Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
5 Cumulative Risk Assessment of Phthalates and Related Chemicals Our understanding of the toxicity of phthalates and the associated underly- ing mechanisms has improved considerably in the last few years. Effects on re- productive development in the male constitute one of the most sensitive end points. Some phthalatesâsuch as DBP, BBP, DEHP, and DINPâare able to disrupt male sexual differentiation by interfering with androgen biosynthesis; this culminates in what has been described as the phthalate syndrome or more generally as the androgen-insufficiency syndrome. Because the chemicals have a similar effect spectrum, it is likely that they act in concert when they occur to- gether. However, not only phthalates can disrupt male sexual differentiation. As discussed in Chapter 3, other classes of chemicals, so-called antiandrogens, are also able to interfere with male development by opposing the actions of fetal androgens in different ways. Antiandrogens can block the effects of fetal andro- gens by antagonizing the androgen receptor (AR) or can reduce concentrations of fetal androgens by inhibiting key enzymes responsible for the conversion of precursor steroids into androgens. Other chemicals exhibit mixed mechanisms, for example, by both inhibiting enzymes and blocking the AR. Thus, there may be considerable potential for cumulative effects of phthalates and other classes of antiandrogens in that any interference with AR-related effects may result in components of the phthalate syndrome. This chapter assesses the empirical evidence of combined effects of sev- eral phthalates, of nonphthalate antiandrogens, and of phthalates and these other antiandrogens. Because of the importance of developmental effects, the over- view focuses almost exclusively on experimental evidence from reproductive- toxicity studies. Many published experimental mixture studies were motivated by an interest in determining the type of combination effect (for example, addi- tive or synergistic) of the agents involved. That effort often required the admini- stration of doses of test chemicals that were associated with measurable effects but were far removed from exposures experienced by humans. What will lend further urgency to calls to conduct cumulative risk assessment is the demonstra- 106
Cumulative Risk Assessment of Phthalates and Related Chemicals 107 tion of combined effects at low doses of each mixture component. For that rea- son, the committee scrutinized the evidence in the literature particularly with respect to low-dose combined effects. After examining the empirical evidence, this chapter considers options for conducting cumulative risk assessment of phthalates and other antiandrogens. First, several questions are addressed to set the stage for considering various approaches. Which phthalates should be subjected to cumulative risk assess- ment? Should other antiandrogens be included? If so, which ones? What criteria should be used to group phthalates and other antiandrogens for cumulative risk assessment? Next, approaches to quantitative assessments of cumulative effects are discussed. For cumulative risk assessments of dioxins and other chemical classes, the toxicity equivalency (TEQ) concept has gained broad acceptance and is in widespread use. Accordingly, this chapter addresses whether the TEQ concept presents a practicable option for cumulative risk assessment of phthalates and other antiandrogens or whether alternative approaches should be adopted. The chapter concludes with a discussion of possible stepped ap- proaches to cumulative risk assessment of phthalates and other antiandrogens. CRITERIA FOR CHOOSING DOSE ADDITION OR INDEPENDENT ACTION AS A DEFAULT EVALUATION METHOD Dose addition and independent action (here used synonymously with re- sponse addition) provide two possible approaches to dealing with the mixture issue. However, when one is faced with the task of evaluating specific mixtures, the issue arises as to whether either of the two concepts is appropriate for the mixture in question and should be chosen for assessment. That question be- comes all the more important when the two concepts produce different predic- tions of mixture effects. However, in only a few cases have dose addition and independent action been evaluated together against the same set of experimental mixture data with the aim of establishing whether either approach produces valid predictions of combined effects (for a review, see Kortenkamp et al. 2007). As pointed out by the U.S. Environmental Protection Agency (EPA 2000), the empirical basis of choosing between dose addition and independent action as a default approach for risk assessment is not strong. The decision in favor of either approach as a default for mixture risk assessment is based largely on perceptions of whether the scientific assumptions that underpin dose addition or independent action are met. For such purposes, the two concepts have been allied to broad mechanisms of combined toxicity, as described below. Dose addition is often stated to be applicable to mixtures composed of chemicals that have a similar or common mechanism of action (EPA 1986, 2000, 2002; COT 2002). However, the original paper by Loewe and Muischneck (1926) contains little that roots dose addition in mechanistic considerations; the idea of similar action probably derives from the âdilutionâ principle, which forms the basis of this approach. Because chemicals are viewed as dilutions of
108 Phthalates and Cumulative Risk Assessment: The Tasks Ahead each other, it may be implicitly assumed that they must act via common or simi- lar mechanisms. In contrast, independent action is widely assumed to be appropriate for mixtures of agents that have diverse or dissimilar mechanisms of action. Al- though it is rarely stated, that assumption probably stems from the stochastic principles that guided the development of the approach. Acting independently is equated with the notion of acting through different mechanisms. By activating differing effector chains, the argument goes, every component of a mixture of dissimilarly acting chemicals provokes effects independently of all other agents that are present, and this feature appears to lend itself to statistical concepts of independent events. Independent action is often held to be the default assess- ment concept when the similarity criteria of dose addition appear to be violated (COT 2002). If âdissimilar actionâ is taken implicitly as the simple negation of âsimilar action,â it is then assumed that independent action must hold (with the further implicit assumption that only two choices are available), even without further proof that the underlying mechanisms satisfy the dissimilarity criterion. Although those ideas are plausible, their application to specific combina- tions of chemicals is far from clear-cut. One major difficulty lies in defining reliable criteria for similarity of mechanisms of action. Often, the induction of the same phenomenologic effect is deemed sufficient for accepting similarity of action. However, that could be inappropriate for some combinations of chemi- cals that operate by distinct molecular mechanisms. At the other extreme of the spectrum of opinion, the similarity assumption might require an identical mo- lecular mechanism involving the same active intermediates. That position, with its strict similarity criterion, may mean that few chemicals qualify for inclusion in mixture-effects assessments and many others that provoke the same response are left out. In effect, that approach would provide an unrealistically narrow per- spective on existing mixtures. A middle position is occupied by the view that interactions with the same site, tissue, or target organ should qualify for similar- ity (EPA 1986, 1989; Mileson et al. 1998). SIMILAR OR DISSIMILAR ACTION: A DEFAULT CONCEPT FOR CUMULATIVE RISK ASSESSMENT OF PHTHALATES AND OTHER ANTIANDROGENS? It is not immediately obvious which criteria should be used to classify phthalates as similarly or dissimilarly acting chemicals. EPA (2000) has recom- mended that decisions about whether to use dose addition or independent action should be based on information about the toxic and physiologic processes in- volved, the single-chemical dose-response relationships, and the type of re- sponse data available. If information about target tissue concentrations is avail- able, such judgments can focus on the toxic mechanism of action within that tissue. With phthalates, external doses, but not target-tissue doses have been used, and in such cases EPA (2000) demands that decisions about similarity of
Cumulative Risk Assessment of Phthalates and Related Chemicals 109 action consider all processes, including uptake, metabolism, elimination, and toxic mechanism. Although there is little detail about the precise uptake mechanisms of phthalates, it is clear that they all undergo hydrolysis to produce phthalate monoesters that are then transported to their site of action. In the case of many monoesters, there is a rapid reduction in fetal testosterone production in the Leydig cells of the testes and a consequent indirect effect through down- regulation of key enzymes important in the transport and conversion of steroid precursors. The resulting impairment of Leydig cell function triggers a decrease in androgen-mediated gene expression. Because androgen action is a key driver in sexual differentiation, disturbance of androgen-mediated development gives rise to profound effects on the male reproductive system (Foster 2005). Phthal- ates can be judged to exhibit a similar mechanism of action, with dose addition the appropriate default assessment approach according to EPA guidelines. How- ever, differences in phthalate metabolism may lead to dissimilarities in the pre- cise toxic mechanism. For example, some metabolites of DEHP (MEOHP and MEHHP) can antagonize the AR (Stroheker et al. 2005), but that is not the case for all metabolites of phthalates. Furthermore, some phthalates can induce some peroxisome-proliferator-activated receptor isoforms, but others lack this ability (Bility et al. 2004). Should the mechanisms of action of those phthalates there- fore be judged to be dissimilar and independent action adopted according to other suggestions in EPA guidelines? EPA (2000) also stipulated that to qualify for dose addition, all dose- response curves should be congruent. That requirement is not met by phthalates. Dose-response studies have revealed a large variety of shapes (Rider et al. 2008; Howdeshell et al. 2008). Does that mean that the concept of dose addition should not be used in connection with phthalate mixtures? The demand for congruent curves may be derived from a misunderstand- ing of the mathematical features of the dilution principle that underpins dose addition. It appears to have been thought that the principle requires a constant proportionality between the doses of chemicals in a mixture that produce a given effect. For chemicals with different potency, a chemical may have to be admin- istered at a dose that is a multiple of another chemicalâs dose to achieve the same effect of a specific size. Although that proposition would lead to dose ad- dition, it does not follow that the dose-response curves of all dose-additive mix- ture components have to be congruent. Congruent curves result only if it is also demanded that the multiple is constant for all effect levels. For example, let 10 dose units of substance A induce an effect of 12 (on an arbitrary scale), and as- sume that only 5 dose units of substance B are required to produce the same effect. In that case, there is a proportionality factor of 2 between the doses of the two chemicals. Congruent shapes of the dose-response curves result if the same proportionality of 2 is preserved at all other effect levels. Thus, in this arbitrary example, 20 dose units of A and 10 dose units of B each would produce the same but larger effect of, say, 17. Now consider mixture effects under the prin- ciple of dose addition. Accordingly, 5 dose units of A and 2.5 of B are expected
