- Email: [email protected]
42 The new EU chemicals policy
s14
Workshop W4. In vitro methods in toxicology
different risks are determined by psychological factors. In contrast, the technical risk estimates traditionally provided by experts do not influence people’s behaviours and responses in the same way as their risk perceptions. For example, a risk that people perceive to be involuntary in terms of their personal exposure is more threatening than one that they choose to take, even if the probability of harm is the same, or even less. Other concerns are very specific to particular hazard domains (for example, consider risk – and related - perceptions associated with transgenic organisms or endocrine disruptors). It is also important to communicate the difference between probability and variability associated with risk estimates. Risk communication with the public about a risk must take account of the actual concerns of the public (for example, potential for negative environmental impact, unintended human health effects, or vulnerable groups within the population). When the public want information about a risk, they prefer a clear message regarding risks and associated uncertainties, including the nature and extent of disagreements between different experts. Furthermore, societal priorities for risk mitigation activities may not align with those identified by expert groups. Dismissing the former as irrelevant may result increased distrust in the motives of regulators and industry, with consequences for public confidence in regulatory activities linked to public protection. Awareness and understanding of public concerns is also the basis for the development of an effective risk communication strategy, as these concerns should be explicitly addressed as part of the communication process. Examples relevant to risk communication and toxicology will be provided. 41
COMMUNICATING THE RISK OF ACRYLAMIDE IN FOOD THE LESSONS LEARNED
Leif Busk. National Food Administration, Uppsala, Sweden The presentation of data demonstrating increased levels of acrylamide in a variety of heated food commodities attracted great media interest in Sweden. This was followed by media coverage at a lower level in other European countries as well as in the United States. Consumers were probably confused by diverging message from different sources. The issue of acrylamide in food is complex. For example, risk characterisation is based on animal data with inherent problems to calculate risk levels for man. The mechanism of action needs further elucidation. The difference between doses that cause carcinogenic effect in animals and those that man is exposed to is lower than for any other single chemical carcinogen in food. Acrylamide is found in a variety of commodities, many regarded as staple foods. We have been exposed to acrylamide for a very long time. Hence, communication with consumers is very difficult since the issue involves both the more ¡normal¡ questions of validity of data in addition to difficult general scientific questions, e.g. the value of animal data versus human data. The issue is further complicated by the fact that a drastic reduction of the exposure would have a profound effect on life style and be extremely costly. This restricts the realistic available management options adding to the communication problem. In Sweden the issue became even more complicated. This was due to an initially hostile media reacting on the wordings in the invitation to the press conference where the data were presented. The workshop presentation will focus on the general problems of communicating complex issues as acrylamide in foods with special emphasis on what can be learned from this specific case.
W4 In vitro methods in toxicology 42
THE NEW EU CHEMICALS POLICY
G. Vollmer. European Chemical Bureau, Ispra, Italy The European Commission develops for the time being the drafts for a new chemicals regulation. It includes registration, evaluation and authorization of dangerous chemicals. This new legislation will not only give more burden to industry and authorities. It will increase the evaluation of use and exposure. It will also give an enormous
importance to alternative tests, which will replace the previous ones. These alternative tests will have to be carried out for some 20 000 chemicals. During the speech, the foreseen alternative methods, the request for further validation, the timetable for the implementation and the impact on industry will be outlined 43
IN VITRO METHODS IN TOXICOLOGY: SKIN IRRITATION
P.A. Botham. Syngenta Central Toxicology Laboratory, Alderley Park, Macclesfield, UK Several alternative in vitro methods for identifying skin irritants have been developed in the last 10 years, the most promising of which use either reconstituted human skin models or animal (e.g. pig or mouse) skin organ cultures. In 1998, the European Centre for the Validation of Alternative Methods (ECVAM) commissioned a pre-validation study of five methods which had successfully met a pre-set challenge of performance using 10 specified reference chemicals and had clearly defined protocols and prediction models. Two reconstituted human skin methods (Epiderm and EPISKIN) and one animal skin model (the mouse skin integrity function test, SIFT) performed well in Phases I and II of the study (intralaboratory reproducibility and protocol transfer) and proceeded to Phase III which assessed reproducibility and predictive ability in three independent laboratories using a set of 20 coded test chemicals (10 irritants and 10 non irritants). Intralaboratory reproducibility was again acceptable for all three methods but inter-laboratory reproducibility was acceptable only for EPISKIN. The predictive ability of all three methods was also inadequate (between 55–58% overall, with Epiderm and SIFT giving particularly high under-prediction rates and EPISKIN a high over-prediction rate). Subsequently, refinements were made to both the Epiderm and EPISKIN methods to improve their predictive ability. This work also showed that it was possible to establish a common protocol for the two methods which gave acceptable under and over-prediction rates. Modification of the statistical analysis of data from the SIFT model similarly improved its predictive ability. In 2003, ECVAM concluded that all three methods could proceed to a full validation study. This will be conducted in two phases and is scheduled for completion by the end of 2004. 44
