the roof gardens - detrafford estates group

FUNDAMENTAL AND APPLIED TOXICOLOGY 36, 1 5 - 2 9 (1997)ARTICLE NO. FA962281

Subacute Toxicity of a Mixture of Nine Chemicals in Rats: DetectingInteractive Effects with a Fractionated Two-Level Factorial Design

JOHN P. GROTEN, ERIC D. SCHOEN,* PETER J. VAN BLADEREN, C. FRIEKE KUPER,

JOB A. VAN ZORGE,| AND VICTOR J. FERON

Division of Toxicology, TNO Nutrition and Food Research Institute, Zeist; * Department of Applied Statistics, TNO Institute of Applied Physics, Delft;and ^Ministry of Housing, Spatial Planning, and Environment, The Hague, The Netherlands

Received February 1, 1996; accepted December 2, 1996

Subacute Toxicity of a Mixture of Nine Chemicals in Rats: De-tecting Interactive Effects with a Fractionated Two-Level FactorialDesign. GROTEN, J. P., SCHOEN, E. D., VAN BLADEREN, P. J.,KUPER, C. F., VAN ZORGE, J. A., AND FERON, V. J. (1997). Fun-dam. AppL Toxicol. 36, 15-29.

The present study was intended (1) to find out whether simulta-neous administration of nine chemicals at a concentration equalto the "no-observed-adverse-effect level" (NOAEL) for each ofthem would result in a NOAEL for the combination and (2) to testthe usefulness of fractionated factorial models to detect possibleinteractions between chemicals in the mixture. A 4-week oral/inhalatory study in male Wistar rats was performed in which thetoxicity (clinical chemistry, hematology, biochemistry, and pathol-ogy) of combinations of the nine compounds was examined. Thestudy comprised 20 groups, 4 groups in the main part (n = 8) and16 groups in the satellite part (n = 5). In the main study, the ratswere simultaneously exposed to mixtures of all nine chemicals[dichloromethane, formaldehyde, aspirin, di(2-ethylhexyl)phtha-late, cadmium chloride, stannous chloride, butyl hydroxyanisol,loperamide, and spermine] at concentrations equal to the"minimum-observed-adverse-effect level" (MOAEL), NOAEL, or1/3NOAEL. In the satellite study the rats were simultaneouslyexposed to combinations of maximally five compounds at theirMOAEL. These combinations jointly comprise a two-level factorialdesign with nine factors (=9 chemicals) in 16 experimental groups(1/32 fraction of a complete study). In the main part many effectson hematology and clinical chemistry were encountered at theMOAEL. In addition, rats of the MOAEL group showed hyperpla-sia of the transitional epithelium and/or squamous metaplasia ofthe respiratory epithelium in the nose. Only very few adverse ef-fects were encountered in the NOAEL group. For most of the endpoints chosen, the factorial analysis revealed main effects of theindividual compounds and interactions (cases of nonadditivity)between the compounds. Despite all restrictions and pitfalls thatare associated with the use of fractionated factorial designs, thepresent study shows the usefulness of this type of factorial designto study the joint adverse effects of defined chemical mixtures ateffect levels. It was concluded that simultaneous exposure to thesenine chemicals does not constitute an evidently increased hazardcompared to exposure to each of the chemicals separately, pro-vided the exposure level of each chemical in the mixture is at mostsimilar to or lower than its own NOAEL.

The bulk of studies to assess the toxicity of chemicalsdeals with exposures to single compounds (Yang, 1994a).Although toxicity studies of single compounds are importantfor obtaining basic toxicological information, man is alwayssimultaneously exposed to a large number of chemicals.With the possible exception of some specific mixtures, it isuncertain how the combined toxicity of these chemicalsshould be assessed or how combined toxicity should be takeninto account in standard setting for the individual com-pounds.

The main problems in the risk assessment of chemicalmixtures are in fact the possible chemical interactions thathamper the prediction of the toxicity of the mixture. Sincethese interactions may occur at various end points in thetoxicodynamic as well as the toxicokinetic phase, the combi-nation toxicologist has the unpleasant task of dealing witha vast scope of chemical interactions. Because of this multi-tude of interactions, each possible chemical interaction can-not be tested individually, and due to the immense numberof combinations involved, systematic and complete toxicitytesting of chemical mixtures is practically impossible. Stud-ies dealing with the toxicology of mixtures are confrontedwith this problem; for instance, Yang and Rauckman (1987)indicated that a 25-chemical mixture of ground water con-taminants has 2U — 1 or 33,554,431 possible combinationsto be tested. A design to test the systemic toxicity of thishuge number of groups is virtually impossible from an ethi-cal, economical, and practical point of view. One way toovercome this problem is to treat the mixture as a singlecompound and to test the mixtures as a whole. This approachhas been advised for mixtures that are not well characterized(Mumtaz et al, 1993), but it has also been applied for as-sessing the combined toxicity of defined chemical mixturesconsisting of nephrotoxicants, pesticides, carcinogens, and/or fertilizers (Jonker et al, 1993; Charturved, 1993; Heindelet al., 1994; Feron et al., 1995a; Ito et al, 1995). In thesestudies an experimental design was chosen, reflecting thenet combined effects of all components in the mixture; tolimit the number of test groups possible, interactive effectsof the components in relation to the effects of individualchemicals were not taken into account.

15 0272-0590/97 $23.00Copyright O 1997 by the Society of Toxicology.

AU rights of reproduction in any form reserved.

Downloaded from https://academic.oup.com/toxsci/article-abstract/36/1/15/1652050by gueston 01 February 2018

16 GROTEN ET AL.

Whether mixtures should be tested in a simple, definedmode or in a complex, undefined mode, the main questionfrom a governmental perspective is whether the exposure tomixtures of chemicals at low, realistic doses is of real healthconcern. Although the number of papers dealing with thetoxicology of mixtures gradually increases, most of thesestudies are restricted to two or three compounds tested inshort-term studies at relatively high, often toxic doses. Thereis a clear lack of information on prolonged, repeated toxicitystudies on combinations of chemicals (more than two) atlow (nontoxic) and subtoxic doses (Krishnan et al., 1991;Heindel et al, 1994). In acute and subacute toxicity studiesin rats it has been shown that combined oral administrationof compounds at the "no-observed-adverse-effect level"(NOAEL) of each of them did not lead to clear additivityor synergism of effects, provided the mechanism of actionof the compounds was dissimilar (Jonker et al., 1990, 1993).In contrast, in a 4-week toxicity study with mixtures con-sisting of four nephrotoxicants with similar mode of action,it was shown that the dose-additivity rule could be applied(Feron et al., 1995b) at dose levels around the NOAEL.These 4-week toxicity studies were interpreted based on thecommon approaches in the assessment of mixtures; for acombination of compounds with a similar mode of action,one might expect dosis-addition, whereas compounds witha dissimilar mode of action may show effect addition (Bliss,1939; Placket and Hewlett, 1952; Mumtaz et al, 1994). Theresults of Jonker et al. (1990, 1993) indicate the unjustifi-ableness of general application of the additivity rule in riskassessment of exposure to chemicals. Since this preliminaryconclusion may have far-reaching consequences for healthrisk assessment, it was deemed desirable to carry out a simi-lar type of study using a combination of compounds highlyrelevant to the general population in terms of use patternand level and frequency of exposure. Such a study shouldbe focused not only on the effect of the mixture as a whole,but also on possible interactive effects between the com-pounds of the mixture.

One way to detect interactive effects, i.e., nonadditiveeffects, between more than two chemicals in a chemicalmixture is to use factorial designs. The use of factorial de-signs, in which n chemicals are studied at x dose levels (x"treatment groups), has been suggested by the U.S. Environ-mental Protection Agency as one of the valuable statisticalapproaches for risk assessment of chemical mixtures (Svens-gaard and Hertzberg, 1994). Very recently a 2s study waspresented to describe interactions between the carcinogenicactivity of 5 polycyclic aromatic hydrocarbons (Nesnow,1994) and a 53 study has been used to identify nonadditiveeffects of three compounds on developmental toxicity (Na-rotsky et al, 1995). Full factorial designs however lead tovery costly experiments, and even if only two dose levels areused, it is already virtually impossible to perform complete,conventional toxicity tests using 2" test groups to identify

interactions between all chemicals of interest. One way todeal with this problem is the use of fractionated factorialdesigns. These fractionated designs still identify most of theinteractions between the compounds and determine whichcompounds are important in causing effects, but have theadvantage that the number of test groups is manageable(Plackett and Burman, 1946; Box et al, 1978). Fractionalfactorial designs have been shown to be an efficient, i.e.,cost effective, approach to identify interactive effects be-tween seven trace elements and the cadmium accumulationin the body (Groten et al, 1991) and to determine structure-activity relationships for 10 halogenated aliphatic hydrocar-bons (Eriksson et al, 1991).

The present study was intended to determine whether si-multaneous administration of nine compounds at a concen-tration equal to the NOAEL for each of them will result ina NOAEL for the combination. The second aim was to testthe usefulness of fractionated factorial models to predictpossible interactions in chemical mixtures without the obli-gation to test large numbers of combinations.

MATERIALS AND METHODS

Materials

Dichloromethane (methylene chloride), cadmium chloride, di(2-ethylhex-yl)phthalate (DEHP), and acetyl salicylic acid (aspirin) were obtained fromE. Merck (Darmstadt, Germany). Paraformaldehyde was purchased fromJanssen Chimica (Beerse, Belgium) and loperamide was generously pro-vided by Janssen Chimica (Beerse, Belgium). Butyl hydroxyanisol (BHA),stannous chloride, and spermine hydrochloride were obtained from Fluka(Basel, Switzerland). All test materials were of analytical grade.

