Time extrapolation in regulatory risk assessment: The impact of study differences on the extrapolation factors.

In human risk assessment, time extrapolation factors (EFs) account for differences in exposure duration of experimental studies. We calculated EFs based on N(L)OEL (no (lowest) observed effect level) ratios, dividing shorter-term by longer-term values. The 'oral' datasets comprised 302 EFs (subacute-subchronic) and 1059 EFs (subchronic-chronic). The 'inhalation' datasets contained 67 EFs (subacute-subchronic) and 226 EFs (subchronic-chronic). The experimental EF distribution oral:subchronic-chronic showed that study parameters like deviation in dose selection and spacing influence mainly the data variance. Exclusion of these influences led to a dataset representing more realistically the difference of N(L)OELs with prolonged treatment. This dataset showed a GM of 1.5, indicating that the impact of a longer treatment period on the study N(L)OEL is on average not high. A factor of 1.5 seemed to be also sufficiently conservative for subacute-subchronic and subchronic-chronic extrapolation (inhalation or oral exposure). EFs for groups of similar compounds did not differ, but for compounds with low and high NOEL values. Relatively toxic compounds (GM 1) might thus not require time extrapolation. Within and between chemical variance was analysed in the dataset oral:subchronic-chronic (GSD 4.8). The variance between chemicals should be considered within extrapolation by selecting an appropriate percentile for which a chemical variance factor is suggested. In risk assessment, often a combination of EFs is required. Our analysis indicates that such a combination will result in an accumulation of non-toxicological variance and therefore unrealistically high EFs. Further evaluations are needed to identify appropriate chemical variance factors for these situations.

Evaluation of route-to-route extrapolation factors based on assessment of repeated dose toxicity studies compiled in the database RepDose®.

The majority of repeated dose toxicity studies are available for the oral route. For risk assessment, however, data are needed from the relevant exposure route, i.e. inhalation or dermal. Instead of conducting additional animal studies, route-to-route (R2R) extrapolation may be performed. To explore uncertainties associated with this approach, we derived extrapolation factors (EF) based on no/lowest effect levels (NOELs/LOELs) in the Fraunhofer RepDose® database. For R2R extrapolation oral-to-inhalation 246 study pairs on 110 chemicals were analyzed. Systemic effects triggered the LOELs in the underlying inhalation studies in 49.2%, local effects in 21.9% and both local and systemic effects in 30.9% of the data pairs. For systemic effects in inhalation studies an EF of 2.2 (95% confidence interval: 1.2-3.1) was derived, for local effects, the EF was 4.4 (95% confidence interval: 2.0-8.6), and the EF without distinguishing local or systemic effects (any EF) was 3.2 (95%, confidence interval: 1.7-5.0). Calculation with LOELs instead of NOELs, exposure duration and intrinsic properties of the chemical (toxicity or physicochemical properties) did not influence the EF significantly. For R2R extrapolation oral-to-dermal 46 study pairs on 28 chemicals were analyzed. An overall EF of 0.4 (95%, confidence interval: 0.2-0.9) was obtained. Here, we found a significant difference of EFs for low and high toxic chemicals. Overall, we conclude that reliable systemic NOELs/LOELs can be obtained for inhalation studies via R2R extrapolation from oral studies. Based on the data for any EF we propose to use an EF of 3, which covers also the uncertainty that unexpected local effects may occur in an inhalation study. For the dermal route, our dataset was too small to allow general conclusions, but the results so far do suggest that the current ECHA guidance is conservative when assuming that dermal absorption is as high as oral absorption.

Inhalation TTC values: A new integrative grouping approach considering structural, toxicological and mechanistic features.

