Brown & Root Environmental600 Clark Avenue, Suite 3

King of Prussia, PA 19406-1433Phone:(610)491-9688Fax:(610)491-9645

TO:

ATTN:

RE: i

WE ARE SENDING YOU Q Attached Q Under separate cover via ___________ the following items:n Shop drawings n Prints n Plans fj Samples n Specifications

Copy of letter Q Change order Q

COPIES DATE NO. DESCRIPTION

v

————————————— tTHESE ARE TRANSMITTED as checked below:

n For approval . Q Approved as submitted Q Resubmit ______copies for approvalQ For your use Q Approved as noted Q Submit _______ copies for distributionQAs requested n Returned for corrections n Return ________ corrected printsn For review and comment n ______________________________________D FOR BIDS DUE _______' _______________. D PRINTS RETURNED AFTER LOAN TO US

REMARKS A/7?-/ ̂"Ĉ'C

COPY TO

FILE: SIGNED

If the enclosures are not as noted, kindly notify us at once. HRc\n9l (\~J

MULTIPLE LOW-LEVEL EXPOSURE

Bach chemical component of a chemical mixture has a uniquepotential to influence the toxicity of other mixture components.The magnitude and capacity of this potential is.rarely known.Despite empirical evidence that supports this assumption, thecomponent-based approach does not take this into consideration.The Hazard Index (HI) or component-based approach, assumes nointeractions between the components of the mixture and assumesdose additivity. The goal of the HI approach is to construct theplausible toxicity iridex that would have been calculated had themixture itself been tested. However, in practice, the guidanceis being applied only to the critical eff.ect of each componentchemical because acceptable exposure levels for major effects ofa chemical are not derived. Thus, toxicity to several targetorgans is not being included in the overall toxicity assessmentprocess.

Limited, yet compelling, data suggests that for noncancerendpoints, adverse effects are unlikely to occur when theindividual components in the mixture are present at levels wellbelow their respective thresholds. Synergistic or antagonisticeffects, not readily predicted from the mechanisms•of action ofthe individual components, are possible when the mixturecomponents are present at levels equal to .or.above theirindividual thresholds. Finally, synergistic carcinogenic effectshave been observed in a,nimal studies of mixtures, even atrelatively low dos.es,.

Toxic responses from exposures to chemical mixtures atenvironmental levels are most likely insidious effects disruptinghomeostasis of the organism. The exposed animals may appeartotally "normal" clinically or based on,conventional toxicologicend points. Such a subclinical state may provide a basis forenhancement of .otherwise mild toxic responses from acuteexposure(s) of chemical, physical, and/or biological agents.Thus, a concept of "generic promoter or -Cotoxicant", " whichincludes the promotional role in carcinogenesis but is applicableto all toxicities, may 'be advanced for chemical mixtures.

METALS

the metallic carbonyls are generally considered some of the safe levels for air concentration areto present more of a hazard than either the in fact based on topical or skin exposure and notmetal or its other compounds. For example, the inhalation exposure. The organic leads andacceptable level for metallic nickel and its sol- thallium are examples of this.uble compounds is 3.0mg/Ma, while that for However, the effects of other metals such asnickel carbonyl is 0.007 mg/M3. The hydrides of nickel, beryllium, and arsenic involve skinthe metals also tend to be more toxic than the changes (Table 18-3) as part of their spectrummetal or the other metallic compounds. This is of toxicity.' Dermal toxicity may occur fromparticularly clear with respect to arsenic and occupational exposure, atmospheric contamina-arsine. It should be noted that in Table 18-2 tion, medicinal use, or from contact with the

Table 18-3. TARGET ORGANS OF METALS

METAL

AluminumAntimonyArsenicBariumBerylliumBismuthBoranesBoronCadmiumChromiumCobaltCopperGalliumGermaniumGoldHafniumIndiumIronLanthanonsLeadLithiumMagnesiumManganeseMercuryMetal hydridesMolybdenumNickelNiobiumOsmiumPalladiumPlatinumRhodiumRubidiumRutheniumSeleniumSilverStrontiumTantalumTelluriumThalliumTin (organic)TitaniumTungstenUraniumVanadiumZincZirconium

GASTRO-INTESTINALTRACT

444

4" 4

4-4 '

4

4

4"4

4

4-' 4

4-

c->̂

RESPIRA-TORYTRACT

4"4444

4

444-

4

44

44

4

4

4

4

4

4

4-

4

4*

.3

CNS

44

4444-

4

44

4-4444

4

44

4

44-4-

4

4

£

CARDIO-VASCULARSYSTEM LIVER SKIN BLOOD KIDNEY

4 4 44 4 - 4

4 44

4 4 - +4 4

4- . 4-4 4 4

+ + - .4

4 . 44

4 4 4 44 44 44- 4

" 4 44 4

4 ' +

4* ' 4-

4 4 - 44

4 4

44 4

4

4 44 4

4-

4' +4 +

44 -f

44- 4

i 3. ) 3 3

BONE ENDOCRINE

4-

4

4

4

4

4-

4

4

4

4

/ flpK^

L

OaB>

It\3O

••

ol

C-a

—

X

_r

-C.

~

5U

— -

|[jamma-BHC (Lindane)

1

X

X

IVieptachkx epoxida

X

sia§"

X

ag

X

i

*

tIB

X

01COXO

X

1

X

8srtoen

X

X

X

^

X

TJIs

n- Hm >

•F oro zOm*

1y

8 , i \-a ($ '•£ i

* 1 1 1 1 1 ^ \

| S 1 I 1 V,f 1 f * I ?1 1 ̂ | I

l i f t * "*"jhif

g f f f t1 § s 1I i " -^.

1 i

j <

^

, —

1 ^4-

4-1 -t • -£. 1 •>

O

-c -,,-4.-C. r-z r- m

mzm >

nmz

LIVESTOCK HEALTH SYMPTOMS

Loose bowels (incomplete digestion?)

Coughing & runny noses

Poor Growth

Legarthy

Incoordination

Breakdown by Target Organs

1) Gastrointestional Tract

2) Respiratory System

3) Central Nervous System

Rrr.VLATORY TO.XI- ' il.' -.- •.--•' f-H \K.M.\COLgOY 22.

Chemical Mixtures: Current Risk Assessment Methodologiesand Future Directions _______

JENNIFER SEED,* RONALD P. BROWN,* STEPHEN S. OLIN, AND JEFFERY A. FORANInternational Life Sciences Institute, Risk Science Institute, 1126 16th Street N.W., Washington, DC 20036

Received March 20, 1995

V~~ " fore, the potential exists for adverse effects to occurSome of the most challenging problems that toxicolo- due~~to-exposure to-the mixture that may be greater

gists confront are determining how biological effects than, less than, or equal to the sum of the effects ofof components in a complex mixture may interact, de- the individual components of the mixture.termining how these interactions affect the overall Historically, regulatory agencies have conducted risktoxicity of the mixture, and determining how to incor- assessments and set standards for individual chemi-porate this information into risk assessments of chemi- cals. Recognizing that this approach may not be ade-cal mixtures. There has been considerable effort in this quately protective, the U.S. EPA published generalarea since the publication of the U.S. Environmental guidelines for risk assessment of chemical mixtures inProtection Agency's B̂ delines for risk assessment of 1986 &.S. EPA, 1986). Since this publication was re-cfaemical mixtare. in 1986. This paper reviews the tor- leaged ŷ has ̂ ^̂ ^ effort made to de_ndnology used to describe chemical interaetiona and yd ^ refine methodolô eg for rigk aS8essment ofthe methodologies that have been developed for con- , . , . , „.. e . ^, ... -,ducting risk assessments of chemical mixtures. Partic- chemiĉ ™*»*f. This paper reviews the existing andular attention is directed towards an examination of ne.wlv developed nsk assessment methodologies forthe applicability and validity of the methods for the nuxtoes and critically examines assumptions in theseassessment of risk posed by exposure to environmen- approaches. In particular, the assumptions of dose- andtally relevant concentrations of chemical mixtures, response-additivity for noncartinogens and carcino-Limited, yet compelling, data are reviewed that sug- gensi respectively, are examined in light of new datagest that for noncancer endpoints, adverse effects are on the effects of low dose exposure to chemical mixtureunlikely to occur when the individual components in components.the mixture are present at levels well below then* re- A number of graphical and statistical methods havespective thresholds. Synergistic or antagonistic ef- been developed that serve as valuable tools for de-fects, not readily predicted from the mechanisms of termining whether an interaction is occurring, and, ifaction of the individual components, are possible so, what the nature of the interaction is. However, thewhen the mixture components are present at levels use of these methods for risk assessment of complexequal to or above their individual thresholds. Finally, mixtures is limited for several reasons. First, thesesynergistic carcinogenic effects have been observed in methods have been developed for relatively "simple"animal studies of mixtures, even at relatively low mixtures containing a small number of components.doses. oimAc.d.̂ prem̂ ________________ Second, many data points are usually required, and for

this reason, the methods are impractical for most toxiceffects. Finally, these methods test whether an interac-

INTRODUCTION tion is occurring within the specific experimental doseX. - range. Extrapolation beyond this range is not possible;Humans are continuously exposed to low levels of therefore a quantitative estimate of risk outside the

chemical mixtures by a variety of routes and for vary- experimental range cannot be achieved. Thus, a de-Lng lengths of time. Interactions may occur among tailed description of these methods is not included inchemicals in the mixture that alter their toxicity; there- this review.

Over the years, a number of different terms have, _ , , . , , . , , , been introduced to describe chemical interactions. The1 To whom correspondence should be addressed. ei-cr „* t i j -L • j. •.• i_ i j i.' Present address: Center for Devices and Radiological Health US6 °f afferent terms to describe interactions has led to

(HFZ-112), U.S. Food and Drug Administration, 12709 Twinbrook some confusion in the literature. The term "interaction"Parkway, Rockville, MD 20857. implies that one agent affects the biological action of

0273-2300/95 * 12.00 76Copyright O 1995 by Acad«nie Press, Inc.All rigfau of reproduction in may form reserved.

