Approaches for Experimental Studies of Mixtures
Many approaches can be used for experimental studies of mixtures, depending on the desired goals (31). Problem definition and the formulation of specific questions are important steps that need to be addressed before selecting a specific approach. If toxic effects and risks posed by a mixture are to be determined, then the toxicologic evaluation of the mixture is carried out by testing the whole mixture through the tier or screening approach. If a causative agent(s) is to be determined to mitigate exposures to the active ingredient of a mixture or to identify the source of pollution, bioassay-directed fractionation is carried out. Finally, if the tools for predictive values are being developed, then toxicologic evaluation of individual components, in various combinations, is carried out to gain toxicologic knowledge about the mechanism and mode of action as well as mechanisms of interactions between the components and within the mixture.
Ideally, all the components of the mixture need to be identified, and their toxicity experimentally determined or obtained from the literature. Several testing protocols can be used to obtain appropriate information, but the actual experimental design depends on the number of chemical components of the mixture and if it is desirable to assess possible existing interactions between chemicals in a mixture. The mixture should be tested both at high effective concentrations and at low realistic concentrations.
Once the data are generated, it is necessary to analyze and interpret them. Many descriptive terms and mathematical, graphical, and statistical models have been used to evaluate the joint toxicity of mixtures. Interactive effects of compounds in mixtures with more than three compounds can be best ascertained with the help of statistical designs such as (fractionated) factorial designs or ray design.
In view of the importance of joint toxicity assessment of human exposure to multiple toxicants, appropriate experimental designs and methods of analysis must be used to support conclusions of additivity, synergy, and antagonism. Whole Mixtures
Whole mixture studies involve exposing test systems to the intact mixture and conducting exposure-response studies to evaluate the nature and magnitude of the hazard associated with exposure. The design of these studies is usually chosen to reflect the net effect of all compounds in the mixture. This approach is applied to study real-life mixtures, such as tobacco smoke, jet fuels, or specially designed mixtures (88).
JP-5 is U.S. Navy's primary jet fuel. It is made up of a collection of hydrocarbons such as paraffins, monocycloparaffins, bicycloparaffins, olefins, alkylbenzenes, and others. The whole mixture of jet fuel JP-5 was administered to groups of 37-50 female C57BL/6 mice at 0, 150, or 750 mg/m by inhalation continuously for 90 d (89). The endpoints evaluated were clinical signs, hematology, blood chemistry, body weight, and histopathological examination of major tissues. No effect on body weight gain was noted. The only remarkable finding in mice was hepatocellular fatty changes and vacuolization at 150 and 750 mg/m . This study was used to derive an MRL value of 3 mg/kg/d (25).
The whole mixture approach is recommended for situations where the mixtures are not well characterized and for mixtures with reasonably stable concentrations. In many cases, a whole mixture approach is advised because it can provide a real-life situation exposure scenario. However, caution should be exercised since in many cases these mixtures do vary in composition from time to time and from one exposure to the next. Without knowledge of the individual effects of each of the components to the response given by the whole mixture, no unique single estimate for risk to exposed populations can be estimated. Formulated Mixtures
A systemic toxicity testing of n components in a chemical mixture would involve 2n-1 experiments to address all possible combinations at one dose level for each component. To include several doses, one must use a more focused design such as a full-factorial design, which involves kn experiments when a range of k doses is applied for each of the n components. In most cases, such mechanistically oriented experiments involve separating the mixture into several components that are studied together in formulated mixtures.
A classical design in the statistical literature for studying toxicologic interaction is a factorial design where each of the chemicals in the mixture is studied at all levels of the other chemicals. Generally, the levels of each factor are evenly spaced so as to cover systematically the dose region of interest. The logic of a factorial design is to support efficiently the estimation of a response surface that includes interaction model parameters (90).
A 5 x 5 x 5 factorial design was utilized to identify nonadditive effects on developmental toxicity in Fischer 344 rats caused by combinations of trichloroethylene (TCE), di(2-ethylhexyl)phthalate (DEHP), and heptachlor (HEPT) (78). The 5 x 5 x 5 full-factorial design was selected to detect binary and tertiary interactions among the chemicals in the mixture. The chemicals were administered by gavage to Fischer 344 rats on gestation days 6-15. Dose levels were 0, 10.1, 32, 101, and 320 mg/kg/d for TCE; 0, 24.7, 78, 247, and 780 mg/kg/d for DEHP; and 0, 0.25, 0.8, 2.5, and 8 mg/kg/d for HEPT. The dams were allowed to deliver, and their pups were weighed and examined postnatally. Of the nine endpoints that were analyzed statistically, six had significant binary interactions. Both synergistic and antagonistic interactions were detected among the three components. Maternal death showed no main effects, but DEHP and HEPT were synergistic. For maternal weight gain on gestation days 6-8, main effects for all three agents were observed, as well as TCE-HEPT synergism and DEHP-HEPT antagonism. Maternal weight gain on gestational days 6-20, adjusted for litter weight, showed main effects for TCE and HEPT, but no interactions. Main effects for all three chemicals were evident for full-litter resorptions and prenatal loss. For full-litter loss, the TCE-HEPT and DEHP-HEPT interactions were antagonistic. Postnatal loss showed DEHP and HEPT main effects but no interactions. Analysis of pup weights on day 1 revealed TCE and DEHP main effects and DEHP-HEPT antagonism; on day 6, DEHP and HEPT main effects, DEHP-HEPT antagonism, and TCE-DEHP synergism were evident. Microphthalmia and anophthalmia incidences revealed TCE and DEHP main effects but no interactions. This extensive examination of a full-factorial design elucidates the complexities of studying and interpreting mixture toxicity. Although the Narotsky study illustrates the utility of full-factorial design to investigate binary and tertiary interaction, the study also used the large number of 2000 pregnant rats experimentally.