110 Phthalates and Cumulative Risk Assessment: The Tasks Ahead to produce an effect of 12. To provoke an effect of 17, dose addition would re- quire application of 10 dose units of A and 5 of B. However, dose addition would apply even if the demand of constant proportionality between the effect doses of the two chemicals is not fulfilled. For example, let 20 dose units of A produce an effect of 17, as before, but assume that B has a lower potency at this effect level, such that 16 dose units are necessary to yield an effect of 17. Dose addition would still be applicable: it can be predicted that 10 units of A and 8 units of B combined should produce an effect of 17. Thus, although the addi- tional requirement of constant proportionality is a precondition for the applica- tion of relative potencies and toxicity equivalence factors (TEFs), it is not neces- sary for the general use of dose addition, which also works with curves of different shapes (Berenbaum 1989; Hass et al. 2007; Howdeshell et al. 2008). Moreover, the existence of congruent dose-response curves for mixture compo- nents does not constitute evidence for or against dose additivity. Given the above discussion, EPA may wish to revise some of its guidance (for example, EPA 2000). The application of the criteria used by EPA (2000) for making choices be- tween dose addition and independent action for phthalate mixtures leaves ambi- guities that cannot readily be resolved without further empirical evidence, which would overrule any such heuristic arguments in any case. The committee concludes that the criteria applied by EPA are too narrow and restrictive because they leave out other chemicals that can disrupt male sex- ual differentiation but in ways that differ in some respects from phthalates (see Chapter 3). With phthalates and other antiandrogens, the case can be made for adopting a physiologic approach to analyzing toxic mechanisms of action with respect to similarity or dissimilarity. If it is recognized that the driver of male sexual differentiation during development is the effect of androgen action, it may be irrelevant whether the hormonesâ effects are disrupted by interference with steroid synthesis, by antagonism of the AR, or by some other mechanism (for example, affecting consequences of AR activation). The resulting biologic effects with all their consequences for male sexual differentiation may be simi- lar, although the molecular details of toxic mechanismsâincluding metabolism, distribution and eliminationâmay differ profoundly in many respects. Judged from such a perspective, a focus on phthalates to the exclusion of other antian- drogens (or other more esoterically acting agents) not only would be artificial but could imply serious underestimation of cumulative risks posed by agents to which there is coexposure. In contrast, the differences in the mechanisms of action of phthalates and other antiandrogens could mean that the independent-action principle is better suited for evaluating the combined effects of chemicals. That issue cannot be decided without considering empirical evidence. Accordingly, the question be- comes, are there data in the recent literature that can help to resolve some of these difficulties?
Cumulative Risk Assessment of Phthalates and Related Chemicals 111 EXPERIMENTAL EVIDENCE OF CUMULATIVE EFFECTS OF COMBINATIONS OF PHTHALATES AND OTHER ANTIANDROGENS This section reviews experimental studies of combined effects of several phthalates, of other antiandrogens, and of phthalates and other antiandrogens. Rather than a comprehensive review of the literature, the primary aim is to ex- amine empirical evidence of combined effects of phthalates and other antiandro- gens. A secondary aim is to assess whether experimentally observed mixture effects agree quantitatively with the additivity expectations derived from dose addition. In dealing with those issues, it is important to recognize that dose addi- tion and independent action often yield identical, experimentally indistinguish- able, or trivially distinct predictions of additive combined effects. However, under some circumstances, the two concepts produce additivity predictions that differ enough to be distinguished experimentally. It is then possible to discern which concept better agrees with observed effects. In such cases, the argument for using the better predictor (dose addition or independent action) as an ap- proximation for mixture risk assessment is strong. For that reason, published data will, wherever possible, be examined in relation to agreement with dose addition or independent action. It is not always straightforward to judge the quality of agreement between experimentally observed data and predictions based on dose addition or inde- pendent action. Although it is frequently possible to distinguish qualitatively which of the two concepts approximates the data better, there are no generally accepted criteria for statistical assessments. One approach is to demand that the predictions overlap with the confidence intervals of the experimental data, but this may lead to an overly strict criteria for agreement. An alternative would be to consider how variations in single-chemical response data affect the uncertain- ties associated with predictions by using boot-strapping methods. In that way, confidence intervals for predictions can be calculated (Hass et al. 2007). The agreement with observations can then be judged statistically by considering the overlap between the confidence intervals of the prediction with that of the ex- perimental data. Frequently, however, data quality and experimental design (or the lack of information presented in the literature) do not allow the use of such approaches. In the absence of generally accepted criteria for assessing agree- ment with predicted additivity, qualitative judgments often have to be made without the use of statistical reasoning. Although that approach may be unsatis- factory, the committee emphasizes that there are currently no practical alterna- tives, owing to a lack of theoretical foundations able to underpin better practice. Combinations of Phthalates Yield Good Evidence of Dose-Additive Effects Howdeshell et al. (2007) examined a binary mixture of DBP and DEHP. Those two phthalates are thought to have a common mechanism of action, but they yield different metabolites. Pregnant Sprague-Dawley rats (six dams per
112 Phthalates and Cumulative Risk Assessment: The Tasks Ahead dose) were exposed to the phthalates during gestation days 14-18 at 500 mg/kg- d each, both singly and in combination. Their male offspring were examined for a wide array of effects typical of disruption of male sexual differentiation, in- cluding changes in fetal testosterone production, changes in anogenital distance, epididymal agenesis, retained nipples, gubernacular agenesis, hypospadias, and number of animals with malformations. Dose addition generally predicted larger effects than independent action, although for some end points the two concepts predicted equal effects. It is not possible to duplicate the dose-addition predic- tions given by the authors, because they were based on unpublished dose- response data on the individual phthalates. However, the authors observed that the responses generally agreed well with dose addition and were higher than the additivity expectations derived from independent action for changes in anogeni- tal distance, epididymal agenesis, and number of malformed males. The study indicates that dose addition provides fairly good predictions of many of the ef- fects that make up the androgen-insufficiency syndrome. Independent action often underestimated the observed responses. Recently, Howdeshell et al. (2008) presented the results of a mixture study of five phthalates in which suppression of fetal testosterone production at gesta- tion day 18 was measured as a result of exposure of pregnant Sprague-Dawley rats. BBP, DBP, DEHP, DIBP, and DPP were combined in a fixed ratio. The committeeâs reanalysis of the raw data revealed that for testosterone reduction, dose-addition and independent-action predictions were generally similar (see Figure 5-1 and Appendix C for further details and analysis). Over a large range of effect levels, the observed reductions in testosterone production agreed well with the responses predicted by either model, although there were small, statisti- cally significant differences between the dose-addition prediction and the ob- served data. Combinations of Antiandrogens Follow the Principle of Dose Addition By using the isobole method (an application of dose addition), Nellemann et al. (2003) found that the fungicides procymidone and vinclozolin, both AR antagonists, additively inhibited testosterone binding to the AR. Administration of a 1:1 mixture to castrated, testosterone-treated male rats led to dose-additive alterations in reproductive organ weights, androgen concentrations, and AR- dependent gene expression. Birkhoj et al. (2004) extended the use of the isobole method to three-component mixtures of the pesticides deltamethrin, methiocarb, and prochloraz. An equimolar mixture of the three additively suppressed AR activation in vitro. When a combination of those three with simazin and tribenuron-methyl was given to castrated testosterone-treated rats, changes in adrenal gland and levator ani weights and in expression of AR-associated genes were observed. The combination of all five chemicals had effects that were not found for the individual pesticides, but whether the effects were dose-additive could not be assessed by the authors.
Cumulative Risk Assessment of Phthalates and Related Chemicals 113 1.2 1 Testosterone (fraction of control) . 0.8 0.6 0.4 0.2 0 0 200 400 600 800 1000 1200 1400 Mixture dose (mg/kg-d) Model Dose addition Independent action Data mean and 90% confidence interval FIGURE 5-1 The committeeâs reanalysis of the combined effects of five phthalates on suppression of testosterone production (Howdeshell et al. 2008). See Appendix C for further details. A mixture of the AR antagonists procymidone and vinclozolin was evalu- ated in the Hershberger assay (reviewed by Gray et al. 2001). Although it was not possible to evaluate the dose-additivity prediction with the information pro- vided, the mixture appeared to exhibit effect addition in percentage reduction of ventral prostate and levator ani weights.1 Hass et al. (2007) examined a mixture of three AR antagonists (vinclo- zolin, flutamide, and procymidone) in an extended developmental-toxicity model in the rat. Disruption of sexual differentiation in male offspring was stud- ied; the end points were changes in anogenital distance (AGD) and nipple reten- tion (NR). On the basis of AGD changes, the joint effect of the three chemicals was predicted well by dose addition, but the observed effects on NR were slightly greater than those predicted by dose addition. In this study, the agree- ment between dose addition and experimentally observed responses was evalu- ated statistically by using boot-strapping methods. Metzdorff et al. (2007) analyzed further the material from the Hass et al. (2007) study by following effects typical of antiandrogen action through differ- ent levels of biologic complexity. Changes in reproductive organ weights and of androgen-regulated gene expression in prostates of male rat pups were chosen as 1 In effect addition, the combined effect of several chemicals is calculated by summing the responses to the individual agents at the doses present in the mixture.