Animals and Maintenance

Albino male rats, Wistar outbred [Crl(WI)WU(BR)], were obtained froma colony maintained under SPF conditions at Charles River Wiga GmbH,Sulzfeld, Germany. At the start of treatment, the rats were 9-10 weeks old.Before the start of the experiment, the rats were housed under conventionalconditions in an animal room four or five males to a cage in suspendedstainless steel cages fitted with wire mesh floor and front. During the experi-ment the rats were housed individually in inhalation chambers in similarstainless steel cages. Animals were rotated weekly, in such a way that theywere kept equally divided over each chamber. The room temperature waskept at 22 ± 20°C and the relative humidity at 40-70%. During the exposurethe animals had no access to food or water. After exposure, the animalswere fed the institute's cereal-based rodent diet or one of the test diets.Diets and drinking water were provided ad libitum. Community tap waterwas supplied from an automatic drinking-water system. Lighting was artifi-cial, with a sequence of 12 hr light and 12 hr dark.

Experimental Design and Treatment

A 4-week oral/inhalatory study was performed in which the toxicity ofcombinations of nine compounds was examined. The nine chemicals wereselected from the data base on 4-week toxicity studies previously performedat our institute. The studies chosen were public accessible. The chemicalsthat were selected represent a combination of compounds highly relevantto the general population in terms of use pattern and level and frequencyof exposure (drugs, food additives, food contaminants, biogenic amines,and industrial solvents). The compounds were not chosen based on their


INTERACTIVE EFFECTS OF NINE CHEMICALS 17

similar mode of action. The study comprised 20 groups, 4 main groups(eight animals/group) and 16 satellite groups (five animals/group).

Main Groups

Rats of the main groups were simultaneously exposed to nine chemicals(dichloromethane, formaldehyde, aspirin, DEHP, cadmium chloride, stan-nous chloride, BHA, loperamide, and spermine) at concentrations equalto the minimum-observed-adverse-effect level (MOAEL), NOAEL, or1/3NOAEL; the fourth group was a control group receiving basal diet notsupplemented with test chemicals and breathing fresh air. The dose levelswere established based on results of previous 4-week studies, which wereperformed under similar test conditions in our laboratories. A brief summaryof the results from these preliminary studies is shown in Table 1.

Satellite Groups

In the 16 groups of the satellite study the rats were simultaneously ex-posed to various combinations of chemicals, all at the MOAEL. The 16groups jointly comprise a two-level factorial design with nine factors. Thus,the present study was a 1/32 fraction of a complete two-level study (i.e.,1/32 X 2*). Each compound is absent in 8 of the experimental groups andpresent in the other 8. For any pair of compounds, 4 of the 16 groupscontain both compounds, 4 groups contain neither, 4 groups contain onlythe first one of the pair, and 4 groups contain only the second compoundof the pair (see Table 2). The combinations of chemicals in the satellitestudy were chosen such that the results would allow analysis of the interac-tions between the nine chemicals (two-factor interactions), but would alsoallow for an optimal analysis between the main effects of the individualcompounds (see Statistical Analysis).

Except for the "all-MOAEL group," the satellite study had five animalsper group. The animals were allocated to the groups by restricted random-ization, using the quintile of their body weight as allocation criterion. Thus,before randomization there were five initial body weight groups. Eachtreatment group had one randomly chosen animal from each of the bodyweight groups. The main study, including the all-MOAEL group, had eightanimals per group. By analogy with the satellite study, there were 8 initialbody weight groups. This randomization was carried out separately fromthe randomization of the satellite study. When complementing results ofthe 15 satellite groups with the all-MOAEL group, we used the animalsfrom the 5 main study body weight groups with a mean closest to the 5satellite study body weight groups. In subsequent analysis, we used for thebody weight groups a blocking variable. We were apparently successful inthat, because the grouping accounted for a substantial part of the variationin many parameters.

Test Diets

Except for formaldehyde and dichloromethane, all test substances wereadministered ad libitum via the diet, at constant dietary concentrations, for4 weeks.

Test diets were prepared by blending the test compounds and cereal-based rodent diet in a Stephan cutter and were then stored in sealed plasticbags in a freezer at -20°C. Twice a week the feeders were refreshed withthe test diets. Analysis of the diets revealed that the actual concentrationof the test compounds at 1/3NOAEL, NOAEL, and MOAEL was as follows(expressed as a percentage of intended value): 75, 84, and 96% for aspirin;95, 96, and 95% for BHA; 90, 86, and 100% for cadmium chloride; 94,87, and 88% for DEHP; 76, 91, and 102% for spermine; and 92, 82, and83% for stannous chloride. Loperamide was only analyzed at the MOAELlevel and concentration was 103% of the intended level. All compoundswere stable during a 14-day storage period.

Inhalatory Exposure

Animals were exposed to formaldehyde and dichloromethane by inhala-tion in H-1000 multitiered inhalation chambers manufactured by Hazleton

c

ioU

3

|

•3S

Isr

3 S3

uac

2

i

00

I8.X

u

H "'I << > , _ t/5

< f ? ̂5 3 ' ^

«

65

H «_•

I!

u

=o o. a. co

_23°

*

Si:

- — v «j

f a M ° : •

j[jj * u gj< E- c <O < o O

x U E x

< <OOJ5

- * 5 « •

g

ill

S ^

siI o.f- JJ

• I !: >

i^llilll-X X X -J

• r^ XX

: X enin c^

2 €

E-

h >•2 *

2D.

i1

&JSfe!2 : £ ui

!8 8-> fN —

!l

H q< I

m a

•5 -o

II

III00 S -g

CO

•n - 2

"H

§n

Q -5•o eU °

c

III111

"2 aO -t8.

i

o

V u n —

si si5 ion o

j a"3 U T3 V

C . c CL•- o a -r•O S -c mcj — o "J

I » S -i J

2 13 < D. >.

a < | i?

g < ° m

E • I oS 3 a«»

| 2 °. oD O *^

E c = as e o =E o-u -e u u. f?^ £ « •-;2>?l 1

ol 1 I.: -E P ?


18 GROTEN ET AL.

TABLE 2Test Groups and Exposure Levels of a 4-Week Toxicity Study with Combinations of Nine Compounds in Male Rats

Formaldehyde Dichloromethane Aspirin CdCl2 SnCl2 Loperamide Spermine(For) (MC) (Asp) (Cd) (Sn) (Lop) (Sper) BHA DEHP

Main groups

4 GroupsControl1/3NOAELNOAELMOAEL

16 Groups"ForSn/MC/Lop/AspCd/MC/Sper/AspSn/Cd/Sper/Lop/ForBHA/MC/Sper/LopSn/BHA/Sper/Asp/ForCd/BHA/Lop/Asp/ForSn/Cd/BHA/MCDEHP/Sper/Lop/AspSn/DEHP/MC/Sper/ForCd/DEHP/MOLop/ForSn/Cd/DEHP/AspBHA/DEHP/MC/Asp/ForSn/BHA/DEHP/LopCd/BHA/DEHP/SperMOAEL*

Satellite groups

Note. Dose levels of MO ART, and NOAEL are given in Table 1. Main groups and satellite groups comprised eight and five animals, respectively." In the satellite study all compounds were dosed at the MOAEL.* Data of five animals of the main study were used for the establishment of the 16th group of the satellite study (for details see Materials and Methods).

Systems Inc., USA. The chambers are illuminated externally by normal thereafter. Food consumption (FC) was measured once every week bylaboratory TL-lighting. The number of air changes was at least 12 per hour. weighing the feeders.The rats were exposed for 6 hr a day, 5 days a week, during 4 weeks,resulting in a total number of 20 exposure days. Dichloromethane was Hematologyevaporated by bubbling pressurized air through the test material. Formalde-hyde gas was generated by dissolving paraformaldehyde in water and vapor- On Days 23 and 24 blood samples were taken from the tip of the tail ofization of this solution under heating. The generated mixture of air with all animals and examined for hemoglobin (Hb) concentration, packed cellformaldehyde and/or dichloromethane was diluted with filtered air from volume (PCV), red blood cell count (RBC), white blood cell count (WBC),the air-conditioning system to obtain the desired test concentrations. The and dirombocytes (THROM) using a Sysmex K-1000 hematology analyzerconcentration of dichloromediane in the test atmosphere was determined (Toa Medical Electronics Co. Ltd., Japan). Prodirombin time (PTT) wasby total carbon analysis using a flame ionization detector (Carlo Erba, Italy). determined using a Normotest kit. Methemoglobine (MHb) and carboxyFormaldehyde concentrations were determined colorimetrically by means hemoglobin (CO-Hb) concentration were determined within 1 hr after sam-of the Hantzsch reaction using an analyzing system of Skalar Analytical pling according to the methods of Brown and Bauer, respectively. The(Breda, Netherlands). Atmosphere samples were taken automatically from mean corpuscular volume (MCV), mean corpuscular hemoglobin (MCH),each of the chambers in an alternating order at a location close to the and mean corpuscular hemoglobin concentration (MCHC) were calculated,animals' site. The sampling lines were heated to avoid condensation of

the test material during sampling and transportation. Each chamber was Clinical Chemistrymonitored about once every hour. Analysis of the test atmosphere showedthat the actual concentrations of dichloromethane were 36 ± 2.1, 103 ± Clinical chemistry was conducted at autopsy in blood collected from the5.0, and 520 ± 20 ppm for 1/3NOAEL, NOAEL, and MOAEL, respectively. abdominal aorta of all rats. Blood was collected in heparinized plastic tubesThe mean actual concentrations of formaldehyde during the 20 exposure and centrifuged at 1250g for about 15 min, using Sure-sep TJ dispensersdays were 0.35 ± 0.03, 1.09 ± 0.1, and 3.1 ± 0.25 ppm for 1/3NOAEL, from General Diagnostics and then analyzed for alkaline phosphataseNOAEL, and MOAEL, respectively. (ALP), alanine aminotransferase (ALAT), aspartate aminotransferase

r DJKP ti (ASAT), gammaglutamyl transpeptidase (GGT), total protein (TP), choles-Observations t e r o l ( C H O L ) tngiycerides (TriGlyc), albumin (ALB), urea, creatinine

Genera] health status of the rats was recorded twice a day. Body weights (CREAT), total bilirubin (Bili-Tot), nonfasting glucose (GLUQ, sodium,(BW) were recorded individually on the first day of the study and weekly potassium, calcium, chloride, and inorganic phosphate. Analyses were per-



formed on a Cobas-Bio centrifugal analyzer using Baker and Boehringerreagent kits. Sodium and potassium were analyzed using an Electrolyte-2analyzer, chloride was analyzed with a Chloro counter.