The feasibility of carrying a full-factorial design with many chemicals rapidly decreases (90). To overcome the usually costly full-factorial designs, statistically less-than-full designs are used. These designs are referred to as fractionated factorial designs. A fractionated two-level factorial study was designed for a combination of nine chemicals in a subacute rat study (91). In the study, an efficient fractionated design for 16 different groups was used as a subset of the full design, which would have required 29 (512) experiments. The combination experiments (satellite part) were composed of a fraction of 1/32 subsets (of the full 512 experiments). The study was intended to find out whether simultaneous administration of nine chemicals at a concentration equal to the "no-observed-adverse-effect level" for each of the chemicals would result in a NOAEL for the combination. A 4-wk oral/inhalation study in male Wistar rats was performed in which the toxicity (clinical chemistry, hematology, biochemistry, and pathology) of combinations of nine chemicals was examined. The study consisted of 20 groups, 4 groups in the main part (n = 8) and 16 groups in the satellite part (n = 5). In the main study, the rats were simultaneously exposed to various combinations of all nine chemicals (dichloromethane, formaldehyde, aspirin, di-(2-ethylhexyl)phthalate, cadmium chloride, stannous chloride, butyl hydroxyanisol, loperamide, and spermine) at concentrations equal to
"minimum-observed adverse-effect level" (MOAEL), NOAEL, or a NOAEL. In the satellite study, the rats were simultaneously exposed to combinations of maximally nine factors ( = 9 chemicals) in 16 experimental groups (1/32 fraction of a complete study). In the main part, many effects on hematology and clinical chemistry were observed at the MOAEL. In addition, rats of the MOAEL group showed hyperplasia of the transitional epithelium and/or squamous metaplasia of the respiratory epithelium in the nose. Only very few adverse effects were observed in the NOAEL group. For most of the endpoints chosen, the factorial design revealed main effects of the individual compounds and interactions (cases of nonadditivity) between the compounds.
Other fractionated designs include ray designs in which mixtures of chemicals under study are evaluated along rays of fixed ratios. In a ray design, for example, for a mixture of three chemicals with fixed ratios, represented by chemicals A, B, and C, a 1:0:0 ratio represents a ray of chemical A alone, while a 1:1:1 ratio represents a ray of equal levels of the three chemicals. A ray design for a small number of chemicals and many mixture rays can support the estimation of a response surface. However, the advantage of a ray design is that it can also be used with a mixture of many chemicals and a few mixture rays (90). The ray design was employed to estimate a response surface of developmental toxicity in rats using data from an earlier study in which a full-factorial design was used (78, 90). Similar to the ray designs, other fractional procedures such as the central composite and Box-Behnken designs use specific regions of the dose-response surface to optimize combinational experimental procedures. Mathematical/Statistical Procedures
Once the data are generated, they need to be analyzed and interpreted. Many descriptive terms and mathematical, graphical, and statistical models have been used to evaluate the joint toxicity of mixtures. In general terms, the purpose of these models is to help interpret data for the entire range of the dose-response surface based on a mathematical/statistical description of the interaction criteria. Thus validated models can also be used to extrapolate from one region to other regions of the dose-response space. The models can also be used for the development of efficient experimental design by considering the cycle of model-experiment procedures to optimize the use of resources and time. The following discussions explain three mathematical and statistical procedures frequently used by scientists interested in combinational toxicology.
Isobolographic Methods An isobole is a contour line that represents equal effects of two agents or more in a mixture. Thus, when the joint effects of various dosages of two agents are plotted, each point of equal response (e.g., ED50, percent lethality, ... , etc.) corresponding to varying doses of both chemicals form the isobole. Isoboles can be used to characterize the nature of the toxicologic interaction. This is done by comparing the isoboles to the line of additivity as shown in Fig. 6.9. The graphical representation of the interactions criteria can also be depicted mathematically as follows: For additivity:
where Ae, B and Ne are the doses of chemical components A, B, and Nthat produce the measured response of each chemical as if it were the only component of the mixture. Ac, Bc, and Nc are the doses of each chemical that produce a similar response when the chemicals are all combined together.
The preceding equations represent the criteria for assessing interaction modes among the different agents in a mixture (92). However, for these equations to be used, individual chemicals in the mixture should have a nonzero response at the given doses (i.e., AQ 0, B * 0, ..., NQ =£ 0). The major disadvantage of the isobolographic methods is the requirement for a large number of experiments to produce the individual isoboles. For example, one can start with doses of chemicals A and B for a binary mixture, if the response is not equal to the one chosen for the isobole, then doses of A and/or B have to change up or down until the fixed response is obtained. This highly iterative procedure is very resources extensive. With a conventional experimental approach, the isobolographic method is tedious and requires extremely large data sets. For instance, 2000-3000 animals were used to generate an isobole to study the interaction between ethanol and chloral hydrate effect on the righting reflex of mice (93). Furthermore, the isobolographic methods can only be applied to chemicals that share similar mechanisms and induce the same endpoint of toxicity.
For chemicals that do not share similar mechanisms, a more general mathematical procedure than isobolographic methods is employed. One such procedure is the median-effect principle (MEP). This method is based on the assumption that dose-response relationships of many physical, chemical, and biological processes, specifically related to ligand-enzyme receptor-site interactions, can be described by a general formula:
Was this article helpful?