Following a request from the European Commission the Scientific Committee was asked to develop principles and guidance for the establishment of protocols for 90-day feeding studies in rodents with whole food/feed. The design of such protocols should be based on the specific properties of food/feed derived from genetically modified plants and other novel food under investigation and in line with the purpose of the study. In view of the multidisciplinary nature of this subject, the task was assigned to the Scientific Committee.
Risk assessment of whole food/feed comprises an integrated approach where information is required on a number of characteristics from various types of tests, including toxicity. Data generated from toxicity testing, whether collected from in vivo or in vitro studies, provide fundamental information to carry out a risk assessment of a food for human consumption, or of a feed for animals.
This guidance further develops the general procedure set out in the OECD Guideline for the Testing of Chemicals – Repeated Dose 90-day Oral Toxicity Study in Rodents (OECD TG 408), and provides specific advice for performing and reporting experiments carried out with whole food/feed. The guidance aims to help applicants in designing, conducting, analysing, reporting and interpreting repeated-dose 90-day oral toxicity studies of whole food/feed in rodents for the purpose of risk assessment. It is however not intended to provide prescriptive experimental test protocols to carry out such an experiment. It is also outside the scope of the guidance to address the question whether and when such testing is warranted for the safety assessment of whole food/feed.
The conduct of the study and documentation should follow good laboratory practice.
Appropriate characterization of the whole food/feed to be tested is required and should include, among others, a description of the source, its composition, the manufacturing process, information on stability and the presence of chemical and/or microbiological contaminants.
Preparation of appropriate test diets is a key element of the experiment with respect to the choice of the diet type, nutritional balance and necessary adjustments, processing, and storage. Since it is often not possible to include whole foods in an amount that will reliably induce toxicity and thus to obtain a dose-response relationship, fewer dose levels but more animals in control and top dose groups should be used to maximise the power of the study. The use of one control group plus two dose levels (high and low) are recommended when testing whole food/feed. The highest dose level of the whole food/feed should not cause nutritional imbalance or metabolic disturbances in the test animal, and the lowest dose level should always be above the anticipated human/target animal intake level.
In accordance with the European Directive 2010/063, the test animals should be housed socially. The guidance recommends that animals of the same sex are housed in pairs. The experimental unit (ExpU) is therefore a cage containing two animals. More than two individuals per cage are also acceptable if justified, in which case the cage containing more than two animals is the experimental unit.
A comprehensive set of endpoints as set out in the OECD TG 408 should be measured during and at the end of the 90-day period, as appropriate; additional endpoints may be considered for the assessment, depending on the nature of the food/feed being tested and the available information. A comprehensive statistical analysis should be conducted including the analysis of the differences between males and females.
A randomised block design is recommended for 90-day toxicity studies when testing whole food/feed. Other designs may be acceptable provided that appropriate justification for using them is given.
The use of a power analysis to estimate a sample size capable of detecting a pre-specified biologically relevant effect size with a specified power and significance level should be done to determine an appropriate sample size. Those objectives have to be specified clearly and translated into effect sizes that are considered by experts as being biologically relevant. The biological relevance of any observed differences should be considered, whether or not they reach the chosen level of statistical significance. This assessment should involve the use of point and interval (e.g. confidence) estimates in addition to the significance level.
In the case of GM food, the inclusion in the experimental design of reference groups, fed with a diet containing commercially available food/feed similar to the test food/feed, in order to estimate the natural variability of endpoints is in general not recommended since this would substantially increase the number of test animals. Historical background data on variations in endpoint values should primarily be obtained from databases available in the actual testing facility or in the public domain. Inclusion of reference groups should be considered if no acceptable historical background data is available.