detectable focus. This paragraph (and particularly the last statement re sensitivities of the analysis) raises serious questions as to the reliability of EPA's cancer risk assessment. No information is provided on the cell radius or minimum detectable focus size assumed in EPA's analysis. In fact, this is the only mention that either of these two quantities are required at all in the analysis. EPA must provide enough detail in its analysis to permit the reader to determine how these values were used in the analysis, how EPA selected those values, what values were selected, and the sensitivity of EPA's risk assessment to those selections. EPA must clearly distinguish between what is being assumed (i.e., what is going into the modeling) and what is being concluded as a result of the modeling. Although EPA's preferred dose response model is linear, it seems clear that a threshold model would provide an equivalent or nearly equivalent description of the data. This is the most important issue in the dose response- modeling and should be thoroughly explored in EPA's analysis. Even if EPA's risk assessment based on the animal data is correct, without additional assumptions regarding the relative sensitivities to dioxin of various types of tumors in humans and animals, the risk assessment based upon animal data only provides estimates of the risk of liver tumors in a single strain of female rats. Therefore, despite limitations in the human data for dioxin, it would have been appropriate for EPA to have conducted a more compre hensive analysis of the human data. EPA's risk assessment based on human data is derived from the published data from three studies. Reliance on these published data necessitated a number of assumptions and approximations by EPA that could have been avoided by use of the raw data from other studies. The Committee also recommends that the data from the Ranch Hand cohort, when published, be considered for inclusion in this analysis. The cancer risk assessment models applied to the human data by EPA are conceptu ally flawed. [Several Members of the Committee contend that the cancer risk model is not particularly flawed, but represents an acceptable approach.] Both the additive and multiplicative models express the cancer mortality rate at a given age as a function of a summary measure of exposure. Clearly, this summary measure should involve only prior exposures and not future exposures. However, the summary exposure measure used by EPA is average lifetime exposure, which involves both past and future exposures. A practical consequence of this model misspecification is that in the case of the additive model, the increased mortality rate under constant exposure is independent of age (e.g., a person has the same probability of dying of a dioxin-induced cancer between the ages of one and two as between the ages of 70 and 71), which is clearly inappropriate. EPA appeared to recognize this and made a compensating ad hoc adjustment to the risk estimate obtained from the additive risk model. The Committee recommends that both the additive and multiplicative models be reformulated to incorporate more biologically plausible summary exposure measures (e.g., cumulative past exposure). In EPA's cancer risk calculations based upon epidemiological data, exposures in dioxin- exposed subcohorts were summarized by the median exposure. The stated reason for choosing the median (over, say, the mean) was that body levels were quite variable and not symmetrically distributed." Neither of these assertions appear to be supported by the draft document. At any rate, these are not appropriate reasons for selecting the median over the mean as a summary measure of exposure, and there appears to be no clear reason stated for preferring the median over the mean. This choice would not be required if the raw data were used in the analysis, which illustrates an advantage of basing an analysis upon the raw data. Given the problems and limitations identified with the analysis, it is not clear that this work added significantly to our understanding of the dose-response for TCDD. Rather than giving a high priority to refining these techniques at the present time, the Committee recommends that EPA review this effort, and communicate clearly the strengths and limita tions of the work. The Agency should evaluate critically the potential for future work in this area to elucidate the dose-response for TCDD in humans. 4.9.2 Use of the RfD in Evaluating Incremental Exposures (Charge Question 18) The question of possibly rejecting of the Reference Dose (threshold) approach for evaluating incremental exposure because of the existing background levels relates also to health Charge questions 14 and 17 (as noted earlier). All three issues are part of a more fundamental problem which has not been addressed and needs to be included to provide balance to the reassessment document. This fundamental issue concerns the basis for the selection of the dose response relationship to be used in assessing the (non-cancer) adverse effects of dioxin. The selection of the dose response function clearly distinguishes the EPA approach from that of some (but not all) other agencies and bodies which have carried out similar risk assessments on dioxin. Although all of these groups used the same toxicologic and epidemiologic data bases as EPA, all except EPA have elected to use some type of threshold and safety factor methodology for their health risk evaluation. Chapter 8 of the reassessment document needs to describe and evaluate this alternative dose response relationship, discuss the approaches and findings of the other relevant agencies, and justify the basis for selecting another approach. Although the Agency concludes (page 8-13 of the reassessment document) that the use of the linear multistage model (LMS) needs to be re-evaluated, the re-evaluation consists of enhancing the LMS approach with a PBPK analysis and a 2-stage analysis rather than a discussion of alterative approaches. A discussion of such alternative approaches, and their results, should also be reflected in the appropriate places in the summary chapter 9. As noted earlier in this report (see sections 