Final Report Summary - SYSTEQ (The development, validation and implementation of human systemic Toxic Equivalencies (TEQs) as biomarkers for dioxin-like compounds)
Executive Summary:
Risk assessment of dioxin-like compounds (DLCs) is using a method in which each compound is assigned a toxic equivalency factor (TEF) by the WHO. This methodology is based on the fact that these DLCs are acting additively when present in mixtures. However, the present TEFs are only intended for estimating risks after oral intake, e.g. by consumption of food products and breast milk. Using these TEF values for human tissues or blood for risk are scientifically unsupported and may be unsound. The major objective of the EU-SYSTEQ project was to establish if separate systemic-TEFs should be established for human risk assessment of DLCs. In addition, this project focussed on possible new biomarkers of exposure and the validity of in vitro data for human risk assessment. At the end of the EU-SYSTEQ project an international expert meeting was organized to discuss and evaluate the results obtained in this project.
The results from EU-SYSTEQ do not warrant the development of separate systemic-TEFs for risk assessment. This is based on the fact that intake and systemic, plasma-based relative effect potencies (REPs) are similar in the two in vivo rodent models within +/- half an order of magnitude. Such an uncertainty range has already been proposed by the WHO as inherent to the present TEF values, which are based on a large number of REPs. Only a few compounds have systemic-REPs that fell outside this uncertainty range and the most likely explanation for this discrepancy can be found in congener-specific body distribution that is suggested to be rodent-specific.
To determine potential novel biomarkers for DLC exposure, expression of other genes, besides the “classical” biomarkers CYP1A1, 1B1 and 1A2, were studied. Gene expression of Ahrr, Tiparp, Aldh3a1, Notch1 and Tnfaip8l3 were identified as potential candidates for novel biomarkers of DLCs. AHRR gene expression was considered to be sensitive in certain human models. Based on the EU-SYSTEQ data, CYP1A1 gene expression and activity remains the most responsive biomarker for humans.
The EU-SYSTEQ data show that for rodents, in vitro data could be an adequate replacement for in vivo systemic data for the estimation of REPs of DLCs. Based on this, it may be suggested that similar relationships between relevant in vitro test systems and in vivo data may also apply in humans for other metabolically persistent compounds.
In vitro experiments with rodent and human models combined with multivariate analysis indicated species-specific response patterns between different DLCs. For some compounds, differences in REPs were greater than one order of magnitude between rodent and human models, but also between the WHO-TEF and human in vitro response. Notably, human in vitro REPs of PCB126 are up to two orders of magnitude lower than the present WHO-TEF. Based on similarities observed between rodent in vitro and in vivo REPs in the EU-SYSTEQ project, a similar human in vitro – in vivo relationship may be expected for this important DLC. Thus, the EU-SYSTEQ human in vitro data provide support to suggest a re-evaluation of WHO-TEFs and development of human-specific TEFs for some congeners, such as PCB 126 and mono-ortho PCBs.
Project Context and Objectives:
Introduction.
Polychlorinated dibenzo-p-dioxins (PCDDs), dibenzofurans (PCDFs) and biphenyls (PCBs) commonly occur in the environment and human food chain. A total of 419 possible congeners exist, from which a significant number of congeners is still detected in humans and wildlife, at levels that might cause long term health effects. Extensive toxicological and exposure studies have been performed during the last decades providing detailed knowledge about the toxicokinetics and toxicodynamics of these compounds. Based on these studies, the human and environmental risk assessment is driven by the toxicologically most relevant PCDDs, PCDFs and PCBs. The structure-activity relationship (SAR) involves basically compounds with halogen substituents on the 2,3,7,8-positions in the PCDD and PCDF molecules, while for PCBs halogen substitution on two adjacent 3,4 or 5 positions with lack of ortho substitution provides a similar dioxin-like mechanism of action. Thus, human and wildlife exposure to dioxin-like compounds involves a complex mixture of congeners sharing a common mechanism of action via the arylhydrocarbon receptor (AhR).
Based on this similarity, the Toxic Equivalency (TEQ) concept was developed for dioxin-like compounds. As such, TEQ values are a sum of the products of a congener-specific toxic equivalency factor (TEF) and a concentration in a matrix, such as human plasma or milk. The TEF value for a specific PCDD, PCDF or PCB is a composite value derived from a range of biomarkers with relative effect potencies (REPs) being congener and endpoint specific. To determine these TEF values, in vivo (semi) chronic studies have been given the highest weight for human and environmental risk assessment. These TEQ values are used as biomarkers to predict possible risks for humans and wildlife.
TEF values and harmonization.
Starting in the early 1990s WHO organized international expert meetings with the objective to harmonize TEFs for dioxin and dioxin-like compounds, including some PCBs. Since 1998 these TEF values have also been differentiated between mammals, birds and fish, with mammalian TEFs being used for human risk assessment. During the latest WHO TEF expert meeting in 2005, a combination of individual relative potency (REP) distributions, expert judgement and point estimates was used to assign or re-evaluate an existing TEF factor. To support such WHO expert meetings, a database with relative potencies from endpoints that can be used as a biomarker (REPs) for dioxin like compounds has been compiled. Besides the re-evaluation of the WHO 1998 TEF values, the WHO TEF meeting in 2005 also evaluated the validity, criteria and correct use of the TEF/TEQ methodology.
At the latest WHO TEF meeting in 2005, it was explicitly concluded that the present WHO-TEF scheme and TEQ methodology for human risk assessment is only intended for estimating intake exposure to dioxin-like chemicals, e.g. by consumption of food products and breast milk. This limitation comes from the fact that most mammalian toxicity studies with these compounds are based on oral intake, often through the diet. A major conclusion from the 2005 WHO expert meeting was that the application of the present TEFs to human tissue or blood samples for risk assessment maybe scientifically unsupported and unsound. Therefore, the use of systemic-TEFs was suggested by the WHO panel as an additional and highly relevant approach, besides the use of the present intake-TEFs.
Are intake- and systemic-TEFs similar?
