This project will spotlight one area out of the huge field of toxicogenomics, i.e. cancer genomics. Within this area it will focus on a special class of risks – the hotly debated potentials of electromagnetic fields (EMF´s) to cause cancer – for two reasons. First, a definitive focal point allows us to address one example in depth and maintain the research complexity to a manageable level. Second, EMF´s should prove to be a good exemplary case because ELF-EMF are classified by IARC as a possible carcinogen (IARC 2002) and new research provides some evidence that RF-EMF´s might cause genotoxic effects (Reflex 2004).

Five research questions will serve as guides for the performance of the project. The first two refer (1) to the gaps in risk assessments and (2) to the expected contributions of toxicogenomics to fill or bridge these gaps. The other three research questions refer to (3) challenges of toxicogenomics-based risk assessments on risk characterization, (4) issues of intuitive toxicogenomics, i.e. how lay people interpret toxicogenomic data, and (5) demands on risk communication.

Research question 1: What gaps and what uncertainties exist in the ELF- and RF-EMF risk assessment?

Most of the reviews on health effects of EMF´s do not focus on existing gaps and uncertainties – usually their focal point is the available scientific evidence for health effects (see IEGMP 2000, SSK 2001, BUWAL 2003, NRPB 2004). However, some gaps are often mentioned: for instance, the lack of a confirmed mechanistic model, the possibility of window effects, the suspected distinctive effect of pulsed EMF´s and the assumed special susceptibility of children (see ICNRIP 2005). These issues as well as others should be systematically compiled. In general, however, a systematic review of the gaps and uncertainties in EMF risk assessment is still missing.

Research question 2: How can toxicogenomics contribute to fill or bridge these gaps and to overcome the uncertainties in the ELF- and RF-EMF risk assessments?

Risk assessment is usually regarded as being comprised of four elements: (1) hazard identification, (2) dose-response assessment, (3) exposure assessment, and (4) risk characterization (NRC 1983). At present, discussions about the opportunities of toxicogenomics to improve risk assessments are focused on (a) the mode of effect, (b) on susceptible populations and sensitive life stages, and (c) on mixtures (EPA 2004). In the context of EMF risk assessment special emphasis should be given to genetic differences influencing phenomena such as DNA repair, chromatin structure, cell cycle control and apoptosis that can modify the response of cells to EMF. Another important aspect is the question whether EMFs can act as co-carcinogens together with other genotoxic and/or non-genotoxic carcinogens (Scarfi 2001).

Research question 3: How can the current practice of cancer risk characterization be improved upon in order to take into account toxicogenomic data?

Toxicogenomics studies suggest that altered gene expressions often are predictive for phenotypic alterations. The key issue is to determine which gene expressions can provide reliable and valid indicators of toxicity at earlier time-points and at lower doses than traditional toxicology parameters. Research question 3 focuses on how to integrate this information into cancer risk characterization, for instance in the IARC system that is mainly based on epidemiologic studies and animal experiments. The various proposals of risk characterization will be taken into account (IARC 1999, EPA 2000, NHMRC 2000, SSK 2001, Neutra et al. 2002, Wiedemann et al. 2005).

Research question 4: How do lay people perceive and interpret toxicogenomic data?

Risk perception[5] studies have shown that lay people´s risk perception differs from experts´ opinion and assessment, which is often a major reason for risk conflicts in society. However, lay people´s risk perception is not only characterized by deficits, cognitive limitations or biases. The important message from the psychometric studies on risk perception is that for lay people, risk is more than (some combination of) ‘exposure to’ and ‘probability of damage’, which are the core parameters of the technical concept of risk. On the contrary, risk perception is a process which is influenced by knowledge, values and feelings and is considerably dependent on the cultural context.

This suggests that lay people will interpret toxicogenomic data differently than experts. So far no scientific studies exist that deal with this research question.

However, for the field of predictive genetic testing, several studies have investigated patients’ reactions to DNA testing. But this line of research has mostly focused on the emotional consequences such as distress or anxiety (e.g. Claes et al. 2005, Shaw et al. 1999) and not on cognitive aspects such as understanding DNA test results (see Denayer et al. 1997, Welkenhuysen et al. 2001). The picture which emerges from these findings is far from being clear. For instance, in their meta-analysis Shaw et al. (1999) found evidence for increased psychological distress, anxiety, and depression in the short-term, but not in the long-term. However, available data are still limited.

Information for answering this question can also be drawn from research on “mental models” which tries to uncover the beliefs of lay people about the causal structure of the real life phenomena (e.g. Jungermann et al. 1988; Morgan et al. 2002), and from studies on “intuitive toxicology” (Kraus, Malmfors & Slovic 1992, MacGregor et al.1999, Mertz et al. 1998, Slovic et al. 1995). The studies revealed that lay people hold beliefs about noxious substances which systematically differ from the experts’ understanding of central concepts of risk assessment (e.g. dose-response relationship, exposure).

Research question 5: How must toxicogenomics-based risk communication be tailored in order to avoid misinterpretations, i.e. false sense of risk or safety?

Risk communication is more than translating expert knowledge into lay people-friendly formats. It should support risk evaluation competencies and should enable reflexive judgements about risk issues. We believe that there is a special need to support efforts to improve these competencies with respect to toxicogenomics, in the general public. A particular challenge for toxicogenomics based risk communication is the understanding of probabilistic information. Studies on the lay understanding of predictive DNA tests have documented the difficulties lay people encounter when they are confronted with this type of information (Claes et al. 2004, Denayer et al. 1997, Lippman-Hand 1979).

There are research findings suggesting that these difficulties in interpreting test results may also lead to a false sense of risk (Gigerenzer 2002) or safety (Denayer et al. 1997, Marteau et al. 1996).

While for the problem of presenting probabilistic information a number of suggestions have been made, e.g. using visualization (Lipkus & Hollands 1999) or natural frequency formats (Gigerenzer & Edwards 2003), which however still require further evaluation, the question of how to best correct improper mental models and beliefs about toxicogemonics must be based on the results of the analyses from Work Package 4 (see description of work packages below).

A completely different but not less important aspect, which is often neglected in risk communication, concerns the ethical and policy issues that are raised by new technological developments. In our project both sides – risk and ethics/policy – should be equally weighted.