114 Phthalates and Cumulative Risk Assessment: The Tasks Ahead end points for extensive dose-response studies. With all the end points, the joint effects of the three antiandrogens were dose-additive. That conclusion is sup- ported by a statistical evaluation of the agreement between dose-addition predic- tions and observations that the study authors conducted by judging overlap of confidence intervals of the prediction and the experimental data. In the examples presented here, the AR antagonists evaluated in the stud- ies are known to induce antiandrogenicity by the same mechanism. Combinations of Phthalates with Other Antiandrogens Also Exhibit Dose-Additive Effects Hotchkiss et al. (2004) investigated a mixture of BBP and linuron, an antiandrogen capable of antagonizing the AR and disrupting steroid synthesis. The combination decreased testosterone production and caused alterations in androgen-organized tissues and malformations of external genitalia. Quantitative additivity expectations based on the effects of the single chemicals were not calculated in this study, so agreement with dose addition or independent action cannot be assessed. However, the combination of BBP and linuron always pro- duced greater effects than each chemical on its own. That result demonstrates that BBP and linuron can act together to produce an effect spectrum typical of disruption of androgen action. Rider et al. (2008) conducted mixture experiments with the three phthal- ates BBP, DBP, and DEHP in combination with the antiandrogens vinclozolin, procymidone, linuron, and prochloraz. The mixture was given to pregnant rats with the aim of examining the male offspring for a variety of developmental effects typical of antiandrogens. Its components have a variety of antiandrogenic mechanisms of action. Vinclozolin and procymidone are AR antagonists, and linuron and prochloraz exhibit a mixed mechanism of action: inhibiting steroid synthesis and blocking the steroid receptor. In calculating additivity expecta- tions, the authors used historical data from their laboratory; however, the studies sometimes had dosing regimens that differed from those used in the mixture experiments. Data on the effects of some individual phthalates were not avail- able. To bridge that data gap for the purpose of computing additivity expecta- tions, it was assumed that the three phthalates were equipotent. Despite some uncertainty inevitably introduced by that assumption, dose addition gave predic- tions of combined effects of the mixed-mode antiandrogens that agreed better with the observed responses than did the expectations derived from independent action. For a number of end pointsâincluding seminal vesicle weights, epidid- ymal agenesis, and NRâthere was reasonable agreement with dose addition. For others, such as hypospadias, the observed effects exceeded the dose-addition expectation. A statistical evaluation of the agreement between dose addition and experimental data was not provided by the study authors, and the committee judged that such an analysis was not possible on the basis of the published data.
Cumulative Risk Assessment of Phthalates and Related Chemicals 115 Nevertheless, independent action led to considerable underestimation of the ob- served combined effects in all cases. Table 5-1 summarizes the mixture studies that allowed quantitative com- parison of observed combined effects with predictions derived from dose addi- tion. The committee notes that the studies revealed a large variety of differently shaped dose-response curves for phthalates acting individually (Howdeshell et al. 2008; Rider et al. 2008) and antiandrogens acting individually (Hass et al. 2007). The studies provide empirical examples in which chemicals with similar mechanisms can have entirely different dose-response curves. COMBINED EFFECTS OF LOW DOSES OF PHTHALATES AND OTHER ANTIANDROGENS When it comes to judging the risks associated with low-level exposures, there are marked differences between the chemical-by-chemical approach to risk assessment and evaluations that take mixture effects into account. Where single- chemical risk assessments might yield the verdict âabsence of risk,â dose addi- tion or independent action might yield the opposite conclusion. An obvious deduction from the dilution principle of dose addition is the expectation that every component at any dose contributes, in proportion to its prevalence, to the overall mixture toxicity. Whether the individual doses of mix- ture components are effective on their own does not matter. The idea can be illustrated by considering a dose-fractionation experiment (see Figure 5-2), where a dose of 4 Ã 10-2 arbitrary dose units produces an effect of measurable magnitude. The same effect will be obtained when the chemical is administered in 10 simultaneous portions of 4 Ã 10-3 dose units, even though the response to each one of those dose fractions is not measurable (or is exactly zero if there is a true dose threshold). If dose addition applies, the same holds when 10 portions of 10 chemicals with identical response curves are used. Thus, com- bined effects should also result from chemicals at doses associated with zero effect (dose thresholds) or even lower doses, provided that sufficiently large numbers of components sum to a suitably high effect dose. Theoretically, the situation described above is not necessarily the case ith independent action where simultaneous exposure to large numbers of chemicals at doses associated with zero effects is expected to produce a zero mixture ef- fect. An experimental assessment of that idea, however, is complicated by the fact that true zero effect levels (dose thresholds), if they exist at doses larger than zero, are difficult to determine empirically. Particularly in the case of mix- tures of a large number of components, that proposition forces clear distinctions between zero effects and small, albeit statistically insignificant effects. For ex- ample, under independent action the combined effect of 100 chemicals, each of which individually provokes a response of 1%, will be 63% of a maximally in- ducible effect. If each of the 100 chemicals produces an effect of only 0.1%,
TABLE 5-1 Mixture Studies of Phthalates and Other Antiandrogens 116 End Point Assay or Organism Mixture Components Assessment Reference Mixtures of phthalates In vivo, suppression of testosterone Sprague-Dawley rats exposed in BBP, DBP, DEHP, DINP, DPP ~ DA or IAa Howdeshell et al. synthesis utero 2008 Mixtures of antiandrogens In vitro, inhibition of androgen- AR CHO cell-based AR reporter Procymidone, vinclozolin = DA Nellemann et al. induced AR activation gene assay 2003 In vitro, inhibition of androgen- CHO cell-based AR reporter gene Deltamethrin methiocarb, = DA Birkhoj et al. 2004 induced AR activation assay (modified) prochloraz, 2 inactive substances In vivo, changes in AR-dependent Castrated testosterone-treated male Procymidone, vinclozolin = DA Nellemann et al. gene expression Wistar rats (extended Hershberger 2003 assay) In vivo, changes in hormone Castrated testosterone-treated male Procymidone, vinclozolin = DA Nellemann et al. concentrations: LH and FSH Wistar rats (extended Hershberger 2003 assay) In vivo, changes in reproductive organ Castrated testosterone-treated male Procymidone, vinclozolin = DA Nellemann et al. weights Wistar rats (extended Hershberger 2003 assay) In vivo, effects on androgen-regulated Male Wistar rats exposed in utero Vinclozolin, flutamide, = DA Metzdorff et al. 2007 gene expression in prostate and postnatally procymidone In vivo, changes in reproductive organ Male Wistar rats exposed in utero Vinclozolin, flutamide, = DA Metzdorff et al. 2007 weights and postnatally procymidone In vivo, changes in AGD Male Wistar rats exposed in utero Vinclozolin, flutamide, = DA Hass et al. 2007 and postnatally procymidone In vivo, NR Male Wistar rats exposed in utero Vinclozolin, flutamide, = DA at low Hass et al. 2007 and postnatally procymidone effective doses; > DA in median effects rangeb
TABLE 5-1 Continued End Point Assay or Organism Mixture Components Assessment Reference Mixtures of phthalates and antiandrogens In vivo, changes in AGD Male Sprague-Dawley rats Vinclozolin, prochloraz, ~ DA Rider et al. 2008 exposed in utero procymidone, linuron, BBP, DBP, DEHP In vivo, changes in NR Male Sprague-Dawley rats Vinclozolin, prochloraz, ~ DA Rider et al. 2008 exposed in utero procymidone, linuron, BBP, DBP, DEHP In vivo, changes in seminal vesicle Male Sprague-Dawley rats Vinclozolin, prochloraz, = DA Rider et al. 2008 weights exposed in utero procymidone, linuron, BBP, DBP, DEHP In vivo, hypospadias Male Sprague-Dawley rats Vinclozolin, prochloraz, > DA Rider et al. 2008 exposed in utero procymidone, linuron, BBP, DBP, DEHP Note: AR, androgen receptor; CHO, Chinese hamster ovary, DA, dose addition; EDx, effective dose at x response level; FSH, follicle-stimulating hor- mone, IA, independent action; LH, luteinizing hormone; NR, nipple retention. a EDx observed / EDx predicted â 1/0.9. b EDx observed / EDx predicted â 1/1.7. 117
118 Phthalates and Cumulative Risk Assessment: The Tasks Ahead 1.8 1.6 1.4 Effect (arbitrary scale) 1.2 1.0 0.8 0.6 1 2 0.4 0.2 0.0 10-3 10-2 10-1 100 Dose (arbitrary units) FIGURE 5-2 Illustration of a âshamâ mixture experiment with chemicals that all exhibit the same dose-response curve. At the low dose to the left (arrow 1, 4 Ã 10-3 dose units), the effect is hardly observable. A combination of 10 agents at that dose (arrow 2, total dose, 4 Ã 10-2 dose units) produces a significant combined effect, consistent with expecta- tions based on dose addition. the expected combined effect will be 9.5%. With the test systems used in toxi- cology, distinguishing such small effects from those seen in untreated controls is practically impossible. It is well established that regulatory toxicology has dealt with the problem of small responses at low doses by using uncertainty factors to approximate zero effect levels for the purpose of estimating âsafeâ exposures of humans. As a starting point for establishing such âallowable,â âacceptable,â or âtolerableâ exposures, no-observed-adverse-effect levels (NOAELs) are used. The NOAEL is the highest dose or exposure at which no statistically or biologically adverse effects can be identified (EPA 1994). It is used as a point of departure for esti- mating tolerable human exposures by dividing by uncertainty factors. A number of shortcomings of NOAELs, however, have been identified. There are problems with a single numerical value adequately reflecting study size and the shape of the underlying dose-response curves (Crump 1984; Slob 1999). NOAELs are not fixed attributes of toxic substances; rather, they reflect features of experimental design. Larger experimental studies will detect effects at lower exposures and thus will yield lower NOAELs (Crump 2002; Scholze and Kortenkamp 2007). To deal with those conceptual problems, the benchmark dose (BMD) has been developed as a statistical tool to determine acceptable exposures to a chemical (Crump 1984). The BMD is a dose that causes a prescribed effect (generally within or close to the experimentally observed range) and is estimated