At present, the TEQ concept for human risk assessment is mainly based on in vivo animal experiments with oral dosage as the principal route of exposure. Consequently, the present human TEFs may only be applicable for exposure situations in which oral ingestion occurs. Nevertheless, these ‘intake’ TEFs are commonly, but possibly incorrectly, used by regulatory authorities to calculate ‘systemic-TEQs in human blood and tissue levels, and consequently considered to be biomarkers for either exposure or effect. Thus, intake- and systemic-TEQs are implicitly assumed to be similar by most, if not all, regulatory authorities, which may cause an important flaw in the present use of this TEQ concept for risk assessment. At the start of the EU-SYSTEQ project there was little experimental data available, which would either reject or accept the use of these ‘intake’ TEFs and TEQs as biomarkers in human blood or tissues. At that time, only a few in vivo studies had addressed the difference between intake- and systemic-TEFs to a limited extent, but these studies indicated possible and significant differences between both types of TEFs. Thus, the present use of TEFs and associated TEQs for human blood and tissue concentrations is not supported by scientific validation.
The application of systemic-TEFs.
The availability of systemic-TEFs to determine systemic-TEQs as a biomarker is essential for many types of risk assessment, because concentrations in human blood or tissues are often used and proven most suitable to determine above-background exposure situations, e.g. by accidental food poisoning or suspected differences in environmental exposure between populations. The question is, if in these situations the use of intake-TEFs will only give a minimal error in the risk assessment process compared to many other uncertainties inherent to the TEF concept. Under the assumption that systemic- and intake-TEQs may be significantly different, the development of systemic-TEFs can play a major role as biomarkers, to determine the degree of exposure in which exposure (lifestyle) and bioavailability (soil, airborne particle) can vary strongly. If needed, the availability of validated systemic-TEFs may also give a stronger quantitative support during poisoning episodes or epidemiological studies with respect to health impact assessment. Another advantage of the proposed systemic-TEF approach is the fact that it allows better comparison between experimental studies and humans, because it normalizes differences in toxic kinetics and metabolism. In figure 4.1.1 A general overview is given of the differences between systemic- and uptake-TEFs and their relevance for the risk assessment process.
Different research aspects of EU-SYSTEQ.
Based on the problem definitions described above, the EU-SYSTEQ project comprised of the following seven defined research parts:
1) Short term in vivo studies with rat and mice to establish differences or similarities for ‘intake’ and ‘systemic’ REPs for the most relevant PCDDs, PCDFs and dioxin-like PCBs for human risk assessment.
2) In vitro studies using both rodent and human systems to determine differences between species and in vivo data.
3) Besides monitoring expression of established AhR target genes, the identification of possible novel biomarkers for exposure in human and rodent models.
4) Multivariate comparative analysis for TEFs derived from rodent and human in vitro systems to identify possible differences in QSARs between species.
5) Comparative analysis of ‘systemic’ TEFs and in vitro TEFs from rodent systems.
Project Results:
DIFFERENT RELATIVE EFFECT POTENCIES FROM IN VIVO RAT AND MOUSE STUDIES.
In this part of the project, intake-REPs and systemic-REPs were compared in female C57BL/6 mice and Sprague Dawley rats based on the administered dose and liver, adipose, or plasma concentrations. 2,3,7,8-tetrachlorodibenzodioxin (TCDD), 1,2,3,7,8-pentachlorodibenzodioxin (PeCDD), 2,3,4,7,8-pentachlorodibenzofuran (4-PeCDF), 3,3´,4,4´,5-pentachlorobiphenyl (PCB 126), 2,3´,4,4´,5-pentachlorobiphenyl (PCB 118), and 2,3,3´,4,4´,5-hexachlorobiphenyl (PCB 156) were used, which represent approximately 90% of the dioxin-like activ¬ity in the human food chain. The non-dioxin-like 2,2´,4,4´,5,5´-hexachlorobiphenyl (PCB 153) was also included as a negative control for dioxin-like compounds. Three days after oral exposure, we calculated intake-REPs and systemic-REPs for hepatic CYP1A1-associated ethoxyresorufin-O-deethylase (EROD) activity and Cyp1a1, 1a2, and 1b1 gene expression in the liver and peripheral blood lymphocytes (PBLs).
In the past, it has been shown, that PCBs used in experimental studies were containing dioxin-like contaminants at levels that could have induced sensitive AhR-mediated effects. For this reason, PCB 118, 156 and 153 were specifically purified before use in the EU-SYSTEQ project. After purification, the remaining toxic equivalent (TEQ) con¬tributions that were present in these three PCBs were consid¬ered to have no influence on the final outcome of our in vivo as well as in vitro results.
Many experimental studies done in the past have reported significant differences in maximal induction between dioxin-like congeners. This phenomenon was also observed in the EU-SYSTEQ project. Many of the dose–response curves in our project did not attain a maximum efficacy or similar Hill slope as the reference compound TCDD. As a result, significant uncertainties in calculating EC50 values could occur. Therefore, REPs in the EU-SYSTEQ project were calculated using a benchmark response (BMR) approach. In our case the dose or concentration needed for a congener to reach 20 % of the TCDD response (BMR20TCDD) was used. The advance of this benchmark approach at the lower part of the dose–response curve is that the lack of agreement in curve shape is less pronounced than when using the EC50-based approach. This BMR20TCDD also presents an exposure situation that is more relevant and closer to the actual human exposure.
In vivo mouse studies.
In the mouse Cyp1a1, 1b1, 1a2 gene expression or activity were used to determine intake-REPs and systemic-REPs in the liver and PBLs. Full details are provided in Van Ede et al. 2013a (1). For comparison of congener-specific REPs across exposure matrices (intake, liver, adipose or plasma), the intake-REP in our mouse study was set to 1. Subsequently, deviations were calculated for vari¬ous systemic-REPs with the same end point (Figure 4.1.2). Two different types of deviations between systemic-REPs and intake-REPs were observed. Based on liver concentrations (wet weight or lipid weight), systemic-REPs of PeCDD, 4-PCDF, and PCB 126 were at most one-third of the intake-REP. In contrast, systemic-REPs of PCB 118 and PCB 156 are up to one order of magnitude higher than their intake-REPs for hepatic endpoints. When systemic-REPs for hepatic effects of PeCDD, 4-PeCDF, and PCB126 were calculated using adipose tis¬sue and plasma concentrations, systemic-REPs were up to one order of magnitude higher than intake-REPs, depending on the endpoint studied. We found the opposite for the systemic-REPs of PCB 118 and PCB 156, which were at most one-third of the intake-REPs. In PBLs, systemic-REPs based on plasma concentra¬tions also deviated from intake-REPs, in a man¬ner similar to that of systemic-REPs of hepatic endpoints based on plasma concentration. Thus, significant deviations from intake-REPs were found for systemic-REPs in the mouse model, as can be seen from figure 4.1.2. However, this difference was usually within half a log unit of the intake-REP, but deviations for some congeners and effects were between a half and one log order of magnitude. It should be noted that such variation is also considered to be of relevance in the WHO-TEF concept, as it is assumed that these TEF values are having +/- half log uncertainty.
