TEF concept

From Testiwiki
Jump to: navigation, search


How should the TEF concept be applied?


Validity and criteria of the TEF concept

During the 2005 WHO reevaluation of the 1998 WHO TEF values, both the general TEF concept and REP criteria were extensively discussed. The criteria for inclusion of a compound in the TEF concept at this meeting were similar to those used at the two earlier WHO expert meetings (Ahlborg et al., 1994; Van den Berg et al., 1998). These criteria are that for inclusion in the TEF concept a compound must

  • show a structural relationship to the PCDDs and PCDFs;
  • bind to the AhR;
  • elicit AhR-mediated biochemical and toxic responses;
  • be persistent and accumulate in the food chain.[1]

It was recognized that the vast amount of literature available in this field provides many examples of uncertainties associated with the determination of REPs. In addition, high variation can sometimes be found in REP values for the same congener and for similar endpoints in different species (e.g., rats vs. mice).

The 2005 WHO reevaluation of the TEF values made extensive use of the review and REP database of Haws et al. (2006) in which a set of criteria was developed to identify, include, or exclude REPs for dioxin-like compounds. Extensive consultations between the compilers of this database with the WHO represented by M. van den Berg and R. E. Peterson took place. However, it must be emphasized that for this 2005 TEF reevaluation, the expert panel used all available REPs, either included or excluded in this database, and made their own assessment (Haws et al., 2006). Studies published since the 1997 reevaluation were also fully evaluated.

When reviewing the database of mammalian REPs for dioxin-like compounds, it was observed that even for the most thoroughly studied congeners like 2,3,4,7,8-pentachlorodibenzofuran (PeCDF) and PCB 126, significant gaps in knowledge exist (Haws et al., 2006). Reasons for significant differences in REPs for the same congener can be caused by the use of different dosing regimens (acute vs. subchronic), different endpoints, species, and mechanisms (e.g., tumor promotion caused by at least two different mechanisms as for monoortho– substituted PCBs), as well as different methods used for calculating REPs. Thus, different methodological approaches used in different studies clearly provide uncertainties when deriving and comparing REPs. If future study designs to derive REPs were more standardized and similar, the variation in REPs when using the same congener, endpoint, and species might be expected to be smaller.

At this 2005 meeting, the "ideal" REP study design was discussed as previous WHO TEF evaluations did not provide sufficient information regarding the criteria that needed to be met to establish an REP value and give an expert panel the greatest degree of confidence in a particular REP. The following general guidelines for a future ideal dose-response study used to determine an in vivo REP resulted from the workshop:

  • A full dose-response curve for both the congener and for 2,3,7,8-TCDD should be determined.
  • The congener and 2,3,7,8-TCDD should be administered by the same route to animals of the same species, strain, sex, and age, and the animals should be housed, fed the same diet, and maintained under the same conditions in the same laboratory.
  • Ideally, the absolute maximal response (efficacy) should be similar for both the congener and for 2,3,7,8-TCDD and their dose-response curves should be parallel, but in practice, this is often not observed for various reasons.
  • If the above dose-response criteria are met, the REP should be calculated by dividing the effective dose 50% (ED50) of 2,3,7,8-TCDD by the ED50 of the congener.
  • If full dose-response relationships are not attained and determination of ED50’s is not possible, lowest observed effect doses or concentrations or benchmark doses could be used to determine the REP. However, such an REP has more uncertainty than if ED50’s were used.

