ACRE Advice

ACRE's response to concerns raised in written representations and submissions associated with the CHARDON LL public hearing and to statements made at ACRE's open hearing relating to the safety assessment of T25 GM maize conducted under Directive 90/220/EEC

Animal Feed Safety Issues

Point 10. It has been suggested at the CHARDON LL hearing that the Advisory Committee on Animal Feedingstuffs (ACAF) has not considered the safety of T25 maize as animal feed.

At the time that the T25 maize application dossier was under review in 1996, ACAF did not exist. At that time ACRE was not constituted to advise in detail on feed safety and so the dossier (like others) was referred to animal feed experts at the Ministry of Agriculture, Fisheries and Food (MAFF). In addition, the Advisory Committee on Novel Foods and Processes was ask for its advice on the safety of T25 for use as food and in animal feed - it was content that the risks were very low. In the absence of objections expressed to ACRE by MAFF and particularly in the light of the ACNFP's views, ACRE was content for a favourable opinion to be forwarded to the European Commission. It is noteworthy that the dossier had been considered in detail by the authorities in France where it was originally submitted and subsequently it has been assessed by other member states and by the commission's Scientific Committee for Plants.

ACAF was established in June 1999 specifically to provide government with independent expert advice on all issues relating to the safety and use of animal feedingstuffs. So in 2000, when ACRE was asked to review T25 maize in the light of views expressed at the CHARDON LL hearing, it asked ACAF for advice on animal feed aspects. ACAF is satisfied that on the basis of the information available, there is nothing to indicate that T25 maize grain or its products pose any more risk to humans or animals than non-GM varieties if used in animal feed.

Point 11. It has been asserted that the 14 day duration of the rat toxicity test does not realistically model toxicity caused over lifelong exposure.

AgrEvo commissioned a repeated dose, oral toxicity test in rats to assess the toxicity of the PAT protein. The body weights, organ weight and food consumption of rats were measured. There were no statistically significant differences in these parameters in rats fed PAT protein compared to rats in control groups. The duration of this study has been criticised as being too short; but this is a recognised procedure whereby test animals are fed 'large' amounts of a specific substance over a relatively short period and monitored for any toxic effects. After two weeks of this regime, there was no apparent effect on the animals indicating that PAT is not acutely toxic.

In ACAF's view, the rat study was not entirely without criticism. Members considered that the origin of the PAT protein fed to rats was obscure and the use of activity units and weight of PAT did not allow the level of exposure to be determined accurately. On request the company clarified these points: PAT protein was synthesised in bacteria (Escherichia coli) and the dosage fed to rats in this study was at least 1000 times greater than dietary exposure from eating GM plants.

There has been some criticism in written representations and submissions that the PAT protein used in the rat toxicity study was synthesised in a bacterium - the suggestion being that PAT protein made in E. coli may undergo post-translational modification causing differences in stability compared to PAT synthesised in plants. However, PAT protein from both sources was passed through a matrix along an electric gradient (gel electrophoresis) and no differences in their respective mobilities were observed. This suggests that no significant post-translational modification of the PAT protein synthesised in the bacterium has occurred. The use of bacteria to generate large amounts of pure recombinant protein is normal practice because the alternative generally requires the production of considerable quantities of plant material.

Point 12. It has been suggested that the chicken feeding study was a poorly designed experiment and that a trend for higher mortality in GM fed broiler chickens gives cause for concern. The nature of the trial has also been criticised i.e. the suitability of feeding T25 maize grain to chickens when the target organism are ruminants that will be fed T25 maize silage.

The chicken feeding study was not part of the application dossier submitted to the French competent authority by AgrEvo in 1995 - it was submitted afterwards to support conclusions on the nutritional equivalence of T25 maize grain with its non-GM counterparts. The UK government did not ask ACRE to assess a report of the chicken feeding study that was circulated to member states by the French in 1997. The chicken feeding study was not considered by ACRE until 2001 when it asked for ACAF's advice on animal feed aspects of the T25 maize dossier in response to criticism raised at the CHARDON LL hearing.

