The main objective of the present study was to test the hypothesis that the increased tumor incidence observed with a mixture of dioxins could be predicted based upon the potency-adjusted dose-additive effect of the individual compounds present within the defined mixture. A key assumption in this approach is that, across the different studies, the shape of the dose–response curves for the increased incidence of each respective neoplasm is fundamentally the same. This was indeed the case, indicating that it is appropriate to describe the relative carcinogenicity of each compound/mixture by the ratio of their ED50
values to that of TCDD. Furthermore, we showed that the observations seen for the mixture for each of the four neoplasms were, in general, consistent with the potency-adjusted dose-additive effects seen individually for TCDD, PCB-126, and PeCDF. Finally, we showed that for hepatocellular adenoma and gingival SCC, the effects seen for the mixture were generally consistent with the use of the WHO TEF values (Van den Berg et al. 1998
) of 0.1 and 0.5 for PCB-126 and PeCDF, respectively. Moreover, although the use of the WHO TEFs for increased incidences of cholangiocarcinoma of the liver and CKE of the lung was statistically rejected, the estimated potency of the mixture relative to TCDD alone for these sites was 0.98 and 1.21, respectively, only marginally different from the expected value of 1.0.
It is important to note that the current WHO TEFs are based on an expert evaluation of individual studies that examined the relative potency of a given chemical to the reference compound, TCDD, which is assigned a potency of 1 (Van den Berg et al. 1998
). TEF values are an order of magnitude estimate of the overall “toxic potency” of a given compound and therefore do not specifically refer to the potency from any single study with a particular end point. By comparison, an RPF is determined for a specific chemical in a single study relative to a specific end point. Consequently, it was expected that the estimated potencies would not be identical to the WHO TEF values. It is noteworthy that although RPF values for other end points used for the derivation of TEFs span several orders of magnitude, in general the RPFs for each compound across different sites varied less than half an order of magnitude.
From these analyses, it is evident that the current WHO TEF value of 0.1 for PCB-126 is an appropriate value. For cholangiocarcinoma, hepatocellular adenoma, and gingival SCC, the optimal potencies of PCB-126 were 0.11, 0.09, and 0.09, respectively. The optimal potency for induction of CKE was 0.20. The increased potency for the mixture appeared to be due to a higher than predicted observed potency of PCB-126 for this site (0.21 compared with its WHO TEF of 0.1). Although use of an overprediction of potency would ultimately be protective of human health when used in a risk assessment setting, an underprediction of risk would be less protective. Additional research is required to understand the pathogenesis of these squamous neoplasms and if this may be related to human lung cancer risk.
For PeCDF, it appears that the current WHO TEF of 0.5 (Van den Berg et al. 1998
) somewhat overestimates its potency for all the analyzed neoplasms. For cholangiocarcinoma, hepatocellular adenoma, CKE, and gingival SCC, the optimal potencies of PeCDF were 0.16, 0.34, 0.30, and 0.26, respectively. This suggests that the current TEF value for PeCDF ought to be reevaluated for its application in quantitative cancer risk assessments. The lower potency of PeCDF observed here is consistent with earlier work on the promotion of altered hepatocellular foci in rat liver in a two-stage initiation–promotion model of hepatocarcinogenesis (Waern et al. 1991
). In that study, the authors estimated that the potency of PeCDF relative to TCDD was approximately 0.1, when based on weekly administered dose after an initial loading dose.
Although it is beyond the scope of this article to fully compare the present TCDD study with previously reported studies of dioxins and PCB mixtures, in general the site specificity of effects from these studies was consistent with those prior studies with TCDD and related compounds. In the feed study of TCDD conducted by Dow Chemical Company, Kociba et al. (1978)
observed increased incidences of neoplasms in the liver, lung, and hard palate. The increased incidence of cholangiocarcinoma that was seen in the present series of studies has not been seen before in cancer bioassays of DLCs or PCB mixtures. A detailed comparison of study design issue and comparative dose–response modeling of data from these studies will be reported separately.
Although the focus of the data reported here was evaluation of carcinogenicity, we have previously reported on the examination of induction of cytochrome P450 data from animals killed at interim time points during the conduct of these studies (Toyoshiba et al. 2004
). In that analysis, we found that in general there was lack of support for common dose–response shape for induction of CYP1A1 and CYP1A2 in the liver and CYP1A1 activity in the lung. Moreover, when modeling the data under the assumption of common shape, there was in general a lack of dose additivity for the mixture. This appeared to be driven for the most part by the dose response for induction of P450 by PeCDF, which showed higher levels of induction of P450 than the other compounds. PeCDF can sequester in the liver at high levels, leading to high body burdens of this compound at higher doses. Given that the induction of these P450s is tightly linked to tissue levels of the compound, the variation in dose response likely reflects differences in short-term pharmacokinetics and pharmacodynamics. In contrast, neoplasia in these studies appears to be a more protracted response that requires longer durations of constant exposure. Hence, pharmacokinetic and pharmacodynamic differences between the compounds may not be as influential on the ultimate dose–response models.
The data presented here support additivity for compounds whose primary mechanism of action is via the AhR. However, it was not designed to address additivity for compounds that may have multiple modes of action that are also included in the current TEF scheme, for example, mono-ortho-PCBs (e.g., PCB-118). In addition, because exposure to PCBs always occurs as a mixture, this study was not designed to address whether the potency of a DLC is affected by non-DLCs such as the di-ortho-PCBs (e.g., PCB-153). To this end, additional studies as part of this evaluation of the dioxin TEF scheme being conducted by the NTP are examining the carcinogenicity of PCB-153 and PCB-118 and also mixtures of PCBs (PCB-126 and PCB-153, and PCB-126 and PCB-118).
In summary, to our knowledge this is the first study that has systematically examined the specific interactions within a mixture of compounds in the context of the chronic rodent carcinogenicity bioassay. The main conclusion from this study is that we cannot reject the hypothesis of potency-adjusted dose additivity for induction of rodent neoplasms for a defined mixture of DLCs. Moreover, the optimal potency of the defined mixture was almost the same as for TCDD alone. These analyses underscore that the use of TEFs and dose additivity for assessing mixtures of persistent AhR ligands is reasonable for cancer risk assessments and is now supported by some experimental evidence.