Mixture extrapolation approaches assessments

In our opinion, the data are sufficiently clear to suggest that when it is not feasible to test the mixture in question, mixture extrapolation is the preferred option compared to no extrapolation. Indeed, all literature observations suggest that applying mixture extrapolation is to be preferred over not applying mixture extrapolation. Technical options for extrapolation are concentration addition, response addition, and the mixed-model approach, of which concentration addition is most often applied. Exceptions may apply in cases that are more specific. For example, when it is clear that 2 compounds precipitate (a situation of no exposure due to chemical interactions in the environment), one should acknowledge this prior to assessing mixture risks by mixture extrapolation approaches. When the data of a study allow, refined conclusions are possible. For example, when the study design is appropriate and the mathematical models are appropriate, researchers are able to discriminate between concentration addition and response addition, and (with sufficient experiment efforts) between these models and the mixed-model approach. 

Higher tier mixture extrapolation approaches such as this are used in practice, by calculating site-specific msPAF values, across a broad range of assessment questions. A practical example is provided by De Zwart (2005), who studied the impacts of pesticide use in The Netherlands. In this study, the specific mode of action of the pesticide was taken into account, as shown in Box 5.2. This analysis resulted in spatiotemporal indicators of relative toxic pressure across The Netherlands (Figure 5.3). 

For the derivation of the PNEC several approaches have been proposed. Generally these can be categorised into three distinct assessments a conservative, a distributional, and a mixture toxicity approach. In conservative approaches, usually the most (realistic) sensitive endpoint such as LC50 or the known no observed effect concentration (NOEC) is taken and divided by an uncertainty factor (10-100). The selected uncertainty factor value depends on the type of endpoint and the number of available data, and is applied to account for laboratory to field extrapolations, species differences in sensitivities, and similar uncertainties. In distributional approaches, a series of, or all available, literature data are taken and a selected cut-off value is applied to the distribution of these data. The cut-off value may be, e.g., the concentration value that will protect 95% of the species (tested). In general, again an uncertainty factor (usually of 10) is then applied to take into account species differences. In the mixture toxicity approach, a similar mode of action is assumed for the assessment of the combined (additive) effect of the mixture. All relevant mixture components are scaled relative to the most potent one. This results in relative potencies for each component. The total effect of the mixture is then evaluated by... 

Because there are various options to account for possible combined effects, including concentration addition alone, response addition alone, or concentration addition and response addition in concert, and motives from study quality, assessment endpoint, mechanistic features, and statistical characteristics and biases, there should be a way to logically choose amongst methods for mixture extrapolation. The logical and pragmatic way to choose amongst alternative approaches is to design and follow a tiered approach (see Chapter 1). 

Tiering is often applied in risk assessment in order to reduce expenditures in time, money, and labor when the assessment requires only simple and possibly conservative output. Table 5.3 provides a suggested tiered approach in mixture extrapolation and is further described in the bulleted list below. The tiering is based on the way that mixture mechanisms are addressed in the approach. It is assumed that issues such as matrix and media extrapolation have been addressed according to the methods described in the pertinent chapters. 

In single-species assessments, the interpretations of mixture assessments tend to be mostly absolute. Hence, risk assessors often focus on particular species and particular compound groups (e.g., risks of PCB mixtures for birds), allowing them to interpret and explain their experimental data to the best of their abilities. On the other hand, many risk assessors apply mixture extrapolation methods to address risks for communities. The applications of SSD-based methods for this evolved fast and now cover a wide set of approaches, ranging from ecological multiple-stress analyses to overall approaches such as life-cycle assessment. Especially in the latter set of approaches, the risk assessor can often allow the method to only yield relative... 

A tiered system for mixture extrapolation is proposed. The lowest tier is based on extrapolation using toxicological point-estimate information such as EC50 values. This translates into the use of toxic units, toxic equivalencies, and similar techniques. The use of the entire concentration-response relationships of the separate compounds is recommended for Tier-2, in conjunction with the use of either concentration or response addition as a modeling approach. In Tier-3, a mixed-model approach can be considered, to more specifically address considerations on toxic modes of action. In the latter case, the approach may be extended to allow incorporation of the responses of different ecological receptors (Tier-4). Research needs have been clearly identified in community-level mixture assessments. 

When there is no guidance to choose the best method, it is always possible to use the window of prediction. Especially for retrospective risk assessments that should result in a decision, 2 or more extrapolation approaches can be used for the same problem (for example, the 2 mixture models), and both results can be compared to the decision criterion. In case both predictions are (by far) lower or higher than the decision criterion, the decision would remain the same, irrespective of the model Only in cases where the window of prediction overlaps with the decision criterion would further work be necessary. 

In addition to the graphic approach for toxicity data and the verification of uncertainty factors, other areas are under study such as route-to-route conversion, high-dose to low-dose extrapolation, approaches to assess the health risk from less-than-lifetime exposures, and refinement of risk assessment approaches for chemical mixtures. All of these areas represent progress in the methods used for risk assessment of single chemicals and chemical mixtures. With the new risk assessment guidelines currently being developed, the U.S. EPA can move forward to better and more consistent health risk assessments. 

There are many extrapolation methods, of different complexities, and with different purposes and suitabilities for prospective and retrospective risk assessments. A compilation of the methods is insufficient to guide the choice of procedures to use when assessors need to conduct risk assessments. Therefore, a practical and pragmatic guide to extrapolations and their everyday use is provided in the last chapter. It defines a general stepwise approach to identifying the types of extrapolation (matrix and media, (Q)SARs, mixtures, etc.) that are most relevant for an assessment problem, and it defines an overall approach to the assignment of tiers. 

---