ACNFP
THE ADVISORY COMMITTEE ON NOVEL FOODS AND PROCESSES
ACNFP Secretariat
Food Standards Agency
Room 515b Aviation House
125 Kingsway
London WC2B 6NH
Tel: +44 (0)20 7276 8595
Fax: +44 (0)20 7276 8564
acnfp@foodstandards.gsi.gov.uk
Statistically valid data to support applications for safety clearance of crop product under EC Regulation on Novel Foods and Novel Food Ingredients 258/97
1998: This paper is intended to promote discussion about the provision of statistically valid data in applications for product approvals under the EC Regulation on Novel Foods and Novel Food Ingredients 258/97 (EC Novel Foods Regulation). The paper presents the ACNFP's conclusions.
Paper for consideration by the European Commission:
Purpose
This paper is intended to promote discussion about the provision of statistically valid data in applications for product approvals under the EC Regulation on Novel Foods and Novel Food Ingredients 258/97 (EC Novel Foods Regulation). This issue has been considered by the UK Competent Authority's competent food assessment body, the Advisory Committee on Novel Foods and Processes (ACNFP), and the paper presents the ACNFP's conclusions.
Introduction
The European Commission's guidelines¹ which accompany the EC Novel Foods Regulation have been welcomed by potential applicants for product approvals under the Regulation. The European Commission's guidelines give a clear indication of the type of data that is needed to form the basis of a safety assessment and the UK Competent Authority has incorporated the relevant sections from the guidelines into its computerised decision tree.
However, the European Commission's guidelines do not include advice on the appropriate sample sizes to be included in data submitted in an application for approval under the Regulation. The ACNFP is concerned that all safety assessments should be based on statistically valid data and has outlined its conclusions in this paper.
Background
The report of a recent FAO/WHO expert consultation on biotechnology and food safety² states that for food safety assessment purposes:
"Substantial equivalence is established by a demonstration that the characteristics assessed for the genetically modified organism (GMO), or the specific food product therefrom, are equivalent to the same characteristics of the conventional comparator. The levels and variation of characteristics (e.g. phenotypic and composition characteristics etc.) in the GMO must be within the natural range of variation for those characteristics considered in the comparator and be based upon an appropriate analysis of data."
Natural variation
The natural variation of a range of measured attributes or comparators may be large because of the wide variation in conditions, such as soil fertility, temperature, light or water availability, under which crops are grown. The issue may be compounded by other effects including agronomic treatments to the growing crop, such as herbicide applications to a herbicide tolerant crop, or post-harvest stress during storage (e.g. for potatoes stored in clamps).
Natural variation may obscure the differences between possible comparators for crops grown under identical conditions. The differences between the comparators of a GM crop and those of its non-GM control grown under identical conditions should be used to construct the criteria which determine whether a GM crop may be considered to be substantially equivalent to its non-GM control. Thus, for example, the mean of a comparator should be within a certain percentage of the control mean. The range of values quoted in the literature to express natural variation should be considered to be of secondary importance.
Specification
The companies which request the safety assessment of GM foods or ingredients are, usually, technology providers and not the processors of the final or intermediate food products or ingredients; they often have no control over the specification of the final food ingredient. Consequently, proximate and detailed chemical analyses of the unprocessed sources of GM foods or ingredients are an important part of the safety assessment. Proximate and detailed chemical analyses required are analyses of fat, fatty acid profile, solvent extracted non-saponifiable matter, protein, amino acid profile, micronutrients, antinutrients, crude fibre, ash, and moisture contents.
Significant differences between the results of proximate and detailed chemical analyses of a GM crop product and its non-GM control may be an indication of secondary or pleiotropic effects and such changes would need to be investigated further.
Confidence limits and sample sizes
The following methodology assumes that the differences in means between the GM crop and its control are independent for each measured comparator. A GM crop is considered to be substantially equivalent to its non-GM control when the difference between the underlying means fulfils some criterion for each measured comparator. Even if a GM crop is not substantially equivalent to its control, it is still possible that the differences between the observed means from a field trial could fulfil all the criteria for equivalence. The likelihood that this may happen can be estimated. If this probability is very small, we can say that there is evidence that the GM crop is substantially equivalent.
The test for substantial equivalence can be performed by constructing confidence intervals for each measured comparator. The test finds the GM crop to be substantially equivalent if each confidence interval is completely within the criterion for that comparator. By convention, 95% confidence intervals are commonly used. However, 99% and 99.9% confidence intervals are also used when greater certainty that the true value of the estimate is within the interval is required. The width of the interval is greater for higher levels of confidence.
Clearly, there are two important criteria which need to be set, namely the appropriate confidence interval to use in the test and the sample size. It is recommended that the Commission seeks EU-wide agreement on the most appropriate confidence interval. It may be considered necessary to alter the confidence interval used in the assessment according to the comparator under consideration and the potential hazard that it presents. It is in the applicant's interest to ensure that an adequate number of samples are analysed to reduce the chances that substantially equivalent GM products are deemed to be not substantially equivalent to their non-GM controls.
Trial sites, years and replicates
For crop varieties which are intended to be grown commercially in the EU, the location of trial sites should be representative of the range of environmental conditions under which the varieties would be expected to be grown. The number of trial sites should be sufficient to allow accurate assessment of agronomic and compositional characteristics over this range. Similarly, trials should be over a minimum number of years to allow adequate exposure to conditions met in nature. It is recommended that the minimum number of years and the minimum number of trial sites should be set by the EU competent authorities and agreed by the Commission. The UK recommends that the minimum number of sites should be six and the minimum number of years should be two.
Replication at each trial site is necessary to ensure that environmental effects at a single location are minimised and to allow the adequacy of statistical models to be checked. To reduce any effect from naturally occurring genotypic variation within a crop variety, the minimum number of replicates used should be not less than three and will need to be appropriate to the crop species under trial. Trials carried out by the National Institute of Botany in the UK, which assesses crop varieties for the UK National List of recommended plant varieties, require three replicates at each trial site for GM maize and four replicates at each trial site for GM spring oilseed rape. To allow effective comparison, the GM crop and its non-GM counterpart should be grown in adjacent plots at the same site and at the same time.
Data from the different sites should not be pooled before being statistically evaluated. Data from each site should be evaluated separately to allow information on each individual site to be provided as part of the dossier.
Conclusions
Statistically valid data is necessary for the effective comparison of a GM crop with a non-GM control. The use of such data will minimise the effect of natural variation and will indicate whether there are any differences between the GM crop and the non-GM control which require further investigation.
The ACNFP recommends that GM crops are trialled under the range of environmental conditions in which it is intended that the crop should be grown commercially. A minimum of six sites should be used, with the number of replicates to be determined for individual crop species. A larger trial, which would provide more information, may be necessary in some cases. Trials should be carried out for a minimum of two successive years.
References
1. Commission Recommendation of 29 July 1997 concerning the scientific aspects and the presentation of information necessary to support applications for the placing on the market of novel foods and novel food ingredients and the preparation of initial assessment reports under regulation (EC) No 258/97 of the European Parliament and of the Council.
2. FAO/WHO (1996) Biotechnology and food safety: Report of a Joint FAO/WHO Consultation, Rome, Italy, 30 September - 4 October 1996. FAO Food and nutrition Paper 61.