Biochemistry

At autopsy the left lateral lobe of the liver was preserved in liquidnitrogen. The tissue was thawed and homogenized using a Teflon pestle in a10 mM Tris-HCl buffer (pH 7.4, 4°C). The palmitoyl-CoA oxidase activity(PalmCoA) was determined in the total liver homogenates after a triplethaw/freeze procedure according to Reubsaet et al. (1988).

Pathology

On Day 28 rats were killed by exsanguination via the abdominal aortaunder ether anesthesia and then examined grossly for pathological changes.The weights of the adrenals, heart, kidneys, spleen, testes, lungs, and liverwere recorded and the ratios of organ weight to body weight (i.e., relativeweight, RW) were calculated. These organs, and in addition esophagus andstomach, were preserved in neutral aqueous phosphate-buffered formalinsolution, embedded in paraffin wax, sectioned at 5 fim, stained with hema-toxylin and eosin, and examined microscopically. The head was removedfrom each of the carcasses and flushed retrograde through the nasopharyn-geal orifice with 10 ml 4% buffered neutral formaldehyde solution. Thereaf-ter, the heads were decalcified and sections of the nose (5 /*m) were preparedas described previously (Cassee et al., 1996b). Nose section levels 2 and3 were examined for histopathological changes. In die main part of thestudy microscopic examination was carried out for all organs; in the satellitepart of the study, only target organs (i.e., liver and nasal passages) wereexamined microscopically.

Statistical Analysis

Main Croups and Satellite Croups

Body weights on Day 28 were evaluated by one-way analysis of covari-ance followed by an application of Dunnett's multiple comparison testOrgan weights and hematology, clinical chemistry, and biochemistry valueswere evaluated by one-way analysis of (co)variance followed by an applica-tion of Dunnett's multiple comparison test. Histopathological data wereevaluated by Fisher's exact probability test.

Satellite Groups

If all possible combinations of the compounds would have been tested,the study would have comprised 29 (=512) combinations of chemicals.However, at most 16 satellite groups could be managed experimentally.Therefore, the satellite part of the study comprised a 1/32 fraction of a two-level factorial design with nine factors (=9 chemicals) in 16 experimentalgroups as described by Box et al. (1978). The use of a fractional factorialdesign implies that we must deal with a complex confounding patternbetween main effects and interactions (this means that main effects arealiased with each other and with interactions). First, the complete aliasespattern in a 1/32 fraction of a 29 design was worked out in code form (A,B, . . . , J) according to Box et al. (1978). The next step was to assignchemicals to the codes and formaldehyde was chosen for code J. Theconfounding pattern of the present study with the aliases between two-factor interactions and main effects is given in Table 3. We intentionallychose formaldehyde aliased with two-factor interactions because at theconcentration levels chosen formaldehyde acts as a local irritant in the noseand we can assume dial there will be no further interaction with systemiceffects of the other compounds in the study. This will greatly help theinterpretation of die aliased pattern since there will be eidier responseswidiout any effect of formaldehyde (omit these effects from the list in Table3) or responses without interactions between two nonformaldehyde effects(omit these effects from the list). Therefore, J = formaldehyde is a goodchoice to interpret the confounding pattern. To analyze the aliasing of first-

TABLE 3Confounding Pattern between Compounds in the Satellite

Groups Using a Fractionated 29 Design with Nine Chemicals in16 Experimental Groups in a 4-Week Toxicity Study

Aliases between main effect and two-factor interactionSn = Sper.LopCd = Lop.ForBHA = Asp.ForDEHP = Cd.ForMC = DEHP.ForSper = Sn.ForLop = Cd.ForAsp = BHA.ForFor = DEHP.MC = Sn.Sper = Cd.Lop = BHA.Asp

Aliases between compounds involved in two-factor-interactionSn.Cd = BHA.MC = Sper.Lop = DEHP.AspSn.BHA = Cd.MC = DEHP.Lop = Sper.AspSn.DEHP = MC.Sper = BHA.Lop = Cd.AspSn.MC = Cd.BHA = DEHP.Sper = Lop.AspSn.Lop = BHA.DEHP = Cd.Sper = MC.AspSn.Asp = Cd.DEHP = BHA.Sper = MC.Lop

Note. Division of test groups and complete aliases pattern in a 33 fractionof 2 ' design was first worked out in code form (A, B, . . . J) according toBox et al. (1978). The next step was to assign chemicals to the codes.Formaldehyde was chosen for code J. Aliases between two- and three-factor interactions are not evaluated in the present model. Abbreviations ofchemical names are shown in Table 2.

order interactions with each other, when such a "string" is statisticallysignificant, we know that at least one of the interactions is active. Whichone? (1) Look at active main effects, (2) look to determine whether thereis an interaction in the string involving two active main effects. If there is(are), this (these) is the most probable one(s). If not, (3) look to determinewhether there are interactions involving one of the active main effects. Inany case, use the opinion of the experts.

Main effects. If all compounds would show effect-addition, die effectof any compound could be calculated by subtracting the mean of the groupsnot containing the compound from die mean of the odier groups. Thegeneral balance in die design ensures diat die effects of die other compoundsare canceled. Thus, me joint effect of the compounds could be decomposedinto nine main effects.

Nonadditivity between the compounds. It is plausible that some (most)compounds will not exhibit effect-addition and in diese cases die termnonadditivity is applied to describe die interaction between the compounds.A measure of nonadditivity is die difference between die effect of a com-pound in the presence of anodier one and die effect of the compound indie odier one's absence. It follows from the properties of the experimentaldesign that the measures of nonadditivity are confounded (i.e., aliased) inseven sets of four confounded first-order interactive effects. In one of diesesets the interactive effects are also confounded with the main effect offormaldehyde. Any interaction of formaldehyde with another compound isconfounded widi a main effect not due to die odier compound. We assumethat cases of interaction (read nonadditivity) of two compounds did notdepend on die presence of a diird one. Thus, diree-factor interactions arenot taken into account.

Statistical significance of die effects concerning body weights, organweights, and hematology, clinical chemistry, and biochemistry values, re-spectively, was evaluated widi t tests according to Fisher. In this approachmultiple / tests are performed only when a preliminary F test turns out tobe statistically significant (5% level). This method is preferred above die


20 GROTEN ET AL.

popular Bonferroni method. The latter method dictates performing singletests at a level of 0.0033 (=0.05/15) to obtain a joint significance level ofthe 15 tests carried out on one and the same response of at most 0.05.Through its restrictive level of significance, the Bonferroni method is lesspowerful dian the Fisher method (Hochberg and Tamhane, 1987).

There were 15 independent tests for each parameter, each test correspond-ing to one of the effects as given in Table 3. Statistical significance in agroup of confounded compounds implies statistical significance of maineffects or two-factor interactions. Which one(s) is a matter of expert inter-pretation as described above or of further research. Since many effects arespecific for one or two of the compounds, expert interpretation in this sensewill hardly be necessary. Thus, we were able to attribute most cases ofstatistical significance of confounded groups as shown in Table 3 to onlyone of the several possibilities [i.e., main effect or two-factor interaction(s)].

For the analysis of the carboxy-Hb findings, the data were reduced to 0= absent and 1 = present. These data, as well as the histopathologicalfindings, were evaluated with stepwise regression analysis under a general-ized linear model using the logit-link and a binomial error distribution(McCullagh and Nelder, 1989). For the satellite groups, we used a stepwiseforward selection procedure and initially tried out main effects only. Maineffects were included in the model whenever the corresponding decreasein deviance exceeded 3.84. In a second step, interactions were fitted aswell. For die reduced data of the main study the following independentbetween-groups comparisons were studied: (1) control versus three dosagegroups, (2) 1/3NOAEL versus NOAEL, and (3) NOAEL versus MOAEL.

RESULTS

All rats survived to scheduled autopsy. No relevant clini-cal signs were observed during the study. The hematologicalparameters PTT, RBC, and MCHC were not affected. Theclinical chemistry of the plasma did not show relevantchanges in inorg-P, citrate, sodium, urea, and chloride. Theseparameters are therefore not presented in the tables.

Main Groups

Compared to the control group, only rats in the MOAELgroup showed growth retardation and a decrease in foodintake (data of Day 28 are shown in Table 4). Combinedexposure of the nine compounds at the MOAEL resulted ina significant decrease in MCV and thrombocyte counts (forhematology parameters see Table 4). CO-Hb concentrationwas significantly increased in the NOAEL and MOAELgroups (3.7 and 11.0%, respectively) and values were similarto those found in the preliminary study in rats that wereexposed to dichloromethane alone (cf. Table 1). A CO-Hbformation of less than 5% is not considered to be of toxico-logical significance (WHO, 1987). There were no other he-matological effects in the main groups.

At the MOAEL, many effects on clinical chemistry (Table5) were encountered such as a decrease in ALP activity, adecrease of the glucose, triglycerides, and cholesterol con-centrations, and an increase in ALAT/ASAT activities, albu-min, bilirubin, and total protein concentrations. In the lowerdose groups only a few statistically significant changes inclinical chemistry were noted: a decrease in ALP activityand triglyceride concentration in the rats of the NOAELgroup and a decrease in bilirubin concentration in the rats

of the 1/3NOAEL group. The change in bilirubin levels wasnot seen in the preliminary studies and there was not a cleardose-effect relationship. Therefore, it is doubtful whetherthis bilirubin effect can be attributed to the treatment.