In vivo rat studies.
In the rat dose–response relationships for hepatic EROD activity and gene expression of Cyp1a1, 1b1, 1a2 and Ahrr and PBLs were determined using intake or administered dose levels and liver, adipose tissue or plasma concentrations. All compounds, except the non-dioxin-like PCB-153, significantly induced hepatic CYP1A1 activity and Cyp1a1, 1b1, 1a2 and Ahrr gene expression. Full details are provided in Van Ede et al. 2013b. A clear distinction in hepatic Cyp1a1, 1b1, 1a2 and Ahrr gene expression was observed between the more potent AhR agonists TCDD, PeCDD, 4-PeCDF and PCB-126 and less potent AhR agonists PCB-118 and PCB-156. In PBLs of the rat, the induction of Cyp1a1, 1b1 and Ahrr genes was up to three orders of magnitude lower compared to those in the liver. The comparative BMR20TCDD concentrations for different congeners were calculated. Based on administered dose, tissue and plasma concentrations, intake-REPs and systemic-REPs could then be calculated. Similar to the mouse, changes between systemic REPs and intake REPs were compared for which each congener-specific intake-REP was set to 1 and deviations from this intake-REP are given for each systemic-REP with the same biological endpoint (see figure 4.1.3). When comparing intake-REPs with systemic-REPs in the rat, deviations of not more than half a log were observed for PeCDD, PCB126 and PCB-156 using the different biological matrices. A clear exception was found for 4-PeCDF for which the systemic-REPs deviated up to one order of magnitude from the intake-REPs based on adipose tissue and plasma concentrations. Gene expression of Cyp1a1, 1b1 and Ahrr were also determined in rat PBLs. Deviations between intake-REPs and systemic-REPs followed a similar congener-specific pattern as observed when using plasma and adipose tissue as biological matrices (figure 4.1.3.).
The same experimental design of the EU-SYSTEQ mouse and rat studies allows a combination of the results from both studies and a comparison of the present WHO-TEF values used for human risk assessment. This comparison is graphically shown in figure 4.1.4. It can be concluded that the intake-REPs and plasma-based, systemic-REPs for PeCDD, PCB-126, PCB-118 and PCB-156 are similar and deviate no more than a half log. A clear exception is found for the plasma-based systemic-REP of 4-PeCDF that is an order of magnitude higher compared to its intake-REP.
In general, these plasma-based, systemic-REPs distribute mostly within the suggested WHO-TEF uncertainty range. However, it should be recognized that the EU-SYSTEQ REPs of mouse and rat are based on a limited number of biological endpoints, whereas the WHO-TEFs consist of REPs that cover a much broader range of endpoints. This possibly provides an uncertainty, which hampers drawing a definite conclusion about the (dis)similarity of systemic-REPs of our study and the WHO-TEFs. Nevertheless, the results of our rat and mouse EU-SYSTEQ studies indicate that the use of WHO-TEFs for blood concentrations in humans gives an uncertainty that mostly falls within the uncertainty range of +/- half a log unit, which is assumed to be present in WHO-TEFs.
THE ROLE OF DIFFERENCES IN DOSE METRICS FOR SYSTEMIC-REPS
WHO-TEF values are based upon administered dose as the exposure metric. Yet, these TEFs are widely applied to assess human risks based on concentrations in human blood serum. Differences in absorption, distribution, metabolism, and excretion can affect the relative potency of a congener, but this is not taken into account when a REP is based on administered dose. The main objective of EU-SYSTEQ project was to establish in vivo systemic-REPs in mouse and rat and compare these with intake-REPs in the same experimental model. The EU-SYSTEQ project used a single oral dose regimen with measurements of tissue concentrations and responses in both hepatic and extra-hepatic biological matrices three days after dosing.
A relevant question related to our short term in vivo studies is their relevance compared with (semi) chronic animal studies, which usually are the major contributors to the derivation of a WHO-TEF value. As part of our EU-SYSTEQ project the tissue distribution data across the tested compounds and dose levels in our three-day studies were compared with data from previous studies with these rodents using both single and sub-chronic dosing regimens. A detailed evaluation and comparison between different studies is given in Van Ede et al. 2013c. In addition, EC50 values for Cyp1a1 gene expression and activity based on hepatic concentrations were compared between the EU-SYSTEQ short-term studies and earlier (semi-) chronic studies with rat and mouse. The patterns of the distribution between liver and adipose tissue based on wet weight for rats and mice across congeners from single and semi-chronic studies are displayed in figures 4.1.5 and 4.1.6. These data show that the liver:adipose tissue ratio changes nonlinear across the dose range used by the various experimental studies with rats and mice. This nonlinear behaviour observed for a number of potent dioxin-like compounds has been attributed to the hepatic induction of CYP1a2 and subsequent binding of dioxin-like compounds to this protein. As a result, the liver:adipose ratio shifts toward a higher value than expected with an increasing dose. In contrast with data from experimental studies with rodents, available distribution data for humans in the general population display little or no hepatic sequestration (Van Ede 2013c).
A second and more important observation for the EU-SYSTEQ project is the fact that the used three day experimental time period resulted in comparable liver:adipose tissue ratios as those found in more long term (semi)chronic experimental studies. Based on this observation, it is concluded that the selected three-day experimental design in the EU-SYSTEQ project also provides relevant information for systemic-REPs that are relevant for longer (experimental) exposure time periods. This conclusion is also supported by the fact that hepatic EC50 values for CYP1A1 induction by TCDD are similar between short term and long-term studies, when expressed as hepatic concentrations independently of the experimental time period. In addition, calculating systemic-REPs based on hepatic concentrations and responses could result in REPs that are not fully applicable to the relevant toxic endpoints and systemic exposure measures in most human studies. Systemic-REPs based on blood, serum or plasma concentrations with an extra-hepatic response may give a more accurate prediction of the relative potency of a congener for humans under environmental exposure conditions.