Cumulative Risk Assessment of Phthalates and Related Chemicals 119 by fitting a regression model to experimental data. Compared with NOAELs, BMDs have the advantage of yielding lower numerical values with data of poor quality. Numerous papers have evaluated the properties of BMDs (summarized in Crump 2002), and the topic has been the subject of a National Research Council evaluation (NRC 2000). Accordingly, the committee felt that an in- depth discussion of the threshold problem in toxicology and the issues surround- ing the use of NOAELs and BMDs as the basis of toxicologic risk assessment was outside the scope of this report. It suffices to say that BMDs have been en- dorsed by EPA as an acceptable replacement of NOAELs whenever appropriate quantitative data are available (EPA 1994). That conclusion is supported by an evaluation of a large database of developmental-toxicity experiments to compare BMD approaches with NOAELs. For continuous response variables, BMDs associated with 5% additional risk produced dose estimates similar to NOAELs (Allen et al. 1994). The issue to be examined here is whether there is evidence that phthalates, in combination with other phthalates or with other antiandrogens, exhibit com- bined effects at doses that are used in risk assessment by regulatory agencies worldwide as points of departure (PODs) for estimating tolerable exposures of humans. Those PODs are typically NOAELs or lower confidence limits of BMDs (BMDLs). A complicating factor is that the majority of combined-effect studies with the chemicals were not carried out with the intention of addressing the low-dose-mixture issue directly. That gap can be bridged by reanalyzing published papers, but the task requires considerations of methodologic issues related to the concept and design of low-dose-mixture studies. Mixture Studies with Doses around Points of Departure for Risk Assessment: Methodologic Considerations A requirement for experimental studies intended to address the issue of mixture effects at doses around PODs for regulatory risk assessment is that such estimates are derived for each mixture component by using the same assay sys- tem (and end point) as chosen for the mixture study, ideally under identical ex- perimental conditions. Ignoring that demand can lead to the inadvertent admini- stration of some or all mixture components at doses exceeding their PODs, which would undermine the aim of the experiment. But delivery of doses smaller than PODs, either by design or by accident, might present problems if the experimental system lacks the statistical power to detect small effects. For example, it would be futile to attempt an experiment in which two agents are combined at one hundredth of their individual PODs. The resulting mixture ef- fect, if it exists, would be too small to be detectable in most cases, and the ex- periment would be inconclusive. Accordingly, a number of criteria can be derived for critical evaluations of experimental mixture studies. First, the effects of individual mixture compo- nents ideally will be determined in parallel with the mixture experiment for the
120 Phthalates and Cumulative Risk Assessment: The Tasks Ahead same end point. In some published studies, that was not done, and single-agent data from similar experimental conditions had to be relied on. Second, in well- designed studies, PODs are estimated for each mixture component, and the ab- sence of statistically significant effects is verified by direct testing. Where that demand was not met, doses without significant effects had to be estimated by regression analysis of dose-response data on the individual chemicals based on similar conditions. Mixture Effects of Combinations of Phthalates and Other Antiandrogens at Doses around Points of Departure The study by Hass et al. (2007) was designed to assess low-dose-mixture effects of AR antagonists in a developmental-toxicity model in the rat. NOAELs for vinclozolin, flutamide, and procymidone were estimated with change in AGD as the end point. The NOAELs in the study were similar to BMDs corre- sponding to effect levels of about 5%. When all three chemicals were combined at doses equivalent to their own NOAELs, reductions in AGD of 50% were ob- served. Quantitatively, the effects agreed well with the responses predicted by dose addition (see Figure 5-3), and the results were supported by a statistical evaluation of the observed data with dose-additivity predictions. Although not designed for such purposes, the experiment by Howdeshell et al. (2008) on suppression of testosterone synthesis after developmental expo- sure to five phthalates indicates that phthalates are able to work together when present at individually ineffective doses. Statistically significant reductions in fetal testosterone synthesis were observed after administration of a total mixture to pregnant Sprague-Dawley rats at 260 mg/kg-d. The mixture contained DPP at 20 mg/kg-d and each other phthalate at 60 mg/kg-d. DPP was tested on its own at 25 mg/kg-d, and the remaining phthalates were examined after single admini- stration at 100 mg/kg-d. At those doses, none of the single phthalates induced effects significantly different from those recorded in unexposed controls,2 al- though the doses in the single-phthalate experiments exceeded those in the mix- ture. Figure 5-4 extends the analysis to phthalate doses that were present in the lowest tested mixture dose of 260 mg/kg-d. That mixture dose produced a reduc- tion in testosterone synthesis that was statistically significantly different from untreated controls. Regression analysis of the dose-response data on the individ- ual phthalates was used by the committee to estimate BMDs and BMDLs (see Appendix C). The BMDL for BBP was estimated as 66 mg/kg-d. Those values were 20 mg/kg-d for DBP, 31 mg/kg-d for DEHP, 10 mg/kg-d for DPP, and 47 2 A simultaneous test of equivalence between the unexposed controls and low doses (100 mg/kg-d for BBP, DBP, DEHP, and DINP and 25 mg/kg-d for DPP) was rejected; this indicated equivalence between the low-dose mean and the control mean according to a 35% rule for equivalence bounds for the ratio of means. The doses in the mixture of 260 mg/kg-d included BBP, DBP, DEHP, and DIBP at 60 mg/kg-d and DPP at 20 mg/kg-d (that is, at doses below those used in the equivalence test).
Cumulative Risk Assessment of Phthalates and Related Chemicals 121 mg/kg-d for DIBP. Those BMDLs approach the doses present in the mixture of 260 mg/kg-d, which yielded statistically significant effects. The committee notes that dose-addition and independent-action predictions were generally similar, although dose addition gave the more conservative predictions (see also Figure 5-1). FIGURE 5-3 Low-dose combined effects of three AR antagonists with changes in AGD as the end point (Hass et al. 2007). Shown are litter means (circles) and mean responses with their 95% confidence intervals (bars with error bars). In all groups, the number of dams was 16, except for FLUT and PRO, in which case eight dams were dosed. When given as individual chemicals, vinclozolin (VZ, 24.5 mg/kg), flutamide (FLUT, 0.77 mg/kg), and prochloraz (PRO, 14.1 mg/kg) did not produce changes significantly differ- ent from those in control males. When combined at those doses (light gray bar, mixture obs), significant effects were observed (p < 0.05) that agreed well with the dose-addition prediction (white bar, mixture pred). NOAELs were estimated by using multiple contrast tests according to Hothorn (2004). The predicted mixture effects were derived from dose- response regression models for individual chemicals by using dose addition.
122 Phthalates and Cumulative Risk Assessment: The Tasks Ahead 100 Testosterone production (% control) 80 60 40 20 0 BBP DBP DEHP DIBP DPP Mix DA IA FIGURE 5-4 Low-dose combined effects of phthalates with suppression of testosterone synthesis as the end point. Shown is the committeeâs analysis of the data reported by Howdeshell et al. (2008). The hatched bars depict the model-predicted mean effects of BBP, DBP, DEHP, and DIBP at 60 mg/kg-d and DPP at 20 mg/kg-d. Error bars show 95% confidence intervals. Given the experimental design of Howdeshell et al., no indi- vidual mean effects would be predicted to be statistically distinguishable from controls (100% dotted line). The mean effect of a mixture of all phthalates at those doses (dotted bar, Mix) is statistically significantly different from untreated controls (the error bar shows the 99% confidence interval of the observed difference between mix and control groups). The expected mean combined effects derived from dose addition (DA, white bar) and independent action (IA, dark gray bar) are also shown, with 95% confidence intervals of the predicted mean. The individual responses to BBP, DBP, DEHP, and DIBP at 60 mg/kg-d and DPP at 20 mg/kg-d were estimated by fitting a nonlinear logistic regression model to the data reported by Howdeshell et al. (see Appendix C). Individual group means were obtained by applying the statistical model in Appendix C independ- ently to each dose or control group. All confidence intervals were obtained with the pro- file likelihood method. The study by Rider et al. (2008) provides some indications of combined effects of phthalates and AR antagonists at low doses. A combination of vinclo- zolin and procymidone (each at 3.75 mg/kg), prochloraz (8.75 mg/kg), linuron (5 mg/kg), and BBP, DBP, and DEHP (each at 37.5 mg/kg) was the lowest tested mixture dose that produced observable changes in AGD. Although the dose-response data on the individual chemicals are of insufficient quality to de- rive doses without observable effects, they nevertheless suggest that the doses are ineffective on their own. Similar conclusions can be drawn from the data provided for effects on NR and on hypospadias.