The present publication describes an integrative grouping concept to derive threshold values for inhalation exposure. The classification scheme starts with differences in toxicological potency and develops criteria to group compounds into two potency classes, namely toxic (T-group) or low toxic (L-group). The TTC concept for inhalation exposure is based on the TTC RepDose data set, consisting of 296 organic compounds with 608 repeated-dose inhalation studies. Initially, 21 structural features (SFs) were identified as being characteristic for compounds of either high or low NOEC values (Schüürmann et al., 2016). In subsequent analyses these SF groups were further refined by taking into account structural homogeneity, type of toxicological effect observed, differences in absorption, metabolism and mechanism of action (MoA), to better define their structural and toxicological boundaries. Differentiation of a local or systemic mode of action did not improve the classification scheme. Finally, 28 groups were discriminated: 19 T-groups and 9 L-groups. Clearly distinct thresholds were derived for the T- and L-toxicity groups, being 2 × 10(-5) ppm (2 µg/person/day) and 0.05 ppm (4260 µg/person/day), respectively. The derived thresholds and the classification are compared to the initial mainly structure driven grouping (Schüürmann et al., 2016) and to the Cramer classification.

Sensitivity of different generations and developmental stages in studies on reproductive toxicity.

Numerous studies on reproductive toxicity are expected to be necessary under the EU program on Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH). Therefore, it is important to analyse existing testing strategies including also the recently implemented extended one-generation reproduction toxicity study (EOGRTS, OECD guideline 443). For this purpose the responsiveness of the different generations and developmental stages in studies on reproductive toxicity is analysed and critical targets of reproductive toxicity are identified by using the Fraunhofer FeDTex database. The F1 generation is identified as most responsive generation in more than 50% of one-generation and multi-generation reproduction studies. Within the F1 generation the adult stage is mostly affected compared to the prenatal or postnatal stage. The target analysis in F1 has revealed alterations in body weight as highly sensitive for all developmental stages. Other important targets are the liver, kidney, testes, prostate, sperm parameters as well as developmental landmarks. The findings in the F2 generation have shown a higher responsiveness than F1 only in 3% of the studies. Although in 29 studies new effects are observed in F2 offspring compared to F1 irrespective of dose levels, overall no severe new effects have emerged that would change classification and labelling and justify an F1 mating. The presented data support the importance of F1 for risk assessment and demonstrate that the study design of the EOGRTS is a suitable alternative to two-generation studies. However, compared to a conventional one-generation study the EOGRTS may identify additional effects but will change risk assessment with respect to NOELs only in rare cases.

Relevance of non-guideline studies for risk assessment: the coverage model based on most frequent targets in repeated dose toxicity studies.

A common challenge for human risk assessment is the quality of the available animal studies. Non-guideline studies are often limited for different aspects of study design and documentation. Within this publication the relevance of a limited scope of examination is discussed. Preliminary analyses of the RepDose database have shown that liver, body weight, kidney and clinical symptoms are frequently affected in oral repeated dose toxicity in rats and mice (Bitsch et al., 2006), while many other targets are seldom affected. As most of these targets are investigated frequently also in non-guideline studies, it is likely that they provide a reliable NOEL, although the full spectrum of endpoints has not been covered. Based on RepDose data we investigate the relevance of individual targets for determining the LOEL and the consequences for risk assessment. The resulting coverage model for subchronic oral rat studies includes up to six targets and an additional assessment factor for LOEL extrapolation. It can be applied to assess the reliability of non-guideline studies with respect to the scope of examination. Furthermore the application of the coverage model to other databases will increase and/or specify the chemical domain and reveal respective targets as well as effects.

Improvement of the Cramer classification for oral exposure using the database TTC RepDose--a strategy description.

The present report describes a strategy to refine the current Cramer classification of the TTC concept using a broad database (DB) termed TTC RepDose. Cramer classes 1-3 overlap to some extent, indicating a need for a better separation of structural classes likely to be toxic, moderately toxic or of low toxicity. Groups of structurally similar compounds of high toxicity in Cramer class 1 and of moderate to low toxicity in Cramer class 3 were identified and reassigned to the appropriate Cramer class according to their observed toxicological potency in in vivo studies. This refinement results in a better discrimination of Cramer classes 1 and 3 and an increased number of substances in Cramer class 2. The TTC values are 8.7 µmol/person/d (class 1), 6.72 µmol/person/d (class 2) and 0.28 µmol/person/d (class 3). Assuming a median molecular weight of 220 g/mol for the compounds of the TTC RepDose DB, the corresponding TTC values are 1930, 1478 and 63 µg/person/d for classes 1, 2 and 3 respectively. The derived thresholds are close to the TTC values initially proposed by Munro with 1800, 540 and 90 µg/person/d for classes 1, 2 and 3 respectively. Additional structural classes are discussed with a view to further refinement of the current Cramer classification scheme.