DIFFERENT RELATIVE EFFECT POTENCIES FROM IN VITRO MODELS.
Within the EU-SYSTEQ project, a wide array of in vitro models has been used to determine REPs for various congeners and for the detection of species- or tissue-specific sensitivities for dioxin-like compounds tested. For these tests, both primary cells systems (hepatocytes, PBLs, keratinocytes) as well as immortalized cell lines have been used.
Rat and human hepatocytes.
20 Different dioxin-like compounds and PCBs were tested in both types of hepatocytes. In rat hepatocytes, sensitive responses were observed and the REPs were determined either by using the EC50 or BMR20TCDD values. In general, the derived REPs were in good agreement with the WHO-TEFs. This was to be expected, since the TEF values have been derived from a substantial number of rat studies in which hepatic effects were selected as major biological or toxicological endpoints. Largest deviations from the WHO-TEFs were found for the weak AhR agonists such as the mono-ortho PCBs. Significant differences in responses towards dioxin-like compounds were observed in human hepatocytes compared with those from the rat. First of all, the sensitivity of response was much less for the human hepatocytes. For example the response induced by TCDD was one to two orders of magnitude more sensitive in the rat hepatocytes. In addition, the response of human hepatocytes towards PCB 126 was at least two orders of a magnitude lower than those observed in rat hepatocytes. Such a low responsiveness was also observed in the human hepatocytes for the mono-ortho PCBs 118 and 156. In figure 4.1.7 the differences in responses between the rat and human hepatocytes are given for the congeners tested.
Rat and human hepatoma cell lines.
Similar to the comparison between rat and human primary hepatocyes, AhR-mediated responses in the rat H4IIE and human HepG2 hepatoma cell-lines were compared. The human HepG2 cells were less sensitive than the rat hepatoma cells. Again, the most notable difference between the rat and human hepatoma cells was observed for PCB 126, for which the response in the latter cell system was again one to two orders lower than that observed in the rat cells. As in the human, primary hepatocytes, the mono-ortho PCBs (e.g. 118,156) did not show a measurable response in the human HepG2 cells.
Rat and human lung cell lines.
In past many in vitro studies on dioxin-like compounds focussed on liver cell systems using CYP1A1 induction and activity as a biomarker. However, in the human liver cell CYP1A1 activity is not dominantly present. In contrast, human lung cells are much more sensitive towards the induction of CYP1A1 activity by dioxin-like compounds. For this reason, the EU-SYSTEQ project also included in vitro tests using human and rodent lung and bronchial cell lines. In vitro REPs were determined using the rat RLE-6TN and mouse MLE-12 lung cell lines as well as the human lung epithelial A549 and bronchial BEAS-2B cell lines. In these lung cell systems, the conventional responses on CYP1a1, CYP1b1 and Ahrr genes were studied to determine in vitro REPs for the selected congeners. In addition, two less-conventional, potentially novel biomarkers for AhR induction, TCDD-inducible poly(ADP-ribose)polymerase (TIPARP) and aldehyde dehydrogenase 3A1 (ALDH3A1), were also included. Based on the induction of these AhR target genes (CYP1A1, CYP1B1, AHRR, TIPARP and ALDH3A1) within rat, mouse and human lung cell models upon treatment with the selected PCDD/Fs and PCBs, REPs were calculated based on BMR20TCDD and EC50 values. Again the human lung cell systems were found to be 1-2 orders of magnitude less sensitive to TCDD than the rodent lung cells. Most REPs of PCDD/Fs were comparable with the WHO-TEF values. However, the REPs of PCB 126 in the human lung cells were one to two orders of magnitude lower than the WHO-TEF. Generally, the mono-ortho PCBs (PCB 118, 156, 169, 189) showed very low or no AhR-mediated activities. Primary lung tissue was also used from the in vivo rat study and the induction of Cyp1a1, Cyp1b1, Ahrr, TIPARP and Aldh3a1 genes by TCDD, PCB 126 and mono-ortho PCBs 118 and 156 was determined and used to calculate intake- and systemic-REPs (Table 4.1.1). The REP derived for the dioxin-like PCB 126 in rat primary lung tissue was in the range of the WHO-TEF value if Cyp1A1 or Aldh3a1 induction was used. The induction of Cyp1B1, Tiparp and Ahrr genes was less sensitive and also resulted in REPs that were approximately one order of magnitude higher than the WHO TEF value. Furthermore, systemic-REPs in rat lung tissue were determined based on plasma levels and found to be approximately similar to intake-REPs for PCB 126. The systemic-REPs for the mono-ortho PCBs 118 and 156 tended to be lower than the intake-REPs, but were still within half a log difference.
CALUX® cell lines of different species.
For a better understanding of species-specific relative effect potencies (REPs), responses of 20 dioxin-like compounds (DLCs) were assessed using chemical-activated luciferase gene expression assays (CALUX®) derived from rat, mouse, guinea pig and human cell lines. These data show that PCDD, PCDF and PCB mediated responses in the human CALUX® cell line again differ significantly from responses in the rat, mouse and guinea pig derived CALUX® cell lines. Not only is the human cell line one to two orders less sensitive for TCDD compared to the guinea pig, rat and mouse. Very significant congener-specific species differences in REPs were also observed between human and rodent CALUX® cell lines. Such a difference was most distinct for the human REP of PCB 126, which was one to two orders lower than those observed for the guinea pig, rat and mouse. In addition, both human CALUX bioassays did not show an AhR-mediated induction for the six mono-ortho PCBs, nor for the non-ortho substituted PCBs 77 and 169. Noticeable, human CALUX® assay derived REPs for 1,2,3,4,6,7,8-HpCDD, 1,2,3,4,7,8-HxCDF and 1,2,3,4,7,8,9-HpCDF were significantly higher compared to the rodent CALUX assays derived REPs. In figure 4.1.8 an overview is given of the ratios of CALUX® derived REPs in these four CALUX® cell lines and WHO TEFs. As can be seen from this figure, most CALUX® REPs of PCDDs and PCDFs are within the proposed uncertainty range of the WHO-TEFs, but significant deviations from this can be observed for the non- and mono-ortho PCBs tested.
Human lymphocytes.