In the original T25 maize dossier submitted by AgrEvo in 1995, conclusions on nutritional equivalence were based on the results of compositional analysis; this is in accordance with accepted practice. ACRE was satisfied that the T25 maize grain is compositionally equivalent to conventional maize grain except for the presence of the PAT protein and that this protein poses a negligible risk to animal or human health.

ACRE accept that broiler chickens are a sensitive test system but are not convinced of their relevance in assessing the 'wholesomeness' of T25 maize since ruminants are the target organism for this product and these will be fed T25 maize silage, not grain. At ACRE's hearing, it was suggested that a laboratory-based study using microbes found in the guts of ruminants would have been more informative than a chicken feeding study. ACRE is concerned that applicants should consider the relevance of any feeding trial that is carried out and its limitations.

However, as the chicken feeding study was submitted for consideration, ACRE is clear that, it should meet the standards expected for information presented in application dossiers. This means that experimental design, including statistical analyses should be sufficient to address questions asked of the study and clear, detailed descriptions of both should be supplied.

The chicken feeding study was conducted in 1996. In 1997, the French circulated a report of the study to other member states. In the UK, MAFF were asked to lead on this (ACAF had not been established at this point)- the report was not sent to ACRE as it was not constituted to advise in detail on animal feed safety. MAFF did not report any objections and the UK agreed with the French that the chicken feeding study did not alter the risk assessment of T25 maize.

During the CHARDON LL public hearing and subsequently at ACRE's open hearing, the relevance of the chicken feeding study, its experimental design and statistical analysis were criticised. Concern was also expressed about mortality rates in birds fed T25 maize grain compared to those fed grain from a non-GM variety. In the light of these concerns, ACRE asked ACAF for its advice on the safety of T25 maize for use in animal feed.

The chicken feeding study was carried out at the Department of Animal and Poultry Sciences at the University of Guelph in Canada. Broiler chickens were chosen as the target species because they provide a very sensitive test system. During the first 18 days of life the body weight of broiler chicks increases15-fold, therefore inadequacies in diet are quickly apparent. The study took place over forty two days and involved 280 male birds, half were fed T25 maize grain and the other half, non-transgenic grain.

An expected mortality rate of 5-8% was quoted for male broilers kept at the research facility at the time of the study. Therefore, under the conditions used in this experiment the expectation was that up to 11 birds would die in each group of 140 birds (8%).

The mortality rate of chickens fed T25 maize grain was within the expected range for chickens reared at the facility (10 dead out of 140 birds, i.e. 7%), whereas the mortality rate of chickens fed non-GM maize was lower than predicted (5 dead out of 140 birds, i.e. 3.6%). Concerns over the mortality rate of birds fed T25 maize have been raised; however, these are within the expected range for chickens reared under the conditions used in this study.

The power of the study to resolve differences in growth rate, if they exist, between chickens fed T25 maize grain and those in the control group is limited. There are opposing views on whether further statistical analysis of the data could increase the power of this study to answer questions on the nutritional equivalence of T25 maize. The differing points of view expressed by statisticians at the hearings are:

(i) If analysed differently, the sensitivity of this study to detect any small difference in the growth rates of T25 maize fed birds and those fed non-GM grain might be increased.

(ii) Flaws in the experimental design of the chicken feeding study would prevent any conclusions from being made, irrespective of whether the data is further analysed or not. The flaws identified were - insufficient replication of treatments (GM and non-GM fed birds) and the lack of a positive control treatment that could demonstrate whether the study has the power to detect a known decrease in growth rate. The study was also criticised for being inadequately described i.e. the detail given was not sufficient to repeat the experiment.

ACRE has asked the company to determine whether reanalysis of the raw data from the chicken feeding study could increase the power of this experiment in comparing the nutritional value of T25 maize grain with that of a non-GM counterpart. This has been done and ACRE is a waiting the company's formal response.