Cumulative Risk Assessment of Phthalates and Related Chemicals 123 NON-DOSE-ADDITIVE COMBINED EFFECTS OF PHTHALATES AND OTHER ANTIANDROGENS Strong evidence of non-dose-additive combined effects suggestive of syn- ergism (relative to dose addition) with phthalates and other antiandrogens is lacking. However, there are some data that indicate toxic interactions (greater than dose-additive effects) when hypospadias and other genital malformations are evaluated as the end points of concern. Rider et al. (2008) found that BBP, DBP, DEHP, vinclozolin, procymidone, linuron, and prochloraz induced more hypospadias than predicted on the basis of dose addition. Because of the as- sumptions that had to be made in their study to bridge some data gaps (see above), it is not possible to say with certainty whether the observations represent a true synergism with respect to dose addition, but the possibility cannot be ruled out. To resolve the issue, it will be important to subject the individual chemicals and their combinations to extensive dose-response studies that spe- cifically investigate hypospadias and other genital malformations. Hotchkiss et al. (2004) tested a combination of BBP and linuron at doses that were ineffective on their own. When they were combined at the given doses, hypospadias, cleft phallus, and other genital malformations were found in about 60% of the male offspring. An assessment of the results in terms of devia- tion from expected additivity is complicated by the lack of dose-response data on the individual chemicals. Such an analysis would reveal whether the ob- served massive increases in malformations represent synergism or are the con- sequence of the low-dose additive effects previously discussed. The frequent extreme steepness of dose-response curves for hypospadias makes the latter ex- planation plausible. Similar considerations apply to the results presented by Christiansen et al. (2008). About 50% of the male offspring showed genital malformations after exposure to a mixture of vinclozolin, flutamide, and procymidone, whereas none individually produced observable genital malformations at the doses used in the mixture as measured under the same conditions for vinclozolin and procymidone and on the basis of published data on flutamide. The potential for non-dose-additive combined effects to occur should be systematically explored. The work should not only focus on combinations of phthalates and other antiandrogens but consider the possibility that chemicals devoid of antiandrogenic activityâfor example, chemicals associated with tes- ticular toxicity, such as cadmiumâmay exacerbate mixture effects. CUMULATIVE RISK ASSESSMENT OF PHTHALATES AND OTHER ANTIANDROGENS: BASIC ISSUES Cumulative risk assessment of phthalates and other antiandrogens cannot be implemented without addressing a number of basic issues. The first is the question of which chemicals to include in mixture risk assessment. The second
124 Phthalates and Cumulative Risk Assessment: The Tasks Ahead is the mixture-effect assessment methods that can accurately predict combined effects. The third is the effects on which cumulative risk assessment should be based. Which Criteria Should Be Used to Group Phthalates and Other Chemicals for Cumulative Risk Assessment? The criterion proposed by EPA (2000) for grouping chemicals for cumula- tive risk assessment is âtoxicological similarity,â which may introduce ambigui- ties when applied to phthalates and other antiandrogens. An inappropriately nar- row interpretation would exclude many chemicals that also produce effects related to the androgen-insufficiency syndrome. Instead, a physiologically based approach for establishing grouping crite- ria for phthalates and other antiandrogens is strongly recommended. The recog- nition that androgen action is the driver of male sexual differentiation during development, with a multitude of underlying molecular mechanisms, implies that phenomenologic criteria should be used for grouping purposes. Thus, the starting point of approaches for grouping should be the physiologic process, not mechanisms or modes of action of the chemicals to be assessed. On the basis of considerations of the physiologic processes, a number of relevant end effects suggest themselves, and these should provide the basis of grouping. Accord- ingly, all chemicals that can induce some or all of the effects that make up the androgen-insufficiency syndrome should be subjected to cumulative risk as- sessment. Table 5-2 lists examples of chemicals that should be grouped with phthalates and considered for cumulative risk assessment. TABLE 5-2 Examples of Chemicals That Should Be Considered for Cumulative Risk Assessment of Phthalates and Other Antiandrogens According to a Physiologically Based Grouping Approach Chemical End Point or Evidence Phthalates: BBP, DBP, DEHP, DIBP, Androgen-insufficiency syndrome, DINP, DPP testosterone-dependent development AR antagonists: vinclozolin, procymidone Androgen-insufficiency syndrome, dihydrotestosterone-dependent development Linuron, prochloraz Androgen-insufficiency syndrome, AR antagonists, suppression of testosterone synthesis 5Î±-reductase inhibitors Androgen-insufficiency syndrome, dihydrotestosterone-dependent development Azole fungicides: ketoconazole, Suppression of testosterone synthesis in vivo, tebuconazole, propiconazole AGD changes in vivo Polybrominated diphenyl ethers AR antagonists in vivo TCDD, some PCBs Suppression of AR expression, AGD changes in vivo
Cumulative Risk Assessment of Phthalates and Related Chemicals 125 There are reports that 2,3,7,8-tetrachloro-dibenzo-p-dioxin (TCDD) can induce reductions in AGD by a mechanism that involves down-regulation of the AR and consequent suppression of AR-dependent genes in reproductive tissues (Ohsako et al. 2002). Reductions in AGD have also been observed with some coplanar polychlorinated biphenyls (PCBs) (Faqi et al. 1998; Rice 1999). Thus, a physiologic approach to grouping antiandrogens for purposes of cumulative risk assessment suggests inclusion of those chemicals. In contrast, the remaining effects typically attributed to the androgen-insufficiency syndrome, such as changes in NR and malformations, have not been found after administration of dioxins or PCBs in reproductive developmental-toxicity studies of rodents. Ex- perimental studies are needed to resolve the issue of combined effects of TCDD, PCBs, and other antiandrogens. Which Approach Should Be Used for Quantifying Cumulative Risks Posed by Phthalates and Other Chemicals? The brief overview of relevant mixture studies of antiandrogens has shown that there is strong empirical evidence of dose addition as an accurate predictor of mixture effects. Independent action often yielded similar quantitative predic- tions but in some cases has led to substantial underestimation of combined ef- fects. The committee could identify no case in which independent action pre- dicted combined effects that were in agreement with experimentally observed responses and at the same time were larger than the effects anticipated by using dose addition. Because the use of relative potencies and the use of TEFs are special ap- plications of the dose-addition concept, such approaches might suggest them- selves as a straightforward way of making quantitative assessments of the ef- fects of phthalates and other antiandrogens. However, application of the relative- potency concept (and a fortiori the TEF concept) requires parallel dose-response curves. If that demand is not met, equivalence factors will vary with the effect levels chosen for analysis. The data provided by Hass et al. (2007), Metzdorff et al. (2007), Howdeshell et al. (2008), and Rider et al. (2008) show clearly that phthalates and antiandrogens exhibit dose-response curves with widely differing slopes and shapes. An additional complication is the fact that dose-response relationships vary widely with the end point chosen for analysis. The relative potency of antiandrogens is not the same for every end point, so it is difficult to assign a global TEF to a specific antiandrogen. Thus, basic requirements for using either the relative-potency or TEF approach are violated, and their use cannot be recommended. Instead, dose addition should be used for quantitative evaluations of the joint effects of phthalates and other antiandrogens. It is a widely held miscon- ception (EPA 2000) that dose addition is applicable only with congruent dose- response curves (for a general discussion, see Gennings et al. 2005 and Kortenk- amp et al. 2007). Although high-quality dose-response data on individual
126 Phthalates and Cumulative Risk Assessment: The Tasks Ahead chemicals are desirable as a basis of predictions about mixture effects over a range of effect levels, dose addition can also be used when only point estimates, such as NOAELs, are available. Defining Points of Departure for Mixtures A mixture that produces dose-additive effects must satisfy the following expression: n ci * â ECx i =1 = 1, (1) i where ci* is the concentration (or dose) of substance i in a mixture that produces a known total effect X and ECxi is the concentration (or dose) of substance i that causes the effect X when applied individually. When the sum of the terms is larger than 1, there is antagonism; when it is smaller than 1, there is synergism. Equation 1 is referred to as the sum of toxic units or the sum of hazard indexes for particular selections of ECxi, such as reference doses. The schematic in Figure 5-5 illustrates how the interrelations play out when the aim is to establish a POD for a mixture of five hypothetical chemicals. In this example, the thin vertical lines associated with the individual dose- response curves in Figure 5-5 represent the BMDLs for each single chemical. Let the BMDLs corresponding to a particular benchmark response (BMR) of chemicals 1-5 be 90, 3.5, 11.8, 17.8, and 3.95 mg/kg-d, respectively. By using dose addition, it is possible to predict the effects of a mixture of all five chemi- cals when the mixture ratios are in proportion to the individual BMDLs (black solid curve in Figure 5-5). The black curve can be used to read off the expected effect of a dose of the mixture equal to the sum of all BMDLs. That procedure shows that the combination of the BMDLs cannot be considered to be without effect because it produces a reduction in response to about 90% in this particular case (black dashed vertical and horizontal arrows in Figure 5-5). To ensure that the mixture effect of the five chemicals is indistinguishable from the effect asso- ciated with the BMDLs of the individual chemicals, the doses of the mixture components have to be lowered. Equation 1 can be used to determine how much lower the doses of all in- dividual components of the mixture must be to ensure that the combined effect corresponds to the BMR. That will occur when Equation 1 is fulfilled for the special case of ECxi = BMDLi. Several permutations of the doses ci* of the chemicals in the mixture that satisfy Equation 1 can be found. One can distin- guish the following two extremes: (1) A single chemical is present at its BMDL. Equation 1 holds only when the doses of all other chemicals are zero.