In the MOAEL group the weights of heart and spleenwere decreased and the liver weight was increased (Table6). The relative weights of all organs (except those of spleenand heart) were significantly higher in rats of the MOAELgroup (data not shown). Relative kidney weights were alsoincreased in rats of the NOAEL and 1/3NOAEL groups(Table 6).

Gross examination at autopsy was essentially negative.There were no treatment-related histopathological changesin the adrenals, heart, esophagus, spleen, testes, lungs, orlarynx. Treatment-related histopathological changes wereseen in the liver and nasal cavity in the MOAEL group, butalso in the NOAEL group (Table 7). In the livers of allrats of the MOAEL group hepatocellular hypertrophy wasobserved. Similar, but much less severe and less frequentchanges in the liver were seen in all rats of the NOAELgroup. In the range-finding study histopathological liverchanges had been observed in rats exposed to DEHP oraspirin at the MOAEL and occurred also in rats exposed tospermine at levels above the MOAEL (cf. Table 1).

The nasal cavity of all rats in the MOAEL group showedvery slight to moderate hyperplasia of the respiratory andtransitional epithelium and/or very slight to moderate squa-mous metaplasia of both epithelia at cross level 2. Surpris-ingly, similar nasal changes were seen in all rats of theNOAEL group, although to a lesser degree (Table 7). Fromthe results of the preliminary studies with the individualcompounds it was clear that such effects on the nasal epithe-lium were to be attributed to formaldehyde and not to anyof the other compounds. However, it was also known thatrats exposed to I ppm formaldehyde alone would not showsuch or any other nasal effects (Woutersen et al., 1987).

A few rats showed focal inflammatory cell infiltrates and/or alveolar hemorrhages in the lungs (data not shown). Theincidence of these lesions was slightly (though not statisti-cally significantly) higher in the animals of the MOAELgroup than in the controls.

Satellite Groups

In the satellite study the rats were exposed at the MOAELonly; compared to the controls, several changes in hemato-logical and clinical data (Tables 4-6) were encountered. Toanalyze main effects of individual compounds and interac-tions between the compounds the data set as shown in Tables4 - 6 was subjected to a factorial analysis as described underMaterials and Methods. With the application of the factorialdesign it was possible to identify main effects of the individ-ual compounds and interaction (cases of nonadditivity) be-tween two compounds.


TABLE 4Mean Body Weights, Food Consumption, and Hematological Findings in Male Rats Exposed for 4 Weeks to Selected Combinations

of Maximally Nine Compounds via the Food or by Inhalation

Control1/3 NOAELNOAELMOAEL

ForSn/MC/Lop/AspCd/MC/Sper/AspSn/Cd/Sper/Lop/ForBHA/MC/Sper/LopSn/BHA/Sper/Asp/ForCd/BHA/Lop/Asp/ForSn/Cd/BHA/MCDEHP/Sper/Lop/AspSn/DEHP/MC/Sper/ForCd/DEHP/MC/Lop/ForSn/Cd/DEHP/AspB H A/DEH P/MC/Asp/ForSn/BHA/DEHP/LopCd/BHA/DEHP/Sper

BW28g

304.3 ± 21.1302.5 ± 14.3297.6 ± 20.7262.8 ± 20.4**

304.5 ± 25.5269.1 ± 17.5267.4 ± 20.8285.0 ± 21.1286.3 ± 14.2270.0 ± 9.7266.6 ± 14.4283.6 ± 19.2270.7 ± 20.2290.0 ± 16.8287.2 ± 16.0264.5 ± 13.9282.1 ± 19.1289.3 ± 13.0289.5 ± 22.3

FC28g/rat/day

20.1 ± 0.419.9 ± 0.219.3 ± 0.818.2 ± 0.6

19.0 ± 1.118.4 ± 0.418.7 ± 0.519.1 ± 0.619.6 ± 0.319.3 ± 0.518.9 ± 0.319.6 ± 0.617.9 ± 0.518.9 ± 0.418.6 ± 0.418.1 ± 0.419.3 ± 0.419.8 ± 0.519.2 ± 0.5

PCVliter/liter

0.439 ±0.0150.434 ± 0.0050.425 ± 0.0090.446 ± 0.006

0.453 ± 0.0090.460 ± 0.0080.427 ± 0.0080.427 ±0.0110.450 ±0.0100.433 ± 0.0040.427 ± 0.0030.414 ± 0.0060.441 ± 0.0070.442 ± 0.0040.418 ± 0.0100.436 ± 0.0090.447 ± 0.0060.422 ± 0.0030.416 ± 0.004

HB mmol/liter

MCVliter/liter

Main groups

9.8 ± 1.29.9 ± 0.39.7 ± 0.6

10.2 ± 0.3

50.9 ± 0.450.7 ± 0.650.5 ± 0.550.1 ± 0.5*

Satellite groups

10.2 ± 0.310.5 ± 0.29.8 ± 0.39.8 ± 0.5

10.4 ± 0.59.9 ± 0.29.7 ± 0.29.5 ± 0.3

10.1 ±0.210.1 ± 0.39.6 ± 0.5

10.1 ± 0.310.3 ± 0.39.7 ± 0.19.5 ± 0.2

50.7 ± 0.351.6 ± 0.348.8 ± 1.4**49.9 ± 0.450.9 ± 0.350.2 ± 0.949.0 ± 0.8**48.7 ± 0.4**51.4 ± 0.551.2 ± 0.848.0 ± 1.0**49.8 ± 0.351.6 ± 0.450.1 ± 0.648.0 ± 0.6**

MCHfmol

.13 ± 0.05

.15 ± 0.01

.15 ± 0.02

.14 ± 0.01

.15 ± 0.02

.18 ± 0 .03"

.12 ± 0.03

.14 ± 0.02

.18 ± 0.01**

.15 ± 0.02

.12 ± 0.01

.12 + 0.03

.18 ± 0.01**

.17 ± 0.02*

.10 ± 0.03

.16 ± 0.03

.19 ± 0.02**

.15 ± 0.02

.10 + 0.01

CO-Hb%

0.8 ± 1.81.1 ± 1.03.7 ± 1.5**

11.0 ± 1.3**

0.0 ± 0.08.0 ± 1.5**9.3 ± 0.9**0.0 ± 0.02.7 ± 2.20.0 ± 0.00.0 ± 0.07.1 ± 2.3**0.7 ± 1.06.4 ± 2.5**3.9 ± 2.70.0 ± 0.03.3 ± 0.50.0 ± 0.00.0 + 0.0

MHb%

.16 ± 0.21

.11 + 0.16

.29 + 0.22

.35 ± 0.50

.64 + 0.38

.33 ± 0.27

.22 ± 0.25

.44 + 0.59

.35 ± 0.21

.45 ± 0.15

.38 ± 0.34

.15 ± 0.41

.31 + 0.20

.32 ± 0.17

.23 ± 0.25

.34 ± 0.43

.33 + 0.46

.26 ± 0.32

.32 ± 0.37

WBClO'/liter

12.6 + 2.611.3 ± 1.511.5 + 1.313.1 + 2.6

11.5 ±12.0 +11.0 ±10.8 ±9.9 ±9.1 ±

11.4 ±

.4

.9

.8

.0

.1*

.0*

.38.3 + 0.4**9.5 ± 3.1*

10.2 + 0.711.0 + 0.710.5 + 1.49.5 ± 0.9*

10.4 + 1.89.6 + 1.6*

THROMlO'/liter

597 ±561 ±549 ±457 ±

571 +438 ±465 +537 +526 +478 +531 ±592 ±474 +552 ±593 ±497 ±543 ±590 ±596 +

317264

135**

2895**35*

10211363344070*567757411970

Note. Only the parameters showing treatment-related changes are included in this table (for abbreviations see Materials and Methods). Values are the means ± SD for groups of eightanimals (main groups) or five animals (satellite groups). Food intake are cage means (5 rats/cage). Although body weight and food intake were recorded weekly, only the mean over 4weeks is presented. The values marked with asterisks differ significantly from the controls [body weight, analysis of covariance ± Dunnett's test; hematology parameters, ANOVA ±Dunnett's test except for CO-Hb and MetHb; Kruskal-Wallis ANOVA ± Mann-Whitney U tests (two-sided): *p < 0.05, **p < 0.01]. To analyze main effects of individual compoundsand interactions between compounds, the data set of the satellite groups was also subjected to factonal analysis (see Materials and Methods) and results are presented in Table 8. Groupcodes: For, formaldehyde; MC, methylene chloride; Asp, aspirin; DEHP, di(2-ethylhexyl)phthalate; Cd, cadmium chloride; Sn, stannous chloride; BHA, butyl hydroxanisol; Lop, loperamide;and Sper, spermine.