Within the EU-SYSTEQ project species-specific differences in potency for Cyp1A1 activity and Cyp1a2, Cyp1b1 and Ahrr gene expression were determined for twenty selected PCDD, PCDF and PCB congeners in primary human PBLs and splenic cells obtained from the mice used in the earlier described in vivo studies. Human peripheral blood is relatively easy to collect, which makes it an interesting matrix for monitoring human health. In addition, changes in AhR-mediated gene expressions in PBLs are commonly used as biomarker of human exposure, despite the uncertainties and inter-individual variability in responses. Furthermore, present concerns regarding responses in human populations upon DLC exposure are more and more focused on extra-hepatic responses. Consequently, studies with respect to human responses that focus on extra-hepatic tissues or matrices such as PBLs are of distinct interest from a human risk assessment point of view. Results show that median human REPs for 1,2,3,4,6,7,8-HpCDD, 4-PeCDF, 1,2,3,4,7,8-HxCDF and 1,2,3,4,7,8,9-HpCDF determined in PBLs were significantly higher compared to REPs in the mouse spleen cells, but lower than the WHO-TEFs that are mainly based on rodent studies. In contrast, the REP for PCB 126 in human PBLs was two orders of magnitude lower than the WHO TEF, again indicating the lower human sensitivity towards this important dioxin-like PCB. Together, these data show congeneric and species-specific differences in REPs for some, but not all dioxin-like congeners tested. These comparative data from human PBLs and mouse murine splenic cells indicate that more emphasis should be placed on human-tissue derived REPs in the assignment of a TEFs for these compounds. These deviations from WHO TEFs as determined with human PBLs are shown in figure 4.1.9.
Human keratinocytes.
In the EU-SYSTEQ project, primary human keratinocytes were used to perform a genome-wide gene expression screen in order to identify novel AhR target genes. Keratinocytes were among others chosen as a model because the skin represents a human target organ for the toxicity of dioxin-like compounds. Genome-gene expression arrays were performed in both undifferentiated (basal) and differentiated terminal keratinocytes. It was shown that these two cell types show significant differences in sensitivity towards dioxin-like AhR ligands. These gene expression arrays were performed both in wild-type cells, as well as in cells where the AhR expression was knocked-down by siRNA- or siRNA-mediated technology. These genetic controls facilitated the identification of direct AhR target genes against the background of genes that were dysregulated in an AhR-independent manner. In addition to the known battery of xenobiotic metabolizing enzymes, which represents valid AhR target genes, also a number of physiologically and toxicologically interesting AhR target genes were identified that may be relevant as future AhR biomarkers for human exposure and toxicity. In figure 4.1.10 the up-regulation of two novel suggested biomarkers for dioxin-like toxicity, Ahrr and Aldh3a1, are shown besides the well-known Cyp1b1 gene induction. The use of siAhR indicates that especially and Aldh3a1 genes are indeed AhR mediated. In addition, concentration-response relationships were determined for the 20 selected PCDDs, PCDFs and PCBs in these human keratinocytes using Cyp1a1 and Ahrr gene expression as endpoints. These experiments showed that the 2,3,7,8-substituted PCDDs and PCDFs all induced Cyp1a1 and Ahrr with EC50 values around E-10 to E-8 M, depending on the number of chlorine atom in the PCDD or PCDF molecule. The mono-ortho PCBs e.g. 118 and 156 did not significantly induce gene expression of Cyp1a1 and Ahrr at levels that were similar of higher than normally found in the human body. Again, the induction of Cyp1a1 and Ahrr by the non-ortho dioxin-like PCB126 was approximately two orders of magnitude lower than TCDD, again confirming a REP in this human cell model is much lower that the WHO-TEF (see figure 4.1.11).
QSAR modeling to determine species differences.
More knowledge is needed on species sensitivity to dioxin-like compounds and PCBs, in particular with respect to differences between humans and experimental animal species. The use of mouse, rat, guinea pig and human recombinant cell lines containing an AhR responsive reporter gene (firefly luciferase, Calux®) in combination with quantitative structure-activity relationship (QSAR) analysis can potentially provide more insight on this issue. A QSAR represents a statistical model that quantifies the relationship between the structures of the compounds and the corresponding biological activity. The model provides a prediction of the biological activity of structurally similar, but untested compounds as well as discovering structural analogies that influence the activity of a group of compounds.
In the EU-SYSTEQ project, the potencies of a set of 20 selected PCDD, PCDFs and PCBs were determined using AhR-dependent Calux® bioassays from rat, mouse and human hepatoma cells, and guinea pig intestinal adenocarcinoma cells (as described earlier). Based on these in vitro data, species sensitivity and variation were examined using relative effect potency concentrations and principal component analysis (PCA). QSAR models were developed to relate these calculated REP values of the tested compounds with the calculated descriptors. Orthogonal projection to latent structures (OPLS) was used to finally predict the REPs for the dioxin-like compounds that have been assigned a TEF value by the WHO. The most significant descriptors of the derived models were identified to study differences in their structure-activity relationships in the tested species.
In figure 4.1.12 it is clearly illustrated that the Calux® human cell line is less sensitive to AhR-activation by TCDD, as well as PCB126, when compared to the rat, mouse and guinea pig Calux® cell lines. When BMR20TCDD ratios were compared between rodent species for all congeners tested, it showed that BMR20TCDD concentrations for guinea pig were one order of magnitude lower for the PCDDs and PCDFs tested and up to two orders of magnitude lower for the PCBs tested, when compared to rat and mouse. In the human Calux® assay, BMR20TCDD concentrations were up to two orders of magnitude higher compared to the rat, mouse and guinea pig Calux® assays for all congeners tested.
The QSAR plots of the predicted values versus the experimentally measured values for all CALUX assays are shown in Figure 4.1.19. Generally, there is a good agreement between the experimental and predicted REP values. While a better agreement was seen for PCDD/Fs for the rat, mouse, and guinea pig models, a low prediction error was achieved for PCBs in the human model as compared with the other species. This is however mainly due to the low activity of PCBs in the human CALUX assays, where of the tested PCBs only PCB126 induced AhR-activity. Furthermore, there is discrimination between PCBs and PCDD/Fs as can be seen from the results of QSAR models (Figure 4.1.13). This may indicate a different mechanism of binding for PCBs, PCDDs and PCDFs. PCBs have been previously reported to bind AhR differently than PCDDs and PCDFs.
