The most serious shortcomings of the Report are in the principle of "substantial equivalence" on which all safety assessment is based.

6.1 The principle is intentionally vague and ill-defined to be as flexible, malleable and open to interpretation as possible

"Substantial equivalence embodies the concept that if a new food or food component is found to be substantially equivalent to an existing food or food component, it can be treated in the same manner with respect to safety (i.e., the food or food component can be concluded to be as safe as the conventional food or food component)." (p.4)

This principle is unscientific and arbitrary, encapsulating a dangerously permissive attitude towards producers, and at the same time offers less than minimalist protection for consumers and biodiversity, because it is designed to be as flexible, malleable and open to interpretation as possible.

"Establishment of substantial equivalence is not a safety assessment in itself, but a dynamic, analytical exercise in the assessment of the safety of a new food relative to an existing food..The comparison may be a simple task or be very lengthy depending upon the amount of available knowledge and the nature of the food or food component under consideration. The reference characteristics for substantial equivalence comparisons need to be flexible and will change over time in accordance with the changing needs of processors and consumers and with experience." (pp. 4-5) In other words, one can choose to compare whatever is the most convenient at a particular time, and for a particular purpose. And if on one set of criteria, the product is not substantially equivalent, a different set of criteria could be used, always to the advantage of the producers.

In practice, the principle allows comparison of the transgenic line to any variety within the species, and even to an abstract entity made up of the composite of selected characteristics from all varieties. That is exemplified in the safety evaluation reported by the company Calgene on several of their products (Redenbaugh et al, 1995). By a judicious use of additional varieties any changes from the control recipient variety could be bracketed. In theory, a genetically engineered line could have the worst features of every variety and still be substantially equivalent. Such comparisons actually conceal significant changes resulting from the genetic modification per se, which should alert conscientious researchers to a more careful characterization of the genetically modified organism.

Bernard Shaw was reputed to have been propositioned by a beautiful though not too bright lady who wanted to have his child so it would have his brains and her looks, but Shaw was said to have discouraged her by pointing out that the child could end up having her brains and his looks instead. So, it is the particular combination of characteristics that makes all the difference. But under the present safety assessment regime, both combinations would be deemed "substantially equivalent". The danger is that particular combinations of nutrients or metabolites might fall within the "equivalent" range determined in this fashion, and yet be anti-nutritional or outright lethal or toxic.

And if that were not enough, producers are assured that, even when products are not substantially equivalent, they can be shown to be substantially equivalent except for defined differences, and "further safety assessment should focus only on those defined differences" (p.8). Lest one is in any doubt, it is stated on p.11 that, "Up to the present time, and probably for the near future, there have been few, if any, examples of foods or food components produced using genetic modification which could be considered to be not substantially equivalent to existing foods or food components." Calgene's genetically engineered Laurate canola oil should, by no stretch of the imagination, be considered substantially equivalent to ordinary canola oil. But, "other fatty acids components are GRAS [Generally Recognized as Safe] when evaluated individually because they are present at similar levels in other commonly consumed oils." Similarly, "substituion of Laurate canola for coconut and palm kernal oils does not raise any safety concerns for intended uses, in part because the major components, the fatty acids laurate and myristate, are identical."(Redenbaugh et al, 1995, p. 43)

In other words, it is already a foregone conclusion that most, if not all the products now and for the forseeable future will be assessed as "substantially equivalent", and if not, then considered GRAS by a judicious choice of a comparator.

It is significant that the Dutch courts have recently ruled Monsanto's genetically engineered soy beans not equivalent in quality to natural soy beans, as was claimed in the advertisement of Albert Heijn, the biggest supermarket chain in the Netherlands. Albert Heijn is itself part of the Dutch multinational, Ahold, which owns supermarket chains in many countries around the world. The complaint was filed by the Dutch Natural Law Party (Storms, 1997).