For studies that are designed to determine REPs, it is clear that in vivo studies have the highest priority because they combine both toxicokinetic as well as toxicodynamic aspects. Therefore, in vivo studies should preferably be used for setting TEFs. Nevertheless, in vitro studies can contribute significantly to establish the AhR-mediated mechanism of action of a compound and explain possible differences in species sensitivity, especially with respect to that of humans versus experimental animal species. For in vitro studies, stricter criteria should be applied as these are from an experimental design point of view usually easier to accommodate than in vivo studies. For in vitro studies, the following experimental design is suggested to determine an REP:

  • Avehicle group and at least four graded concentrations of a congener and four graded concentrations of 2,3,7,8-TCDD should be selected.
  • For congener and 2,3,7,8-TCDD treatment groups, three of these concentrations should elicit a response that falls between the EC20 and EC80 for the congener and for 2,3,7,8-TCDD.
  • At least one concentration should elicit a maximal response (EC100), and the concentration-response curves should be parallel.
  • The REP should be based on the EC50 of 2,3,7,8-TCDD and the EC50 of the congener.

In general, 2,3,7,8-TCDD has been used as the reference compound of choice, but in several studies, PCB 126 has been used instead of 2,3,7,8-TCDD. Based on available data from the literature, it was concluded that PCB 126 could indeed be used as a reference compound in rat studies with an REP value of 0.1. Recent studies have confirmed this value for multiple endpoints (Toyoshiba et al., 2004; Walker et al., 2005). However, it should be examined in more detail if the same REP for PCB 126 is applicable as a reference compound for mouse studies. The REP values for some endpoints such as enzyme induction tend to be significantly lower in mice than in rats (Birnbaum and DeVito, 1995; DeVito et al., 2000; Harper et al., 1993; van Birgelen et al., 1996a). In this respect, it should be noted that mice studies in which PCB 126 was used as a reference compound were excluded from the database and from further consideration because of other methodological reasons (Haws et al., 2006).[1]

Literature data also indicate that the PCB 126 REP for enzyme induction in human cell systems, including primary hepatocytes, breast cancer cell lines, and primary lymphocytes, may be one or two orders of magnitude lower (van Duursen et al., 2003; Zeiger et al., 2001). In addition, the apparent binding affinity of 2,3,7,8-TCDD to the human AhR is generally 1/10th that of the AhR of the more sensitive rodent species, but significant variation among individual humans occurs (Ema et al., 1994; Harper et al., 2002; Poland et al., 1994; Ramadoss and Perdew, 2004; Roberts et al., 1990). It has been suggested that on average, humans are among the more dioxin-resistant species, but the human data set is too limited to be conclusive (Harper et al., 2002; Okey et al., 2005). A study with AhR-humanized mice may indicate lower responsiveness toward toxic effects of 2,3,7,8-TCDD (Moriguchi et al., 2003). Taken together, this information warrants more research into REP values in human systems to establish if the present TEFs based on rodent studies are indeed also valid for humans.

Additivity is an important prerequisite of the TEF concept, and this aspect was revisited in detail by the 2005 expert panel. It was concluded that results from recent in vivo mixture studies with dioxin-like compounds are consistent with additivity and support the TEF approach (Fattore et al., 2000; Gao et al., 1999; Hamm et al., 2003;Walker et al., 2005). Gao et al. (1999) studied the REP and additivity of 2,3,7,8-TCDD, 1,2,3,7,8-PeCDD, and 1,2,3,4,7,8-HxCDD in a rat ovulation model; their results confirmed both parallel dose-response curves and mixture additivity. Fattore et al. (2000) measured hepatic vitamin A reduction in rats after subchronic dietary exposure to a low-dose mixture containing 1,2,3,7,8-PeCDD, 2,3,4,7,8-PeCDF, and 1,2,3,6,7,8-HxCDF to test additivity. The effects of this mixture showed that the predicted results based on WHO 1998 TEFs were approximately twofold higher. Hamm et al. (2003) studied a mixture of nine dioxins, furans, and coplanar PCBs and looked at developmental reproductive endpoints in rats, comparing results of the mixture to that of 2,3,7,8-TCDD alone. The results showed that the experimental estimated TEQ was within a factor of two of that predicted from the WHO 1998 TEFs. A mixture study from the National Toxicology Program was also examined by the expert panel, and again the results generally supported additivity and parallel dose-response curves for complex and long-term neoplastic and nonneoplastic endpoints (Walker et al., 2005).