Taken these data together, it is evident from these data that AhR-mediated responses in the human Calux® cell line differ significantly from responses in other rodent species derived Calux® cell lines. The human Calux® cell line is obviously less sensitive for TCDD. Also apparent congener-specific differences in potency were observed between human and rodent CALUX cell lines. This is most clearly reflected by a lower human REP for PCB 126 (0.003) compared to guinea pig (0.2) rat (0.07) and mouse (0.05). Surprisingly, human-derived REPs for 1,2,3,4,6,7,8-HpCDD, 1,2,3,4,7,8-HxCDF and 1,2,3,4,7,8,9-HpCDF were significantly higher compared to rodent derived REPs. Principal Component Analysis (PCA) indicated that REPs derived from rat, mouse, and guinea pig revealed an induction pattern similar to each other and to the WHO TEFs compared to human REP values. The most influencing chemical descriptors in the human model were clearly different from the rodent models, indicating different ligand-receptor interactions between human and rodents. These in vitro and in silico derived data from the EU-SYSTEQ project could be used as a basis for a better understanding of species variations and development of risk assessment tools of dioxin-like compounds and PCBs.
In vitro versus systemic REPs.
With respect to possible differences between intake-REPs and systemic-REPs, toxicokinetics and metabolism play a major role. The influence of the latter two factors is usually minimal in the in vitro assays described above for persistent organic pollutants such as PCDDs, PCDFs and PCBs. Thus, the question can be raised if in vivo derived systemic-REPs are more closely related to in vitro derived REPs. In figure 4.1.14 an overview is given for the intake-, systemic- and in vitro-REPs that have been obtained during the EU-SYSTEQ studies for PeCDD, 4-PeCDF and PCB 126. These three congeners represent a significant part of the total amount of TEQs in human exposure via food in the EU. As can be seen for the PeCDD and 4-PeCDF, these data indicate that indeed the in vivo systemic-REPs are close to the obtained in vitro REPs from the EU-SYSTEQ project. Such a similarity is less distinct for PCB 126, but remarkably the in vitro REPs for this congener are falling better in the +/- half log uncertainty range of the WHO TEF. For the other in vivo tested congeners a comparison between the in vivo and in vitro was not possible, as dose-response curves from the in vivo studies were too incomplete for these comparisons. Nevertheless, the data for PeCDD and 4-PeCDF provide indications, albeit limited, that in vitro REPs are more close too systemic-REPs than intake-REPs.
IDENTIFICATION OF NOVEL BIOMARKERS FOR DIOXIN-LIKE EXPOSURE.
Dioxin-like compounds evoke a broad spectrum of biochemical and toxic responses, i.e. enzyme induction, dermal toxicity, hepatotoxicity, immunotoxicity, carcinogenicity as well as adverse effects on reproduction, development, and the endocrine system in laboratory animals and in humans. Most, if not all, of the aforementioned responses, are mediated by the AhR. In the EU-SYSTEQ project the elicited biochemical effects of a selection of dioxin-like compounds and the non-dioxin-like PCB 153 were examined in mouse (in vivo) and human liver cell models (in vitro).
Mouse in vivo studies.
Three short-term in vivo studies were carried out aiming to characterize the alterations in hepatic gene expression. Based on the results obtained from these mouse studies, the seven test compounds were categorized into three classes; those which are 'pure' AhR ligands (TCDD, PeCDD, 4-PnCDF, and PCB 126) or solely CAR inducers (PCB 153), and the ones which are AhR/CAR mixed-type inducers (PCB 118, PCB 156). Moreover, the analysis of hepatic gene expression patterns after a single oral dose of either TCDD or PCB 153 revealed that the altered genes fundamentally differed (See figure 4.1.15). In Table 4.1.2 the ten highest up and down-regulated genes in the liver of female mice after TCDD exposure are presented. These formed the basis for further in vitro studies for new sensitive biomarkers in the EU-SYSTEQ project.
AhR knockout versus wild type mice.
A transgenic mouse study was performed with the genotypes Ahr-/- and Ahr+/+ and a single dose of TCDD. The aim of this mouse study was to analyze AhR-dependent and -independent effects in female Ahr knockout and wild type mice after a single dose treatment with TCDD. Microarray analysis of the TCDD-treated Ahr+/+ mouse livers revealed that a large number of genes associated with the lipid metabolism were either up- or down-regulated. The significantly increased liver weights together with the distinct alterations in mRNA expression of genes associated with lipid metabolism in female Ahr+/+ mice suggest the development of a hepatic steatosis by TCDD. Furthermore, various genes linked to the xenobiotic metabolism were up regulated in female Ahr+/+ mice, but not in Ahr-/- mice. Cyp1a1 was by far the highest up regulated gene by TCDD in Ahr+/+ mice, followed by Cyp1b1 and Cyp1a2. From the group of additional selected target genes Tiparp, Hsd17b2, and Ahrr were also up regulated in Ahr+/+ mice. Unexpectedly, administration of TCDD resulted in the up and down regulation of a large number of hepatic genes in female Ahr-/- mice. However, the pattern of significantly up and down regulated genes in Ahr-/- mice distinctly differed from Ahr+/+ mice. Thus, TCDD treatment led to alterations of a large amount of genes in the absence of the AhR. These observations should be further evaluated in the future to determine if these altered AhR –independent genes also contribute to TCDD-mediated toxicity in mice and possibly other species. In figure 4.1.16 the number of up and down-regulated genes in Ahr-/- and Ahr+/+ C57Bl/6 are given, including the number of genes expressed in both types of mice.
Gene expression in human hepatic cells.
In the EU-SYSTEQ project also human liver cell models (primary human hepatocytes and HepG2 cells) were used to assess the effects mediated by dioxins and dioxin-like compounds. The liver represents a primary target tissue in which TCDD has been shown to elicit carcinogenic and tumor promoting effects in rodents and these types of studies have formed an important basis for the risk assessment of dioxin-like compounds, especially in the USA. Thus, in vitro experiments with primary human hepatocytes and the human hepatocellular carcinoma cell line (HepG2) can provide important information to identify, validate, and establish quantifiable novel biomarkers for dioxin exposure and AhR-dependent modes of action.