Cumulative Risk Assessment of Phthalates and Related Chemicals 127 (2) Equation 1 holds when all chemicals are present at their individual BMDLs divided by the number of the other effective chemicals in the combina- tion, five in the present example. Thus, with the individual BMDLs of chemicals 1-5 of 90, 3.5, 11.8, 17.8 and 3.95 mg/kg-d, respectively, Equation 1 holds with (90/5)/90 + (3.5/5)/3.5 + (11.8/5)/11.8 + (17.8/5)/17.8 + (3.95/5)/3.95, which resolves to 1/5 + 1/5 + 1/5 + 1/5 + 1/5 = 1. In other words, a combination of chemicals 1-5 of 18, 0.7, 2.36, 3.56, and 0.79 mg/kg-d, respectively, should pro- duce less than the BMR (black vertical arrow in Figure 5-5). Joint effect of sum Sum of five single Combined effects curve of five single BMDLs BMDLs divided for mixture ratio in doses by five proportion to BMDLs 120 100 Effect (% of untreated controls) 80 60 40 20 0 100 101 102 103 Dose (mg/kg-d) BMDLs of each of the Sum of five single five single components BMDLs FIGURE 5-5 Schematic to illustrate the derivation of a point of departure for a mixture dose, here the lower confidence limit of a benchmark dose (BMDL). Shown are the dose- response curves for five single hypothetical chemicals (thin curves) and their correspond- ing BMDLs (thin vertical lines). In this hypothetical case, 100% equals the effect seen in untreated controls. The solid black curve shows the combined effects of a mixture of all five chemicals with mixture ratios proportional to their individual BMDLs. The sum of the single BMDLs (vertical dashed arrow) will exceed the effect associated with the BMDLs, the so-called benchmark response (horizontal black dashed arrow). To achieve the benchmark response for the mixture (black vertical arrow), the individual BMDLs of all components have to be lowered by a factor of 5. Other combinations may also reach a combined zero effect, and these are accessible by calculating new mixture effect curves corresponding to the chosen mixture ratios. If one component is present at its BMDL, the mixture is without effect only when the doses of all other components are zero.
128 Phthalates and Cumulative Risk Assessment: The Tasks Ahead The example presented here can be extended to the case where all chemicals are present in an arbitrary mixture. In that case, the sum in Equation 1 will be less than 1 provided that every chemical is present at less than its BMDL/n, so that in all cases such a mixture will have an effect less than the BMR. When the aim is to assess the effects of joint existing exposures, the pro- cedure is slightly different. Equation 1 can be used, but this time with ci* repre- senting the doses of the individual chemicals that are in specific exposure sce- narios. As before, ECxi represents the individual BMDLs of each chemical i. If the sum of the toxic units ci*/ECxi is less than 1, the joint effect of the mixture must be less than the BMR at least under the dose-addition hypothesis. Which Effect Outcomes Should Form the Basis of Cumulative Risk Assessment? PODs, such as BMDs and NOAELs, for individual chemicals in mixtures are important elements of cumulative risk assessment. They can be the basis of reference values for cumulative effects, which can be used for risk assessment or standard-setting. Reference values for individual chemicals are estimated in rela- tion to specific effect outcomes and toxic end points. However, although the specific effects produced by phthalates and other antiandrogens show common- alities, there are differences. The responses seen after disruption of androgen action during development depend on whether the effects of dihydrotestosterone or those of testosterone are compromised. Although there is overlap in the spec- trum of effects resulting from exposure to phthalates and other antiandrogens, some responses are specific to disruption of testosterone action, and others are seen only after blocking of dihydrotestosterone action. For example, none of the AR antagonists suppresses testosterone synthesis, and they have weaker effects in disrupting the development of testosterone-dependent tissues, such as the epididymis. Conversely, phthalates are less effective in disrupting reproductive development that depends on dihydrotestosterone and causing malformations, such as hypospadias, that result from that type of disruption. To make a common grouping of phthalates and other antiandrogens prac- ticable, it is necessary to deal with the fact that not all relevant agents produce all aspects of the androgen-insufficiency syndrome. The committee has consid- ered several ways of addressing that situation. One option is to recognize that the induction of any of the effects of the androgen-insufficiency syndrome is symptomatic of disruption of androgen action. Therefore, the androgen- insufficiency syndrome should be dealt with as a whole. This approach makes it necessary to aggregate the various qualitatively different components of the syndrome into one common measure. Because the array of effects produced by phthalates and other antiandrogens shows a degree of overlap, an alternative option is to focus on effects common to the chemicals and to base cumulative risk assessment on the most sensitive common outcome. Those two options and their implications are detailed in the following sections.
Cumulative Risk Assessment of Phthalates and Related Chemicals 129 Option 1: Dealing with the Syndrome as a Whole The qualitatively different component effects of the androgen- insufficiency syndrome can be aggregated by noting for each experimental sub- ject as to whether any of the observed end points signify some degree of toxic- ity. For example, one might say that any malformation, an AGD deviating by two standard deviations from the mean of that in the unexposed control group, or an organ weight below a specified weight would indicate that the subject ex- perienced toxicity.3 A usual analysis of dose-response data could then be con- ducted by using those measures of toxicity for each experimental subject. Al- though the method incorporates multiple end points associated with the androgen-insufficiency syndrome, it assumes equal levels of toxicity associated with each dichotomized end point in classifying each subject as demonstrating toxicity or not. In this way, PODs, such as BMDs and NOAELs, for single chemicals can be estimated and used as input for deriving reference values for combinations of antiandrogens. A variation would be to develop a scoring method that incorporates the set of end points while adjusting for the degree of toxicity related to each end point. For example, a method commonly used in the engineering literature known as desirability functions (see Appendix D) could be used to define a toxicity score for each of the component end points of the syndrome on a unitless scale be- tween 0 (most toxicity) and 1 (no toxicity). On the basis of each end-point- specific toxicity score, an overall composite toxicity score can be constructed. Coffey et al. (2007) have described the development of such an overall compos- ite score for the many outcomes measured in toxicologic studies. Appendix D shows how this approach can be used productively for the assessment of antian- drogenic chemicals. Option 2: Focusing on the Most Sensitive End Point of the Androgen- Insufficiency Syndrome It appears that in neonatal male rats NR is the most sensitive common ef- fect of phthalates, AR antagonists, and chemicals that act via a mixed mecha- nism of action. Thus, NR in rats could be chosen as a common end point, and the BMDs or NOAELs for single chemicals could form the basis of cumulative risk assessments of phthalates and other antiandrogens. However, phthalates induce reductions in testosterone synthesis at lower doses than required for changes in NR. Therefore, risks posed by phthalates might be underestimated if cumulative risk assessment is conducted in relation to NR in rats. Comparative dose-response studies would need to be conducted to 3 The committee is not endorsing any specific value to signify toxicity; that is a matter for further evaluation.
130 Phthalates and Cumulative Risk Assessment: The Tasks Ahead evaluate the degree to which hazard assessments based on NR in rats could un- derestimate the risks associated with phthalates. If the use of NR in rats as the evaluation end point is insufficiently protec- tive, the most sensitive individual end point for each of the groups of antiandro- gens (for example, reduction in testosterone synthesis for phthalates or NR for AR antagonists) could be chosen for the estimation of reference values, which in turn are used to derive cumulative effects as a sum of toxic units or hazard in- dexes. However, this approach violates one of the preconditions of dose addi- tionâthe induction of the same effect. Nevertheless, it can be used from a pragmatic viewpoint, considering that there is a precedent for it in the use of hazard indexes. Estimation of Cumulative Effects: Animal-to-Human Extrapolation An issue for consideration with the above options is whether aggregation for cumulative effects should be conducted with animal data and then extrapo- lated to the human or whether reference values for human exposure, such as a tolerable daily intake (TDI), should be derived first for individual chemicals and aggregation for cumulative risks then carried out with TDIs in a second step. Those alternatives are depicted schematically in Figure 5-6. The first approach (Figure 5-6A) is suitable for almost all the options out- lined above for dealing with the different end points of the androgen- insufficiency syndrome except perhaps when the most sensitive individual end point is used to derive reference values for single chemicals. In that case, it might be more appropriate to estimate TDIs first (Figure 5-6A); this allows addi- tional flexibility by giving the opportunity to adopt tailor-made uncertainty fac- tors for each individual chemical during the estimation of single chemical TDIs. Because the aggregation for cumulative effects is carried out with individual TDIs, the use of different end points from animal studies for the estimation of TDIs can be accommodated. The procedure sketched in Figure 5-6B is also compatible with all the options for dealing with the component effects of the androgen-insufficiency syndrome that were discussed above. STEPPED APPROACHES TO CUMULATIVE RISK ASSESSMENT OF PHTHALATES AND OTHER ANTIANDROGENS A corollary of the dose-addition principle is that cumulative effects are to be expected even when all mixture components are present at doses below their zero effect levels for the individual components if a sufficiently high number of relevant chemicals are combined at sufficient doses. The demonstration that exposure to individual chemicals may be below some risk-criterion level for the individual components is uninformative. To estimate risks that stem from cumu- lative exposure to phthalates and other antiandrogens, information on the nature
Cumulative Risk Assessment of Phthalates and Related Chemicals 131 Tolerable cumulative exposure Exposure (human) Uncertainty factor Reference value for mixture Dose (animal) Reference values for single chemicals A Tolerable cumulative exposure Exposure (human) Tolerable exposures for single chemicals Uncertainty factors Dose (animal) Reference values for single chemicals B FIGURE 5-6 Aggregation for cumulative effects and animal-to-human extrapolation. The horizontal black arrows represent dose axes for the effects of antiandrogens in ani- mals and humans. (A) Aggregation for cumulative effects (large black circle on âdose [animal]â axis) is carried out at the level of animal-derived reference values (open small circles on âdose [animal]â axis). The calculation of mixture effects is symbolized by the parabolic lines. A reference value for cumulative effects in animals (black circle) is then combined with an uncertainty factor to derive tolerable cumulative exposures for humans (large white circle on âexposure [human]â axis). (B) Reference values for single chemi- cals (open small circles on âdose [animal]â axis) are combined with uncertainty factors to derive individual tolerable exposures for humans first (black circles on âdose [human]â axis); these values are then used to derive standards for cumulative risk assessment (large white circle on âdose [human]â axis).