In summary, the chicken feeding study was designed to consider unexpected, undefined characteristics of T25 maize grain that could impact on its nutritional status. There is no evidence from this study that T25 maize grain fed to animals would pose an increased risk compared with conventional maize grain. ACRE has reservations about the usefulness of this study in the risk assessment of T25 maize as animal feed. ACRE welcomes as much relevant information in application dossiers as possible, however data submitted must meet expected standards as previously described. To this end, ACRE has asked that the company determine what potential this study has to compare the nutritional status of T25 maize grain with its non-GM counterparts in chickens.

ACAF is satisfied that on the basis of the information available, there is nothing to indicate that T25 maize grain or its products pose any more risk to humans or animals than non-GM varieties if used in animal feed ³³.

The company have commissioned a dairy cattle feeding study with T25 maize and will present the data to the French competent authorities when completed. As is standard practice, ACRE will review new information generated in this trial and update the risk assessment accordingly.

Point 13. It has been asserted that composition analyses have shown statistically significant differences between T25 and non-GM maize varieties.

While the principle of substantial equivalence has received wide international support it has also been subject to criticism. In some cases there appears to be a mistaken perception of how substantial equivalence is applied. It is not a safety assessment in its self; substantial equivalence is a starting point for identifying differences between a transgenic variety and an appropriate comparator that has a history of safe use. The equivalence of a transgenic variety's chemical composition, including a range of nutritionally important parameters and any naturally occurring toxins, is compared to its conventional counterpart. The next step would then be to determine the relevance of any differences that are found; this may involve further detailed tests. Substantial equivalence is therefore a starting point in risk assessments and it is only one part of a safety assessment. Other aspects may include toxicology experiments, comparisons between the amino acid sequence of transgenic proteins and that of known allergens and characterisation of transgene insertion sites in host genomes.

The use of compositional analysis is considered an indicator of nutritional status. At ACRE's hearing it was suggested that additional feeding studies are necessary to establish nutritional equivalence between T25 maize and its non-GM counterparts - this is discussed in point 14.

The Compositional Equivalence of T25 Maize Grain

Compositional data for T25 maize grain are within the normal expected range for commercially grown maize varieties. This supports the view that T25 maize grain is compositionally equivalent to these varieties apart from its synthesis of PAT protein. A detailed risk assessment of PAT was conducted separately and includes assessments of potential toxicity and allergenicity. ACRE is satisfied that any risk posed by the PAT protein to animal or human health is negligible.

Some statistically significant differences were observed between T25 maize grain and its non-GM counterpart but these values are within the accepted values for conventional maize lines. Variation in the chemical composition of maize principally arises because it is a hybrid and breeders intentionally use parental lines with as much genetic distance as possible. In addition, agronomic and environmental conditions also affect composition. Because of this, animal feed experts in ACAF felt that differences between T25 maize and its non-GM counterpart were unlikely to be significant because livestock already experience subtle differences in the composition of maize.

There has been some disagreement as to the compositional equivalence of T25 maize grain based on data submitted to the CHARDON LL hearing. This centred on whether linolenic and arachidic fatty acid contents in T25 maize grain were within the range expected for non-GM maize varieties grown commercially. ACAF are content that T25 maize is compositionally equivalent to non-GM maize. Further analyses conducted since the original composition data was submitted support this ³⁴.

The relevance of chemical analyses on maize grain for establishing nutritional equivalence has been questioned since it is silage that will be fed to ruminants.

The Compositional Equivalence of T25 Maize Silage

ACAF experts considered that current data for T25 maize silage is not sufficient for conclusions about equivalence to be made, however, the data presented give no cause for concern. ACRE agrees with ACAF's view that sufficient additional analytical data for T25 maize silage should be provided to demonstrate whether it is substantially equivalent to silage from conventional maize varieties. Alternatively data from a feeding study made with dairy cattle could be used to demonstrate the wholesomeness of the silage and a lack of any effect on milk production and composition. Milk production data is usually taken as a relatively sensitive indicator of body condition.