TABLE 5Mean Values of Clinical Chemistry in Plasma and Biochemistry in Liver of Male Rats Exposed for 4 Weeks to Selected Combinations

of Maximally Nine Compounds via the Food or by Inhalation


For

Sn/MC/Lop/AspCd/MC/S per/AspSn/Cd/Spcr/Lop/ForBHA/MC/S per/LopSn/BHA/Sper/Asp/ForCd/BHA/Lop/Asp/ForSn/Cd/BHA/MCDEHP/Sper/Lop/AspSn/DEHP/MC/S per/ForCd/DEHP/MC/Lop/ForSn/Cd/DEHP/AspBHA/DEHP/MC/Asp/ForSn/BHA/DEHP/LopCd/BHA/DEHP/Sper

GLUCmmo I/liter

9.52 + 0.838.96 ± 0.709.47 ± 0.398.28 ± 0.32"

7.94 + 0.428.03 ± 0.928.01 ± 0.878.49 ± 0.488.48 ± 0.607.71 ± 0.88"8.61 + 1.268.09 + 0.449.66 ± 1.058.22 + 0.638.47 ± 0.888.48 + 0.938.44 ± 0.508.78 ± 0.618.60 ± 0.73

ALP

U/liter

306 + 50278 ± 16254 ± 45°227 ± 2 1 "

281 ± 26234 ± 1 7 "268 ± 34*180 ± 6*226 ± 3 5 "210 ± 1 7 "221 ± 1 0 "174 ± 9 "323 ± 29222 + I I "244 ± 20*°236 + 1 3 "246 ± 1 4 "184 ± 7 "204 ± 1 4 "

ALATU/liter

47 ±43 ±47 ±59 ±

46 ±47 ±81 ±53 ±48 ±44 ±88 +46 +52 ±43 ±80 ±52 ±49 +44 ±69 ±

7

4

8

6 "

4

9

11**7

8

7

7 "4

6

17

9 "

4

3

8

7 "

ASATU/liter

70 ± 765 ± 771 ± 1082 ± 7*

70 ± 77 1 + 686 ± 1375 ± 765 ± 370 ± 1196 ± 11*65 ± 777 ± 1271 ± 1688 + 14*80 ± 1068 ± 1269 ± 872 ± 7

TP

g/literALB

g/liter

Main groups

61 ± 161 + 16 0 ± 158 + 2 "

Satellite

61 ± 256 ± 1 "57 ± 1**59 + 362 + 258 ± 158 + 260 ± 257 ± 1 "59 ± 159 + 356 + 160 + 261 + 162 ± 2

32 + 132 + 032 ± 134 + 1 "

; groups

32 +32 +33 ±32 +33 ±33 ±33 ±33 ±33 ±33 +34 +33 ±34 +34 ±34 + • •

Bili-Tot/jmol/liter

2.0 ± 0.41.5 ± 0 . 3 "1.8 ± 0.22 5 + 0.4*

1.7 + 0.22.3 + 0.12.8 ± 0.21.8 + 0.31.8 ± 0.32.0 ± 0.32.0 ± 0.41.8 ± 0.22.6 ± 0.41.2 ± 0.2**1.5 ±0.12.6 + 0.3*2.0 ± 0.41.2 ± 0 . 2 "1.5 + 0.4

CHOLmmol/liter

.65 ± 0.10

.64 + 0.1162 + 0.09.40 ± 0.11**

82 + 0.1936 + 0.12".28 + 0.10".64 + 0.09.78 + 0.10.53 + 0.09.59 ± 0.06.86 + 0.06.29 ± 0 .08".44 + 0.07*.40 ± 0.07**.36 ± 0.06".53 + 0.17.91 ± 0.14**.75 + 0 12

Tnglycmmol/liler

0.96 + 0.200.89 + 0 230.68 + 0.20*0.43 + 0.14**

0.90 + 0.160.54 ± 0.11**0.50 + 0.12"0.62 + 0.05**0.65 i 0.12**0.59 ± 0.15**0.48 ± 0.1 !••0.64 ± 0.17"0.47 ± 0 12**0.49 + 0.20**0.43 + 0.08**0.50 ± 0 1 3 "0.67 ± 0 .15"0.58 + 0.16"0.75 + 0.25*

CREATjimol/liter

22 ± 120 ± 122 ± 123 + 1*

21 ±25 +24 ±20 +20 i23 +24 ±20 ±25 ±20 +23 ±25 + 225 ± 219 + 2*19 + 1*

PalmCoA/xmol/min/ml

0.44 ± 0.080.45 + 0.070.53 ± 0.091.60 + O.52**

0.34 + 0.020 88-+ 0.23*0.73 ± 0.340.39 + 0 100.38 + 0.060.64 ± 0.250.75 + 0.160.35 + 0.031.35 ± 0 45**1.22 ± 0.15**0.92 ±0.16*1 44 + 0.28"1.36 ± 0.32"0.59 ±0.170 63 + 0.16

Note. Only the parameters showing treatment-related changes arc included in this table (for abbreviations see Materials and Methods). Values are the means ± SD for groups of eightanimals (main groups) and for groups of five animals (satellite groups). The values marked with asterisks differ significantly from the controls [ANOVA ± Dunnett's test (two-sided): *p< 0.05, **p < 0.01]. To analyze main effects of individual compounds and interactions between compounds the data set of the satellite groups was also subjected to factorial analysis (seeMaterials and Methods) and the results are presented in Table 8. Group codes: For, formaldehyde; MC, methylene chloride; Asp, aspirin; DEHP, di(2-ethylhexyl)phthalate; Cd, cadmiumchloride; Sn, stannous chloride; BHA, butyl hydroxanisol; Lop, loperamide; and Sper, spermine.



TABLE 6(Relative) Organ Weights of Male Rats Exposed for 4 Weeks to Selected Combinations

of Maximally Nine Compounds via the Diet or by Inhalation


ForSn/MC/Lop/AspCd/MC/Sper/AspSn/Cd/Sper/Lop/ForBHA/MC/Sper/LopSn/BHA/Sper/Asp/ForCd/BHA/Lop/Asp/ForSn/Cd/BHA/MCDEHP/Sper/Lop/AspSn/DEHP/MaSper/ForCd/DEHP/MOLop/ForSn/Cd/DEHP/AspBHA/DEHP/MaAsp/ForSn/BHA/DEHP/LopCd/BHA/DEHP/Sper

Wadrenalsg

0.049 ± 0.0050.052 ± 0.0050.052 ± 0.0050.048 ± 0.005

0.053 ± 0.0030.048 ± 0.0030.049 ± 0.0030.055 ± 0.0080.051 ± 0.0050.048 ± 0.0030.048 ± 0.0050.050 ± 0.0050.048 ± 0.0030.051 ±0.0080.050 ± 0.0030.047 ± 0.0080.048 ± 0.0050.055 ± 0.0050.051 ±0.005

Wkidneysg

2.22 ±0.192.34 ± 0.112.35 ± 0.142.24 ±0.17

2.12 ±0.202.24 ± 0.212.12 ± 0.252.17 ± 0.232.16 ± 0.112.34 ± 0.212.19 ± 0.082.19 ± 0.212.19 ± 0.312.20 ± 0.202.20 ± 0.252.26 ± 0.152.29 ± 0.222.25 ± 0.142.25 ± 0.25

RWkidneysg/kg

Wspleeng

Main groups

7.29 ± 0.267.73 ± 0.31*7.91 ± 0.36**8.52 ± 0.21**

0.534 ± 0.0340.526 ± 0.0330.509 ± 0.0330.464 ± 0.033**

Satellite groups

7.37 ± 0.258.25 ± 0.16**7.90 ± 0.26**7.54 ± 0.157.51 ± 0.188.62 ± 0.34**8.17 ± 0.20**7.68 ± 0.337.99 ± 0.34**7.44 ± 0.217.24 ±0.168.10 ± 0.16**8.07 ± 0.43**7.69 ± 0.107.72 ± 0.27*

0.52 ± 0.050.48 ± 0.020.47 ± 0.020.50 ± 0.030.48 ± 0.040.48 ± 0.050.46 ± 0.040.49 ± 0.050.46 ± 0.050.50 ± 0.050.49 ± 0.030.46 ± 0.050.48 ± 0.050.50 ± 0.050.48 ± 0.05

Wheartg

1.03 ± 0.030.98 ± 0.020.95 ± 0.020.91 ± 0.02**

1.01 ± 0.140.92 ± 0.090.88 ± 0.071.01 ± 0.050.96 ± 0.050.92 ± 0.050.92 ± 0.030.97 ±0.110.94 ±0.110.98 ± 0.080.95 ± 0.140.87 ± 0.080.98 ± 0.140.97 ± 0.080.97 ±0.11

Wliverg

10.9 ± 0.9710.95 ± 0.8111.17 ± 1.0511.51 ± 1.03*

10.46 ± 0.8110.18 ± 0.449.85 ± 1.129.79 ± 1.0

10.61 ± 0.5510.07 ± 1.2010.06 ± 0.9210.68 ± 1.3710.55 ± 1.2311.23 ± 1.4010.84 ± 0.9810.43 ± 0.7012.36 ± 0.67*11.88 ±0.7012.13 ± 1.04*

Wlungg

.26 ± 0.10

.24 ± 0.05

.25 ± 0.08

.17 ± 0.08

.22 ± 0.05

.13 ± 0.11

.09 ± 0.08

.17 ± 0.11

.22 ± 0.08

.16 ± 0.05

.16 ± 0.10

.20 ± 0.11

.20 ± 0.14

.18 ± 0.08

.19 ± 0.11

.13 ±0.08

.23 ± 0.05

.22 ± 0.11

.19 ± 0.11

Note. Values are the means ± SD for groups of eight animals (main groups) or five animals (satellite study). The values marked with asterisks differsignificantly from the controls [ANOVA ± Dunnett's tests (two-sided): *p < 0.05, **p < 0.01]. To analyze main effects of individual compounds andinteractions between compounds the data set of the satellite groups was also subjected to factorial analysis (see Materials and Methods) and results arepresented in Table 8. Group codes: For, formaldehyde; MC, methylene chloride (dichloromethane); Asp, aspirin; DEHP, di(2-ethylhexyl)phthalate; Cd,cadmium chloride; Sn, stannous chloride; BHA, butyl hydroxyanisol; Lop, loperamide; and Sper, spermine.