B R 5021*73 _

CHEMICAL MIXTURES RISK ASSESSMENT METHODOLOGIES

TABLE 1Terminology for Chemical Interactions

Term Definition . Synonymi

Additivity The combined effect of two agenta is equal to the sum of the effects of each Summationagent given alone Independence

IndifferenceZero interactionSimple similar actionUnconditional independence

Synergiim The combined effect of two agents is greater than the sum of the effect* of Potentiationeither agent given alone Coalitivity

InteractionUnisynergismAugmentationSensitizationSupra-additionIndependent lynergismDependent synergumDegradative gynerpamGreater than additiveUnisynergismCosynergismSuperadditionConditional independence

Antagonism The combined effect of two agents is less than the sum of each agent given Interactionalone Depotentiation

Besensiti rationInfra-additionNegative synergyLess than additiveSubadditionInhibitionAntergismCompetitive antagonismNoncompetitive antagonismUncompetitive antagonismAcompetitive antagonism

Not*. Source: Calabres« (1991).

another agent. Two basic types of interactions, syner- cosynergism, potentiation, and coalitivity, are used togism and antagonism, can be denned (Calabrese, 1991). describe particular synergistic interactions. Cosyner-A synergistic effect occurs when the combined effect of gism occurs when two agents enhance the tenacity oftwo agents is greater than the effects of each agent each other. Potentiation occurs when one agent en-given alone, while an antagonistic effect occurs when hances the toxicity of another agent, but does not affectthe combined effect of two agents is less than the sum the toxic endpoint itself. Coalitivity occurs when theof the effects of each agent given separately. It is im- combination of two agents is very toxic, whereas theportant to recognize that these terms do not describe action of the agents individually is ineffective. For thethe mechanism of interactions or the quantitative as- purposes of this review, the terminology outlined bypecta of the interaction; the terms simply describe Calabrese (1991) will be used, and wherever possible,whether the outcome is different from what would be three terms will be relied upon: synergism, antago-expected if the agents were administered individually, nism, and additivity.Alternatively, one agent may not affect the biologicalaction of another agent, and in such cases the term METHODOLOGIES FOR CHEMICAL MIXTURES"noninteraction" would apply. An example of noninter- RISK ASSESSMENTaction is additivity; an additive effect occurs when thecombined effect of two agents is equal to the sum of In 1986, the U.S. EPA developed a framework forthe effects of each agent given alone. conducting risk assessments of chemical mixtures (U.S.A list of terms that are frequently used as synonyms EPA, 1986; Mumtaz and Hertzberg, 1993; Mumtaz et

for synergism, antagonism, and additivity is provided al., 1993), This framework described three methodsin Table 1. Three of the common terms for synergism, that could be used to conduct a quantitative risk assess-

ment depending on the availability of data on the mix- used to e.-nhiuut.- the cancer risk •-.(".-.-vf.Tai diesel emis-ture and on the components of the mixture. The first sions (Albert et al., 1983; Lcwtas. 1955'. but it has notand most preferable method is to use data obtained received enough further use to determine its wicfe-from testing the actual mixture to assess the potential spread applicability.for health effects associated with exposure to the mix- The lack of quantitative data on all but a few chemi-ture. These data are used in the risk assessment as if cal mixtures limits the use of the methods describedthe mixture was a single agent. However, this method above. The third method that the U.S. EPA recom-has been used to assess the risk of relatively few mix- mends for conducting a risk assessment of chemicaltures due to the lack of data on the specific mixture of mixtures, the "component-based" method, is thereforeregulatory concern. With the exception of diesel ex- used most often. This method involves assessing thehaust (Albert et al., 1983; Lewtas, 1985) and some mix- risk posed by exposure to the individual components oftures of groundwater contaminants (Germolec et al., the mixture. Data from risk assessments of individual1989; Chapin et al., 1989; Hong et al., 1992; Heindel et compounds are then used to estimate the risk of theal., 1994; Simmons et al., 1994), few studies have been mixture of interest by applying a dose additivity modelconducted on well-defined chemical mixtures. for systemic toxicants and a response additivity modelAn important limitation in the use of information on for carcinogens.

the specific mixture for risk assessment is that complex Dose additive approaches are based on the assump-mixtures can undergo changes with time, and often tion that the components of a mixture behave as if theyvary in composition from source to source. For example, were dilutions or concentrations of each other. It isin assessing the toxicity of a complex mixture such as also assumed that the mixture components act via adiesel exhaust, the toxicity of a particular sample may common toxicological mechanism. The two most ac-be assessed. However, chemical interactions may occur cepted dose additive approaches are the hazard indexamong the components in the sample, altering its com- (HI) approach and the toxicity equivalency factor (TEF)position; therefore a sample tested at one point in time approach. In both approaches, the doses of the mixturemay differ from a sample tested later in time. In addi- components are first normalized to a common scale andtion, the exact composition of complex mixtures such then summed. In the HI approach, the doses are stan-as diesel exhaust will vary depending on the source, dardized using health-based values such as acceptablewhich may also affect the toxicity. These factors may daily intakes (ADIs) or reference doses (RfJDs). In thelimit the utility of data derived from the specific mix- TEF approach, the dose of each component of the mix-ture for risk assessment purposes. ture is normalized against the dose of one of the compo-The second most preferable method outlined by the nents, usually the most potent compound, to derive a

U.S. EPA is to use data obtained from testing one or relative potency for each. The relative potencies aremore mixtures of composition similar to the mixture of then summed to estimate the toxic or carcinogenic po-interest. A similar mixture is defined as "a mixture tency of the mixture. These approaches, as well ashaving the same components but in slightly different methods to refine them, are described below.ratios, or having several common components but lack-ing in one or more components, or having one or more Hazard Index Approachadditional components." The obvious challenge in the ^ ^ fa ft hazarduse of this method comes in determining how similar ̂ ^̂ 05̂ hazard quotient of each compound ina surrogate mixture is to the mixture of interest The ̂ mixture ̂.^ .g j to ̂ dose ^ *degree of similarity is evaluated on a case-by-case basis d̂ed fa ^ maximum acceptable dose for thaf com.and may be based on several factors including test data pQund> guch &g an ̂ ^ ^^ quotients for eachand chemical and biologic properties mixture component are then added together to calcu-Use of the "similar mixture" method requires the late ̂ m a^ foUows.

availability of in vivo data on the similar mixture sothe toxicity or carcinogenic potency of the mixture of „interest can be estimated. However, in vivo toxicity HI = 'or carcinogenicity data are available for relatively few ,-.complex mixtures. One alternative that may be usedwhen assessing the carcinogenic potential of several where n is the number of components of the mixturemixtures in a particular class is the relative potency (U.S. EPA, 1986; Mumtaz and Hertzberg, 1993; Mum-or comparative potency method based on in vitro data taz et al., 1993).(Albert et al., 1983). When the relationship between in This method is most appropriately used for mixturevitro data and in vivo data can be established for one components that induce the same toxic effect in themixture, it is theoretically possible to estimate the in same organ system via the same mechanism of actionvivo carcinogenic potency of similar mixtures based on since dose additivity is assumed to be valid only underthe results of in vitro assays. This method has been these conditions. Although it may be possible to make

A R 50 21* 75

CHF.Mit'A! V.iXTl'KK.- !:'..-K \--:--:.'.. -. .,„.... ..

this assumption For relatively simple mixtures oFchem- uncertainty tucu.r^f 1. which is equivalent to the addi-ically or toxicologically related substances, the assump- tivity assumption, -.vould be used when there are datation may be violated For complex mixtures oF com- indicating that a synergistic response is unlikely orpounds that act on multiple organ systems by various when the concentration oF the mixture components ismechanisms. Multiple mechanisms and target organs low. An uncertainty Factor oF 10 would be used whenare more likely when dealing with complex mixtures less is known about the potential For interactions tosuch as contaminated groundwater or soil at a hazard- occur or iF relatively high concentrations oF the mixtureous waste site. In these situations, the U.S. EPA recom- components are present.mends that a separate hazard index be calculated for While the potential exists for synergistic interactionseach toxic endpoint of concern. to occur among components of a complex mixture, it isThe hazard index should be viewed as an approxima- also possible that antagonistic responses will be ob-

tion of the noncarcinogenic risk posed by exposure to served. Since the approach proposed by the NRC (1989)the mixture and not as an absolute risk value. This only incorporates additive and synergistic responses,concept is explained in the U.S. EPA (1986) guidelines it may overestimate the risk from exposure to somefor conducting chemical mixtures risk assessments: complex mixtures, especially those containing low lev-_ , , , , , ., , ,... , . . .. els of constituent compounds. A different scheme hasThe hazard index provides a rough measure of likely toxicity , , , . *,, ,. , _ . . ,1nnn*and require, caution interpretation. The hazard index is only recently been developed by Mumtaz and Durkin (1992)a aurâ rical Indication of the nearnesa to acceptable limits of to account for both synergistic and antagonistic inter-exposure or the degree to which acceptable exposure levels are actions in the derivation of the hazard index. In thisexceeded. approach, an alphanumeric weight-of-evidence classi-

There is considerable debate regarding the values for ncation is assigned to a binary mixture that takes intothe hazard index that represent an unreasonable risk, account a number of factors including whether thatGenerally, it is assumed that there may be increased interaction is likely to be additive, antagonistic, or syn-risk when the hazard index is greater than 1. However, ergistic; an understanding of the mechanism by whichfactors such as the sensitivity of the endpoint and size the interaction may occur; the potential toxicologies!of the uncertainty factor used to derive the hazard quo- significance of the interaction; the duration and se-tient for each compound should be taken into account quence of exposure; whether in vitro or m tnuo datain deciding whether a hazard index represents an un- were used to describe the interaction; and whether thereasonable risk. toxicity data were collected using a route of exposureAlthough the lack of data on specific interactions re- similar to the route expected under environmentally

quires that the additivity assumption be invoked for relevant situations. Numbers or letters are ascnbed tomost chemical "mixtures, the U.S. EPA guidelines en- each of these factors as shown in Table 2.courage the use of data on interactions (synergism, an- T"6 features oF specific interactions can be character-

. tagonism) when available. Several modifications of the ized usini this method. For example, selenium isstandard hazard index approach have been proposed known to inhibit the toxic effects of cadmium uponto incorporate data on interactions. For example, the coexposure. Features that would be important in theSubcommittee on Mixtures of the NRC Safe Drinking weight-of-evidence classification of this interaction in-Water Committee (NRC, 1989) recommended using an clude: U>the mechanism by which this interaction oc-additional uncertainty factor, when appropriate, to de- curs is fairly wel1 understood; (2) data from other re-

! rive the hazard index for a mixture to protect against lated compounds are consistent with the proposedI potential synergistic interactions. The hazard index mechanism; (3) the toxicological significance of this in-I would then be calculated as: teraction can be demonstrated; and (4) in vivo data on

this interaction are available from long-term studies! s j that utilized the same route by which humans are

HI *• (UF) X ——' • likely to be exposed. Using the letters and numbers! i-i ""•*' that correspond to each of these features (Table 2), an

alphanumeric weight-of-evidence classification for thei The NRC subcommittee urged that such an approach effect of selenium on the toxicity of cadmium can be

be .flexible and recommended the use of uncertainty represented as

80TABLE 2

Classifications Used in the Weight-of-Evidence Scheme

Determine if the interaction of the mixture is additive t=), greater than additive (>), or less than additive (ra.A.2.