132 Phthalates and Cumulative Risk Assessment: The Tasks Ahead of the chemicals in the mixture, the magnitude of exposures to the individual chemicals, their potency, and their number is required. Thus, knowledge about the prevalence and quantities of other chemicals that might contribute to the risk in question is critical. Incomplete information about this aspect of exposure as- sessment will introduce considerable uncertainty and the potential for underes- timating risks. The example illustrated in Figure 5-5 highlights the ways in which the presence of other chemicals that produce the effect of interest determines the extent to which threshold levels for single compounds may have to be corrected to ensure that a mixture is without effect. The larger the number of effective chemicals, the larger the downward correction of the single thresholds may have to be to guarantee safety. What is the number of antiandrogens that might con- tribute to disrupting male sexual differentiation? Recent screening efforts for AR antagonists provide some first clues. Ko- jima et al. (2004) examined 200 pesticides for their ability to antagonize the AR. Of the 200 tested compounds, 66 were found to be active. Vinggaard et al. (2008) screened 397 chemicals for AR antagonism and identified 178 active ones, of which 17 had a potency higher than or similar to that of flutamide. The authors developed a global quantitative structure-activity-relationship model that predicted that 8% of the chemicals would be active AR antagonists. Those efforts suggest that a large number of chemicals might be active in vivo AR antagonists capable of disrupting male sexual differentiation. Possibly because of toxicokinetic influences that prevent the buildup of suitably high concentrations in target tissues, some in vitro antagonists fail to show effects in in vivo models. However, insufficient information is available about correlations between in vitro and in vivo antiandrogens. Conclusive data that might help to resolve the issue are not likely to become available soon, not least because the testing of candidate chemicals in in vivo developmental-toxicity models is ex- tremely time-consuming and expensive. The uncertainties and knowledge gaps call for appropriately conservative approaches that incorporate default assumptions about the likely number of antiandrogens that might contribute to human exposure scenarios. The commit- teeâs proposal to deal with that would be adoption of a stepped approach as fol- lows. Step 1: Cumulative Risk Assessment with Incorporation of Default Assumptions about the Likely Number and Potency of Unidentified Antiandrogens In a first screening step, cumulative risk assessment of phthalates, AR an- tagonists, and mixed-mode antiandrogens could be carried out by making allow- ance for unidentified antiandrogens. That could be achieved by using the toxic unit (hazard index) Equation 1, as follows: Human exposures to each chemical are represented by ci*, and ECxi are estimates of tolerable daily exposures. To
Cumulative Risk Assessment of Phthalates and Related Chemicals 133 take account of unidentified antiandrogens of unknown potency, a default num- ber of âplaceholderâ toxic units (for example, 10-100) can be added. That re- quires some assumptions about the potency and prevalence of the unknowns. A reasonable first approximation would be to expect ECxi around the median of the TDIs for established in vivo antiandrogens, with ci* equal to ECxi divided by the total number of toxic units in the equation. If the sum of toxic units obtained in this way is 1 or smaller, the cumula- tive risks posed by phthalates and related chemicals can be regarded as quite low. If, however, the procedure yields a value larger than 1, risk-reduction measures may be advised. Alternatively, the assessment can be refined. Step 2: Cumulative Risk Assessment of Phthalates and Other Antiandrogens The above procedure is repeated by including AR antagonists and mixed- mode antiandrogens that have known in vivo activity but without making as- sumptions about unidentified antiandrogens. If this step signals risks, risk- reduction measures may have to be considered. Alternatively, a refined step considering only phthalates may be included. It is also possible to conduct the stepwise procedure in reverse order, be- ginning with phthalates and antiandrogens that have established in vivo activity. If the first risk-assessment step does not indicate risks, the assessment broadens to assume worse-case scenarios. STANDARD-SETTING In some regulatory settings, it may be deemed desirable to derive exposure standards for phthalates and other antiandrogens that take account of cumulative exposures. In such cases, tolerable daily exposures to individual chemicals could be corrected downward by incorporating an additional âmixture uncertainty fac- tor.â The additional uncertainty factor would have to take account of the number of chemicals to which simultaneous effective coexposure is deemed likely. CONCLUSIONS A major challenge to conducting cumulative risk assessment is choosing an approach to predict mixture effects. However, evidence from the recent peer- reviewed scientific literature shows not only that phthalates produce mixture effects but that the effects are often predicted well by using the dose-addition concept. That is also true for other classes of antiandrogens and for combina- tions of phthalates with such antiandrogens. Although a variety of molecular mechanisms are at play, dose addition provided equal or better approximations of mixture effects compared with independent action (when such comparisons
134 Phthalates and Cumulative Risk Assessment: The Tasks Ahead were performed). In no example in the literature did independent action produce a mixture-effect prediction that proved to be correct and differed substantially from that produced with dose addition. The evidence that supports adoption of a physiologic approach is strong. Experimental evidence demonstrates that toxic effects of phthalates and other antiandrogens are similar despite differences in the molecular details of the mechanisms, including metabolism, distribution, and elimination. The criteria recommended by EPA (2000) for guiding decisions between dose addition and independent action appear too narrow when applied to phthal- ates and other antiandrogens, particularly those requiring similarity in uptake, metabolism, distribution, and elimination and congruent dose-response curves for application of dose addition. The requirements are not met by combinations of phthalates with other antiandrogens, but the dose-addition principle applies. The case for using dose addition as an approximation for mixture risk assess- ment of phthalates and other antiandrogens is strong. When risks posed by low-level exposures need to be evaluated, there are substantial differences between the single-chemical approach and cumulative risk assessment. There is good evidence that combinations of phthalates and of other antiandrogens produce combined effects at doses that when administered alone do not have significant effects. In some cases, those doses are similar to those used as PODs to estimate tolerable human exposure. The results highlight the problem that may arise when PODs for individual chemicals are used as the basis of human-health risk assessment in situations in which exposure to other chemicals with similar effects also occurs. The results emphasize the necessity of conducting cumulative risk assessment of phthalates and other antiandrogens to assess risks posed by exposure to mixtures of these compounds. Assessments based solely on the effects of single phthalates and other antiandrogens may lead to considerable underestimation of risks to the developing fetus. In this chapter, the committee has provided recommendations on various aspects of conducting cumulative risk assessment. The recommendations were designed specifically to deal with phthalates and antiandrogens. However, the conceptual framework that the committee has used is generic and lends itself to dealing with other groups of chemicals, provided that the relevant toxicologic data are available. REFERENCES Allen, B.C., R.J. Kavlock, C.A. Kimmel, and E.M. Faustman. 1994. Dose-response as- sessment for developmental toxicity. II. Comparison of generic benchmark dose estimates with no observed adverse effect levels. Fundam. Appl. Toxicol. 23(4):487-495. Berenbaum, M.C. 1989. What is synergy? Pharmacol Rev. 41(2):93-141. Bility, M.T., J.T. Thompson, R.H. McKee, R.M. David, J.H. Butala, J.P. Vanden Heuvel, and J.M. Peters. 2004. Activation of mouse and human peroxisome proliferator- activated receptors (PPARs) by phthalate monoesters. Toxicol. Sci. 82(1):170-182.