Evaluation of time extrapolation factors based on the database RepDose.

In chemical risk assessment for many substances only short-term animal studies are available for the evaluation of long-term human exposure. Therefore usually extrapolation factors (EF) are used to extrapolate NOAELs from existing short-term studies to NOAELs for long term exposure. In this report time EFs are derived, based on NOEL/C or LOEL/C ratios (short term N(L)OEL/long term N(L)OEL) from the large datasets of the database RepDose (www.fraunhofer-repdose.de) on repeated dose toxicity for oral or inhalation administration. Within a tiered approach several sources of variability, e.g. use of LOEL/C ratios or differences in dose spacing were analyzed and if needed subsequently excluded. The reduction of data variability resulted in "final" EFs datasets, which are as far as possible based on compound-specific, time-dependent differences in toxicity. For distribution functions of oral repeated dose toxicity studies characterised by GM, GSD and 90th percentiles the following data are obtained: subacute-to-subchronic - GM 1.3, GSD 2.4, 90th 4.0, subacute-to-chronic - GM 3.4, GSD 3.7, 90th 18.2, and subchronic-to-chronic - GM 1.4, GSD 2.1, 90th 3.6. The number of data for inhalation exposure is limited, but with regard to systemic toxicity the derived EFs confirm the respective oral EFs.

Evaluation of inhalation TTC values with the database RepDose.

The thresholds of toxicological concern (TTCs) define limit values for substances of unknown toxicity below which dietary intake is considered to be of no concern to human health. The TTC concept has already been used for risk assessment of e.g. food contaminants or flavoring substances and is in discussion to be applied to other classes of compounds such as cosmetic ingredients, household products, non-relevant metabolites in drinking water, and impurities in pharmaceuticals. The present publication aimed to evaluate whether the current TTC concept can also be applied to define limit values for inhalation exposure, using a data set of 203 industrial chemicals from the database RepDose. It has been shown, that the NOEC values in classes 1, 2, and 3 are distributed over six orders of magnitude resulting in a considerable overlap between the distribution curves for the three classes. Inhalation thresholds for Cramer classes 1 (compounds likely to be of low-toxicity), 2 (compounds likely to be of moderate toxicity), and 3 (compounds suspect for high toxicity) were analyzed close to the approach described by Munro for oral TTCs. The 5th percentiles NOEC of Cramer classes 1-3 result in thresholds of 1.5×10(-3) ppm for Cramer class 1 and 2.2×10(-5) ppm for Cramer class 3. A threshold could not be derived for class 2 because of the small number of compounds available. If calculated as body doses, the inhalation thresholds for classes 1 and 3 (71 and 4 µg/person/d, respectively) are considerably lower than the oral thresholds derived by Munro (1800 and 90 µg/person/d). It has been shown that one reason for this difference is the high sensitivity of the respiratory tract to local effects. In a next step, the values obtained were further refined. If organophosphates or compounds with structural alerts for genotoxicity are excluded, the TTC in Cramer class 1 increases, whereas the TTC in Cramer class 3 remains the same. Based on these analyses two inhalation TTCs for non-genotoxic compounds are proposed: 3.6×10(-3) ppm (180 µg/person/d) for Cramer class 1 and 2.4×10(-5)ppm (4 µg/person/d) for Cramer class 3.

REPDOSE: A database on repeated dose toxicity studies of commercial chemicals--A multifunctional tool.

A database for repeated dose toxicity data has been developed. Studies were selected by data quality. Review documents or risk assessments were used to get a pre-screened selection of available valid data. The structure of the chemicals should be rather simple for well defined chemical categories. The database consists of three core data sets for each chemical: (1) structural features and physico-chemical data, (2) data on study design, (3) study results. To allow consistent queries, a high degree of standardization categories and glossaries were developed for relevant parameters. At present, the database consists of 364 chemicals investigated in 1018 studies which resulted in a total of 6002 specific effects. Standard queries have been developed, which allow analyzing the influence of structural features or PC data on LOELs, target organs and effects. Furthermore, it can be used as an expert system. First queries have shown that the database is a very valuable tool.

