The Health Effects Institute

HEI Communication 4
Background for the Complex Mixtures Program

INTRODUCTION

It is HEI's mission ultimately to determine and provide a quantitative measure of the human health effects of exposure to vehicle emissions at concentrations relevant to actual ambient exposure. Our understanding of the toxicity of such environmental pollutants, even when they are studied in isolation, is incomplete. This problem is compounded by the fact that environmental pollutants rarely, if ever, occur as single substances, but are part of larger mixtures, which often contain thousands of compounds; these are known generally as complex mixtures. The environment, especially in heavily industrialized regions, contains a wide variety of such mixtures. Typical examples of mixtures commonly found in ambient air include cigarette smoke, toxic waste, emissions from industrial sources, and evaporative emissions or products from the combustion of fossil fuels.

Individual components of complex mixtures may be associated with toxic responses, ranging from short-term toxicity to effects with longer-term consequences including mutagenicity and carcinogenicity. However, it is not enough to simply understand the toxicity of individual components in the mixture. The toxicity of one pollutant may be affected by the presence of other pollutants (Elashoff et al. 1987; Simmons and Berman 1989). This may be attributed to a number of factors, including the effects of one substance on the pharmacokinetics or metabolic transformation of a second substance.

Within recent years, many research efforts have focused on identifying and characterizing specific complex mixtures. As analytic approaches have become more sensitive and specific, it has been possible to better identify and quantify components of individual mixtures. Less has been done to develop a systematic and more general approach that could be used to identify in a few steps the toxic components of a number of mixtures and to better understand their toxicologic interactions and effects.

The need for such a systematic approach was the basis of HEI's decision to develop RFA 90-6 "Theoretical Approaches to the Health Effects of Complex Mixtures". This RFA was intended to support several theoretical studies that might, in turn, lead to the development of better analytic approaches.

 

PURPOSE AND GOALS OF STUDIES FUNDED UNDER RFA 90-6

Scientific Background   To assess the toxic effects of a complex mixture, it is necessary to characterize toxicologically active components of the mixture and to understand the interaction of these components with other mixture components.

In general, two broad analytic approaches have been used. One approach is based on partial resolution of a naturally occurring mixture into mixtures of fewer constituents (a break-down approach). The other is based on the artificial construction of mixtures of compounds (a build-up approach). Some aspects of each approach are described in a report of the National Research Council (1988). In the first approach (Lewtas 1988, 1990; Schuetzle and Daisey 1990), various analytical techniques are used to separate complex mixtures into individual components or fractions of fewer components. During separation, the components or fractions are assayed individually for toxicity, generally using short-term tests such as those for mutagenicity. Toxic fractions are then further fractionated. This process is usually referred to as a bioassay-directed fractionation approach. In the second approach (Chapin et al. 1989; Germolec et al. 1989; Yang et al. 1989), pure compounds are combined to form defined mixtures of as few as 2 and as many of 25 to 30 components. These mixtures are then tested for toxicity, generally using short-term tests as biological end points.

Each approach has limitations. One problem of the bioassay-driven approach is that materials may be lost or chemically modified during analysis. This is of special concern if the toxic material is present in very small amounts or is not separated easily from other components. A second problem involves the bioassay-driven strategy often used in analytic schemes. Bioassay-driven separations are specifically directed toward the toxicologically active fractions and usually do not focus on those that are toxicologically inactive. However, the inactive fractions may contain substances that modify the activity of compounds in the active fractions; if the compounds in the inactive fractions remain undetected, their toxicological role in the mixture remains unknown. Aspects of this problem have sometimes been addressed by "spiking" isolated fractions with a known toxic chemical such as benzo[a]pyrene (Kaden et al. 1979).

The build-up approach offers the possibility of relating interactions to a mixture of known chemical composition. However, this approach uses mixtures that are artificially constructed; thus, the specific interactions may not represent interactions found in naturally occurring mixtures, in part because the proportions of individual constituents in the constructed mixtures are unlikely to represent proportions in a "naturally occurring" complex mixture.

To determine chemical interactions in a complex mixture is a major challenge. One significant limitation is the large number of sample combinations that would need to be tested. In theory, if the quantitative and qualitative composition of a mixture were known, all components could be combined and all combinations tested for synergism. However, because of the amount of testing required, such an approach is not feasible. For example, to screen only one dose made up of two components of a mixture of 100 individual components, approximately 5,000 different combinations of binary mixtures would be required. This "simple" approach only would identify interactions at specified proportions; and because it would fail to identify interactions at other concentrations, it would not determine which interactions are local (occurring only at certain proportions) and which are global (occurring at all proportions). Furthermore, this "simple" approach would neither mimic all of the in vivo processes that ultimately affect toxicity nor examine all possible toxicologic effects.

Specific Goals of Funded Studies   In considering how to advance the understanding of component interactions in complex mixtures, the HEI Research Committee felt that a major stumbling block was the practical impossibility of testing all component combinations to assess interaction. The Research Committee suggested that a fruitful approach might be to simplify the number of components to be tested by dividing the mixture into submixtures on the basis of functional groups that might be related to particular mechanisms of toxicity. Then it would be possible to do a preliminary assessment of interactions, without necessarily identifying the toxic components, by looking at interactions between the submixtures. An ideal analytic approach to complex mixtures would be adaptable to a number of different mixtures, and should be designed to separate mixtures into groups according to a common characteristic in a minimum of analytical steps. Thus separated, mixtures could be tested to determine whether or not the common characteristic is a major determinant of toxicity. If the separation process yielded fewer fractions (rather than many), determination of interactions would be more feasible. It was with this in mind that proposals for theoretical approaches to the analysis of complex mixtures were solicited. These proposals were asked to respond to one or both of the following objectives:

It was intended that successfully developed proposals would produce publishable theoretical papers. It was not required that papers describe an approach to a specific complex mixture; however, investigators were asked to develop a theoretical approach that could be tested experimentally using a complex mixture of known composition made up of approximately 100 different chemicals.

Thirteen applications were received. These were evaluated by an ad hoc panel of investigators with expertise in chemical analysis of complex mixtures, bioassay techniques, and statistics; and by the HEI Research Committee. Four applications were funded; three involved the development of analytic approaches to better identify toxic components, and the fourth involved the development of methods to determine interactive effects. Although investigators were not required to include biological methods for testing, one study did address the development of such methods. Total cost for the four theoretical studies was $166,627. Periods of work ranged from three months to one year.

In addition to the funded studies, HEI held a workshop on synergy and independent action to which a number of scientists and mathematicians with expertise in statistical analysis were invited. The workshop was intended to stimulate further thinking in this area. Although the workshop identified some potentially fruitful areas, participants concluded that there was no obvious "next step" to be taken in this field and that further thought was needed. Participants agreed on the value of integrating statistical, biological, and chemical approaches in research efforts.

The Investigators' Reports were received at HEI in the winter and spring of 1993 and reviewed by outside technical reviewers and by the Review Committee at its October 1993 meeting. During review of the Investigators' Reports, the Review Committee and the investigators had the opportunity to exchange comments and to clarify issues in the reports. Because the three analytical studies have a similar focus, and because the analytical studies and the statistical study are complementary, the Review Committee recommended publishing the set of studies together in an HEI Communication format.

There are four reports included in the publication. Three address analytical approaches to identifying toxic components:

and one describes statistical approaches to analysis of interaction:

Home | About HEI | What's New? | Newsletter | HEI International | Publications
Research | Funding | Meetings | Contact HEI | Links | Search | Site Map
Copyright © 2010 Health Effects Institute. Please send comments to webmaster@healtheffects.org