B.I.

B.2.

B.2.

4.s«

C.I.

5.To!.

B.I.

B.I.

B.2.

B.2.

=1.B.2.

6.Xyl.

B.I.

B.I.

B.I.

B.2.

=1.B.2.

FIG. 1. Qualitative interaction matrix for example compounds. Source: Reproduced, with permission, from Mumtaz and Durkin (1992).

AR5021477

"•• ••• M'.XTL'KK- :.:-K X-̂ K.-̂ MENT METHODOLOGIES 01

'I ABLE 3 tors" depending on the conditions of the study. Second,Data Quality Weighting Factors for inherent in the weight-of-evidence approach La the aa-

Wcight-of-Evidence Scheme sumption that the risk posed by exposure to a complex—————————————————————— mixture can be estimated by the sum of the interactionsDtunnme if the interaction of the mutur* U additive i D - 0). occurring between pairs of complex mixture compo-gre.ur th.n .dd.uv. (D - i). or less than add.twe CD - -11- nents ĵ ^ the effect of one compound on the tox-

Mechanistic Information Weigh? jcity of another compound in the mixture may be modi-I Direct Mechanistic Data i.o fied b the presence of the other components of the

11L S3SSiS«£!:MeSt̂ u III mixture.ForexamPle,itcanbedemonstratedthatsele.Tojdcolojical Significance nium antagonizes the toxic effects of cadmium, as dis-A. Interaction directly related to toxkity i.o cussed above. However, it is unclear how seleniumB. Significance of interaction can inferred 0,71 would affect the toxicity of cadmium in the presence of

ModUUr?fiC*n

>KKP KT A!.

K 4 .is then multiplied by the concentration of each conge-Uclativc Ilangc of Toxicities of PCDD and PCDF ner in the mixture to yield the toxicity equivalentCongeners Compared to That of 2.3,7,8-TCDD (TEQ). The toxicity or carcinogenicity of the mixture is

from tne sum Of the TEQs.. Congener /n m-o responses In vitn responses The TEF approach hag been fairiy successful in as.

237 8-utraCDD 10 1.0 sessing the toxicity of mixtures bf PCDDs and PCDFs.l̂ iaj.S-penuCDD 0.59-0.053 0.64-0.07 For example, Eadon et al. ( 1986) characterized the com-1,2,3,6,7,8-hexaCDD 0.16-0.0152 0.5-0.005 position of an extract of soot that contained a complex1,2,3.7,8,9-hexaCDD o.H-0.016 0.009 mixture of PCDDs and PCDFs and quantified theia&SSSD S£T8 S0 -°15 amount of each PCDD and PCDF congener found inoctaCDD

... ..,.- RISK ASSESSMENT METHODOLOGIES 83

TABLE 6 may be present as pesticide residues on fruits and vege-Estimates of Toxicity Equivalence Factors for Five tables. In contrast to the situation with the PCDDa,

Organophosphatc Insecticides Using the NOAEL for PCDFs and PCBS, in vivo toxicity studies have beenChlorpyTifoa aa the Reference Standard________ conducted for the five pesticides. The TEF approach

N O A F I w a s used in this case as a means of expressing thep , ,, NOAEL Ratio » TEF cholinesterase inhibition potency of each pesticide rela-

tive to the potency of chlorpyrifos, wnich the committeeAc*ph«u NA NA decided was the most potent cholinesterase inhibitorChlorpyrifo* 0,03 mg/kg bw/d«y 1 of the five. The TEF was denned as the ratio of the no-n, .u ' Hijm«ni.(aE inhibition AM*!* n« observed-adverse-effect-level (NOAEL), or lowest-ob-Dlmethoate 0.06 msc/kg bw/dmy 0.03/0.05 » 0.6 . , _. , i/T/-\XT-IT \ r \. t. . iRau ChE inhibition served-adverse-effect-level (LOAEL), for each pesticideDiaulfoton NA ' NA to the NOAEL (or LOAEL) for chlorpyrifos, shown inEthion 0.05 ag/kg bw/d»y 0.03/0.05 ̂ 0.6 Tables 6 and 7, respectively. Then, the committee used_________Hum»na. ChE inhibition__________ PDA residue data, which were obtained from 1988 and,,.„...... . . , VT >• in k n ;i 1989 compliance, surveillance, import, and domesticNote. R*prinUd with permission from National Research Council r ' , ' r '(1993) sampling, to estimate the amount of residue of each of

the five pesticides on eight foods (apples, oranges,grapes, beans, tomatoes, lettuce, peaches, and peas)

TEF is derived. Therefore, the predictive ability of the and in three fruit juices (apple, orange, and grape).TEF approach is dependent upon the endpoint and spe- Residues for each pesticide were then converted todes selected when measuring the potency of the mix- chlorpyrifos equivalents by multiplying the concentra-ture. For example, Nagao et al. (1993) examined cleft tion of each pesticide on the food items by the potencypalate formation in NMRI mice following admjnistra- values shown in Table 7; the TEFs based on the ratiotion of PCDD and PCDF mixtures. The potency of the of LOAELs were used since TEFs based on NOAELsPCDD mixture was accurately predicted when I-TEFs were not available for two of the pesticides (Table 6).(NATO/CCMS, 1988) were used for the PCDD conge- There are several factors to keep in mind regardingnera; however, the cleft palate-inducing potency of a the approach by the NRC committee to develop TEFsPGDF mixture was overpredictsd when the I-TEFs for for these organophosphate pesticides. First, TEFs de-the PCDF congeners were used. Another limitation of rived from the ratio of LOAELs were used for all fivethe TEF approach is that the TEF values are derived pesticides. It is apparent that somewhat different re-from short-term or subchronic in vivo experiments or suits may have been achieved if TEFs based on thefrom in vitro bioassays. Therefore, any differences in ratio of NOAELs had been used. For example, a TEFtoxicokinetics are largely ignored. Third, only Ah-re- of 0.6 was derived for ethion using the ratio of NOAELsceptor-mediated effects are used in the TEF approach (Table 6), whereas a TEF of 1.33 was derived using thefor PCDDs, PCDFs, and PCBs. Many of the coplanar ratio of LOAELs (Table 7). Second, the dose levels usedPCBs elicit a number of effects that are not mediated to determine the LOAEL were from the "critical study,"by the Ah receptor (reviewed by Ahlborg et al., 1992). as defined by the EPA for the derivation of the RfD;Therefore, the TEFs will not be predictive for all toxicresponses associated with these compounds. Neverthe-less, this approach has been a valuable "interim" tool TABLE 7to assess the toxicity of mixtures of several important Estimates of Toxicity Equivalence Factors for Fiveclasses of compounds, including PCDDs and PCDFs Organophosphate Insecticides Using the LOAEL for(NATO/CCMS, 1988; Ahlborg, 1989; Ahlborg et al., Chlorpyrifos aa the Reference Standard1992; BeUin and Bames, 1989; Kutz et al., 1990; Safe, ~1990, 1993), PCBs (Ahlborg et al., 1994; Barnes et al.,1991; U.S. EPA, 1987,1991; Safe, 1990,1993) and poly- Pest'adecyclic aromatic hydrocarbons (Nisbet and LaGoy, 1992; Acephata 0.12 mg/kg bw/day o.yo.12 = 0.83U.S. EPA, 1993). . Rats, ChE inhibitionThe TEF approach has also been applied to assess Chlorpyrifos 0.10 mg/kg bw/day o.vo.io = i.o

the toxicity of organophosphate pesticides The NRC ^̂ ££3$̂ *" om25 - 0.40Committee on Pesticides in the Diets of Infants and RaUi ChE inhibitionChildren (NRC, 1993) recently used a TEF approach Disulfoton 0.04 mg/kg bw/day 0.1/0.04 - 2.5to estimate the aggregate risk to children from dietary Rats, ChE inhibitionexposure to a mixture of pesticides. The committee ex- Ethion °;075 °>ĝ « bw/day 0.1/0.075 = 1.33

• j c. i_ i_ «. t- • j - • «1 ,A-~~ Humans, ChE inhibitionattuned five organophosphate pesticides including .—————————————————————————————————acephate, chlorpyrifos, dimethoate, disulfoton, and eth- ôte. Reprinted with permission from National Research Councilion. All five pesticides are cholinesterase inhibitors and (1993).

84 ?KF.I) KT At..

the EPA defines the critical study as that study show- assumed to be equal to the sum of the upper confidenceing effects'at the lowest dose level. However, in some levels of the individual compounds, fn most cases, thecases (chlorpyrifos and ethion) the critical study was a magnitude of the difference in the cancer risk estimateshuman study, and in other cases (acephate, dimetho- calculated by these various methods will be greatestate, and disulfoton) the critical study was a rat study, for mixtures of equipotent compounds. However, evenAlthough in all cases cholinesterase inhibition was the for mixtures of equipotent compounds, the differencesendpoint of concern, the critical dose levels from human in joint risk estimated by summing the upper 95% con-studies should be compared with caution with dose lev- fidence levels and estimated by the methods of Kodellels from rodent studies. and Chen (1994) are not great.