A final equation to describe the value of the parameter inany particular mixture in terms of the variables (i.e., com-pounds) tested was defined (Box et al., 1978) as

Variable,,^ = Mean + 1/2 * EffectA * A

+ 1/2 * EffectAB * A • B, etc.,

where Variable,^ is the total value for the variable in anyparticular mixture chosen; Mean is the overall mean from16 experimental groups; EffectA is the mean effect of com-pounds A, B, etc.; EffectAB is the interactive effect betweencompounds A and B; and A and B have values of either ± 1or —1, i.e., presence or absence of compound A and com-pound B. The final equations for all relevant hematologicaland clinical parameters are shown in Table 8. We will exem-plify the analysis of Table 8 by way of the parameters Palm-CoA and ASAT.

With respect to the PalmCoA activity in the liver, thefactorial analysis revealed that two compounds (aspirin andDEHP) were able to induce PalmCoa activity, whereas BHAslightly reduced the activity. Moreover, there was a rather

slight and unexpected, but significant (/? < 0.01) interactionbetween BHA and DEHP, which resulted in a decreasedtotal PalmCoa activity. Figure la illustrates the interactiveeffect of BHA and DEHP and shows how this interactioncan be visualized from the two-by-two plot of the effect ofthe individual compounds. In the figure the interactive effectbetween two compounds is indicated by the absence of paral-lel lines. In summary, the factorial analysis resulted in thefollowing equation:

PalmCoA = 0.85 + 0.4 • DEHP + 0.3 * Asp

- 0.1 * BHA - 0.07 * BHA * DEHP.

Regarding the parameter ASAT activity the analysis resultedin the following equation:

ASAT,^ (units/liter) = 75.4 - 2.49 * Sn + 5.32 * Cd

+ 2.79 * Lop + 3.66 * Asp + 2.37 * Cd * Lop

- 2.47 * Sn * Cd.


24 GROTEN ET AL.

TABLE 7Type and Incidence of Histopathological Changes in the Nasal Cavity (Level 2) and Liver of Rats Exposed for 4 Weeks

to Combinations of Nine Chemicals via the Diet or by Inhalation

Hyperplasia of respiratory epitheliumVery slightSlightModerateTotal

Hyperplasia of transitional epitheliumVery slightSlightModerateTotal

Squamous metaplasia of respiratory/transitional epitheliumVery slightSlightModerateTotal

Inflammatory cell infiltrationVery slightSlightModerateTotal

Hepatocellular hypertrophyVery slightSlightTotal

Aggregates of RES cells and necrotic hepatocytesSlight

Perivascular mononuclear cell infiltrateSlight

Control

Nasal cavity

0000

1001

0000

1304

Liver

000

0

0

1/3 NOAEL

0101

0202

2002

1304

000

1

0

NOAEL

0011

4228*

1023

0112

314

0

2

MOAEL

0448*

0448*

3317*

1315

448*

0

1

Note. Eight animals were examined per group. No other treatment-related changes in any of the examined organs were found compared to controls(Fisher's exact test, *p < 0.05).

This means that there were three compounds able to increasethe ASAT activity (cadmium chloride, aspirin, and lopera-mide). Thus, rats exposed to one of these three chemicalsshowed an increased ASAT compared with rats which werenot exposed to these chemicals (rats in 8 of 16 groups).One compound (stannous chloride) was able to decrease theASAT activity. Two cases of significant interactions wereidentified, namely the interaction between cadmium chlorideand loperamide (Cd.Lop) and between stannous chloride andcadmium chloride (Sn.Cd). In the former case the effect ofthe combination of the two compounds was 4.74 units/literhigher than could be expected on the basis of summation ofthe effects of the two single compounds cadmium chlorideand loperamide (i.e., Cd.Lop = +4.74); in the latter case,the effect of the combination of the two compounds was 4.94units/liter smaller than expected on the basis of additivity ofstannous chloride and cadmium chloride (i.e., Sn.Cd =—4.94). Figures lb and lc exemplify how the interactive

effects between Cd and Lop and between Cd and Sn shouldbe interpreted. Here again, the interaction is indicated bythe absence of parallel lines between the effect of the twocompounds.

For every random selection of mixtures from the ninecompounds tested it is possible to predict the overall effectfor any particular parameter with the final equations asshown in Table 8. For instance, for animals exposed to cad-mium chloride (viz. Sn absent, Cd present, and Asp absent),the body weight on Day 28 can be estimated to be 279.2 -2.4 * - 1 - 3.4 * 1 - 10.0 * - 1 = 288.2 g.

In the case of the parameters ALP and PCV the finalequation should be regarded with caution due to the fact thatnumerous (possible) aliases are present between the identi-fied two-factor interactions. Due to this large number ofinteractions we cannot interpret the confounding pattern andit is rather difficult to identify the main effects of the mixture(cf. Tables 3 and 8).



TABLE 8Final Equations for Body Weights, Food Intake, and Selected Clinical and Hematological Parameters

of the Satellite Groups in Terms of Main Effects and Two-Factor Interactions

BW28= 279.23 - 2.37*Sn - 3.44*Cd - 10.01*AspFI28= 18.89 + 0.31 *BHA - 0.33*AspWBC= 10.39 - 0.42*BHA + 0.45*Lop + 0.43*Lop*BHA + 0.43*Sn*LopHB= 9.95 - 0.17*Cd + 0.08*Asp + 0.11*Sn*Cd - 0.14»Cd*MCPCV= 434.54 - 8.84*Cd - 3.56*BHA + 4.14*Asp + 3.99*Sn*Cd - 3.74*Sn*BHA + 3.57*Cd*DEHPMCV= 50.0 ± 0.19*Sn - 0.98*Cd - 0.17*BHA + 0.32*Asp + 0.40*Sn*Cd + 0.26*Sn*BHAMCH= 1.15 ± 0.005*Sn - 0.02*Cd + 0.005*MC + 0.005*Asp - 0.01*Sn*Cd - 0.005*Cd*MCALB= 33.19 + 0.36*BHA + 0.61*DEHPALP= 229.75 - 19.84*Sn - 8.71*Cd - 16.75*BHA + 5.54*DEHP + 13.50*Asp + 7.63* Sn*Cd + 9.5*Sn*BHA

+ 7.79*Cd*BHA + 5.79*Sn*DEHP + 4.84*Cd*DEHPBili-Tot= 1.95 + 0.09*Cd - 0.11*BHA + 0.07*Sper + 0.39*Asp - 0.12*BHA*Asp + 0.14*DEHP*AspCREAT= 22.0 - 0.34*BHA + 0.32*MC + 1.96*AspCHOL= 1.56 - 0.025*Cd + 0.11*BHA - 0.05*DEHP - 0.06*MC - 0.045*Sper - 0.15*Asp + 0.025*Sn*Cd + 0.025 *BHA*DEHPGLUC= 8.39 + 0.22*DEHP + 0.21 *LopTHROM= 529.71 - 14.84*Sper - 39.61*AspMHb= 1.33 - 0.07* MCPTT= 40.41 - 0.56*BHA + 0.57*DEHP + 0.65*MC + 0.81 *AspASAT= 75.39 - 2.49*Sn + 5.32*Cd + 2.79*Lop + 3.66*Asp + 2.37*Cd*Lop + 2.47*Sn*CdALAT= 56.40 - 7.58*Sn + 9.8*Cd + 2.63*Lop + 2.6*Asp - 5.53*Sn*CdTP= 58.95 - 0.45*Sn + 1.05*BHA - 1.43*Asp + 0.35*BHA*DEHPNA= 72.75 + 0.51*MC + 0.46*Lop + 0.62*Asp - 0.44*Sn*DEHPTriglyc= 579.87 - 36.88*DEHP - 33.38*MC - 52.13*Lop - 54.38*Asp + 33.38*DEHP*Asp + 47*MC*AspPalmCoA= 0.85 + 0.4*DEHP + 0.3*Asp - 0.1*BHA - 0.07*BHA*DEHPWadren= 0.05 - 0.002*Asp + 0.01 *BHA*DEHPWkidney= 2.21 - 0.03*Cd + 0.03*BHARWkidney= 7.9 + 0.1*Sn + 0.02*BHA + 0.3*AspWspleen= 0.48 - 0.11 *AspWheart= 0.94 - 0.28*AspWliver= 10.63 - 0.17*Cd + 0.44*BHA + 0.55*DEHP - 0.16*lop + 0.18*BHA*DEHPWlung= 1.18 - 0.015*Cd - 0.02*Asp + 0.02*Asp*DEHP

Note. The equation for two compounds A and B in a mixture showing significant main and interactive effects can be described as Variable,^ = Mean+ 1/2 EffectA * A + 1/2 * Effect,, * B + 1/2 * EffectAB * A • B, where Variable^ is total value of the variable of any particular mixture chosen, Meanis total overall mean from 16 experimental groups, Effect is main effect compound A, EffectAB is interactive effect between compounds A and B, andA and B indicate the presence of compounds A and B (1 or - 1 , i.e., present or absent). With the equation it is possible to determine the overall valueof the parameter in any particular mixture in terms of main effects and two-factor interactions. Only significant main effects and two-factor interactionsare given (p < 0.05). If the parameter does not show significant interactions, the chemicals behave in an effect-additive way.

The CO-Hb formation and the histopathological findingswere evaluated in a similar way (see Materials and Methods).As expected, the CO-Hb formation was mainly due to thepresence of dichloromethane and no interactions with othercompounds were observed. There was a significant increaseof the incidence of hepatocellular hypertrophy in rats treatedwith aspirin and DEHP (p < 0.01). Also, rather unexpect-edly, it appeared that rats exposed to cadmium chlorideshowed a slight increase in the incidence of liver cell hyper-trophy (p < 0.05). In the liver there were no significantcases of nonadditivity. In the nose of rats exposed to formal-dehyde there was a clear increase in the incidence of totalepithelial hyperplasia and squamous metaplasia comparedto rats not exposed to formaldehyde (p < 0.01). Factorialanalysis also revealed that in the absence of formaldehyderats exposed to dichloromethane showed a slightly increasedincidence of epithelial hyperplasia (p < 0.05). In the pres-ence of formaldehyde most of the rats showed epithelial

hyperplasia independent of the presence or absence of di-chloromethane (data not shown).