[Reference values for indoor air: dearomatized hydrocarbon solvents (C(9)-C(14))]. / Richtwerte für die Innenraumluft: Aromatenarme Kohlenwasserstoffgemische (C9-C14).

To protect public health the German Joint Working Group on Indoor Guidelines of the Federal Environmental Protection Agency and the States' Departments of Health is issuing indoor air guideline values based on a fixed procedure published in 1996. Regarding dearomatized hydrocarbon solvents/white spirits (DAWS--CAS-No. 64742-47-8, 64742-48-9, 64742-88-7, 64741- 65-7) no human data are available. From animal studies, neurotoxicity, developmental toxicity and reproductive toxicity were identified as critical endpoints. For risk evaluation the Hass et al. (2001) study was used as the pivotal study. Based on effects at 4680 mg DAWS/m(3) for the endpoint developmental toxicity, the lowest adverse effect level for chronic exposure is assessed as 400 mg DAWS/m(3). By applying an interspecies factor of 10, an intraspecies factor of 10 and an additional factor 2 referring to the special physiology of children (higher breath rate compared to adults) a so-called health hazard value of 2 mg DAWS/m(3) indoor air and a so-called health prevention value of 0.2 mg DAWS/m(3) are obtained.

Mutagenicity testing with transgenic mice. Part I: Comparison with the mouse bone marrow micronucleus test.

As part of a larger literature study on transgenic animals in mutagenicity testing, test results from the transgenic mutagenicity assays (lacI model; commercially available as the Big Blue(R) mouse, and the lacZ model; commercially available as the Mutatrade markMouse), were compared with the results on the same substances in the more traditional mouse bone marrow micronucleus test. 39 substances were found which had been tested in the micronucleus assay and in the above transgenic mouse systems. Although, the transgenic animal mutation assay is not directly comparable with the micronucleus test, because different genetic endpoints are examined: chromosome aberration versus gene mutation, the results for the majority of substances were in agreement. Both test systems, the transgenic mouse assay and the mouse bone marrow micronucleus test, have advantages and they complement each other. However, the transgenic animal assay has some distinct advantages over the micronucleus test: it is not restricted to one target organ and detects systemic as well as local mutagenic effects.

Structure-based thresholds of toxicological concern (TTC): guidance for application to substances present at low levels in the diet.

The threshold of toxicological concern (TTC) is a pragmatic risk assessment tool that is based on the principle of establishing a human exposure threshold value for all chemicals, below which there is a very low probability of an appreciable risk to human health. The concept that there are levels of exposure that do not cause adverse effects is inherent in setting acceptable daily intakes (ADIs) for chemicals with known toxicological profiles. The TTC principle extends this concept by proposing that a de minimis value can be identified for many chemicals, in the absence of a full toxicity database, based on their chemical structures and the known toxicity of chemicals which share similar structural characteristics. The establishment and application of widely accepted TTC values would benefit consumers, industry and regulators. By avoiding unnecessary toxicity testing and safety evaluations when human intakes are below such a threshold, application of the TTC approach would focus limited resources of time, cost, animal use and expertise on the testing and evaluation of substances with the greatest potential to pose risks to human health and thereby contribute to a reduction in the use of animals. An Expert Group of the European branch of the International Life Sciences Institute-ILSI Europe-has examined the TTC principle for its wider applicability in food safety evaluation. The Expert Group examined metabolism and accumulation, structural alerts, endocrine disrupting chemicals and specific endpoints, such as neurotoxicity, teratogenicity, developmental toxicity, allergenicity and immunotoxicity, and determined whether such properties or endpoints had to be taken into consideration specifically in a step-wise approach. The Expert Group concluded that the TTC principle can be applied for low concentrations in food of chemicals that lack toxicity data, provided that there is a sound intake estimate. The use of a decision tree to apply the TTC principle is proposed, and this paper describes the step-wise process in detail. Proteins, heavy metals and polyhalogenated-dibenzodioxins and related compounds were excluded from this approach. When assessing a chemical, a review of prior knowledge and context of use should always precede the use of the TTC decision tree. The initial step is the identification and evaluation of possible genotoxic and/or high potency carcinogens. Following this step, non-genotoxic substances are evaluated in a sequence of steps related to the concerns that would be associated with increasing intakes. For organophosphates a TTC of 18microg per person per day (0.3 microg/kg bw/day) is proposed, and when the compound is not an OP, the TTC values for the Cramer structural classes III, II and I, with their respective TTC levels (e.g. 1800, 540 and 90 microg per person per day; or 30, 9 and 1.5 microg/kg bw /day), would be applied sequentially. All other endpoints or properties were shown to have a distribution of no observed effect levels (NOELs) similar to the distribution of NOELs for general toxicity endpoints in Cramer classes I, II and III. The document was discussed with a wider audience during a workshop held in March 2003 (see list of workshop participants).