As an alternative to summing the potencies of indi-Response Additive Methods vidual compounds, modifications can be made to sto-

A 11-.- j i • 11 j • chastic dose-response models to estimate the potencyAresponseadditzvemodel 1S generaUy used .ncancer ^^ ^ y ^J. arerisk assessment of chemical mixtures (U.o. br A, 1986; ., ,, . , ° _. : .. . , ,.„ , , TT , , ,ono it _i * i moo\ available for dose-response extrapolation includingMumtaz and Hertzberg, 1993; Mumtaz et at., 1993). ,. ,,. ,. .. , . ,. j u- * j inn_ • • i u -ii-i.lv. I -riv. linear, multiplicative, logistic, and multistage models.The principal assumption is that the components of the _„ Xron n •** *«• n. j r n. T IT- m i• ̂ -* it. * * •* u * u j-«- 4- The NRC Committee on Methods for the 7n V«K> Test-mixture act on the same target site, but by different . ,0 , ,,. , /̂ m̂ i«nn\ j i 3, . f ,. ., f .t . . , . , ingof Complex Mixtures (NRC, 1988) developed a gen-mechanisms of action; therefore, the toxicological re- 5,. , f,.,. jnui u ji ui. • • i.r., . . , .. ?,,„ . erahzed additive model that can be used to obtain risksponses of the organism to each component in the mix- ,. . - . , .. ,, P.,.* , .v ., B,, j x j ID • estimates of a mixture of carcinogens regardless of theture (rather than the doses) are summed. Response is , , , . .. . .generally expressed as risk, which is calculated as the raodfel ™? *° ̂^product of dose and carcinogenic potency, usually the ̂ -^ "?' • f ' f?upper 95% confidence interval on the potency. This re- tested with the epidemioogical data olationship is assumed to be valid only when the dose- exammed «>e individual and joint effecte of tobaccoresponse curve is -linear at low doses. The risk of a smolong and xiramum exposure. The model was ablemixture of compounds is estimated as the sum of the *° £"*"*the "sk of ̂

CHEMICAL MlXTl'RK- K'.SK ASSKS.-MKNT V- :: !«•'>. .;,i. :\'r ~

TABLE 8Upper-95%-Confidence Limits on Extra Cancer Risk for Selected Chemicals

at Concentrations of 1 ugAiter in Drinking Water

Statistical method of UCL

Chemical

A. ChlorobenzeneB. HexachlorobetueneC. TrichlorotlhyleneD. 1,1,1-Trichloroethane

Mlxtur*

AandB

A and C

A.B.C

A, B, C, D

Mixturemethod

Sum UCLsProposedSum UCLsProposed *Sum UCLsProposedSum UCLsProposed

Directlikelihood

2.14 x 10-'1.92 x 10-'2.74 X ID'73.70 X 10-'

Directlikelihood

2.14 X 10r»2.01 X 10-'4.88 X 10-' '4.30 x 10-'2.41 X 10-*

2,45 X 10"*

Directbootstrap

1.91 X ID'71.50 X 10'*2.79 X 10-'3.78 X 10-'

Directbootstrap

1.69 X 10-*1.63 X 10'*4.70 X ID'14.01 X 10'71.97 X 10-*1.83 X 10'*2.01 X 10-'1.84 X 10-*

ED01» bootstrap

1.91 X 10-'1.50 X 10'*2.79 X 10"'3.78 x 10'*

ED,,bootstrap

1.69 x 10-*"1.65 X 10'*' 4.70 X 10-'4.03 X 10-'1.97 X 10-'1.86 X 10'*2.01 x 10-'1.88 X ID'4

Note. Source: Reproduced, with permission, from Kodell and Chen (1994).i

i

grounds, to yield greater-than-additlve responses, it of the events represented by the clonal two-stageia likely that these effects will he seen only at rela- model of carcinogenesis, low-dose exposure to com-tively high doses. In fact, a number of cases of pounds that affect the various stages of the processgreater-than-additive carcinogenic responses are can be modeled by using the assumption of additivity,well documented (e.g., aflatoxin and hepatitis B; vi- even when theoretical principles suggest that the as-rua in liver cancer). Kodell and colleagues (1991) ar- sumption of additivity is violated at higher doses.gue that the low-dose response to two genotoxic Other mechanistic models also can be applied to theagents will approximate additivity, even though, chemical mixtures problem for carcinogens (Cohen andfrom a theoretical perspective, a multiplicative rela- Ellwein, 1991). Ultimately, the challenge will be to de-tionship is expected. Further, multiplicative or su- velop a full enough understanding of the complex bio-pramultiplicative effects expected between an initia- logical interactions and their dose dependencies totor and a promoter or two promoters may not occur predict carcinogenic responses over a wide range ofat low doses because of a threshold for the effect of doses.the promoter. Therefore, based on an understanding Clevenger e« al. (1991) have developed a pharmaco-

dynamic model to describe the interaction occurringbetween a substrate and an inducer or inhibitor of the

TABLE 9 mixed function oxidase (MFO) enzyme system basedInteraction for Simple Exposures on relationships between dose, elimination rate, and

duration of effect. Data on the effect of piperonyl butox-Cardnogen Carcinogen jde On the ability of hexobarbital to induce sleep was

1 n °n used to test the model. Although the model was applica-InitUtor Initiator Additive ble for analyzing this particular system, its generalCompleter Completer Additive use for risk assessment purposes is limited for severalInitiator Complete Multiplicative reasons. First, as the model is currently written, itsPromoter £">«»*" SupramuitipUcative use u restricted to two Comp0unds. Second, the modelInitiator Promoter Multiphcatwe-supramultiphcative . „ „ _ „ , . f. ' .Initiator and defines effect as the duration of the effect. This is aInitiator Completer Supra-additive submultiplicative common definition in pharmacology, but an uncommon

Initiator and definition (or representation) in toxicology. Finally, theInitiator Promoter Suprm-additive supramultiplicative model is able to predict a response when the concentra-PromoUr "Smpleif Supramultiplicative *«"» ofr \ach compound is known. However, extrapola-————————————————————————————————— tion of the response beyond the experimental range isNott. From Kodell et al. (1991). not possible.

AR502I*82

86 : . —- — --:- -.-TOXICITY STUDIES OF COMPLEX MINTl/RllS .: - :• -:••• • • - -• . :. .; •: :sv threshold lies abort

ADMINISTERED AT LOW DOSES the NOAHL. ifu-r: .tu-.r.-r-v i-ilccls would not be antici-pated when each component of the mixture is present

Central to the methods described in the previous sec- . at levels well below their respective NOAELs. Con-tion are the assumptions of dose- and response-additiv- verscly, a range of effects would be anticipated whenity for noncarcinogens and carcinogens, respectively, each component of the mixture is present at a levelThere is concern that these methods may be inadequate near, equal to, or greater than its individual threshold.because potential antagonistic and synergistic interac- These kinds of scenarios have'begun to be tested.tions are not taken into account. Recently, a number Recently, several studies were conducted to examineof studies have been conducted to assess the toxicity of the relationship between the development of adversecomplex mixtures at environmentally relevant doses, systemic effects and the dose of the mixture compo-While most of these studies do not directly assess the nents relative to their NOAELs and LOAELs (Grotenvalidity of the additivity assumption, the results have et al., 1994; Jonker et al., 1990, 1993a,b, 1994). In theimportant implications regarding the likelihood that first of this series of studies (Jonker et al., 1990), asignificant interactions will occur at low-dose levels. mixture of eight arbitrarily chosen compounds (sodiumThe U.S. EPA (1986) suggested that the dose-addi- metabisulphate, mirex, loperamide, metaldehyde, di-n-

tive approach is appropriate for noncarcinogens when octyltin chloride, stannous chloride, lysinoalanine, andmixture components are present at low doses, and potassium nitrate) was administered to rats for fourwhen the mixture components induce the same toxic weeks. When the dose of the constituent componentseffect in the same organ system via the same mecha- was 5 to ̂ of their respective NOAELs, no adversenism of action. This seems appropriate if low dose is effects were noted. When the constituent compoundsdenned in relationship to the threshold level for non- were present at a concentration that approximatedcancer endpoints. If a mixture contains a variety of their respective NOAELs, some effects were occasion-compounds that act via a common mechanism, but each ally noted, but they tended to'be of minor toxicologicalindividual compound is present at a level slightly be- significance. In contrast, when the constituents werelow, or at its threshold, an additive effect is likely to present at their LOAEL values, a range of responsesoccur; however, the likelihood decreases as component was observed. Several effects including growth retarda-concentrations decrease below their respective thresh- tion and liver damage were more severe than seen witholds. In contrast, if a mixture contains a variety of com- the individual compounds at the same dose levels.pounds that do not act on a common target or by a Other effects including changes in the weight and mor-similar mechanism, or if each similarly acting com- phology of the thymus were less severe in this studypound is present at a level well below its threshold, than after exposure to di-n-octyltin chloride alone.an additive effect would not be expected. When each Some effects such as necrosis of the pancreatic acinarcomponent in a mixture is present at a level well below cells and hypertrophy of the zona glomerulosa of theits respective threshold, it is also questionable whether adrenals were observed when the compounds were ad-synergistic or antagonistic interactions would be antici- ministered individually at doses equivalent to thepatsd. In these cases, application of the additivity as- LOAEL, but were not observed at all when adminis-sumption and/or methods to account for synergistic and tered at the same dose levels in the mixture. In addi-antagonistic interactions would not portray risk accu- tion, a number of effects such as reduced numbers ofrately. ' corpora lutea and increased numbers of multinucleatedSeveral steps are necessary to evaluate the validity giant cells in the epididymides were observed after ex-

of the hypothesis that adverse effects are unlikely when posure to the mixture that had either not been seenthe components of a mixture are present at levels well when the compounds were administered individuallybelow their respective thresholds. The first step is to or had only been seen at doses much higher than theestablish the threshold for the endpoint of concern for LOAEL.each mixture component. Although it is rarely possible A subsequent study examined the toxicity of a mix-to establish the precise threshold, the threshold for ob- ture of nine arbitrarily chosen compounds adminis-servable adverse effects should lie somewhere between tered to rats for 4 weeks (Groten et al., 1994). Seventhe NOAEL and the LOAEL. The second step is to of the compounds were administered in the diet andassess the toxicity of the mixture using the same exper- included aspirin, cadmium chloride, stannous chloride,imental protocol used to assess the toxicity of the indi- loperamide, spermine, butylated hydroxyanisole, andvidual components of the mixture. Finally, the relation- di(2-ethylhexyl)phthalate. In addition, formaldehydeship between the development of adverse effects follow- and dichloromethane were administered by inhalationing exposure to the mixture and the dose of the mixture for 6 hr per day, 5 days a week. Preliminary resultscomponents relative to their respective NOAELs and of this study indicate that when the dose of the nineLOAELs can be established. If adverse effects do not constituent components was 3 of their respectiveoccur when the dose of each component is well below NOAELs, no adverse effects occurred. When the dose