APPLICATION OF IN SILICO TOOLS TO HUMAN HEALTH SAFETY ASSESSMENTS
Joanna S. Jaworska. Procter & Gamble, Eurocor, Central Product Safety Brussels, Belgium There is a renewed interest in (Quantitative) Structure Activity Relationships ((Q)SARs) owing to various societal and political pressures such as the drive to reduce animal testing and the proposed changes in EU chemical management regulations. Applications of (Q)SARs that are considered range from decision support in early phases of product development to regulatory decision making such as in risk assessment or classification. (Q)SARs are simplified (mathematical) representations of complex chemical-biological interactions and consequently (Q)SAR predictions are potentially more uncertain than the underlying test data. This imposes limitations on the acceptable use of (Q)SAR in chemical management and decision-making. Approaches to determine the acceptability of (Q)SAR predictions have been developed in the past, but because of their breadth and generality they have not been widely applied or respected by either (Q)SAR users or developers. As a consequence, decision making on the basis of existing models must be done with care and is subject to expert opinion as there is currently no framework for QSAR use and following it lack of confidence in these predictions. In general, (Q)SARs for environmental endpoints are founded on relatively larger quantitative databases with some mechanistic understanding, whereas the ability to predict local and systemic effects in humans is compromised by a lack of data and a limited understanding of the underlying mechanisms. To date, for human health endpoints, the models are often poor because the endpoints are expressed through many different mechanisms, are receptor mediated, are multi-stage processes comprising adsorption, distribution, metabolism and excretion (ADME) and are site specific. For certain human health endpoints, biological data are also not fully
Workshop W4. In vitro methods in toxicology
different risks are determined by psychological factors. In contrast, the technical risk estimates traditionally provided by experts do not influence people’s behaviours and responses in the same way as their risk perceptions. For example, a risk that people perceive to be involuntary in terms of their personal exposure is more threatening than one that they choose to take, even if the probability of harm is the same, or even less. Other concerns are very specific to particular hazard domains (for example, consider risk – and related - perceptions associated with transgenic organisms or endocrine disruptors). It is also important to communicate the difference between probability and variability associated with risk estimates. Risk communication with the public about a risk must take account of the actual concerns of the public (for example, potential for negative environmental impact, unintended human health effects, or vulnerable groups within the population). When the public want information about a risk, they prefer a clear message regarding risks and associated uncertainties, including the nature and extent of disagreements between different experts. Furthermore, societal priorities for risk mitigation activities may not align with those identified by expert groups. Dismissing the former as irrelevant may result increased distrust in the motives of regulators and industry, with consequences for public confidence in regulatory activities linked to public protection. Awareness and understanding of public concerns is also the basis for the development of an effective risk communication strategy, as these concerns should be explicitly addressed as part of the communication process. Examples relevant to risk communication and toxicology will be provided. 41
COMMUNICATING THE RISK OF ACRYLAMIDE IN FOOD THE LESSONS LEARNED
Leif Busk. National Food Administration, Uppsala, Sweden The presentation of data demonstrating increased levels of acrylamide in a variety of heated food commodities attracted great media interest in Sweden. This was followed by media coverage at a lower level in other European countries as well as in the United States. Consumers were probably confused by diverging message from different sources. The issue of acrylamide in food is complex. For example, risk characterisation is based on animal data with inherent problems to calculate risk levels for man. The mechanism of action needs further elucidation. The difference between doses that cause carcinogenic effect in animals and those that man is exposed to is lower than for any other single chemical carcinogen in food. Acrylamide is found in a variety of commodities, many regarded as staple foods. We have been exposed to acrylamide for a very long time. Hence, communication with consumers is very difficult since the issue involves both the more ¡normal¡ questions of validity of data in addition to difficult general scientific questions, e.g. the value of animal data versus human data. The issue is further complicated by the fact that a drastic reduction of the exposure would have a profound effect on life style and be extremely costly. This restricts the realistic available management options adding to the communication problem. In Sweden the issue became even more complicated. This was due to an initially hostile media reacting on the wordings in the invitation to the press conference where the data were presented. The workshop presentation will focus on the general problems of communicating complex issues as acrylamide in foods with special emphasis on what can be learned from this specific case.