Primary human liver cells represent the closest in vitro model to the human liver as these express the complete spectrum of hepatic drug-metabolizing enzymes. Furthermore, the comparison with results obtained from the immortalized human hepatocellular carcinoma cell line HepG2 will provide essential insight about similarities and differences between untransformed cells and immortalized tumor-derived cell lines. The number of affected genes in primary human hepatocytes is displayed in figure 4.1.17. As can be seen there is limited overlap in genes that are expressed by exposure to TCDD, PeCDD, 4-PnCDF and PCB 126. This is a surprising result, as these four potent dioxin-like compounds are often considered quite similar with respect to mechanism of action. The number of up- or down-regulated genes that overlap with the reference compound TCDD decreased in the order PeCDD > 4-PnCDF > PCB 126. This observation is of particular significance when trying to identify novel biomarkers for human exposure to dioxin-like compounds as a group. If the overlap between TCDD, 1-PnCDD, 4-PnCDF, and PCB 126 was analyzed only 25 up regulated common genes were identified. This limited number of gene overlap does restrict significantly the number of potential novel biomarkers for this group of four dioxin-like congeners, which present a significant part of the total amount of these compounds in human food and human body burden (See figure 4.1.18). Among the highest commonly up regulated genes are members of the cytochrome P450 superfamily CYP1A1, 1A2, and 1B1, which confirm the specificity of these genes as biomarkers for dioxin-like compounds. Other enzymes involved in xenobiotic metabolism are also highly up regulated such as TIPARP, and ALDH3A1. Induction of these two genes has also been studied in a number of other in vitro studies that were done within the framework of the EU-SYSTEQ project. In table 4.1.3 these 25 commonly up regulated genes are presented.
The human HepG2 cell line is frequently used as an in vitro model for the human liver. Based on results with the human hepatocytes (see above) the mRNA expression levels of the five potential AhR target genes (Aldh3a1, HSD17b2, Tiparp, Cd36 and Ahrr) and established AhR-dependent Cyp1a1, Cyp1a2 and Cyp1b1 were determined in HepG2 hepatoma cells. TCDD was able to induce Aldh3a1, Cyp1a1 and Cyp1b1, but did not induce HSD17b2, Tiparp, Cd36 and Ahrr gene expression indicating significant differences between this cell line and human hepatocytes. To obtain a better idea about the possible differences in overall gene expression after TCDD exposure in the HepG2 cell line and human hepatocytes microarray studies were also done. Surprisingly, there was a very limited overlap between the gene expressions in both in vitro human hepatic cell systems, with only 26 out of 345 commonly expressed (See figure 4.1.19). The results question the validity of the HepG2 cells to study dioxin-like toxicity as a surrogate for the human liver.
RELATIVE EFFECT POTENCY ESTIMATES IN AN EXPOSED HUMAN POPULATION.
We sought to determine relative effect potencies (REPs) for systemic human concentrations of dioxin-like mixture components. Although many studies using human cell lines or primary cells have been published to date, human in vivo data that may contribute to the TEF concept have not been published previously. In an attempt to fill this gap, we examined cross-sectional data on relationships between individual mixture components in a population exposed to a mixture of organochlorines and two thyroid outcomes, i.e. thyroid volume and free thyroxine (FT4), and CYP1A1 and 1B1 mRNA levels in blood lymphocytes. We used a benchmark concentration and a regression-based approach to compare the strength of association between each dioxin-like compound and the thyroid and the cytochrome P450 endpoints in 320 adults residing in an organochlorine-polluted area of eastern Slovakia. Based on these results, we estimated the REPs of PCDD, PCDF, and DL-PCB congeners in adult humans (Table 4.1.4). The Pearson correlation between all our REP data and WHO TEFs was y=0.753x-0.0125; R2=0.6604; r=0.813; p<0.001. Our REPs calculated from thyroid end points realistically reflect human exposure scenarios because these are based on chronic, low-dose human exposures and on biomarkers reflecting body burden. Compared with previous results, our REPs suggest higher sensitivity to the effects of dioxin-like compounds.
EU-SYSTEQ EXPERT MEETING.
At the end of the EU-SYSTEQ project an Expert Meeting was organized to discuss the use systemic toxic equivalency factors for dioxin and PCBs based on the results obtained in this project. During this meeting the results of EU-SYSTEQ were discussed and evaluated in order to establish whether or not separate systemic TEFs for these compounds can and should be used in future risk assessment and management processes. The following questions were brought forward to a number of external dioxin experts1 and the partners of the EU-SYSTEQ project:
1) Do the EU-SYSTEQ data provide additional sensitive and novel biomarkers for effects and exposure compared with the more traditional ones like CYP1A1 in rodents and humans?
Answer and Conclusion: Within the EU-SYSTEQ project, the expression of other genes besides CYP1A1, 1B1 and 1A2 have been studied and identified for various congeners and various experimental models (in vitro and in vivo). AhRR (AhR repressor), TIPARP, ALDH3A1, Notch1 and TNFAIP8L3 were identified as potential candidates for novel biomarkers across a range of congeners. From these candidates, AhRR expression was considered to be sensitive in certain human models tested, representing a primary AhR response. Among the novel candidates, depending on the tissue, AhRR is comparable in sensitivity with CYP1A1 and 1B1. AhRR, TIPARP and ALDH3A1 are induced in an AhR-dependent manner and may be involved in AhR-mediated biological effects. As such they may be used as biomarkers of systemic exposure and/or effect. Based on the data provided by the EU-SYSTEQ project CYP1A1 remains the most responsive biomarker for human exposure (See table 4.1.5)
2a) Do the EU-SYSTEQ results provide sufficient support for the development of “systemic” REPs/TEFs or is the their deviation from “intake” REPs/TEFs too limited in view of other uncertainties in the TEF concept?
2b) If “systemic” TEFs need to be developed, are the EU-SYSTEQ results already sufficient to establish interim values for these “systemic” TEFs?
Answers and Conclusions: The results from EU-SYSTEQ do not warrant the development of separate “systemic” TEFs for the congeners tested considering that intake and systemic plasma REPs are similar in the two in vivo rodent models tested. An exception is found for 4-PeCDF for which the plasma-based REPs are higher than those based on intake. This might be related to sequestration of 4-PeCDF in rodent liver due to the degree of induction of and binding to CYP1A2. It is currently unknown if this represents the human background situation. Whilst the same direction of shift is to be expected in humans, the magnitude is less certain.