-.ft.R502U83

of the constituent cump»mutr- wa~ uquai to thoir n.- pre.-i-i:' :: .•••,'•!- ux-ll below their individual NOAELs.spective NOAELs. a few treatment-related efTects oc- and then.-foro well below their thresholds. However.curred including hyperplasia of the nose, decreased some adverse effects, albeit minor, can occur even whenkidney weight and alkaline phosphatase levels, and the mixture components are present at a level equiva-increased liver weight. However, when the dose of the lent to their respective NOAELs, regardless of thecompounds was equal to their respective LOAELs, a mechanism by which the components exert their toxicvariety of effects occurred ranging from antagonism to effect. Furthermore, a range of interactive effects isadditivity to synergism depending on the endpoint. possible when coexposure occurs to mixture compo-The chemicals used in the studies discussed above nents present at a level equivalent to their individual

were arbitrarily chosen with respect to target organ LOAELs.and mechanism of action. Two subsequent studies were The National Toxicology Program (NTP) and othersconducted to examine the effects of a mixture con- have recently conducted several studies of complextaining compounds known to act on the same target mixtures, but did not conduct studies of the individualorgan, but differing in their mechanisms of action. Four components of the mixture. Since they were not con-nephrotoxic compounds, mercuric chloride, potassium ducted to address the question of low level exposuredichromate, d-limonene, and hexachloro-l,3-butadi- within the framework outlined above, it is not possibleene, were administered to rats in a single-dose study to examine directly the validity of the hypothesis thatwith sacrifice 24 hr later (Jonker et al., 1993a) or over adverse effects will not occur if the components of aa 4-week period (Jonker et al., 1993b). In these studies, mixture are present at levels below their individualthere were either no effects when the dose of the con- thresholds. However, it is possible to use the NTP mix-stituent components was | of their respective NOAELs ture studies to examine the hypothesis indirectly byor minimal effects when the doses were equal to the selecting other benchmarks of toxicity for the individ-individual NOAELs. However, subacute (4-week) expo- ual components of the mixtures: Appropriate bench-sure to a mixture containing the compounds at a con- marks for comparison are the NOAEL and LOAEL forcentration equivalent to their respective LOAELs pro- a particular endpoint. In the absence of these data,duced renal toxicity in male rats that was more severe other benchmarks are the NOAEL and LOAEL of thethan that observed after administration of any of the "critical study" that the U.S. EPA used to establishcompound* individually. Acute (24-hr) exposure to this the reference dose (RfD) for the particular chemical ofsame mixture resulted in a range of effects from antag- interest. Although the design of the "critical study"onism to additivity to synergism depending on the end- used to derive the RfD is not necessarily comparablepoint. to the design of the studies conducted by the NTP, pre-Although each of the compounds used in the studies sumably the NOAEL of the "critical study" represents

conducted by Jonker et al. (1993a,b) were nephrotoxic, the lowest NOAEL known.they exerted this effect via different mechanisms. To For the purposes of further testing the hypothesismore fully test the validity of the additivity assump- that adverse effects are unlikely when the componentstion, Jonker et al. (1994) conducted a 4-week study in of a mixture are at levels below their individual thresh-rats of the toxicity of four compounds (trichloroethyl- oldsv the EPA's Integrated Risk Information Systemene, tetrachloroethylene, HCBD, and 1,1,2-trichloro- (IRIS) was searched to find the RfD and the correspond-3,3,3-trifluoropropene) thought to exert their effect via ing NOAEL and LOAEL for each of the chemicals inthe same toxicological mechanism (initial conjugation the mixtures tested by the NTP. These values werewith glutathione in the liver, followed by transport of used to determine the likelihood that the thresholds ofthe conjugate to the kidney and formation of reactive the components of the mixtures tested by the NTP weremetabolites in proximal tubular epithelial cells). Pre- exceeded. The toxicity data obtained from the NTPliminary results indicate that combined exposure to the studies and the benchmarks for the individual compo-four compounds at their respective NOAELs produced nents of the tested mixtures are discussed below.a slight increase in kidney weight, relative to controls. In the first series of studies conducted by the NTP,However, combined exposure to the four compounds at the toxicity of a mixture of 25 chemicals that are oftena dose equivalent to one-half of then* individual found in contaminated groundwater near hazardousLOAELs resulted in clear nephrotoxicity, as measured waste sites (Yang et al., 1989) was examined. Chapinby alterations in multiple endpoints. This study did not et al. (1989) prepared 1, 5, and 10% dilutions of theinclude doses of mixture components that were less mixture in drinking water, and investigated the effectsthan their respective NOAELs. of these mixtures on spennatogenesis in B6C3Fj miceThe results of these studies point out the importance after 90 days of exposure. There were no effects on

of denning "low dose" with respect to the threshold for mean body weight, no histological effects in the liver,each mixture component. In addition, these results are kidney, testis, epididymis, or seminal vesicles, or on theconsistent with the hypothesis that adverse systemic absolute weights of these organs. Epididymal spermeffects are unlikely if the components in a mixture are number and testicular spermatid count were not af-

88 -.-.r.U ET AL.

fected by treatment. The only effect noted \vas an in- (1989) exposed female B6C3Fi mice to a 0.2, 2, or 20%crease in relative kidney weight in the high-dose group drinking water solution for 14 days, and to a 1, 5, or(10% solution). 10% solution for 90 days. In the 14-day study, theThe potential renal and hepatic toxicity of the same high-dose animals (20% solution) exhibited signs of im-aundwater mixture, as well as the effect of exposure munotoxicity, including suppression of granulocyte-the mixture on carbon tetrachloride (CC\4) toxicity, macrophage colony formation, suppression of antigen-

was investigated by Simmons et al. (1994). Fifteen induced antibody-forming cells, and resistance to chal-groups of F344 male rats received one of the following lenge with an infectious agent. In the 90-day study,three drinking water treatments for 14 days: (1) deion- there was evidence of immunotoxicity in the mid- (5%ized water ad libitum, (2) 10% drinking water solution solution) and high-dose (10% solution) groups. In addi-of the groundwater mixture ad libitum, or (3) restricted tion, all treated groups showed signs of mild microcyticdeionized water. At the end of the 14-day treatment anemia. Similar results were reported by Hong et al.period, the rats were gavaged with 0, 0.0375, 0.05, (1992). These investigators exposed female B6C3Fi0.075, or 0.15 ml CCl̂ /kg bw in corn oil. The restricted mice to a 1,5, or 10% dilution of the mixture in drinkingwater control was added because reduced food and wa- water for 2.5 to 31.5 weeks. There was evidence of im-ter consumption were noted in an initial study of the munotoxicity and myelotoxicity (suppression of granu-10% drinking water solution. Rats receiving the 10% locyte-macrophage progenitor cells and erythroid pro- jsolution exhibited significant reductions in food and genitors) in the mid- (5% solution) and high-dose (10% (water consumption, and a significant increase in rela- solution) groups after 15.5 weeks of exposure. jtive kidney weight. However, the water restricted Although the results of these studies appear to refutegroup also exhibited a significant increase hi relative the hypothesis that adverse effects are unlikely whenkidney weight, suggesting that the increased kidney the components in a chemical mixture are at levelsweight observed in the group that received the 10% below their individual thresholds, it is important tosolution may be due to reduced water and food intake, note that the endpoints assessed in the studies are notIn addition, both the water-restricted and 10% solution always assessed in the toxicity studies used to derive agroups exhibited small, but statistically significant, RfD. Second, a RfD has not been determined for severalchanges in several serum parameters; the authors components of the mixture known to cause immunotox-questioned the biological significance of these changes, icity and myelotoxicity including benzene and aroclorThere was no evidence of hepatic toxicity in the three (Table 11), so a comparison of threshold levels with theprater treatment groups in the absence of CC1< treat- actual dose level used in the studies is not possible.

ent, and all three groups exhibited a dose-related in- A second series of studies conducted by the NTP ex-crease in hepatotoxicity following CCi, exposure. How- amined the toxicity of two drinking water mixtures ofever, the response of the groups receiving the 10% solu- pesticides and fertilizers. One mixture simulated ora-tion was similar to that of the water restricted groups, taminants found in California and included the pesti-suggesting that most of the CCl̂ -induced hepatotoxic- cides aldicarb, atrazine, dibromochloropropane, 1,2-di-ity was due to restricted water and food intake. chloropropane, ethylene dibromide, and simazine, andA comparison of the concentration of each chemical the fertilizer ammonium nitrate. The other mixture

in the NTP mixture (at the 10% dilution level) with the simulated contaminants found in Iowa and includedNOAEL or LOAEL of the critical study used to derive the pesticides alachlor, atrazine, cyanazine, meto-the RfD, as well as the RfD, was then made (Table 10). lachlor, and metribuzin and the fertilizer ammoniumThe RfDs of 2 of the 17 chemicals for which an RfD nitrate. Heindel et al. (1994) conducted two studies, awas available, arsenic and cadmium, were based on continuous breeding reproductive toxicity study withhuman data, and therefore the NOAELs should not be Swiss CD-1 mice and a developmental toxicity studycompared to the rodent data of Chapin et al. (1989) and with Sprague-Dawley rats. The animals were admin-Simmons et al. (1994). The RfDs of the remaining 15 istered the California or Iowa mixtures in drinking wa-chemicals were based on rodent data, and in all cases ter at three dose levels, IX, 10X, and 100X, where IXthe doses administered in the Chapin et al. (1989) and was the median concentration of each pesticide compo-Simmons et al. (1994) studies were well below the nent as determined in groundwater surveys in the twoNOAEL of the critical study used to define the RfD. states. In the reproductive toxicity study, neither mix-The results of this comparison are consistent with the ture caused any clinical signs of toxicity, changes inhypothesis that adverse effects will not result from ex- food or water consumption, or body weight in either F0posure to a mixture of chemicals provided that the con- or Fj mice at any dose level. There were no treatmentcentration of each component in the mixture is well related effects on fertility, measures of spermatogene-below its individual threshold. sis, epididymal sperm concentration, percentage motileGermolec et al. (1989) and Hong et al. (1992) investi- sperm, percentage abnormal sperm, or testicular or

gated the potential immunotoxic and myelotoxic effects epididymal histology. In the developmental toxicityof this same drinking water mixture. Germolec et al. study, the pregnant dams were exposed to the mixtures

