Risk assessments in Food Safety and Health

Biometris develops quantitative methods for:

• risk assessment for human health and the environment
• decision making and risk/benefit analysis

Specific applications in the areas of:

• pesticides and other chemical substances
• novel foods such as genetically modified organisms (GMOs) and nanomaterials

For this we use:

• tiered approaches, from simple deterministic to fully probabilistic models
• implementation in stand-alone software or web-based tools

Science-based risk and benefit assessment

It is difficult to assess true health benefits and risks directly because of the complex chain from primary agricultural products (where most measurements are made) to total (daily) intake by individuals over all food products. Also the health effects on individual consumers are often difficult to quantify precisely using epidemiological or toxicological data. Current risk assessment methodology is mostly limited to single substances and simple variables. There is insufficient integration of the exposure assessment and hazard characterisation in the risk assessment process. Nevertheless, risk managers have to take calculated decisions, taking all risks and benefits into account.

Therefore, science-based risk and benefit assessment and rational derivation of recommendations is one of the top priorities of European food policy, both in the areas of traditional substances (e.g. pesticides) as in the area of novel foods such as GMOs. More realistic assessments can lead to bring both health and economic benefits.

New science based risk asssessment tools

At Biometris we use probabilistic methods to develop new science based risk asssessment tools. For a truly science-based risk assessment it is necessary to integrate as many aspects of the food chain as possible into a dietary exposure modelling framework. The framework should be able to handle irregular/limited data, to extrapolate to unobserved high-risk events, and to combine multiple risk (or benefit) factors. Furthermore, there is a need in toxicogenomics to find biomarkers of (internal) exposure and link these to dietary (external) exposure and to biomarkers of effects. In GMO safety assessment further guidance is needed on the application of statistical methods. Development of data analysis and modelling approaches has been recognised as essential considering the exploding data quantities in the field of food safety.

Practical problems to be solved

Biometris focuses on case studies where practical risk assessment problems are to be solved, but the purpose is to develop general methodology. Examples of recent research are:

• Developing an extensive framework for probabilistic exposure assessment for chronic (long-term) and acute risks and benefits: the Monte Carlo Risk Assessment (MCRA) consists of modelling software which can be connected to an international e-platform of distributed databases.
• Integrating exposure and effect modelling: new methods have been developed which allow the integration of exposure and hazard characterisation (generally based on dose-response studies). This has led to the introduction of individual margins of exposure and the model is currently applied to data from the European e-platform.
• Probabilistic health impact assessment: recently methodology has been developed which allows integration of multiple compounds, multiple effects and multiple health impact levels into a single framework. The concept of probabilistic health impact assessment is based on combining distributions of individual margins of exposure.
• Risk-benefit: quantification of health impact from multiple effects (beneficial or hazardous) due to changed exposure to multiple substances.
• Uncertainty analysis: all risk assessment methodology developed at Biometris has a strict division between (intrinsic) variability within populations and uncertainty due to lack of knowledge/ measurements. This approach can help risk managers in their decision making and to quantify the major sources of uncertainty. Research efforts can then be directed to further decrease the uncertainty.
• Modelling rare events: application of extreme value models in exposure assessment and integrated food risk assessment.
• Toxicogenomics: exploration of possibilities to use microarray or profiling data as biomarkers of exposure and/or biomarkers of effect.
• Methods for GMO safety assessment: contributions to advancing appropriate statistical modelling techniques (e.g.  proof of hazard vs. proof of equivalence).

Research for authorities and companies

The major market for this type of research are risk/benefit assessors/managers at the national and European authority level. Additionally, food chemical industry follows the leads given them by these authorities. Both parties can gain by applying new risk assessment models (e.g. less risk, better understood risks, better integration of new knowledge, comparable measures for different effects). Results are brought to the market in various forms, e.g. as research underlying internet softwareapplications such as the Monte Carlo Risk Assessment (MCRA) system, and stand-alone applications such as the newly developed Integrated Probabilistic Risk Assessment (IPRA) model.

Biometris collaborates within Wageningen University & Research with the Department of Human Nutrition, on a national scale with RIVM, and with partners in EU projects, EFSA and ISO working groups. This enables a scientifically leading role in national and international collaborations with RIVM, NVWA, EFSA, European Commission and the agrochemicals industry.