Chemical mixture experiments

Chemical mixture experiments have distinct characteristics that can preclude the use of traditional statistical analysis techniques. Mixture experimentation often poses unique exploratory questions that can be answered efficiently and economically with non-traditional statistical techniques. General statistical guidelines stress the importance of design, preliminary studies, action levels of variables, graphics, and appropriate statistical testing. Fractional ctorial and Simplex designs are just two of many statistical tools that are useful for analyses of mixture experiments. 

Traditional confirmatory statistical methods are not always the best option for analysis of chemical mixture experiments. A mixture experiment is defined as one where the response is dependent upon the relative proportion of ingredients comprising the mixture rather than the absolute amount of different ingredients that influence a response (/). Mixture analysis data can t be analyzed using conventional statistical techniques. Traditional methods assume independent results for each ingredient, an assumption which is violated by the use of relative mixtures. 

There are also a number of powerful NMR experiments that yield detailed information about the structure of pure organic molecules. However, as with other regions of the spectrum (see Figure 14.4), chemical mixtures produce spectra that are combinations of all of the absorption features of all of the components in the mixture. The H NMR spectrum of a mixture of toluene, hexanoic acid, and octanal is shown in Figure 14.6. This spectrum shows the unique absorptions of each of these functional groups and also illustrates that mixtures of compounds give spectra containing the absorption bands of all the components. If an impure sample had such a spectrum, it would not be known... 

They did not start from reagent grade chemicals. The experiments involved the in-vitro bleaching of native rhodopsin in solution. They proposed that they caused the disassociation of rhodopsin into a protein and retinaldehyde in accordance with the conventional wisdom of the time. This wisdom conflicted somewhat with the earlier ideas of Kuhne but became the basis of the current conventional wisdom. They claimed to have removed the native retinene, and then recombined colorless rhodopsin-protein, free of all native retinene, with synthetic retinene, in high concentration. This was accomplished by letting the mixture set in the dark. The pH of these solutions was not described as a function of time. They demonstrated that the material after setting in the dark for 60 minutes exhibited a rise in extinction coefficient of 2 1 over a similar sample without the added retinene,. Both samples exhibited a peak absorption near 500 nm following the experiments. It is not clear why the reference sample showed any absorption at 500 nm if it was truly free of all native retinene,. 

Symyx Library Studio software provides an environment for the specification of one, several, or hundreds of experiments. Chemicals, mixtures, and process parameters are defined and then combined in a recipe-style user interface. 

Figure 3.2 shows the outcome of a mixture experiment with toxicants that disrupt algal reproduction by diverse mechanisms (Faust et al. 2003). The mixture was carefully composed to contain chemicals with widely differing modes of action. In this case, IA was the concept that produced the prediction that best reflected the observed effects of the mixture. CA led to an overestimation of the experimentally observed responses. 

Figure 3.4 Illustration of a sham mixture experiment with chemicals that all exhibit the same dose-response curve. At the low dose to the left (arrow, 4 X 10 3 M), the effect is hardly observable. A combination of ten agents, at this dose (total dose 4 X 10 2 M) produces a significant combination effect, in line with expectations following dose addition.

The term toxic unit (TU) plays an important role in mixture concentration-response analysis. It is defined as the actual concentration of a chemical in the mixture divided by its effect concentration (e.g., c/EC50 Sprague 1970). The toxic unit is equivalent to the hazard quotient (HQ), which is used for calculating the hazard index (HI Hertzberg and Teuschler 2002). The term hazard quotient is generally used more in the context of risk assessment (see Chapter 5 on risk assessment), and the term toxic unit is used more in the context of concentration-response analysis, and therefore the latter term is used here. Toxic units are important for 2 reasons. First, toxic units are the core of the concept of concentration addition concentration addition occurs if the toxic units of the chemicals in a mixture that causes 50% effect sum up to 1. Second, toxic units can help to determine which concentrations of the chemicals to test when a mixture experiment needs to be designed. 

In each of these disciplines several more specific goals can be identified (see text box). Because of this, a diverse set of approaches has been developed for analyzing and assessing the toxicities of chemical mixtures, which can be grouped into 3 major classes 1) mixture experiments in which the toxicity of the mixture is characterized without making any effort to connect it to the toxicities of the components 2) whole mixture approaches, that is, inferring from mixture effects the toxicity contributions of the individual components and 3) component-based approaches, that is, inferring from the mixture components their joint toxicity. 

In this paragraph, we discuss aspects of mixture experiments that need attention while analyzing and assessing the data. These aspects may be endpoint, test organism, or chemical specific ... 

A corresponding normal distribution is available for multiresponse data, that is, for interdependent observations of two or more measurable quantities. Such data are common in experiments with chemical mixtures, mechanical structures, and electric circuits as well as in population surveys and econometric studies. Modeling with multiresponse data is treated in Chapter 7 and in the software of Appendix C. 

It is important to confirm the identity of pesticide residues convincingly. Some methods, such as TLC, paper chromatography, or p-values share the same physical property of partition in achieving separations of mixtures. They do not give independent evidence for the identity of a compound. Similarly, GLC retention times for a compound on different stationary phases are often highly correlated. Thus, the choice of confirmatory techniques should be carefully made. Although powerful methods such as GC/MS are being studied, there is a need for simpler operations—for instance, the formation of chemical derivatives. Experiments with aldrin and dieldrin have revealed a number of reactions which are convenient for the confirmation of residues of these compounds. 

Permeation-skin-gas chromatography (GC)/MS A silastic membrane was coated onto a fiber to be used as a permeation membrane. The MCF was immersed in the donor phase to partition the compounds into the membrane. At a given partition time, the MCF was transferred into a GC injector to evaporate the partitioned compounds for quantitative and qualitative analyses. This technique was developed and demonstrated to study the percutaneous permeation of a complex mixture consisting of 30 compounds. Each compound permeated into the membrane was identified and quantified with GC/MS. The standard deviation was less than 10% in 12 repeated permeation experiments. The partition coefficients and permeation rates in static and stirred donor solutions were obtained for each compound. The partition coefficients measured by this technique were well correlated (Pf — 0.93) with the reported octanol/water partition coefficients. This technique can be used to study the percutaneous permeation of chemical mixtures. No expensive radiolabeled chemicals were required. Each compound permeated into the membrane can be identified and quantified. The initial permeation rate and equilibrium time can be obtained for each compound, which could serve as characteristic parameters regarding the skin permeability of the compound. 

In another experiment (Figure 3), a collaborative effort between the EPA and the NTP, we examined the effects of pretreatment with the 25-chemical mixture of groundwater contaminants for 14 days on the hepatotoxicity of carbon tetrachloride in male Fischer 344 rats (39). 

---