AR502U&5

••-,v - •, .\--r-~vKN: vr ;I-MIMILOGIES ayTABLE 10

Comparison of the Concentrations of the 25 Groundwater Contaminants in the Chemical Mixture with TheirNOAELS, LOAELS, and Reference Doses

Component

AcetoneAroclor 1260Arsenic trioxide*BenzeneCadmium acctaU hydrate'Carbon tttractilorideChlorobenzeneChloroformChromium chloride hexahydrate'1,1-DSchloroethane1,2-Dichloroethanel,l-Dichloro«thylene1 ,2-t»-ort4-dichlon>ethyleneDKa-ethylhexyDphthalat*EthylbenieneLead acetate trihydrale'Mercuric chloride*Methylene chlorideNickel acittte tetrahydrate'PhenolTetrachloroethyleneToluene1,1,1-TrichloroethaneTrichloro«thyleneXylenes

10% Solution*

4.36ND0.941.255.060.0340.010.643.570.143.470.0450.26ND0.0356.930.0453.40.622.880.300.720.210.730.145

mg/kg/day

NOAEL4

100.0ND0.0008ND0.0050.7119.0ND

1468.0NDNDND17.0ND97.1NDND5.855.060.014.0223.0NDND

179.0

LOAEL'

500.0ND0.014NDND7.054.512.9NDNDND9.0

175.019.0291.0NDND52.650.0120.0 '71.0 •446.0NDND

357.0

Rrrr10.1ND

0.0003ND

0.0050.00070.020.011.0NDND

0.0090.020.020.1NDND

0.060.020.60.010.2NDND

2.0

• Value* from Chapin tt al. (1989).* NOAEL of the itudy used to derive the reference dose (IRIS, 1994).' LOAEL of the ttudy .used to derive the reference dose (LRIS, 1994).* Reference dose (IRIS, 1994).' The concentrations shown are for the metal ion, not the salt.

during gestation days 6-20. Neither mixture caused The RfJDs for aldicarb and ammonium-nitrate arematernal or developmental toxicity. based on human data, and therefore should not be com-To compare the (Jose levels of the individual compo- pared to the rodent data of Heindel et al. (1994). How-

nents in each mixture with an estimate of their individ- ever, there are rodent reproductive and developmentalual thresholds, the IRIS database was searched to de- toxicity data available for aldicarb that provide an esti-ternxine whether a RfD had been established for each mate of the thresholds for comparison. A rat two-gener-of the chemicals. RfDs were available for 4 of the 8 ation reproductive toxicity study of aldicarb estab-pesticides present in the California mixture, and for 4 lished a NOAEL of 0.4 mg/kg/day (IRIS, 1994). Al-of the 5 pesticides present in the Iowa mixture; a RfD though different species were used in the two studies,was also available for the fertilizer component, ammo- the lack of adverse effects in the continuous breedingnium nitrate (Table 11). It should be noted that repro- study of the mixture suggests that the levels of aldicarbductive and/or developmental toxicity data were avail- in the California mixture (0.061 mg/kg/day in the 100Xable for these compounds, but in no case were these dose level) were well below the effect threshold. A ratendpoints determined to be the "critical" endpoint. dietary developmental toxicity study established aTherefore, the critical study used to define the RfD NOAEL of 1 mg/kg/day, and a rat gavage develop-would presumably be below the threshold for reproduc- mental toxicity study established a NOAEL of 0.125tive and developmental effects. The dose levels of the mg/kg/day (IRIS, 1994). Thus, the value of 0.042 mg/four pesticides in the 100X Iowa mixture were well kg/day in the 100X group was well beloŵ the thresholdbelow the respective NOAELs of the study used to de- for developmental toxicity.rive the RfD (Table 11). In the California mixture, the No reproductive toxicity studies have been conducteddose levels of atrazine, simazine, and aldicarb sulfone for nitrate, but a rat 3-generation reproductive toxicitywere well below the respective NOAELs used to derive study of sodium nitrite in drinking water establishedthe RfDs (Table 11). a NOAEL of 2D mg nitrite-nitrogen/kg/day. The EPA

SEED ET AL.

TABLE 11Comparison of the Concentrations of the Pesticide/Fertilizer Components in the California and Iowa

Mixtures with Their NOAELS, LOAELS, and Reference Doses

Component

California mixtureAldicarbAldicarb sulfoneAldicarb sulfoxideAmmonium nitrate'Atrazinel,2-Dibromo-3-chloropropane1,2-DichloropropaneEthylene di bromideSimazine

Iowa mixtureAlachlorAmmonium nitrate'AtrazineCyanazineMetoiachlor .Metribuzin

100X"

•0.0610.0610.06135.70.010.00020.0920.0180.006

0.01731.00.00950.00760.00760.012

100X»

0.0420.0410.04125.00.00680.00010.0630.0110.004

0.01223.30.0060.0050.0050.007

mg/kg/day

NOAEL'

0.010.11ND1.63.5NDNDND0.52

1.01.63.5ND15.02.5

\

LOAEL'

0.0250.58ND

1.8-3.225.0NDNDND '5.3

3.01.8-3.225.0ND

150.037.5

————— j ————————

Rfir

0.0010.001ND1.60.035NDNDND0.005

0.011.60.035ND0.150.025

I• These values reflect the estimated daily intake for F0 male mice in the continuous breeding study (Heindel et al., 1994). Daily intake I

for F0 female, and Ft males and females were within ± 10%. i* These values reflect the estimated intake for pregnant dams in the developmental toxicity study (Heindel et al., 1994). !' The NOAEL from the study used to derive the reference dose (IRIS, 1994). , I4 The LOAEL from the study used to derive the reference dose (IRIS, 1994).' Reference dose (IRIS, 1994)./Expressed as mg nitrate-nitrogen/kg/day.

surned that a maximum of 10% of nitrate is converted issue of low level exposure by examining the carcino-ni trite and calculated a NOAEL of 200 mg nitrate- genie potential of mixtures in which the individual com-

nitrogen/kg/day (IRIS, 1994). It is therefore likely that ponents are at levels that do not result in tumor forma-the levels of 36 mg nitrate-nitrogen/kg/day present in tion when administered alone. For example, Ito et al.the 100X California mixture and the 31 mg nitrate- (1991) administered five heterocyclic amines to rats innitrogen/kg/day present in the 100X Iowa mixture were their diet for 6 weeks at doses 5 and gth of the dosewell below the threshold for reproductive effects in ro- known to result in liver tumor formation for each indi-dents. A rat gavage developmental toxicity study of vidual compound. There was a slight increase of gluta-sodium nitrate and potassium nitrate established a thione S-transferase placenta! form (GST-P) positiveNOAEL of 41 mg nitrate-nitrogen/kg/day. Therefore, foci in the livers of rats exposed to the mixtures com- ;the levels of 25 mg nitrate-nitrogen/kg/day in the 100X pared to rats exposed to the individual compounds. The !California mixture, and 23.3 mg nitrate-nitrogen/kg/ authors concluded that the results were consistent withday in the 100X Iowa mixture were below the thresh- additive effects or slight synergism. •———^old. The results of these comparisons are consistent In a subsequent study, Hasegawa et al. (1994a) ad- •with the hypothesis that adverse effects will not occur ministered 10 heterocyclic amines, individually at /if the components of the mixture are present at levels doses ̂ of that known to be carcinogenic or in combina- ' ibelow their respective thresholds. tion at ̂ and -^ the carcinogenic dose, to rats in theirIn contrast to the situation with noncancer end- diet for 6 weeks. A significant increase in the number

points, it is generally assumed that a threshold does of GST-P positive foci was noted in rats exposed to thenot exist for most genotoxic carcinogens. Therefore, de- ^ dose of the mixture, which was consistent with a ifining "low dose" in reference to a threshold is not possi- synergistic effect. The number of GST-P positive foci \ble. Many studies of the potential interaction among for the rats exposed to the 355 dose level did not differcarcinogenic components of mixtures have been con- significantly from control. Similarly, Hasegawa et al.ducted, but the vast majority of these studies employed (1994b) observed a synergistic effect for intestinal tu-doses of the mixture components known to increase mor development in rats exposed to a low dose (| or gthe tumor incidence. Only a few studies of chemical of the carcinogenic dose) of five heterocyclic amines. ]' /\iixtures have been conducted that have addressed the Warshawsky et al. (1993) reported that application

• -..;• At. MIXTURES RISK ASSESSMENT METHODOLOGIK.S 91

of a 0,1% dose of a mixture of five polycyclic aromatic other chemical classes has been limited for several rea-hydrocarbona (anthracene, chrysene, pyrene, fiuoran- sons, including a lack of clear understanding of thethene, and phenanthrene) to the backs of mice for 104 mechanism of action and the lack of a clear correlationweeks resulted in a tumor incidence of 23%. Coexpo- among in vivo and in vitro assays. Other methodologiessure of mice to the same complex mixture and to 0.001% such as the pharmacodynamic model proposed by Clev-benzo(a)pyrene (BaP), a dose that produced no tumors enger et al. (1991) have been developed to incorporatewhen thia compound was applied singly to mice, re- specific mechanistic data. This is a valuable methodol-sultcd in a 47% tumor incidence after 104 weeks. If the ogy for the specific biological systems that it was de-reaponae was additive, one would have expected the signed for, but different methodologies will be requiredtumor incidence upon coexposure to be equivalent to for different biological systems.the response seen upon exposure to the mixture with- Other efforts have been directed toward modifyingout BaP, indicating that the carcinogenic response at the component-based method for mixtures risk assess-low doses of BaP under these conditions was syner- ment proposed by the U.S. EPA (1986), which assumesgistic. dose additivity for systemic toxicants and response ad-Takayaraa et al, (1989) conducted a long-term study ditivity for carcinogens. Several modifications have