Given that "substantial equivalence" can be interpreted in the widest possible sense, and if not, by a judicious choice of comparator, the product can be considered as GRAS, it is difficult to imagine which remaining products cannot pass muster.

The Report recognized that "products could be developed which could be considered to have no conventional counterpart and for which substantial equivalence could not be applied." (p.11). For example, "products derived from organisms in which there has been transfer of genomic regions which have perhaps been only partly characterized."(p.11) This gives the impression that such are hypothetical cases that might arise in future.

But that is not so. The Report has failed to point out that at least one such transgenic organism already exists: Tracy, the sheep engineered with a large segment of the human genome - most of which contains unknown sequence with unknown functions - to produce huge quantities of alpha-antitrypsin in her milk (Colman, 1996). Tracy and her clones may be walking incubators for cross-species viruses to arise by recombination between human and sheep viral sequences. All genomes contain endogenous proviral sequences, and recombination between endogenous and exogenous viral sequences are already implicated in several kinds of animal cancers (see Ho, 1997, Chapter 13). One might think that the Report would treat such cases with extra caution. Not so.

We are assured that even if a food or food component is considered to be not substantially equivalent, producers need not despair, for "it does not necessarily mean it is unsafe and not all such products will necessarily require extensive testing."(p. 12). The Report is clearly preparing the grounds for slipping those products through a regulatory framework that is already worse than toothless.

Further on, in Section 6.6 on "Food organisms expressing pharmaceuticals or industrial chemicals"(p.19), there is the telling statement, "The Consultation recognised that, generally, the genetically modified organism would not be used as food without prior removal of the pharmaceutical or industrial chemical" (p. 19). That is a prelude to serving up the rest of Tracy and the "elite herd" cloned from her, or more likely, superannuated "pharm" animals and any failed transgenic experiment, whatever, as meat for our dinner tables. Transgenic technology is very inefficient and generates a lot of transgenic wastes - the large numbers of failed experiments. Such "foods" from transgenic wastes may be sources of exotic, cross-species food-borne viruses, as mentioned earlier. Furthermore, they will be exempt from safety assessment if the Report is to be taken seriously. A similar category of transgenic waste could be the left-over carcasses of pigs engineered for xenotransplantation.

All the signs are that the producers are handed carte blanche to do as they please for maximum profitability, with the regulatory body acting to allay legitimate public fears and opposition.

The procedure for establishing substantial equivalence, described in less than three pages in the 27-page Report (pp. 6-8), comes under two headings: background information on the characterization of the modified organism and actual determination of substantial equivalence, or characterization of the food product itself.

One glaring omission in the background information is the propensity of the transgenic organism for generating pathogenic viruses by recombination (and whether experiments have been carried out to investigate this propensity). This information is highly relevant for assessing impacts on biodiversity as well as food safety, in view of our current knowledge that superinfecting viruses may be generated from many transgenic plants at high frequencies and that insecticidal recombinant viruses may attack human liver cells (see notes 12-16). There is also disturbing new evidence that viral DNA can survive digestion in the gastrointestinal tract of mice, with large fragments getting into the blood stream and into many kinds of cells (Schubbert et al, 1994).

Likewise, information on the stability of transgenes, and potential for mobility of introduced genes, which are mentioned on p. 6 of the Report, ought to be based on data collected over a number of generations, documenting the stability of the insert as well as expression of the transgenes and the transgenic line in successive generations, so that both consumers and farmers can have confidence in quality control. In a paper presented at a WHO workshop, the author states, "The main difficulty associated with the biosafety assessment of transgenic crops is the unpredictable nature of transformation. This unpredictability raises the concern that transgenic plants will behave in an inconsistent manner when grown commercially."(Conner, 1995, p.27) In general, the inheritance of genetically engineered traits are non-Mendelian in subsequent generations (Schuh et al, 1993) necessitating clonal propagation. Earlier this year, 60,000 bags of genetically engineered canola seeds, enough for planting 600 000 acres, had to be recalled after they were sold in western Canada, because an unexpected gene, not yet approved for market, turned up in the seeds. The seeds were bred and sold by Limagrain, under licence from Monsanto. If the transgenic plants had been monitored for genetic stability of both the transgenes and the transgenic line in successive generations, as they should have been, and careful records kept, those seeds would never have reached the market. This incident also indicates the necessity for product seggregation, clear labelling and post-market monitoring as part of the condition for market approval.