Thus, results in these recent mixture studies could be predicted rather well with the WHO 1998 TEFs, within a factor of two or less. This degree of accuracy was somewhat surprising in view of the complicated experimental situation present in subchronic toxicity studies, where congener-specific toxicodynamics and kinetics are intermingled and can influence the final outcome. In addition, the WHO 1998 TEFs were derived from a range of REPs using different biological models or endpoints and were therefore estimates with an order of magnitude uncertainty (Van den Berg et al., 1998).

Process used to determine TEF values: point estimates, expert judgment, and probabilistic distribution

Both the WHO 1993 and 1997 TEF reevaluations used point estimates derived by expert judgment from a wide range of REPs (Ahlborg et al., 1994; Van den Berg et al., 1998). In the 2005 TEF reevaluation, it was decided by the expert panel to use the REP database from Haws et al. (2006) for initial assessment of a TEF value. This recently published database and applied criteria were a refinement of the criteria and database that were developed to support the two previous WHO TEF reevaluations (Ahlborg et al., 1994; Van den Berg et al., 1998). The criteria for inclusion or exclusion of an REP in this database (Haws et al., 2006) were accepted by the expert panel. These criteria can be summarized as follows[1]:

  • At least one test congener and a valid reference compound must have been included in the study or the reference compound must have been included in an identical experiment from the same laboratory, but in another study.
  • The endpoint must have been an established AhR-mediated response known to be affected by both the test congener and the reference compound.
  • In the REP database, in vivo and in vitro studies were separated.
  • Repetitive endpoints (i.e., measures of the same biological response) were identified in all studies in the database, and the most representative REP value was retained for reevaluation of a TEF.
  • Those studies that used only a single-dose level of either the test and/or reference compound were filtered out of the REP database and not used in the TEF reevaluation process.
  • Results from non-peer–reviewed studies were not used in reevaluating a TEF value and consequently did not contribute to the distribution of REPs for individual congeners.
  • REPs based on biological responses that were statistically significant were included in the 2005 REP database and contributed to the distribution of REPs for individual congeners used to reevaluate TEFs. However, when there was a very limited data set for an individual congener, the panel also considered biological responses that were not statistically significant as part of the overall expert judgment in reevaluating a TEF value.
  • REPs based on quantitative structure-activity relationship studies were included in the REP database.

When using this database, the primary focus of the TEF reevaluation was on in vivo studies (Haws et al., 2006). In vitro studies were only used for support in those situations where no or very few in vivo REP data were available. For in vitro REPs, only established AhR-mediated responses were used to assign REP values.

During the TEF reevaluation, the expert panel considered using REP distributions available from the REP database (Haws et al., 2006) when reevaluating a TEF value. However, the REP distributions in this study are unweighted (Haws et al., 2006), and it was decided that establishing a weighting criteria for REPs generated in different types of studies (in vivo, in vitro, chronic, acute, etc.) was not feasible at this meeting. In addition, it was concluded that REP distributions for a specific congener in this database could not be used to derive a TEF value because a fixed percentile would have to be used as a cutoff point. Such an approach would be like using a single point estimate, but with lower biological or toxicological relevance. This is because all types of in vivo studies (acute, subchronic, etc.) and different endpoints have been combined, and associated REP distributions are shown as a single box plot. Thus, with only unweighted distributions of REPs available, a final expert judgment in the TEF reevaluation process involving the type and quality of the study had preference over the unweighted REP distributions (Haws et al., 2006). Nevertheless, it was recognized by the expert panel that in the future weighted REP distributions could be used for derivation of TEF values, but establishing consensus values for these REP weighting factors would require additional expertise.

