Limitations of currently available in vitro oestrogenicity bioassays for effect-based testing of whole foods as the basis for decision making.

The idea that previously unknown hazards can be readily revealed in complex mixtures such as foods is a seductive one, giving rise to the hope that data from effect-based assays of food products collected in market surveys is of suitable quality to be the basis for data-driven decision-making. To study this, we undertook a comparative study of the oestrogenicity of blinded cereal samples, both in a number of external testing laboratories and in our own facility. The results clearly showed little variance in the activities of 9 samples when using a single method, but great differences between the activities from each method. Further exploration of these findings suggest that the oestrogenic activity is likely an inherent part of the natural food matrix which the varying sample preparation methods are able to release and extract to differing degrees. These issues indicate the current poor suitability of these types of datasets to be used as the basis for consumer advice or food decision-making. Data quality must be improved before such testing is used in practice. [ABSTRACT FROM AUTHOR]