Under background information, it is also crucial to include the upstream and downstream effects of transgenic promoter and enhancer sequences, as well as the presence of genetic elements in the host that might compromise the stability of the transgenes.

A further serious omission in the background information is the explicit requirement to disclose the presence of marker genes, especially antibiotic marker genes which are considered in Section 6.7.

Under "characterization of the food product", we are told it entails "molecular characterization", "phenotypic characterization" and "compositional analysis". While the latter two categories are elaborated subsequently, "molecular characterization" has mysteriously disappeared. Nowhere is it specified which methods of molecular characterization are required, nor which molecular information should be established. That happens to be crucial for identifying unintended effects. In consulting a previous document which reports on a WHO Workshop on the principle of substantial equivalence, molecular characterization is left very vague. it refers to "the inserted DNA"; "the level and mechanism of expression of the protein", which is considered to be "more important than knowing the gene copy number". In other words, the inserted DNA sequence need not be well-characterized at all. It then mentions "the level and function of the introduced gene product in the plant may be useful in judging substantial equiva lence", again implying that the function of the gene product need not be known as a condition for safety approval. If the gene(s) and gene product(s) transferred are well understood, however, the safety evaluation can then "focus on the safety of the expression product and/or changes brought about by the expression product." This is an open endorsement of a totally inadequate, reductionist safety assessment that ignores effects on the system as a whole, especially in the longer term.

In effect, no molecular characterization of the product whatsoever is required. Not even the level of expression of the introduced transgene(s) or marker gene(s) need to be ascertained, much less the effects of promoters and enhancers on neighbouring genes, as judged by the samples of papers presented in the WHO workshop on substantial equivalence. If one happens to know what has been transferred, then safety assessment can focus only on the gene product and its effects. So the two main categories of characterization of the food product are simply, phenotypic characteristics - agronomic, morphological and physiological and compositional comparison - key nutrients and toxicants which are known to be inherently present in the species.

Although the Report recognizes the possibility of "indirect consequences" (p. 4) and that "assessment of the safety of genetically modified organisms must address both intentional and unintentional effects that may result as a consequence of the genetic modification of the food source." (p. 5). These are limited to phenotypic changes that are readily apparent, and alterations in the concentrations of major nutrients or increases in the level of natural (known) toxicants. There is thus no specific requirement to test for unintended effects, per se.

Similarly, while it is stated that "attention must be paid to the impact of growth conditions on level of nutrients and toxicants...attention must be paid to the impact of different soils and climatic conditions." (p. 5) These are not elaborated further, and certainly not required for safety assessment recommended in the Report.

The range of tests which are actually carried out, as exemplified by WHO's Workshop Report on applying the principle of substantial equivalence, are not sufficiently discerning to pick out unintended effects. Unless there are gross morphological or phenotypic changes, there is no need to look for them. And even when there are gross abnormalities, the product can still be assessed to be "substantially equivalent". One paper presented in the WHO workshop reports, "Field trials on the transgenic lines used in these studies showed marked deformities in shoot morphology and poor tuber yield involving a low number of small, malformed tubers during field trials...These changes were attributed to somaclonal variation during the tissue culture phase of transformation....Despite these marked morphological abnormalities, virtually no changes n tuber quality attributes were detected..."(Conner, 1995, p.30). So much for the discerning power of the tests carried out.