3) Are “systemic” REPs from in vivo studies closer to “in vitro” REPs than to “intake” REPs based on the EU-SYSTEQ results?
Answer and Conclusion: The EU-SYSTEQ data show that in vitro data could be an adequate replacement for in vivo systemic data for the estimation of REPs for the compounds tested. It may be hypothesized that similar relationships between relevant in vitro test systems and in vivo data may apply for other metabolically persistent compounds as well.
4) Do the EU-SYSTEQ data provide sufficient support to conclude that the relative potencies of some congeners are sufficiently different in humans compared to the WHO TEFs, which are mainly based on rodent studies?
Answer and Conclusion: Multivariate analysis indicated different response patterns between rodent and human in vitro models. For some compounds the differences in REPs were greater than one order of magnitude between rodent and human models, but also between WHO-TEF and human in vitro response. Notably, the human in vitro REPs of PCB126 are up to two orders of magnitude lower than the present WHO-TEF. The current WHO-TEF (0.1) is almost exclusively derived from long-term rodent in vivo studies. Based on similarities observed between rodent in vitro and in vivo REPs, a similar human in vitro – in vivo relationship may be expected. The combined data for PCB126 from literature and the EU-SYSTEQ project indicate that a human specific TEF for PCB126 is justifiable. Additionally, the human in vitro EU-SYSTEQ data provide support to suggest re-evaluation of WHO-TEFs for some other congeners for the human situation.
Potential Impact:
SOCIO-ECONOMIC IMPACT AND SOCIETAL IMPLICATIONS OF EU-SYSTEQ
Systemic TEFs?
The major objective of the EU-SYSTEQ project was to establish if there was a societal need for the specific development of systemic toxic equivalency factors (TEFs) for the human risk assessment. In order to establish, if specific (sensitive) populations have been exposed to unacceptable high levels of dioxin-like compounds (DLCs), there are basically two approaches available to apply the TEF methodology. The first approach is the assessment of the oral ingestion via e.g. food sources and it requires determination of individual human exposure to these sources. In this process differences in life-style and genetic make-up can influence the variability in body burden between human individuals strongly. In practice it can includes a variability range that can easily stretch one order of magnitude in exposure and associated body burdens. The second approach is the actual measurement of the individual body burden of DLCs via biological matrices such as blood, tissues and human milk. The latter approach provides a more accurate identification of differences between individuals and populations.
The present TEF methodology has been derived from in vivo experimental studies using mainly oral dosages, which provided a range of relative effect potencies (REPs). From these REPs a compound specific uptake-TEF values have been derived, e.g. by the World Health Organization (WHO) that are used for the human risk assessment. With respect to this application of uptake-TEFs an important fundamental question addresses the validity of their use for biological matrices like blood or tissues. At the start of the EU SYSTEQ project the use of these uptake-TEFs for systemic situations had not been properly validated.
The EU-SYSTEQ project showed with two different animal models that the use of uptake-TEFs for systemic situations (systemic-TEFs) generally does not include a deviation, which is outside the assumed +/- half order of magnitude around a TEF value, as suggested by the WHO. Consequently, the results of the EU-SYSTEQ project indicate, that the use of uptake-TEFs for systemic situations do not require further extensive toxicity testing with DLCs. It is concluded that the present intake-TEFs can be used for human risk assessment if applied to e.g. blood and tissues by organizations like the EU, WHO and US-EPA
Novel biomarkers for DLCs
The EU-SYSTEQ project also studied other potential biomarkers for exposure and toxicity of DLCs. This was done in a number of in vivo studies with genetically modified mouse, but also with a large range of primary human cells from the liver, skin and lymphocytes, and range of immortal cell lines from human and animal origin using gene-arrays. These experiments indicated a number of potential novel biomarkers that may be of use in the future risk assessment. However, the classical biomarker, CYP1A1 enzyme induction, remained the most sensitive endpoint for exposure to DLCs. Therefore, it is concluded that there is no need to develop further novel endpoints as biomarkers for DLC exposure and risk assessment
Are in vitro data closer to systemic toxicity data?
Within the EU-SYSTEQ project a wide range of in vitro experiments was performed to determine REPs of different DLCs. These in vitro data were compared with systemic- and uptake-REPs from the in vivo studies with mouse and rat that were also done in this project. This comparison was performed, because it was hypothesized that in vitro data would actually be closer to systemic toxicity data than data derived based on oral uptake. Frequently, aspects like gastro-intestinal uptake and metabolism cause differences in responses observed in in vitro models compared with in vivo studies. These physiological factors are usually missing in in vitro models. In the EU-SYSTEQ model there was an opportunity to test this hypothesis for three potent DLCs from which sufficient in vivo dose-response information was available for the rodent experiments. Although results are not fully equivocal, it was found for two out of three DLCs that in vitro results were closer to the systemic toxicity data compared with those based on oral uptake. Although this observation provides further support for the use of in vitro models in the risk assessment process, further studies are also necessary to elucidate the compound specific differences.
Applicability of animal TEFs for human risk assessment.
Within the SYSTEQ project it was studied to which extent the animal derived toxicity data were actually representing the human sensitivity. Comparative studies using primary cell types or cell lines from humans and rodents were done with a wide range of different PCDDs, PCDFs and PCBs. It was shown that in vitro responses of PCDDs and PCDFs derived from rodent cell models reasonably predicted the REPs in human in vitro systems. However, large deviations were found between human and rodent in vitro responses for a number of dioxin-like PCBs (PCB 126 and the mono-ortho PCBs). In all human cell models used in the EU-SYSTEQ project, dioxin-like responses of these PCBs were much lower than those observed in comparable rodent cell models. These PCBs present a significant part of the human exposure to DLCs, e.g. via food of animal origin or human milk. Thus, these EU-SYSTEQ results with dioxin-like PCBs clearly indicate that the human risks of DLCs via food may be largely overestimated. This observation has significant socio-economic implications.
List of Websites:
Contact details
EU- SYSTEQ Coordinator: Prof. dr. Martin van den Berg
Institute for Risk Assessment Sciences (IRAS), Utrecht University, PO Box 80177, NL 3508 TD Utrecht, The Netherlands |
T + 31 30 2535265; M 31 6 245 76590; e-mail: mailto:m.vandenberg@uu.nl
Website www.systeqproject.eu
e-mail info@systeqproject.eu