Cumulative Risk Assessment of Phthalates and Related Chemicals 135 Birkhoj, M., C. Nellemann, K. Jarfelt, H. Jacobsen, H.R. Andersen, M. Dalgaard, and A.M. Vinggaard. 2004. The combined antiandrogenic effects of five commonly used pesticides. Toxicol. Appl. Pharmacol. 201(1):10-20. Christiansen, S., M. Scholze, M. Axelstad, J. Boberg, A. Kortenkamp, and U. Hass. 2008. Combined exposure to anti-androgens causes markedly increased frequencies of hypospadias in the rat. Int. J. Androl. 31(2):241-248. Coffey, J., C. Gennings, and V.M. Moser. 2007. The simultaneous analysis of discrete and continuous outcomes in a dose-response study: Using desirability functions. Regul. Toxicol. Pharmacol. 48(1):51-58. COT (Committee on Toxicity of Chemicals in Food, Consumer Products and the Envi- ronment). 2002. Risk Assessment of Mixtures of Pesticides and Similar Sub- stances. London, UK: Her Majestyâs Stationary Office [online]. Available: http://cot.food.gov.uk/pdfs/reportindexed.pdf [accessed 28 June 2008]. Crump, K.S. 1984. A new method for determining allowable daily intakes. Fundam. Appl. Toxicol. 4(5):854-871. Crump, K.S. 2002. Critical issues in benchmark calculations from continuous data. Crit. Rev. Toxicol. 32(3):133-153. EPA (U.S. Environmental Protection Agency). 1986. Guidelines for the Health Risk As- sessment of Chemical Mixtures. EPA/630/R-98/002. Risk Assessment Forum, U.S. Environmental Protection Agency, Washington, DC. September 1986 [on- line]. Available: http://www.epa.gov/ncea/raf/pdfs/chem_mix/chemmix_1986.pdf [accessed July 23, 2008]. EPA (U.S. Environmental Protection Agency). 1989. Risk Assessment Guidance for Superfund, Volume I. Human Health Evaluation Manual (Part A). Interim Final. EPA/540/1-89/002. PB90-155581. Office of Emergency and Remedial Response, U.S. Environmental Protection Agency, Washington, DC. December 1989 [online]. Available: http://rais.ornl.gov/homepage/HHEMA.pdf [accessed July 22, 2008]. EPA (U.S. Environmental Protection Agency). 1994. Methods for Derivation of Inhala- tion Reference Concentrations and Application of Inhalation Dosimetry. EPA/600/8-90/066F. Office of Research and Development, U.S. Environmental Protection Agency, Washington, DC [online]. Available: http://cfpub.epa.gov/ ncea/cfm/recordisplay.cfm?deid=71993 [accessed Oct. 16, 2008]. EPA (U.S. Environmental Protection Agency). 2000. Supplementary Guidance for Con- ducting Health Risk Assessment of Chemical Mixtures. EPA/630/R-00/002. Risk Assessment Forum, U.S. Environmental Protection Agency, Washington, DC. Au- gust 2000 [online]. Available: http://www.epa.gov/NCEA/raf/pdfs/chem_mix/ chem_mix_08_2001.pdf [accessed June 10, 2008]. EPA (U.S. Environmental Protection Agency). 2002. Guidance on Cumulative Risk As- sessment of Pesticide Chemicals That Have a Common Mechanism of Toxicity. Office of Pesticide Programs, U.S. Environmental Protection Agency, Washing- ton, DC. January 14, 2002 [online]. Available: http://www.epa.gov/oppfead1/trac/ science/cumulative_guidance.pdf [accessed June 10, 2008]. Faqi, A.S., P.R. Dalsenter, H.J. Merker, and I. Chahoud. 1998. Effects on developmental landmarks and reproductive capability of 3, 3â, 4,4â-tetrachlorobiphenyl and 3,3â4,4â,5-pentachlorobiphenyl in offspring of rats exposed during pregnancy. Hum. Exp. Toxicol. 17(7):365-372. Foster, P.M. 2005. Mode of action: Impaired fetal Leydig cell function â effects on male reproductive development produced by certain phthalate esters. Crit. Rev. Toxicol. 35(8-9):713-719.
136 Phthalates and Cumulative Risk Assessment: The Tasks Ahead Gennings, C., W.H. Carter Jr., R.A. Carchman, L.K. Teuschler, J.E. Simmons, and E.W. Carney. 2005. A unifying concept for assessing toxicological interactions: Changes in slope. Toxicol. Sci. 88(2):287-297. Gray, L.E., J. Ostby, J. Furr, C. J. Wolf, C. Lambright, L. Parks, D.N. Veeramachaneni, V. Wilson, M. Price, A. Hotchkiss, E. Orlando, and L. Guilette. 2001. Effects of environmental antiandorgens on reproductive development in experimental ani- mals. Hum. Reprod. Update 7(3):248-264. Hass, U., M. Scholze, S. Christiansen, M. Dalgaard, A.M. Vinggaard, M. Axelstad, S.B. Metzdorff, and A. Kortenkamp. 2007. Combined exposure to anti-androgens exac- erbates disruption of sexual differentiation in the rat. Environ. Health Perspect. 115 (Suppl. 1):122-128. Hotchkiss, A.K, L.G. Parks-Saldutti, J.S. Ostby, C. Lambright, J. Furr, J.G. Vanden- bergh, and L.E. Gray Jr. 2004. A mixture of the âantiandrogensâ linuron and butyl benzyl phthalate alters sexual differentiation of the male rat in a cumulative fash- ion. Biol. Reprod. 71(6):1852-1861. Hothorn, L. 2004. A robust statistical procedure for evaluating genotoxicity data. Envi- ronmetrics 15(6):635-641. Howdeshell, K.L., J. Furr, C.R. Lambright, C.V. Rider, V.S. Wilson, and L.E. Gray Jr. 2007. Cumulative effects of dibutyl phthalate and diethylhexyl phthalate on male rat reproductive tract development: Altered fetal steroid hormones and genes. Toxicol. Sci. 99(1):190-202. Howdeshell, K.L., V.S. Wilson, J. Furr, C.R. Lambright, C.V. Rider, C.R. Blystone, A.K. Hotchkiss, and L.E. Gray, Jr. 2008. A mixture of five phthalate esters inhibits fetal testicular testosterone production in the Sprague-Dawley rat in a cumulative, dose- additive manner. Toxicol. Sci. 105(1):153-165. Kojima, H., E. Katsura, S. Takeuchi, K. Niyama, and K. Kobayashi. 2004. Screening for estrogen and androgen receptor activities in 200 pesticides by in vitro reporter gene assays using Chinese hamster ovary cells. Environ. Health Perspect. 112(5):524-531. Kortenkamp, A., M. Faust, M. Scholze, and T. Backhaus. 2007. Low level exposure to multiple chemicals: Reason for human health concerns? Environ. Health Perspect. 115(Suppl. 1):106-114. Loewe, S., and H. Muischnek. 1926. Ãber Kombinationswirkungen I. Mitteilung: Hilfsmittel der Fragestellung. N-S. Arch. Exp. Pathol. Pharmakol. 114:313-326. Metzdorff, S.B., M. Dalgaard, S. Christiansen, M. Axelstad, U. Hass, M.K. Kiersgaard, M. Scholze, A. Kortenkamp, and A.M. Vinggaard. 2007. Dysgenesis and histo- logical changes of genitals and perturbations of gene expression in male rats after in utero exposure to antiandrogen mixtures. Toxicol. Sci. 98(1):87-98. Mileson, B.E., J.E. Chambers, W.L. Chen, W. Dettbarn, M. Ehrich, A.T. Eldefrawi, D.W. Gaylor, K. Hamernik, E. Hodgson, A.G. Karczmar, S. Padilla, C.N. Pope, R.J. Richardson, D.R. Saunders, L.P. Sheets, L.G. Sultatos, and K.B. Wallace. 1998. Common mechanism of toxicity: A case study of organophosphorus pesti- cides. Toxicol. Sci. 41(1):8-20. Nellemann, C., M. Dalgaard, H.R. Lam, and A.M. Vinggaard. 2003. The combined ef- fects of vinclozolin and procymidone do not deviate from expected additivity in vi- tro and in vivo. Toxicol. Sci. 71(2):251-262. NRC (National Research Council). 2000. Toxicological Effects of Methylmercury. Washington, DC: National Academy Press. Ohsako, S., Y. Miyabara, M. Sakane, R. Ishimura, M. Kakeyama, H. Izumi, J. Yonemoto, and C. Tohama. 2002. Developmental stage-specific effects of perinatal 2,3,7,8
Cumulative Risk Assessment of Phthalates and Related Chemicals 137 TCDD exposure on reproductive organs of male rat offspring. Toxicol. Sci. 66(2):283-292. Rice, C. 1999. Effect of exposure to 3, 3â4, 4â,5-pentachlorobiphenyl (PCB 126) throughout gestation and lactation on development and spatial delayed alternation performance in rats. Neurotoxicol. Teratol. 21(1):59-69. Rider, C.V., J. Furr, V.S. Wilson, and L.E. Gray Jr. 2008. A mixture of seven antiandro- gens induces reproductive malformations in rats. Int. J. Androl. 31(2):249-262. Scholze, M., and A. Kortenkamp. 2007. Statistical power considerations show the endo- crine disrupter low dose issue in a new light. Environ. Health Perspect. 115 (Suppl. 1):84-90. Slob, W. 1999. Thresholds in toxicology and risk assessment. Int J. Toxicol. 18(4): 259- 268. Stroheker, T., N. Cabaton, G. Nourdin, J.F. Regnier, J.C. Lhuguenot, and M.C. Chagnon. 2005. Evaluation of anti-androgenic activity of di-(2-ethylhexyl)phthalate. Toxi- cology 208(1):115-121. Vinggaard, A.M., J. Niemala, E.B. Wedebye, and G.E. Jensen. 2008. Screening of 397 chemicals and development of a quantitative structure-activity relationship model for androgen receptor antagonism. Chem. Res. Toxicol. 21(4):813-823.