Inhalation toxicity of mineral particles: critical appraisal of endpoints and study design.

Many of the mineral particles that are of concern in regard to lung toxicity are poorly soluble particles (PSPs). They include biopersistent mineral fibers and dusts containing crystalline silica. The preparation of well-defined test particles of respirable size range and their characterization are an essential step that may require more time and effort than the toxicity study itself. For toxicity studies with mineral particles, an investigation of the toxicokinetics is recommended. Such an investigation will yield information that will help to interpret the results if dust overload conditions occur. For mineral particles such as crystalline silica and mineral fibers, an important endpoint is their potential carcinogenicity. The following parameters are important for the design of chronic toxicity studies, and for the prediction of severe chronic effects: lung retention of inhaled materials for assessing the accumulation of particles, persistent inflammation in lungs, persistent proliferation of epithelial lung cells, progressive fibrogenicity, and genotoxicity in the lung cells. These endpoints should indicate whether the materials investigated are of concern in the health effects on exposed humans, and in the effects of the mineral particles for which chronic studies may be required. In addition, this paper focuses on the effects of PSPs combined with fibers, and on the strategies for investigating the potential carcinogenicity of quartz-containing dusts.

IPCS conceptual framework for evaluating a mode of action for chemical carcinogenesis.

The International Programme on Chemical Safety (IPCS) is leading an activity to harmonize approaches to cancer risk assessment as a part of its larger project on the Harmonization of Approaches to the Assessment of Risk from Exposure to Chemicals. Through a series of workshops and the evaluation of case studies, a number of key components of risk assessments relating to harmonization were identified: transparency, terminology, weight of evidence, flexibility, and accessibility/communication. A major impediment to harmonization identified in the consideration of weight of evidence was the evaluation of mode of action. To address this need, a conceptual framework was developed, based on the general principles involved in considering the chemical induction of a specific tumor in animals. This is based partly on the Bradford Hill criteria for causality as modified by Faustman et al. (1997) for developmental toxicity. The framework is described in this paper followed by a worked example. It is recognized that the framework addresses only one stage in the overall characterization of hazard to humans of chemical carcinogens. Another important but separate step is the assessment of relevance to humans. This is a priority area for future work in this project.

Vinyl chloride: still a cause for concern.

Vinyl chloride (VC) is both a known carcinogen and a regulated chemical, and its production capacity has almost doubled over the last 20 years, currently 27 million tons/year worldwide. According to recent reports it is still a cause for concern. VC has been found as a degradation product of chloroethylene solvents (perchloroethylene and trichloroethylene) and in landfill gas and groundwater at concentrations up to 200 mg/m(3) and 10 mg/L, respectively. Worldwide occupational exposure to VC still seems to be high in some countries (e.g., averages of approximately 1,300 mg/m(3) until 1987 in one factory), and exposure may also be high in others where VC is not regulated. By combining the most relevant epidemiologic studies from several countries, we observed a 5-fold excess of liver cancer, primarily because of a 45-fold excess risk from angiosarcoma of the liver (ASL). The number of ASL cases reported up to the end of 1998 was 197 worldwide. The average latency for ASL is 22 years. Some studies show a small excess risk for hepatocellular carcinoma, and others suggest a possible risk of brain tumors among highly exposed workers. Lung cancer, lymphomas, or leukemia do not seem to be related to VC exposure according to recent results. The mutation spectra observed in rat and human liver tumors (ASL and/or hepatocellular carcinoma) that are associated with exposure to VC are clearly distinct from those observed in sporadic liver tumors or hepatic tumors that are associated with other exposures. In rats, the substitution mutations found at A:T base pairs in the ras and p53 genes are consistent with the promutagenic properties of the DNA adduct 1,N(6)-ethenoadenine formed from VC metabolites. Risk assessments derived from animal studies seem to overestimate the actual risk of cancer when comparing estimated and reported cases of ASL.