Bayer CropScience will submit further compositional data for both grain and silage and this will be reviewed as part of a continuing risk assessment.

The company also propose to submit the details of a feeding study in dairy cattle that will be used to examine the wholesomeness of T25 maize silage and any effect on milk production and composition.

Point 14. Concern has been expressed that no tests were carried out to check that T25 maize is safe for livestock

The fact that no feeding studies have been carried out with T25 maize in cattle or sheep has been claimed to represent a weakness in the risk assessment process. ACAF has made its position of animal feed trials clear - it recognises the value of animal feeding studies on target species, particularly for determining nutritional adequacy but does not require that animal tests are carried out in all cases. Instead, ACAF promote the use of appropriate techniques that provide the detail and quality of information required by members to assess feed safety.

ACAF is not of the view that feeding studies with T25 silage must be done. Members considered that if compositional and agronomic equivalence can be demonstrated and if a detailed risk assessment of the transgenic protein provides no indication that it is harmful to livestock then feeding studies are not essential. In the case of T25 maize grain, chemical analyses have demonstrated compositional equivalence with non-GM, commercially available counterparts and there is no evidence from a variety of studies that the PAT protein poses a risk to livestock (i.e. its biochemical characteristics, its fate in gastric juices and the lack of similarity to any known toxin or allergen). However, further data are required to establish whether T25 maize silage is substantially equivalent to non-GM varieties grown commercially (refer to Point 12).

The use of compositional studies to assess the nutritional value of T25 maize to ruminant animals has been criticised. The point made was that chemical analyses, including measurements of fibre content, are a poor predictor of the digestibility of plant matter and that lowered digestibility affects the nutritional impact of feed by reducing the availability of substrates to bacteria in the rumen. However, compositional data are used as an indicator of nutritive impact with non-GM varieties, hence direct animal feeding studies are not required. It is also important to note that there is no evidence from chemical analysis data or from studies with the PAT protein that T25 maize is unsafe for use as an animal feed.

Point 15. There has been criticism of a digestion study carried out on the PAT protein

It has been suggested that a digestion study involving the PAT protein did not accurately model the conditions likely to be found in animals eating the GM maize because the pH of stomach juices rises above the highest pH used in the study when food is ingested.

However, it must be recognised that acid digestion studies such as the one carried out here are only an indicator of how rapidly a protein might be degraded/ inactivated in the stomach and they are not designed, nor were they intended to be a detailed simulation of an animal's digestive system. ACAF was content that these studies supported the conclusion that PAT protein derived from oilseed rape and T25 maize is inactivated in vitro by gastric juices. Measurements of relative PAT activity in simulated human gastric fluid also supports this conclusion ³⁵.

Point 16. It has been suggested that livestock show a preference for eating non-GM maize.

ACRE has discussed the anecdotal nature of some of the concerns raised in the written representations and submissions for the CHARDON LL hearing. One claimed that some animals had shied away from eating GM crops. ACRE noted that it was difficult to answer these points in the absence of any tangible reason or mechanism for why this should be. ACRE was not aware of any peer reviewed scientific data but would welcome the submission of relevant papers or evidence at any time.

33 ACAF's advice on the safety of T25 maize for use in animal feed was issued to ACRE on 5 September 2001.

34 For example, please refer to data from Flachowsky et al.2000 submitted to ACRE's open hearing by Professor Phipps: www.defra.gov.uk/environment/acre/hearing200202/pdf/t25_phipps.pdf. This data has been accepted for publication in a peer-reviewed journal.

35 Wehrmann A., Van Vliet A., Opsomer C., Botterman J. and Schulz A. (1996). The similarities of bar and pat gene products make them equally applicable for plant engineers. Nature Biotech. 14: 1274 - 1278.