Computer models predicting the effect of chemicals on humans

Friday 30 Mar 12

In future, mathematical computer models will help researchers to assess the risks associated with combining chemical substances in foods. The models will help to improve how we assess the action of chemical substances in our bodies. These are the perspectives of a PhD project which Trine Klein Reffstrup has completed at the National Food Institute, Technical University of Denmark.

Every day we are exposed to a wide range of chemical substances from the foods we consume. But what happens when the different substances meet and are combined in so-called chemical cocktails in our organs, tissue and blood? Do the chemicals destroy each other? Do they become more toxic? Are they absorbed by the body? Do they react in the same way in our various organs?

In her PhD project, Trine Klein Reffstrup has studied how mathematical computer models can be implemented as tools to assess the cumulative health effects of chemical substances.

Unknown combination effects

“Today we don’t know precisely what happens when chemicals impact the individual organs simultaneously. However, we do know that the chemicals can act in combination which leads to a weaker or stronger toxic effect in the body than you would expect from what is known about the individual substances in a mixture,” says Trine Klein Reffstrup from the National Food Institute.

“We therefore need to be able to assess in advance the toxic effects and the harmful risks, and to know the beneficial and non-beneficial combined actions. My hope is that, using the computer models, we will within the foreseeable future be able to calculate the effects,” says Trine Klein Reffstrup.

The more we know about the combination effects and the appurtenant risks, the faster and more effectively the authorities can react if harmful values are detected in a food product. In the PhD project, a so-called biosimulation model is tested for the insecticide chlorpyrifos, which has been used to control pests in agriculture, gardens and homes, and which is often found in fruit and vegetables which are checked in Denmark. The next step is to develop a model where two chemicals are involved at the same time so it is possible to simulate the combination effects.

Realistic models

Developing computer models to describe how the chemical substances are distributed throughout the body, how they are transformed and excreted and which effects they have is an advanced process. The models incorporate numerous constants and parameters, which stem from data resulting from experiments with rats, or they can be estimated using QSAR (Quantitative Structure - Activity Relationship). These results can subsequently be transferred to a ‘human-scale’ model in order to simulate reality as closely as possible.  

The plan is to use QSARs to supply the computer models with relevant properties (parameters). This is done if experimental values are not known for the substances which will be used in the models, but where data for similar substances exists. This is a further development of the QSAR area at The National Food Institute. Here, considerable experience and good results have been gained by using QSAR methods for calculating and predicting the properties and health effects of individual substances, knowledge which is being applied in environmental consultation.

The PhD project revealed numerous pitfalls and problems in trying to develop reliable models.

“My provisional studies and results are, however, so promising that I am continuing my work with developing the models at the National Food Institute. The National Food Institute is therefore allocating more resources to ensuring that we have effective tools in future for assessing the risk of harmful health effects from chemical substances in our foods,” says Trine Klein Reffstrup in conclusion.

Read more

Read Trine Klein Reffstrup’s PhD project: Evaluation of methodologies for risk assessment of combined toxic actions of chemical substances and establishment of PBTK/TD models for pesticides (PDF).