W4 In vitro methods in toxicology 42
THE NEW EU CHEMICALS POLICY
G. Vollmer. European Chemical Bureau, Ispra, Italy The European Commission develops for the time being the drafts for a new chemicals regulation. It includes registration, evaluation and authorization of dangerous chemicals. This new legislation will not only give more burden to industry and authorities. It will increase the evaluation of use and exposure. It will also give an enormous
importance to alternative tests, which will replace the previous ones. These alternative tests will have to be carried out for some 20 000 chemicals. During the speech, the foreseen alternative methods, the request for further validation, the timetable for the implementation and the impact on industry will be outlined 43
IN VITRO METHODS IN TOXICOLOGY: SKIN IRRITATION
P.A. Botham. Syngenta Central Toxicology Laboratory, Alderley Park, Macclesfield, UK Several alternative in vitro methods for identifying skin irritants have been developed in the last 10 years, the most promising of which use either reconstituted human skin models or animal (e.g. pig or mouse) skin organ cultures. In 1998, the European Centre for the Validation of Alternative Methods (ECVAM) commissioned a pre-validation study of five methods which had successfully met a pre-set challenge of performance using 10 specified reference chemicals and had clearly defined protocols and prediction models. Two reconstituted human skin methods (Epiderm and EPISKIN) and one animal skin model (the mouse skin integrity function test, SIFT) performed well in Phases I and II of the study (intralaboratory reproducibility and protocol transfer) and proceeded to Phase III which assessed reproducibility and predictive ability in three independent laboratories using a set of 20 coded test chemicals (10 irritants and 10 non irritants). Intralaboratory reproducibility was again acceptable for all three methods but inter-laboratory reproducibility was acceptable only for EPISKIN. The predictive ability of all three methods was also inadequate (between 55–58% overall, with Epiderm and SIFT giving particularly high under-prediction rates and EPISKIN a high over-prediction rate). Subsequently, refinements were made to both the Epiderm and EPISKIN methods to improve their predictive ability. This work also showed that it was possible to establish a common protocol for the two methods which gave acceptable under and over-prediction rates. Modification of the statistical analysis of data from the SIFT model similarly improved its predictive ability. In 2003, ECVAM concluded that all three methods could proceed to a full validation study. This will be conducted in two phases and is scheduled for completion by the end of 2004. 44
APPLICATION OF IN SILICO TOOLS TO HUMAN HEALTH SAFETY ASSESSMENTS
Joanna S. Jaworska. Procter & Gamble, Eurocor, Central Product Safety Brussels, Belgium There is a renewed interest in (Quantitative) Structure Activity Relationships ((Q)SARs) owing to various societal and political pressures such as the drive to reduce animal testing and the proposed changes in EU chemical management regulations. Applications of (Q)SARs that are considered range from decision support in early phases of product development to regulatory decision making such as in risk assessment or classification. (Q)SARs are simplified (mathematical) representations of complex chemical-biological interactions and consequently (Q)SAR predictions are potentially more uncertain than the underlying test data. This imposes limitations on the acceptable use of (Q)SAR in chemical management and decision-making. Approaches to determine the acceptability of (Q)SAR predictions have been developed in the past, but because of their breadth and generality they have not been widely applied or respected by either (Q)SAR users or developers. As a consequence, decision making on the basis of existing models must be done with care and is subject to expert opinion as there is currently no framework for QSAR use and following it lack of confidence in these predictions. In general, (Q)SARs for environmental endpoints are founded on relatively larger quantitative databases with some mechanistic understanding, whereas the ability to predict local and systemic effects in humans is compromised by a lack of data and a limited understanding of the underlying mechanisms. To date, for human health endpoints, the models are often poor because the endpoints are expressed through many different mechanisms, are receptor mediated, are multi-stage processes comprising adsorption, distribution, metabolism and excretion (ADME) and are site specific. For certain human health endpoints, biological data are also not fully