DISCUSSION

Exposure at the MOAEL

The main part of the study revealed that combined expo-sure to nine compounds at the MOAEL of the individualcompounds resulted in a wide range of adverse effects vary-ing from signs of general toxicity (growth depression, re-duced food intake), changes in hematological values (de-creased MCV, increased CO-Hb formation, decreased num-ber of thrombocytes), changes in biochemical/clinicalparameters (decreased glucose, triglyceride, cholesterol, andALP levels; increased ASAT, ALAT, and PalmCoA levels),to increased relative organ weights and histopathologicalchanges in nose, liver, and lungs. Based on the toxicity data


26 GROTEN ET AL.

aDEHP

BHAabsent

MOAEL

mean

absent

0.54

0.50

0.52

MOAEL

1.22

0.88

1.05

mean

0.88

0.69

0.79

bLop

Cd

absent

MOAEL

mean

absent

69.65

75.55

72.60

MOAEL

70.50

85.85

78.18

mean

70.07

80.70

75.39

cCd

Snabsent

MOAEL

mean

absent

70.10

70.05

70.07

MOAEL

85.65

75.75

80.70

mean

77.87

72.90

75.39

FIG. 1. Two-factor interactions (p < 0.05) for the parameters PalmCoa (a) and ASAT (b and c). Interactions are calculated from the two-by-two-plot of the individual compounds shown on the left. Statistical analysis was performed with the data of the 16 experimental groups of the satellite study.Overall mean effects of the individual compounds can be calculated from the margins of the tables. Values in the tables are expressed as ^mol/min/mlfor PalmCoA activity and as units/liter for ASAT activity. The figure on the right illustrates the two-factor interactions: Interactive (nonadditive) effectsare indicated by the absence of parallel lines. Details on the analysis are provided under Materials and Methods.

of the individual compounds, indeed we expected to seemost of these effects also in the combination. A few effectsseen in the subacute toxicity studies with the individual com-pounds had disappeared in the combination, whereas someeffects not seen in the range-finding studies appeared in thecombination. For example, the increase in relative weightsof kidneys, lungs, testes, and adrenals had not been seen inthe preliminary studies or at levels higher than the MOAELof the individual compounds. Although the increase in therelative weight of the adrenals, testes, and lungs may berelated to the growth retardation (Feron et cd., 1973; Oishiet ai, 1979), the increased liver and kidney weights areconsidered of toxicological significance. With respect to thekidney weight, this view is strongly supported by the dose-related increase in relative kidney weight also seen at thelower dose levels.

The decrease in plasma cholesterol and triglyceride levels

in the MOAEL group cannot be explained by the findingswith the individual compounds because these parameters hadnot been included in all of the preliminary studies and wereunaffected when they had been measured. Most likely theeffect on cholesterol and triglycerides was caused by theperoxisome proliferator DEHP which has been shown todecrease plasma cholesterol and triglyceride levels in rats(Dirven et ai, 1990). In the present study both DEHP andaspirin induced acyl-CoA oxidase activity (the first enzymeof the peroxisomal /3-oxidation of fatty acids), and a de-creased lipid metabolism seems therefore obvious. Again,we could not verify this assumption because in the studieswith DEHP and aspirin alone, the glycerides and cholesterolplasma contents were not measured. Factorial analysis ofthe satellite study indeed showed a decreased cholesterolplasma level due to the presence of aspirin and DEHP, al-though it must be mentioned that rats treated with methylene



chloride or spermine also showed slightly decreased choles-terol levels (cf. Table 8).

The interpretation of the results in terms of additive ornonadditive effects is hampered, because the overall toxicityof the individual chemicals cannot be ascribed to one specificmechanism of action. For aspirin, loperamide, and spermineshowing rather aspecific effects in the present study, theinterpretation of the data heavily depends on the toxicologi-cal end points measured. For rather characteristic effectssuch as peroxisome proliferation by DEHP and aspirin, theeffect of formaldehyde on the nasal epithelium, and CO-Hbformation by dichloromethane, for all of which we assumeddissimilar modes of action, we predicted absence of interac-tions and thus response addition. Indeed, in the combinationthese effects were expressed unaltered, reflecting the concept"independent joint action" (Mumtaz et al, 1994).

A thorough analysis to determine whether the adverseeffects of the combination of the nine compounds at theMOAEL were less or more pronounced than the effects ofthe individual compounds was not carried out on the basisof the results of the main part of the study. The reason isthat the results of this part had to be compared with theresults of the previous tests on the individual compounds(cf. Table 1) which unavoidably were obtained under slightlydifferent experimental conditions. Moreover, several of theaffected parameters had not been included in each of thestudies with the individual compounds, which renders aquantitative comparison unreliable. However, in generalterms the main part of the study allows the far from spectacu-lar conclusion that simultaneous exposure to chemicals ineffective (toxic) doses resulted in effects qualitatively andquantitatively similar or dissimilar to those of the individualcompounds. A comparable conclusion has been drawn fromour previous studies in rats with combinations of chemicalswith different target organs (Jonker et al., 1990) or with thesame target organ (kidneys), but with different modes ofaction (Jonker et al., 1993).

A more accurate analysis of the interactive effects couldbe carried out, using the results from the satellite part of thestudy in which a fractional factorial design (l/32nd fractionof a complete design) was applied. Undoubtedly, the con-founding pattern would have been less complex and interpre-tation of the data would have been much easier when a 1/16th fraction or a l/8th fraction had been applied. However,one must realize that the number of test groups in such caseswill increase to unmanageable numbers, viz. 32 and 64,respectively. For the same reason we restricted the analysisto one dose level, the MOAEL of the individual chemicals,although a three-level factorial analysis would have pro-duced relevant information about dose-response curves ashas been shown previously (Cassee et al., 1996a; Narotskyet al., 1995). This also implies that extrapolation of casesof nonadditivity to lower dose regions must be done withgreat care. If the risk assessor attempts to model the complete

dose-response surface of the mixture, another approachmight be followed in which the interest is focused on depar-ture from additivity of the mixture as a whole, rather thanon the identification of specific interactions. This approachsuggested by Berenbaum (1985) utilizes single-chemicaldose-response information in addition to the observed re-sponses induced by the particular combinations of interest.Clearly, the advantage offered by such an approach becomesapparent upon comparing the number of experimental groupsemployed to that required by the use of factorial designs(Gennings, 1994). However, the fractionated factorial ap-proach allows more accurate hypotheses as to critical interac-tions and identifies compounds in the mixture of particulartoxicological importance. A drawback of fractionated facto-rial designs is the possible occurrence of inferences thatresult from a high level of aliasing and prior knowledgeinput needed to be able to interpret the results. In the presentstudy the treatment combinations were selected in a waythat allowed the interpretation of the results using the deci-sion process chosen. Thus, to some extent prior knowledgeis needed to correctly design these types of studies.

The term nonadditivity was used to describe a statisticalinteraction between two compounds. The term is not meantto designate an antagonistic or synergistic effect, becausewithout knowledge about the dose-response curves of theindividual compounds pure antagonism or synergism cannotbe established. It is almost impossible to rule out the possibil-ity that observed interactions are not simple additive effectsbecause full response addition cannot be achieved at doselevels that already have induced maximum or close to maxi-mum responses (binomial or logarithmic dose—responsecurves). For these reasons the interactions for the parametersASAT and PalmCoa as exemplified in Fig. 1 must be com-pared with the dose-response curves established in the pre-liminary studies. Nevertheless, the fact that statistical inter-actions were found indicates that the combined effect of twocompounds was not a simple summation of effects of theindividual compounds. This may be regarded as a warningsignal; a careful analysis and maybe further studies of theinteractions found are indicated, especially in those caseswhere the combined effect is more pronounced than expectedon the basis of full response addition.

Exposure at the NOAEL

Only a few adverse effects were encountered in theNOAEL group: hyperplasia and metaplasia of the nasal epi-thelium, hepatocellular hypertrophy, decreased plasma tri-glyceride concentrations, altered ALP enzyme activities, andincreased (relative) kidney weight. Obviously, combined ex-posure to compounds at their individual NOAELs can resultin a significant effect of the combination which is in linewith previous experimental findings (Jonker et al., 1990;National Toxicology Program, 1993; Yang, 1994b).

The effect on the nasal epithelium can be attributed to


28 GROTEN ET AL.

formaldehyde. However, a series of studies has shown that1 ppm formaldehyde is a NOAEL (Appelman et al, 1988;Woutersen et al, 1987; Wilmer et al, 1987). The findingthat 1 ppm formaldehyde in the mixture is not a NOAELcould not be explained. Studies to unravel this remarkableobservation are in progress.

Except for a slight increase in the relative kidney weight,it was shown that combined exposure below the NOAELproduced no toxic effects. The effect on the kidney weightwas also observed in the 1/3NOAEL group, showing thesensitivity of this rather unspecific toxicological parameter.The satellite part of the study revealed that aspirin, BHA,or SnCl2 may increase the kidney weight (Table 8). In thecase of aspirin and stannous chloride this might be due to thedecreased body weights. Furthermore, there were no cases ofpotentiating interactions. Therefore, the effect of the mixtureof all nine compounds at the MOAEL could be well ex-plained by summation of the effects of BHA, aspirin, andSnCl2 (cf. Table 8). A similar joint (additive) action of thesecompounds might also explain the increase in relative kidneyweight at the NOAEL. Comparable effects on kidney weightwere also found in the mixture studies of Jonker et al. (1990).A sound explanation for this intriguing and consistent findingis lacking.

In spite of a few observed changes, most end points werenot affected at the NOAEL. Similar conclusions were re-ported by Jonker et al (1990), Heindel et al. (1994), andSeed et al. (1995). Therefore, the most important practicallesson from these studies is that in general combined expo-sure to the single chemicals does not constitute an evidentlyincreased hazard, provided the exposure level of each chemi-cal is similar to or lower than its own NOAEL.