4.5.2 and 4.5.3), dioxin is not an initiator and thus is not a complete carcinogen. Dioxin is a non-genotoxic promotor which acts at least in part via the Ah receptor and exhibits numerous U-shaped dose responses (Kociba et al., 1978; Pitot et al., 1987; Maronpot et al., 1993; Teegaurden et al., 1995; Fang et al., in press). Thus, the document cannot ignore a possible threshold dose-response relationship and claim to be comprehensive in its presentation. As noted in the Committee's response to health question 14 (section 4.6.1), the available information on TCDDs suggest that use of the benchmark approach, rather than the reference dose, is probably more appropriate. This approach has been recommended in several previous SAB reports (SAB, 1990; SAB, 1995). The EPA, along with the International Life Sciences Institute (ILSI), has sponsored workshops on this topic, and various EPA staff are among the most progressive and knowledgeable experts in the use of this methodology. The Committee (with the exception of one Member) agrees that, in concept, the reference dose is not designed to evaluate the risk from incremental exposures (however, if background exposures are not accounted for in the population from which data are obtained for calculating a reference dose, the resulting reference dose may represent doses in excess of that background). Although EPA s current methodology for cancer risk assessment allows one to assess risks from incremental exposures, the RfD methodology is not well-suited for this particular use. The Committee recommends that EPA work towards developing and implementing a methodology that would allow the assessment of non-cancer risk resulting from incremental exposures. 4.9.3 Continuum of Response Postulate (Charge Question 19) EPA postulates a continuum of response from events seen at low doses that are not toxic but cause the subsequent development of toxic effects. This idea is expressed several times in the document, but it is not supported by a full discussion. As it stands now, the basis for this statement regarding a continuum of responses is unclear. The statement is far too general and could be taken as implying that all (or any) early changes will necessarily lead to ultimate toxicity. The statement is only defensible in reference to a limited number of specific case examples, but cannot be taken as universally proven. Until a full mechanism of action has been mapped out, the reassessment's position remains unproved in general. The state ment should not be presented as a "postulate (which is widely accepted as a universal truth not requiring proof) but as a current hypothesis (subject to change as new data are discov ered). The specific case developed by EPA to support this hypothesis was the binding of ligand to the Ah receptor, which is then assumed to lead to all, or most, of the toxic responses. The possibility that the Ah receptor system may be a sensing pathway to protect the cell, not an integral part of the machinery associated with the toxic response of a cell to TCDD, is not considered. [One Member of the Committee asserts that The Ah receptor system is probably part of the normal cellular second messenger regulating systems and its inappropriate activation by dioxin (and dioxin-like compounds) can lead to assorted toxic outcomes through a variety of pathways. Any suggestion that this inappropriate activation has positive outcomes is strictly speculative at this time. ] Only in the mechanism of action chapter is there any suggestion that this association may not be universally accepted. However, Ah receptor binding may not be the ultimate mecha nistic step in all responses. It is not proven for all possible toxic endpoints; the associa tion is strongest for enzyme induction in mice, but may not hold in other species. Different strains of rat show remarkably diverse sensitivities to dioxin while possessing active Ah receptors. Although enzyme induction in mice is the classic Ah receptor mediated response, there may be other responses that do not involve Ah receptor. For example, although some immuno-responses are Ah associated in mice, others are appar ently not. [One of the Committee s immunologist participants believes that The evidence for receptor independence from this data is insufficient to contradict the majority of evidence for receptor dependence."] Ah receptor binding may not be a rate-limiting response. The recent development of knock out mice lacking functional Ah receptors may help clarify these points. EPA needs to leave itself some flexibility so that the assessment can remain valid in light of future discoveries. The EPA response at the public review meeting suggested that Ah receptor binding could be considered as necessary but not sufficient. Other events may well also be needed for toxicity to occur. Because of this, TEFs should be based solely on Ah receptor data only when other data are unavailable. There are a variety of individual effects yielding a likelihood of response, a cascade of events. The Committee is not taking the position that EPA is wrong in its view that Ah receptor binding is a critical early event leading to eventual toxicity, but rather that the Agency stated the idea too strongly and without sufficient consideration of toxic immune system effects that have not been shown to be Ah receptor associated. Alternative mechanisms that have been suggested in the published literature were not considered in the document. The evidence for Ah receptor- linked effects (and their role in toxicity), and (for balance) the evidence suggesting a lack of involvement of Ah receptor binding in some effects, should be discussed in the document to support acceptance of this continuum as a current hypothesis. Further more, the continuum theory needs further explanation, some discussion of the merits and limitations, and an indication of the acceptance of this idea in the scientific community. EPA needs to be more flexible in its statements, to allow adaptability to scientific evidence that may be developed in the future. 