There were no metabolic profiles done by routine techniques such as High Pressure Liquid Chromatography (HPLC), nor two-dimensional gel electrophoresis to scan for unintended expression of genes - again, another routine technique. The compositional analyses reported are limited to uninformative amino-acid profiles, or to known components present at levels greater than 0.1%, or 0.01% at best. And, as mentioned earlier, the arbitrariness of the comparator will already hide any changes due to the transferred gene(s) per se, which should alert researchers to unintended effects. Instead, the tests are aimed specifically at intended effects only, and if anything, to conceal secondary, unintended effects as much as possible.

The hazard of unintended effects is already well-attested to by the US epidemic of eosinophils-myalgia syndrome in 1990, resulting in more than 1500 affected and 37 deaths, which is linked to the consumption of L-tryptophan produced by a genetically modifed strain of Bacillus amyloliquefaciens (Mayeno and Glich, 1994). Several trace contaminants identified on HPLC have been implicated in pathogenesis.

A metabolite, methylglyoxal, was found to accumulate at toxic, mutagenic levels in yeasts engineered with multiple copies of one of several yeast glycolytic enzymes to increase the rate of fermentation (Inose and Murata, 1995). Recently, tobacco plants genetically engineered to produce the gamma-linoleic acid, also unexpectedly produced octodecatetraenoic acid, a substance previously unknown in natural tobacco plants (Reddy and Thomas, 1996). In the absence of a metabolic profile on the product, unintended toxic metabolites might have easily escaped notice in safety assessment.

It is equally important to check for unintended gene products being produced, which will not be revealed by routine amino-acid analyses of total lysates, as is done by Calgene for canola meal (Redenbaugh et al, 1995). A minimum requirement should be a two-dimensional gel electrophoretogram of the total proteins (Ho, 1996). Even then, minor modifications in a proportion of the proteins may not be detectable, which may change the properties of the proteins involved. For example, a proportion of the recombinant porcine and bovine somatotropins synthesized in E. coli were found to contain the abnormal amino acid e-N-acetyllysine in place of the normal lysine, only when reversed-phase HPLC analyses were carried out (Voland et al, 1994).

Key questions on the allergenic potential of transgenic foods are raised by the recent identification of a brazil-nut allergen in soybean genetically engineered with a brazil nut gene (Nordlee et al, 1996). It is possible to test for known allergens, as in the case of the Brazil-nut soybean, but not for allergenicity to proteins completely new to the foods involved, as acknowledged in the Report (p.14). It is significant that allergenicity in plants is thought to be linked to proteins involved in defence against pests and diseases (Franck and Keller, 1995). Therefore, transgenic plants engineered for resistance to diseases and pests may have a higher allergenic potential than the unmodified plants. One major novel protein is the insecticide produced by the gene from Bacillus thuringeinsis (Bt), now incorporated into a range of transgenic crop plants, which had never contained them before. Nevertheless, the producers were able to claim substantial equivalence by pointing to its "comparability" (not identity !) "to one of the proteins contained in the commercial microbial formulations that have been used commercially since 1988" (Fuchs et al, p.66). One important characteristic of an allergen is that it resists digestion in the stomach (gastric digestion). According to a recent publication (Astwood et al, 1996), known allergens were stable for 60 mins., whereas non-allergens were fully digested within 15 secs. While one study claimed the Bt protein was readily digestible (Fuchs et al, 1995), another report showed that it failed to be completely digested under gastric conditions after two hours (Noteborn and Kuiper, 1995). In both cases, we are assured that the protein is safe. In view of the recent discoveries that predators eating pests which have ingested the Bt toxin in transgenic crop-plants are also harmed (Bigler and Keller, 1997; Hawkes, 1997), it is irresponsible to assume that the toxin is safe for human beings.