The WHO expert panel decided that a combination of these unweighted REP distributions, expert judgment, and point estimates would be used to reevaluate a TEF. Figure 1 shows the unweighted distribution of in vivo and in vitro REPs and WHO 1998 TEF values for PCDDs, PCDFs, non-ortho PCBs, and mono-ortho PCBs (Haws et al., 2006). These unweighted REP distributions were used to start the selection and decision process for a TEF reevaluation. The 75th percentile of the in vivo REP distribution for an individual congener was used as an initial decision point to review the WHO 1998 TEF for that congener. If the WHO 1998 TEF was below the 75th percentile of the in vivo REP distribution, the data driving this TEF value was extensively reevaluated. If the WHO 1998 TEF value was above the 75th percentile, a quick review was done regarding the decision made at the 1997 WHO meeting with respect to those studies that had been driving the 1998 WHO TEF value. In addition, results of new studies conducted after 1997 or old information missed at the 1997 WHO meeting were evaluated to determine if these would influence the WHO 1998 TEF value for that congener. Based on the combined information, a possible new TEF value was considered. Special attention was also given to validity of WHO 1998 TEF values that were near or higher than the 90th percentile, e.g., 1,2,3,7,8-PeCDF. Thus, the above TEF reevaluation process provided a way both to increase as well as to decrease a TEF value. Figure 2 illustrates the decision scheme used at the expert meeting for the initial reevaluation process of the TEFs. For transparency, the expert judgment process and rationale used by the expert panel for a possible newly assigned WHO 2005 TEF value is explained in the next paragraph. This is followed by subsequent paragraphs devoted exclusively to each congener reevaluated.

As in previous WHO TEF consultations, it was decided by the expert panel to use a stepwise scale for assigning TEFs values. However, instead of assigning TEFs in the increments used previously (0.01, 0.05, 0.1, etc.), it was decided to use half order of magnitude increments on a logarithmic scale at 0.03, 0.1, 0.3, etc. As a result, all (non)revised 2005 WHO TEFs were fitted on a logarithmic scale. This decision to assign TEFs as half order of magnitude estimates may be useful in describing, with statistical methods, the uncertainty of TEFs in the future. Thus, as a default, all TEF values are assumed to vary in uncertainty by at least one order of magnitude, depending on the congener and its REP distribution. Consequently, a TEF of 0.1 infers a degree of uncertainty bounded by 0.03 and 0.3. For a TEF value of 0.3, a degree of uncertainty bounded by 0.1 and 1 was used. Thus, the TEF is a central value with a degree of uncertainty assumed to be at least ± half a log, which is one order of magnitude. However, it should be realized that TEF assignments are usually within the 50th to 75th percentile of the REP distribution, with a general inclination toward the 75th percentile in order to be health protective. However, the latter approach was also influenced by the type and quality of study, e.g., single versus multiple dose, that could not been discerned from the REP distributions shown in Figure 1. This more conservative and health protective approach practically means that for a TEF value the likelihood of a half-log error too low is less than the likelihood of half a log error too high. Due to the new ‘‘spacing’’ to express TEFs on a half-log scale, it was also necessary in the final review process to evaluate each individual TEF value for those congeners for which there were no new data available.


See also


  1. 1.0 1.1 1.2 Martin Van den Berg, Linda S. Birnbaum, Michael Denison, Mike De Vito, William Farland, Mark Feeley, Heidelore Fiedler, Helen Hakansson, Annika Hanberg, Laurie Haws, Martin Rose, Stephen Safe, Dieter Schrenk, Chiharu Tohyama, Angelika Tritscher, Jouko Tuomisto, Mats Tysklind, Nigel Walker, and Richard E. Peterson: The 2005 World Health Organization Reevaluation of Human and Mammalian Toxic Equivalency Factors for Dioxins and Dioxin-Like Compounds. Toxicological Sciences 93(2), 223–241 (2006) doi:10.1093/toxsci/kfl055 . Cite error: Invalid <ref> tag; name "vandenberg2006" defined multiple times with different content Cite error: Invalid <ref> tag; name "vandenberg2006" defined multiple times with different content