Introduced gene	Description/phenotype	Number of products
	Insect resistance
cry1(A)b and cry 1(A)c	truncated Btk endotoxin	4
	Herbicide tolerance
pat and bar	glufosinate ammonium (Basta) tolerance	5
cp4 epsps and gox	glyphosate (Roundup) tolerance	3
	Antibiotic resistance markers
nptll (syn.aph(3)-ll)	kanomycin/neomycin resistance	5
nptlll (syn aph(3)-llla)	amikacin tolerance	1
aadA (syn. ant(3)-1a)	streptomycin/spectinomycin resistance	2
bla	ampicillin resistance	1
	Male sterility/fertility
barnase and barstar	ribonuclease and ribonuclease inhibitor	2
	Sense/antisense gene silencing
pg (partial sense)	delay in fruit softening	1
gbss (antisense)	starch with reduced amylose content	2

  Consideration of any hazards posed by transgenics involving the genes listed in Table 1 has been protracted and drew on historical data such as the absence of detectable toxic effects of the Btk toxin after 30 years use as an insecticide. Thus the FDA and various other national advisory bodies have had nearly 10 years to consider and investigate these product types. Since these GM crops were also the first commercial products, they have been the subject of investigation by many other interested parties and, as a result, a considerable body of data exists on which to base a safety assessment.

  It is unlikely that this slow rate of introduction of new GMOs for assessment will continue. A scan of the data bases produced by the competent authorities in most OECD countries shows that a very large range of traits have been successfully incorporated into crops which are now undergoing field trials. These triats range from protection against various forms of pests to the production of industrial feedstock. Field trials involving 56 different plant species have been completed in the USA alone (compared with 12 in the UK). Trials may involve a single modification to one plant species or, in the case of the more important crops, many different types of transgenic plants. The 450 trials in the USA between 1989 and 1997 made with GM potatoes, for example, encompassed at least 32 different traits. As a consequence, the range of genetic modifications in crops and numbers of new products likely to be seeking regulatory approval in the future will be far greater than those already considered and will inevitably challenge the existing safety assessment and risk management procedures.

  Although Companies seeking to introduce a GM product are required to demonstrate its safety, it is very unlikely that any Company will provide more information than is indicated as necessary in the guidelines which accompany the regulations. Thus the value of information provided to government and to its advisory bodies is dependent on the nature of the information sought. Guidelines based on our limited experience to date need amplifying to deal with all the issues raised by the far greater diversity of genetic traits likely to seek release in the near future. This will be most apparent in the consideration of human safety and three areas now need additional safeguards to provide the level of assurance necessary to allow a transgenic plant to enter the food chain.

8.1 Quality and rigor of existing criteria

  Most companies seeking approval for the release of GM crops in Europe have first sought approval in the USA. Therefore the dossiers being submitted reflect the procedures of the FDA. We conclude that practices currently considered acceptable and promoted by the FDA are not rigorous enough for future use. One of the key issues in establishing safety is the ability of the digestive tract to digest the protein encoded by the transgene. The FDA approved tests of gastric and intestinal survival represent a "best case" situation and do not reflect the digestive capacity of the very young, the elderly and that segment of the population unable to produce stomach acid. Secondly, the difficulty in extracting sufficient protein from a transgenic plant for testing, has led to an "accepted" practice of making tests on the protein from the same gene expressed in a different host, usually a bacterium. It is well known that a protein expressed in a different host will undergo different post-translational modification and will not possess the same biological and physical properties. Extrapolating from the tested behaviour of an isolated protein produced in a bacterium to predicting the behaviour of the same protein when it is an integral part of the transgenic plant is unsound and could lead to premature conclusions about safety.

8.2 Substantial equivalence

  The concept of substantial equivalence is a useful framework in which to consider effects which do not directly relate to the transgene or its product, but which may have been caused by the process of introducing the foreign genetic material. Conventionally, substantial equivalence is established by comparing data on the composition of the transgenic plant grown at several locations over two or more seasons with the same data from the unmodified parent line grown alongside the transgenic line. However, establishing substantial equivalence, depends on correctly specifying and then measuring those elements of concern. Much of the data produced is simply a measure of gross composition which is difficult to interpret. At what levels do changes to protein, carbohydrate or fat concentrations pose an additional risk to human health and why? There is obvious value in measuring the concentrations of known natural toxicants such as glycoalkaloids in potato or glucosinolates in brassicas, since safe levels for these compounds have been established. The routine analytical procedures currently used, however, would be unlikely to detect, toxic metabolites which accumulated because the introduction of transgenic material silenced an existing plant gene and disrupted a metabolic pathway.

  The concept of substantial equivalence would be better served by the use of techniques which make no assumptions but which attempt to measure the construct as a whole. We propose that three relatively new technical approaches would allow a more discerning analysis than at present.

—  Differential display methods can detect differences in mRNA induced by other genes as well as the transcribed gene.

—  Proteomics to detect differences in total protein expression would allay concern about the whether the transcribed gene has led to different protein structures because of additional post-transcriptional changes, eg. by the addition of carbohydrate units by virtue of these processes occurring in a new plant host.

—  Metabolic profiling techniques they could lead to a clear documentation of amplified or suppressed metabolic pathways which would not even be considered in current testing systems. Analyses of metabolic pathways should be mandatory when an introduced gene is known to code for an enzyme involved in the production of plant secondary metabolites (the epsps gene for example).

8.3 Immune and hormonal status

  Existing assessment methods require tests for chronic and acute toxicity. However these are invariably made with the isolated protein produced by the transgene and not with transgenic plant itself. This is because of the difficulty of achieving the higher test concentration required in the diets of test animals. While such tests do provide a measure of security, they offer a poor screen for more subtle effects or those which may have a relatively long gestation. One such issue is that of allergenicity. Existing assessments rely heavily on comparing the similarity of the transgenic protein with known allergens (ie whether the sequence homology is or is not the same as the known allergens). This assumes, however, that the allergy depends only on an epitope with a continuous sequence and, secondly, that all allergens are known. Neither is a comfortable assumption. However there are no widely accepted methods for assessing allergenic potential. We understand that the Royal Socieity is considering the issue of allergies in relation to GMOs so there may well be new proposals for screening or research strategies emerging from their review.

  A second set of issues relate to intestinal secretions and hormonal changes, ie endocrine and exocrine functions which are not addressed in the current guidelines for GMO assessment. Although not particularly relevant to those crops already approved for release, these issues are likely to become of far greater concern in the future. Many of the genes now being considered for introduction to provide insect resistance depend for their action on disrupting the digestive function of the pest (Table 2). Some of the enzyme inhibitors and lectins being considered may produce similar effects in mammals. In addition, they are known to be highly resistant to degradation in the digestive tract when produced in their natural host and, if absorbed, may have effects on many aspects of metabolism, including the immune and hormonal systems. This was the issue being addressed by Dr Pusztai in his research.

  Unless the development of assessment tools parallel and keep pace with developments in plant genetic engineering, it will not be possible for advisory bodies to provide secure advice on the safety to humans of future GM plants seeking release in the UK and Europe. Industry claims that over stringent requirementsfor evidence of safety will stifle development of a vital technology has limited validity and then only in the short-term. It is better to start with rigorous requirements for safety evaluation and then relax the specifications in the light of experience rather than having to tighten regulations once evidence of damage to human health has occured.