of a mixture of 40 carcinogenic compounds that target been proposed to allow for the incorporation of data ondifferent organs, including the liver, intestine, thyroid, interactions. For example, Mumtaz and Durkin (1992)urinary bladder, and skin. The dose of each component proposed a weight-of-evidence approach to account forof the mixture corresponded to jjth of the TD50 (dose both synergistic and antagonistic interactions in thethat produced tumors in 50% of the exposed animals), derivation of the hazard index. The NRC (1989) recom-At the end of the 2-year exposure period, there was an mended that an additional uncertainty factor be usedincrease in the incidence of follicular cell thyroid tu- in the derivation of the hazard index when necessarymora and neoplastic liver nodules in the exposed rats, to account for synergistic interactions. A basic premiseHowever, the authors could not determine whether the of these modifications is that the,additivity assumptionresponse was additive or synergistic. is not always valid and that there is a need to account

for interactions. This is based on studies of the interac-DISCUSSION AND CONCLUSIONS tions among binary mixture components which often

deviate from additivity. However, caution must b« ex-There has been considerable effort invested in devel- ercised when incorporating data on interactions of bi-

oping methodologies for conducting risk assessments nary pairs in risk assessments. First, most interactionof chemical mixtures. Because of the differences in the data are derived from studies using a single high dosecomplexity of chemical mixtures found in the environ- of one or both agents. Therefore, the observed interac-ment, the diverse components of chemical mixtures, tions are specific for the doses studied, and interactionsour inconsistent understanding of the mechanism by at those dose levels should not be assumed to predictwhich an interaction may occur among mixture compo- effects at other doses. Second, interaction data fromnents, and the varying availability of data, it is becom- studies of binary pairs may not be predictive of theing apparent that no one approach will suffice for con- interaction that would occur in a more complex mix-ducting risk assessments of chemical mixtures. As a ture. Third, few studies have addressed the temporalresult, many of the methodologies have been developed nature of interactions, which is an important compo-ln response to specific risk assessment problems, and nent in determining potential toxicity. Finally, the lim-the most successful have been those that incorporate ited number of binary pairs that have been studied aremechanistic data where available. For example, the not representative of the universe of possible binaryTEF approach was developed because of a need to con- pairs, since the components are usually selected be-duct risk assessments of chlorinated dioxins and chlori- cause an interaction is expected, based on prior infor-nated dibenzofurans. This methodology was therefore mation. Thus, generalization of interaction data ob-designed as an interim risk assessment approach for tained from binary pairs, and incorporation of suchStructurally similar compounds, which operate on the data into a quantitative risk assessment of chemicalsame target tissues by similar mechanisms. The TEF mixtures containing more than two components hasapproach has been a valuable "interim" tool to assess not only been difficult, but may be of questionable va-the toxicity of mixtures of several important classes lidity.of compounds, including PCDDs and PCDFs (NATO/ When developing risk assessment methodologies forCCMS, 1988; Ahlborg, 1989; Ahlborg et al., 1992; Bellin chemical mixtures, it would be useful to determine theand Barnes, 1989; Kutz et al., 1990; Safe, 1990,1993), conditions under vrhich the additivity assumption isPCBs (Ahlborg et al, 1994; Barnes et aL, 1991; U.S. valid, and conditions under which synergistic and/orEPA, 1987, 1991; Safe, 1990, 1993) and polycyclic aro- antagonistic interactions are likely to occur. The U.S.matic hydrocarbons (Nisbet and LaGoy, 1992; U.S. EPA (1986) suggested that additivity is likely whenEPA, 1993). However, the use of this methodology for mixture components are present at low doses and when

92 " SEED ET AL.

the components induce the same toxic effect via the of the mixture are present at levels wtii ':•••'•.•»: wirsame mechanism of action. However, the definition of respective thresholds, it is prudent to anticipate exccp-"low dose" was vague. tions to the rule. In addition, these results should beThere are two sets of studies that are particularly extrapolated cautiously to humans. First, humans may

informative in this regard: the TNO studies and the be more or less susceptible to the toxic effects seenNTP studies. Studies of chemical mixtures conducted in the animal studies. In addition, and perhaps moreat the TNO Nutrition and Food Research Institute in importantly, humans may be exposed to many morethe Netherlands (e.g., Groten et al., 1994; Jonker et compounds than those used in the animal studies in-al., 1990, 1993a,b, 1994) underscore the importance of eluding alcohol, therapeutic drugs, and environmentaldefining "low" levels in reference to the threshold of chemicals. These coexposures may increase or decreasethe individual components of the mixture. Studies were the toxkity of mixtures to humans.conducted both with mixtures of components that act Clearly, more research is needed to evaluate the haz-via a common mechanism and with mixtures of compo- ards and risks from exposures to mixtures of chemicals.nents .that do not act via the same mechanism (and Several animal studies demonstrate the potential fortherefore do not meet the criteria for the additivity synergistic carcinogenic effects upon coexposure to veryassumption). These studies neither support nor refute low doses of mixture components. Methodologies needthe additivity assumption when the mixture compo- to be developed to account for potential interactions.nents are present at levels well below the NOAEL for For noncancer endpoints, attention should be directedthe individual components. However, they do indicate toward investigations of the relationship between tox-that there was no discernible toxic response until the icity of a mixture and the levels of the individual corn-dose levels of the individual components approached or ponents relative to their individual NOAELs. In partic-exceeded their individual thresholds. Regardless of the ular, mixtures containing compounds that are thoughtmechanism of action of the components of the mixture, to act via a common mechanistn should be examined;a variety of interactions were observed when the indi- indeed, the question of what constitutes a "commonvidual components were present at levels equivalent mechanism" is still a thorny issue. Questions like theseto their individual LOAELs and therefore above their need to be addressed before progress can be made inintfividual thresholds. --=x improving the methodologies for risk assessment ofThe studies conducted under* the auspices of the NTP \chemical mixtures.

on the toxicity of well-defined complex mixtures s ofdrinking water contaminants (Chapin' et at., „ 1989; / ACKNOWLEDGMENTSHeindel et al., 1994; Simmons et al., 1994) were not/specifically designed to examine the relationship be/ T06 authors thank the members of the ILSI North America Resi-tween the toxicity of the mixture and the dose levels of !Jue Tf̂ ^ C°m̂ tee for ̂ interest- ̂ 'pM comments and.,.,...,•' , ... , ., . . ,. .-, /, financial support of the protect. The secretarial support of Deborahthe individual components relative to their individual Wilson ̂ ^^ Gasperjis ̂^y appreciated.thresholds. However, these studies generally suppdft]the hypothesis that adverse effects are unlikely when! REFERENCESthe mixture components are present at levers wel]_be- \low their individual thresholds.* _ __ \ '" ;

CHEMICAL MIXTURES RISK ASSESSMENT MKTHODOl.' ••','.: -

Dibtnzo-p-Dioxinj and -Dibenxofurant (CDDs and CDFs). 2nd ed, Jonker. D., Jones, M. A.. Van B.jtf'.-rvn. I' -! . •'. .uur.-.fn. K A . '...U.S. Environmental Protection Agency, Washington, DC. - H. P., and Feron.V. J.U993a>. Acute >-.M ilr.i.,xicity of a combm.i-

Calabres*. E. J. (1991). MultipU Chemical Interactions. Lewis Pub- n̂̂ f0?; nephrotoxicanU in rats cornered with the touciiy ofItahera Inc CheUea MI individual compound*. Food Chan. Toxtcol. 31(1), 45-52.

Chapin, R. E.'. Phelps,' J. L,. Schwetz, B. A., and Yang. R. S. H. J• U993b). Subacute (4-wk) oraUox.c.tv̂ of a combinationcontaminant*. III. Male reproduction study in B6C3F, mice. Fund. ^ t«u nephrotoxins in rats: Comparison -with the toxiaty of theAppl Toxicol 13 388-398 individual compounds. Food Chem. Tox:co(.,31(2), 125-136.

Chen, J. J., Gaylor, D. W., and Kodell. R. L, (1990). Estimation of J™ker' D., Woutersen R A., and Feron. V J. (1994). The Additivitythe joint risk from multiple-compound exposure based on single- Awttmption Tested for Cpmbinatunsof Similarly Acting Nephro-compound experiment*. RUk Anal. 10(2), 285-290. «mcan* Poster presentaUon at the Second Annual HERLSympo-

_. „ _ „ „, „ „,,-,. *. n iium, Chemical Mixtures and Quantitative Risk Assessment,Clevenger. M. A., Putzrath, R. M., Brown. S. L., Gmevan, M. E., Rale- h NC November 7_10, 1994.D«ro«a, C.T., and MumUz.M.M. (1991). Risk assessment of mix- „.,„_. , _,, . T „„«., D j • • • uturw: A model bated on mechanisms of action and interaction. In Kodel}' "•>• ™d 9hen' J" Jr' (19?4)' ̂""nf consf™i*s't° ln "skJR«* AnoJyiu. Projjxcfc and Opportunftbs (C. Zervo., Ed.), PP. estimation for mixtures of carcinogens. Risk Anal. 14, 327-332.293-303 Plenum, New York. Kodell, R. L., Krewski, D., and Ziellnski, J. M. (1991). Additive mnd

Coh«n, S. M,, and Ellwein. L, B. (1991). Genetic errora, cell prolifera- multiplicative relative risk in the two-stage clonal expansion modelUon. md circinogeneri,. Cancer ft.* 51, 6493-6505. of «ranogenesis. Risk And 11(3). 483-490.

Eadon, G., Kaniln.ky, L. S., Silkworth. J. B., Aldous, K., Hilker, D., ^ F' W'' B,«™!f • ̂ G. Bretthauer E W. Bottimore, D. P , andOTC«fe. P., Smith, R.. Gierthy, J., Hawley, J., Kirn. N., and De- Gnem H. 1990). The mtemahonal toaaty equivalency factorcaprio, A. (1986). Calculation of 2.3,7.8-TCDD equivalent concen- (l'™ me*od for estimatmgnsks associated vnth exposjires totrationa of complex environmental contaminant mixtures. Envi- comPl«x ml̂ "r̂ of±XlnS and relaUd coraP°unda- ̂oacoJ. Envi-ron. Htdth PuZptA 70, 221-227. ron- Chem- 26' "-109'

Germolec, D. R., Yang. R. S. H., Ackermann, M. F., Rosenthal, Lê '-J-aSSS). Development of a comparative potency method forO. J., B«man G. A., BWr, P., â d Luater. M. I. (1989). Toricology ci?ncer "s.k as,sesf̂ ent of ̂P ex. ̂J," ,û inf •̂ rtLteTn '"»tudi« of a chimical mixture of 25 groundwater contaminanta U. olvo "* tn vttro *«»***?*• **>»*• *?*• ««**• ». 193-203.Itamunoauppreasioa in B6C3F, mice. Fund. AppL Traicol. 13, Mumtaz, M. M., and Durkin, P. D. (1992). A weight of evidence377-387. approach for assessing interactions in chemical mixtures. Tcadcol.