ACKNOWLEDGMENT

This work was supported by a grant from the Ministry of Housing, SpatialPlanning, and Environment, The Hague.

REFERENCES

Appelman, L. M., Woutersen, R. A., Zwart, A., Falke, H. E., and Feron,V. J. (1988). One-year inhalation toxicity study of formaldehyde in malerats with damaged or undamaged nasal mucosa. J. Appl Toxicol. 8, 8 5 -90.

Berenbaum, M. C. (1985). The expected effect of a combination of agents:The general solution. / . Theor. Biol. 114, 413-431.

Bliss, C. I. (1939). The toxicity of poisons applied jointly. Ann. Appl. Biol.26, 585-615.

Box, G. E. P., Hunter, W. G., and Hunter, J. S. (1978). Statistics for Experi-menters. Wiley, New York.

Cassee, F. R., Groten, J. P., Schoen, E. D., and Feron, V. J. (1996a). Sensoryirritation to mixtures of formaldehyde, acrolein and acetaldehyde in rats.Arch. Toxicol. 70, 329-337.

Cassee, F. R., Groten, J. P., and Feron, V. J. (1996b). Changes in the nasalepithelium of rats exposed by inhalation to mixtures of formaldehyde,acetaldehyde and acrolein. Fundam. Appl ToxicoL 29, 208-218.

Chaturved, A. K. (1993). Toxicological evaluation of mixtures of ten widelyused pesticides. J. Appl. Toxicol. 13, 183-188.

Dirven, H. A. A. M., van den Broek, P. H. H., and Jongeneelen, F. J. (1990).Effect of di(ethyl hexyl)phthalate on enzyme activity levels in liver andserum of rats. Toxicology 65, 199-207.

Eriksson, L., Jonsson, J., Hellberg, S., Lindgren, F., Skagerberg, B., Sjo-strom, F., and Wold, S. (1991). A strategy for ranking environmentallyoccurring chemicals. Part HI. Multivariate quantitative structure-activityrelationships for halogenated aliphatics. Environ. Toxicol. Chem. 9,1339-1351.

Feron, V. J., de Groot, A. P., Spanjers, M. T., and Til, H. P. (1973). Anevaluation of the criterion "organ weight" under conditions of growthretardation. Food Chem. Toxicol. 11, 85-94.

Feron, V. J., Woutersen, R. A., Arts, J. H. E., Cassee, F. R., de Vrijer, F.,and van Bladeren, P. J. (1995a). Safety evaluation of the mixture ofchemicals at a specific workplace: Theoretical considerations and a sug-gested two-step procedure. Toxicol. Lett. 76, 47-55.

Feron, V. J., Groten, J. P., Jonker, D., Cassee, F. R., and van Bladeren, P. J.(1995b). Toxicology of chemical mixtures: Challenges for today and thefuture. Toxicology 105, 415-427.

Gennings, C. (1994). An efficient experimental design for detecting depar-ture from additivity in mixtures of many chemicals. In Chemical Mixturesand Quantitative Risk Assessment. Abstracts of the Second Annual HERLSymposium, November 7-10, 1994, Raleigh, NC.

Groten, J. P., Sinkeldam, E. J., Muys, T., Luten, J. B., and van Bladeren,P. J. (1991). Interaction of dietary Ca, P, Mg, Mn, Cu, Fe, Zn, and Sewith the accumulation and oral toxicity of cadmium in rats. Food Chem.Toxicol. 29, 249-258.

Heindel, J. J., Chapin, R. E., Gulati, D. K., George, J. D., Price, C. J., Marr,M. C , Myers, C. B., Barnes, L. H., Fail, P. A., Grizzle, T. B., Schwetz,B. A., and Yang, R. S. H. (1994). Assessment of the reproductive anddevelopmental toxicity of pesticide/fertilizer mixtures based on confirmedpesticide contamination in California and Iowa groundwater. Fundam.Appl. Toxicol. 22, 605-621.

Hochberg Y., and Tamhane, A. C. (1987). Multiple Comparison Proce-dures. Wiley, New York.

Ito, N., Hasegawa, K., Imaida, K., Kurat, Y., Hagiwara, A., and Shirai, T.(1995). Effect of ingestion of 20 pesticides in combination at acceptabledaily intake levels on rat liver carcinogenesis. Food Chem. Toxicol. 33,159-163.

Jonker, D., Woutersen, R. A., van Bladeren, P. J., Til, H. P., and Feron, V. J.(1990). 4-Week oral toxicity study of a combination of eight chemicals inrats: Comparison with the toxicity of the individual compounds. FoodChem. Toxicol. 28, 623-631.

Jonker, D., Woutersen, R. A., van Bladeren, P. J., Til, H. P., and Feron,V. J. (1993). Subacute (4-wk) oral toxicity of a combination of fournephrotoxins in rats: Comparison with the toxicity of the individual com-pounds. Food Chem. Toxicol. 31, 125-136.

Krishnan, J., and Brodeur, J. (1991). Toxicological consequences of com-bined exposure to environmental pollutants. Arch. Compl. Environ, Stud-ies 3.

McCullagh, P., and Nelder, J. A. (Eds.) (1989). Generalized Linear Models,2nd ed. Chapman and Hall, London.

Mumtaz, M. M., Sipes, G., and Yang, R. S. H. (1993). Risk assessment ofchemical mixtures: Biologic and toxicologic issues. Fundam. Appl. Tox-icol. 21, 258-269.

Mumtaz, M. M., DeRosa, C. T., and Durkin, P. R. (1994). Approaches andchallenges in risk assessments of chemical mixtures. In Toxicology ofChemical Mixtures (R. S. H. Yang, Ed.), pp. 565-598. Academic Press,San Diego.

Narotsky, M. G., Weller, E. A., Chinchilli, V. M., and Kevlock, R. J. (1995).Nonadditive developmental toxicity in mixtures of trichloroethylene,



(di(2-ethylhexyl)phthalate, and heptachlor in a 5 X 5 X 5 design. Fundam.Appl. Toxicol. 27, 203-216.

Nesnow, S. (1994). Mechanistic linkage between DNA adducts, mutationsin oncogenes, and tumorigenesis of carcinogenic polycyclic aromatichydrocarbons in strain A/J mice. In Chemical Mixtures and QuantitativeRisk Assessment. Abstracts of the Second Annual HERL Symposium,November 7-10, 1994, Raleigh, NC.

National Toxicology Program (1993). Toxicity studies of pesticides/fertil-izer mixtures. Administered in drinking water to F344/N rats and B6C3F1mice. NTP Technical Report Series No. 36, NTP Publication 93-3385.

Oishi, S., Oishi, H., and Hiraga, K. (1979). The effect of food restrictionfor 4 weeks on common toxicity parameters in male rats. Toxicol. Appl.Pharmacol. 47, 15-22.

Plackett, R. L., and Burman, J. P. (1946). The design of optimum multifac-tonal experiments. Biometrika 33, 305-339.

Plackett, R. L., and Hewlett, P. S. (1952). Quantal responses to mixturesof poisons. J. Roy. Stat. Soc. B 14, 141-163.

Reubsaet, F. A. G., Veerkamp, J. H., Bukkens, S. G. F., Trijbels, J. M. F.,and Monnens, L. A. H. (1988). Acyl-CoA oxidase activity and peroxi-somal fatty acid oxidation in rat tissues. Biochim. Biophys. Acta 958,434_442.

Seed, J., Brown, R. P., Olin, S. S., and Foran, J. A. (1995). Chemical mix-tures: Current risk assessment methodologies and future directions. Reg.Toxicol. Pharmacol. 22, 76-94.

Svensgaard, D. J., and Hertzberg, R. C. (1994). Statistical methods for thetoxicological evaluation of the additivity assumption as used in the envi-ronmental protection agency chemical mixture risk assessment guideline.

In Toxicology of Chemicals Mixtures (R. S. H. Yang, Ed.), pp. 599-640.Academic Press, San Diego.

U.S. Environmental Protection Agency (1986). Guidelines for the healthrisk assessment of chemical mixtures. Fed. Reg. 51, 34014-34025.

Wilmer, J. W. G. M., Wouterscn, R. A., Appelman, L. M., Leeman, W. R.,and Feron, V. J. (1987). Subacute (4-week) inhalation toxicity study offormaldehyde in male rats: 8-Hour intermittent versus 8-hour continuousexposures. J. Appl. Toxicol. 7, 15-16.

WHO (1987). Air Quality Guidelines for Europe. WHO regional publica-tions, European series No. 23.

Woutersen, R. A., Appelman, L. M., Wilmer, J. W. G. M., Falke, H. E., andFeron, V. J. (1987). Subchronic (13-week) inhalation toxicity study offormaldehyde in rats. J. Appl. Toxicol. 7, 43-49.

Yang, R. S. H., and Rauckman, E. J. (1987). Toxicological studies of chemi-cal mixtures of environmental concern at the National Toxicology Pro-gram: Health effects of groundwater contaminants. Toxicology 47, 1 5 -34.

Yang, R. S. H., Hong, H. L., and Boorman, G. A. (1989). Toxicology ofchemical mixtures: Experimental approaches, underlying concepts, and

some results. Toxicol. Lett. 49, 183-197.

Yang, R. S. H. (Ed.) (1994a). Toxicology of chemical mixtures derivedfrom hazardous waste sites or application of pesticides and fertilizers. InToxicology of Chemical Mixtures, pp. 99-117. Academic Press, SanDiego.

Yang, R. S. H. (Ed.) (1994b). Introduction to the toxicology of chemicalmixtures. In Toxicology of Chemical Mixtures, pp. 1-10. Academic Press,San Diego.


the roof gardens - detrafford estates group

Documents