Commentary on the application of (Q)SAR to the toxicological evaluation of existing chemicals.

For ethical and financial reasons it is impossible to perform thorough toxicological testing for all of the more than 100,000 substances registered in the European Inventory of Existing Substances. It was therefore investigated whether the application of (quantitative) structure-activity relationships (QSAR) with commercially available computer programs could predict the toxicological profile and help identify those substances requiring priority toxicological testing. Whereas predictions with respect to complex endpoints such as carcinogenicity, chronic toxicity and teratogenicity are still disappointing, more reliable predictions should be forthcoming in the immediate future for sensitisation, mutagenicity and genotoxicity endpoints.

Ela 1.0--a framework for life-cycle impact assessment developed by the Fraunhofer-Gesellschaft. Part A: The conceptual framework.

The Fraunhofer-Gesellschaft has sponsored the development of a conceptual and flexible, computer aided tool to perform the impact assessment within LCA (life cycle assessment) for technical products and processes. The developed general framework "Ela 1.0" (environmental loads analysis) consists of four elements: the selection of appropriate impact categories, the categorization of emissions and wastes leaving the systems as well as of resource and energy consumption, the characterization and an analysis of the results of the impact assessment. The latter compares the product-based emissions with the total of emissions of a region such as Germany, the EU or OECD countries. The framework Ela 1.0 considers the environmental categories: global warming, ozone depletion, resource and energy consumption, wastes, eutrophication (including COD and BOD as measured parameters), acidification, ecotoxicity, ozone formation and human toxicity. The latter categories are handled by listing of precursors for ozone formation, and by listing of emissions scored according to their human hazard potential. The options, possibilities and limitations of the conceptual framework are presented in part A of a series of publications.

Chlorinated dibenzodioxins and dibenzofurans (PCDD/F) in blood and human milk of non occupationally exposed persons living in the vicinity of a municipal waste incinerator.

The concentrations of chlorinated dibenzodioxins and dibenzofurans (PCDD/F) in human blood and in milk from non-occupationally exposed persons living in the vicinity of a municipal waste incinerator were determined. As compared to background levels in the general population in Germany the results give no indication of an enhanced body burden of PCDD/F. This is in agreement with earlier investigations in the same area, showing normal background concentrations in soil, fruit and vegetables. In conclusion, no direct health hazard related to PCDD/F-emissions from a local municipal waste incinerator may be expected.

Searching for information on toxicological data of chemical substances in selected bibliographic databases--selection of essential databases for toxicological researches.

By using information from printed and online database guides, 18 online bibliographic databases (BD), which cover literature on toxicology were selected from 5 hosts. A search for literature containing information on three selected chemicals was carried out with each of the databases, and the number of documents relevant to toxicology found in them was compared by computer-assisted analysis. Some databases yielded very little information pertinent to toxicology, while others provided a considerable amount. In addition, the databases contained numerous duplicates (references common to more than one database). Most of the relevant documents could be obtained using only 8 of the 18 BDs selected. These databases are: Biosis Previews (BIOSIS), Chemical Abstracts (CA), Chemical Safety Newsbase (CSNB), Excerpta Medica (EMBASE), National Institute for Occupational Safety and Health (NIOSH), Scisearch, Toxicology Information Online (TOXLINE) and the former Toxicology Literature (TOXLIT).

Searching for information on chemical substances in selected biomedical bibliographic databases.

A method was developed which allows effective searching for information on chemical substances in databases. Several searches for chemicals in bibliographic databases were carried out to analyse the method of indexing chemical names. The recall rates of documents were compared to evaluate information resources as well as searching strategies. Recall rates of documents searched with the CAS Nos. were compared to those searched with substance name. It turned out that searching for substances is most specific and fastest with CAS Nos. They should always be used whenever possible. However, this is not sufficient in many BDs, making an additional search using chemical names necessary. Specific search options that have to be considered for each database are reported.