GroUn, J. P., Kupw, F. C., Schoen, E. D.. Van Bladeren, P. J., and IruL Health- 8(6)' S77-406-Feron, V. J. (1994). Subacute toricity studies of a combination of Mumtaz, M. M., and Hertzberg, R. C. (1993). The status of interne-nine chemical, in raU. Poster presentation at the Second Annual ti°ns dat» » risk assessment of chemical mixtures. In HazardIIERL Symposium. Chemical Mixtor*s and Quantitative Risk Ass«ssm

94 _..,_.__,_, _-,_... _.SEF,n K7 Ai,..._- .

Safe, S. (1993). Development of bioassays and approaches for the for CsV.m.iun? ruks associated with exposures to mixtures of clii- •risk assessment of 2,3,7,8-tetrachlorodibenzo-p-dioxin and related rinau-d dibcnzo-p-dioxina and -dibenzofurans iPCDDs andcompounds. Environ. Health Perspecl. 101 (Suppl. 3). 317-325. PCDKsi. EPA/625/3-87/012.

Simmons J E Yang, R. S. H., Svendsgaard, D. J.. Thompson.. U.S. Environmental Protection Agency (1991). Workshop report onM. B Seely. J. C., and McDonald, A. (1994). Toxicology'studies of toxicity equivalency factors for polychlorinated bipheny! cogeners.a chemical mixture of 25 groundwater contaminants: Hepatic and Risk Assessment Forum. EPA/625/3-91/020, U.S. EPA. Washing-renal assessment, response to carbon tetrachloride challenge, and ton- DC.influence of treatment-induced water restriction. J. ToxicoL Envi- U.S. Environmental Protection Agency (1993). Provisional guidanceran. Health 43, 305-325. f°r quantitative risk assessment of polycyclic aromatic hydrocar-

™ , o „ „ j n. < • n /,nom o u- .- bons. EPA/600/R-93/089.Takayama, S., Hasagawa, H., and Ohgaki. O. (1989). Combination ... , , , , „ , , , „ , „• ^ r, ,.„«„. ̂ ,effects offoAy carcinogens administered at low doses to male rats. Warshawsky. D., Barkley. W.. and Bjngham E (1993 . Factors af-Jpn J Cancer Res 80 732-736 fectmg carcinogenic potential of mixtures. Fund.AppL Toxicol. 20,

3/6—382.U.S. Environmental Protection Agency (1986). Guidelines for the y f R- g H>> Q̂ ,̂ T- ĵ Browni R. D.f Chatham, A. T., Arneson.health risk assessment of chemical mixtures. Fed. Reg. 51(185). D. W., Buch.nan, R. C., and Harris. R. K. (1989). Toxicology stud-34014-34025. iea of a chemical mixture of 25 groundwater contaminants. I.

U.S. Environmental Protection Agency (1987). Interim procedures Chemistry development Fund. Appl. Toxicol. 13, 366-376.

AR502l*9l

J. Clean Technol.. Environ. Toxicol.. & Occup. Med., Vol. 6. No. 2.1997 189

RISK ASSESSMENT FOR CHEMICAL MIXTURES: FINE-TUNING THEHAZARD INDEX APPROACH

M. M. MUMTAZ/ K. A. POIRIER,' AND J.T. COLMAN*

'Agency for Toxic Substances and Disease RegistryPublic Health Service

U.S. Department of Health and Human ServicesAtlanta, Georgia

tProctor & GambleSharon Woods Technical Center

Cincinnati, Ohio.

'Syracuse Research CorporationSyracuse, New York

In the absence of whole mixture toxicity testing data, risk assessments are performedfor most chemical mixtures, using the potency-weighted addition of the toxicity ofthe individual components of the mixtures. This approach, the hazard index (HI) orthe component-based approach, assumes no interactions between the componentsof the mixture and assumes dose additivity. An HI is calculated as the sum of thehazard quotients, that is, the ratio between the exposure level and an acceptableexposure level of each individual component of the mixture. The goal of the HIapproach is to construct the plausible toxicity index that would have been calculatedhad the mixture itself been tested. However, in practice, the guidance is being appliedonly to the critical effect of each component chemical because acceptable exposurelevels for major effects of a chemical are not derived. Thus, the toxicity to severaltarget organs is not being included in the overall toxicity assessment process. Inthis research study, we investigated the feasibility of developing target-organ toxicitydoses (TTDs)for chemicals found at hazardous waste sites and their applicabilityas alternatives to the acceptable exposure levels (ALs) currently available incalculating the HI for other than "critical" effects. TTDs were developed using an

!. Address all correspondence to: M. M. Mumtaz, Ph.D., Agency for Toxic Substances and DiseaseRegistry, Public Health Service, U.S. Department of Health and Human Services, 1600 Clifton Road E-29Atlanta, GA 30333. Tel.:(4O4)639-6300. Fax:(404)639-6315. E-maD:MGM@ATSOD3.EM.CDC.GOV.

2. Abbreviations: AL, acceptable exposure level; HI, hazard index; HQ, hazard quotient; LOAEL.lowest-observed-adverse-effect level; MRL, minimal risk level; NOAEL, no-observed-adverse-effect level;NPL, National Priorities List; PEL, permissible exposure limit; RAGS, Risk Assessment Guidance forSuperfund; RfD, reference dose; TLV, threshold limit values; TTD, target organ toxicity dose.

3. Key words: chemical mixtures, hazard index, risk assessment4. Panly presented at Tri-Service Conference on Application of Advances in Toxicology to Risk

Assessment, Wright-Patteison Air Force Base, OH, May 1992 and the Society of Toxicology Annual MeetingNew Orleans, LA, 1994.

5. The views expressed in this paper are those of the authors and do not necessarily reflect the viewsand policies of the U.S. Department of Health and Human Services or the U. S. Environmental ProtectionAgency.

Journal of Clean Technology, Environmental Toxicology, and Occupational Medicine, Vol. 6, No. 2 DP 189-204Copyright O1997 Princeton Scientific Publishing Co., Inc.

ISSN: 1052-1062

190 Mumta-etal.

approach analogous to that used for minimal risk level (MRL) and reference dose(RfD) derivations. Data to derive TTDs are available for various environmentalpollutants, we compared the results of a risk assessment of a simulated mixtureemploying the published methods and the TTD method proposed in this paper. Thepublished methods, as practiced, underestimate or overestimate the mixture's toxicity.The use of the TTDs allows evaluation of toxicity to multiple target organs and thusthe realization of the actual intent of the mixtures guidelines (USEPA, 1986). Thus,TTDs enhance the utility and consistency of the application of the chemical mixturesguidance towards a fuller and more realistic characterization of risk.

INTRODUCTION

Mixtures, as defined by the Environmental Protection Agency (USEPA, 1986), are "anycombination of two or more chemical substances regardless of source or of spatial or temporalproximity." Mixtures include chemicals stored or disposed of in the same area and released as amixture to the environment (e.g., at Superfund sites). In the absence of data on the specificmixture of concern or a similar mixture, the guidelines recommend that the risk assessment bebased on the components of the mixture and that due consideration be given to the potentialinteractions between the components. If sufficient quantitative data regarding interactions areavailable, an interaction model is to be used to combine risk assessments. In the absence ofsufficient data regarding interactions, the mixtures guidelines (USEPA, 1986) recommend anapproach for systemic toxicity assessments based on dose additivity. The first step in this processis calculating a hazard quotient (HQ) for each component of the mixture. The HQ is the ratio ofthe actual exposure or monitored level (E) to the acceptable exposure level (AL) of a givenchemical. The hazard index (HI) is calculated as the sum of the HQs for all the known componentsof the mixture:

HI = Et/ALi+E2/AL2...+Ei/AL, '* •

•where: .

£. = the actual exposure level to the /th toxicantAL. = the acceptable exposure level for the /th toxicant

= the hazard quotient of the /th toxicant

The ALs could be the minimal risk levels (MRLs) derived by the Agency for Toxic Substancesand Disease Registry (ATSDR), the reference doses (RfDs) derived by the U.S. EnvironmentalProtection Agency (USEPA), and other values such as the threshold limit values (TLVs) derivedby the American Conference of Governmental Industrial Hygienists (ACGIH).

The American Conference of Governmental Industrial Hygienists (ACGIH, 1995) and theOccupational Safety and Health Administration (OSHA, 1993) recommend a similar additiveapproach usine the ratio of the exposure concentration to the exposure limit (TLV), or permissibleexposure limit (PEL), for each chemical and sum of the ratios of all the chemical components of

J. Clean Technoi, Environ. Toxicol., & Occup. Med., Vol. 6, No. 2,1997 191

the mixture. If the sum of the ratios > 1, for example 1.78, the exposure limit for the mixture isexceeded. ACGIH specifies that this approach applies only when the components in a mixturehave similar toxicologic effects, whereas OSHA does not restrict the approach to chemicals withsimilar effects. The National Institute for Occupational Safety and Health (NIOSH, 1976) adopteda similar approach in recommending exposure limits for methylene chloride when carbonmonoxide was also present because of the known additivity of the two chemicals with regard toformation of carboxyhemoglobin. The NIOSH (1992) recommendations are based oncarcinogenicity.

Because it is based on the assumption of dose additivity, the HI approach is most appropriatelyapplied to chemicals that cause the same effect by the same mechanism of action (USEPA, 1986,1989; ACGIH, 1992). In other words, the chemical components are considered as dilution? ofone another, that is, they have similar dose-response curves but their potencies vary. The goal ofthe HI approach is to approximate the toxicity index that would have been calculated had themixture itself been tested. However, the mixtures guidelines were being applied only to thecritical effect of each component chemical even though the chemical might cause other effects aswell. Thus, the toxicity to several organs was not being assessed.

The hazard index is a numerical indication of the nearness to acceptable limits of exposure or theextent to which acceptable limits are exceeded for the mixture. As the hazard index approaches 1,concern for the potential hazard of the mixture increases. If the hazard index >I, the concern isthe same as if the exposure for an individual chemical exceeded the acceptable level by the sameproportion, particularly if one particular compone