4.10 Use of Toxicity Equivalence Factors (TEFs) (Charge Question 20) In general, the Committee agrees that the use of a TEF is a valid approach provided that the contribution to the TEQ from: a) TCDD; b) other dioxins and furans; and c) coplanar PCBs are explicitly stated. However, when assessing the toxicity of complex mixtures that are not well-defined, the Committee believes that presenting the results using alternative methods may be warranted whenever possible. It must be noted however, that other than the suggestion (see Section 4.13 below) to apply TEQs separately for 2,3,7,8-TCDD, other dioxins and furans, and co-planar PCBs, as well as for environmental mixtures as a whole, the Committee has no specific proposals for such alternative methods. Although the assessment acknowledges some of the uncertainties associated with the use of TEFs/TEQs, these issues have not been satisfactorily addressed. Since the TEQ approach has been used throughout the assessment document, and many of its conclusions (e.g.,, the position that levels 10-100 times over background pose a possible human health hazard) hinge on the validity of the TEF values and assumptions used, the Committee advises EPA to include a peer-reviewed appendix that will comprehensively review EPA's use of the TEF/TEQ approach in the exposure and health assessment documents. This appendix should clearly outline the assumptions and TEF values used throughout the documents as well as address the following issues: The reassessment document acknowledges that dioxin-like compounds other than TCDD represent greater than 90% of the calculated TEQ value in some instances. Conse quently, the applicability of using TEFs for mixtures containing PCBs with partial agonist and antagonist activities should be addressed. The EPA assumes that there is additivity among TEQs calculated from the TEF values for 7 of the 75 dioxins, 10 of the 135 dibenzofurans, and 13 of the 209 PCBs. However, these dioxin-like congeners constitute a small percentage of the total congeners present in an environmentally relevant mixture. Therefore, the EPA should address the issue of possible interactions, since there is evidence that non-dioxin-like PCBs antagonize several biochemical (e.g. enzyme induction) and toxic responses (e.g., teratogenic and immunotoxic effects) elicited by more potent congeners. Possible synergies should also be considered. New data, which became available since the release of the document have resulted in adjustments to several of the TEF values. The Committee suggests that a comprehensive review of all TEF values be summarized within the appendix for each congener that has previously been assigned a value. In addition, EPA should clearly state the species and responses (e.g., ligand binding, enzyme induction, immunotoxicity) used to derive the TEF value. Finally, since TEF values can vary dramatically based upon the species and response examined, EPA should justify the TEF value that has been selected for evaluating human risk. EPA should document clearly the studies that demonstrate additivity among dioxins, dibenzofurans and PCBs and that the TEF/TEQ approach accurately predicts both short-term (e.g., immunotoxic effects, enzyme induction) and long-term (e.g., carcinogenic, teratogenic, reproductive effects) responses elicited by these complex mixtures. In the event there are insufficient data demonstrating the applicability of the approach for specific toxic endpoints, EPA should justify its position for the use of the TEF/TEQ approach. Although the reassessment recognized that humans are not overly sensitive or resistant to the effects of dioxin-like compounds relative to other species, there is a lack of discussion regarding the differences between human and rodent receptors. Such information should be included in the appendix. EPA should also outline how this information is taken into account when assigning TEF values to congeners. Also, recent studies indicate that there are differences in congener uptake, metabolism, elimination, and storage. This should be acknowledged followed by a discussion of how this information is taken into account when assigning TEF values to congeners. The reassessment document lacks discussion regarding naturally occurring dioxin-like compounds such as benzo[a]pyrene and indole[3,2-b]carbazole. The EPA should comment on the exclusion of these compounds. These agents are not persistent or bioaccumulative, but the precursor of indole[3,2-b]carbazole is present at high levels in the diet, and this constant level of exposure should be reported. Furthermore, there is (very preliminary) evidence that some of these compounds (e.g., indole[3,2-b]carbazole) may have an anticarci nogenic effect (Bjeldanes et al., 1991; Bradlow et al., 1991; Liu et al., 1994; and Wattenberg et al., 1978). As mentioned in the health assessment document (p 9-70), A more detailed description of these issues is contained in U.S. EPA (1989). This referenced document (TITLE, EPA/625/3-89/016), once updated and modified to address the issues listed above, and peer reviewed, could be used as a basis for the appendix and therefore, could reasonably accommodate the Committee's above suggestions. In addition, a balanced comprehensive review should clearly state the assumptions and limitations of the TEF/TEQ approach as well as highlight the areas that warrant further investigation. Examination of the Summary of the Public Comments related to the Exposure and Health documents (EPA/600/6-88/005Ca, Cb, Cc and EPA/600/BP-92/001a, 001b, 001c) indicates that EPA is already aware of the concern surrounding this issue (i.e. the statement that This issue was the single most addressed issue among all of the comments on the risk characterization on page 25 of the Health Comments Summary). Therefore, the addition of a peer reviewed appendix dealing with the TEF/TEQ approach should satisfy several of the concerns raised, or at least indicate which items are still points of contention. Please continue to Section 4.11. |
|||||||||||||||
|
The Cast | The Tragedy
| The
Comedy | The
Ending | News
| Library
| Seating
Plan | Top
|
Home |
CALL (812) 333-8888 FOR MORE INFORMATION OR E-MAIL info@copa.org | |||||||||||||||
The Coaltion Opposed to PCB
Ash in Monroe County, Inc. is a nonprofit organization. | |||||||||||||||