We accept that no safety assessment system is foolproof. A case in point is the rigorous testing that goes on with pharmacological products. It is estimated that despite such rigorous testing, 3% of the products approved for market turned out to have such harmful effects that they have to be withdrawn, while an additional 10% have sufficiently harmful side-effects that limited use has to be recommended (Suurkula, 1997). This underlines the importance of seggregation, clear labelling and post-market monitoring of the health and other impacts of genetic engineered foods. Labelling is a matter of traceability and should be a scientific requirement, not only a consumer option.

Antibiotic resistance marker genes are not mentioned until p. 15, under "Section 6.2 Gene transfer from genetically modified plants", where it is stated that "Their continued use in plants remains critical to the production of genetically modified plants. The Consultation therefore focused on these particular marker genes." However, all it did was support the conclusions of a previous, 1993 Workshop "that "there is no recorded evidence for the transfer of genes from plants to microorganisms in the gut" and that there are no authenticated reports of such bacterial transformation in the environment of the human gastrointestinal tract." These conclusions are not based on actual experiments that have been done to ascertain if these transfers occur. It is a case of interpreting 'the absence of evidence' as 'evidence of absence'.

We are told that the first conclusion was "based on the judgement that transfer of antibiotic resistance would be unlikely to occur given the complexity of steps required for gene transfer, expression, and impacts on antibiotic efficacy." The steps are listed, the first of which is the most crucial, "the plant DNA would have to be released from the plant tissue/cells and survive in the presence of the hostile environment of the GI tract, including exposure to gastric acid and nucleases" (p. 16). But that is untrue. In the course of digestion, plant DNA will be released from the plant cells, and, there is already evidence that large fragments of viral DNA can survive digestion in the gastrointestinal tract of mice (Schubbert et al, 1994). So, it is possible that vector DNA, which carries the antibiotic resistance marker genes may also resist digestion. The question is whether bacteria in the gut can be transformed by the DNA, and there is an urgent need for experiments to be done to answer this question, in view of the wealth of new evidence, since 1993, on the ease with which transformation occurs in all other environments (see below). Most of the old assumptions supporting the previous judgement that transfer is unlikely may be superseded by the new findings.

Because gene-transfer vectors are already extensively modified, with sequence homologies to a wide range of species, and to resist restriction, they may successfully integrate into many bacterial genomes. It is practically impossible to design vectors that prevent horizontal transfer. Furthermore, sequence homology is not required for integration into chromosomes or plasmids, homology only makes it more likely to occur. The assumption that antibiotic resistance marker genes under plant promoters "would not be expressed in a microorganism" (p. 16) is dangerous, as so few bacterial promoters are characterized. While some antibiotic resistance marker genes are placed under bacterial promoters, as in the Ciba-Geigy transgenic maize, there are special mobile genetic elements in microorganisms, called integrons, which carry an enzyme catalyzing the integration of antibiotic resistance genes into specific sites where the integrated genes are then provided with ready-made promoters for expression (Collis et al, 19 93). The Report also fails to take account of the ease with which recombination can occur following horizontal gene transfer, whereby any missing promoter for the gene(s) may be regained.

Horizontal gene transfer has been demonstrated between bacteria in the gut of animals as well as human beings since the 1970s (Anderson, 1975; Freter, 1986; Doucet-Populaire, 1992). That means gene transfer from genetically modified microorganisms must definitely be considered in safety assessment of genetically modified microorganisms, as appears to be recommended by Section 6.3 of the Report, "Gene transfer from genetically modified microorganisms" (pp. 17-18). It is stated that "The Consultation affirmed the recommendation from the 1990 FAO/WHO joint consultation ... regarding genetically modified microorganism including: 1) that vectors should be modified so as to minimize the likelihood of transfer to other microbes; and 2) selectable marker genes that encode resistance to clinically useful antibiotics should not be used in microbes intended to be present as living organisms in food." (p. 18).

However, as stated above, artificial vectors are already extensively modified, modified vectors are often unstable (Old and Primrose, 1996), and may be much more prone to mobilize and to recombine (Allison, 1997; Ho, 1997; Ho et al, 1997). An additional problem of antibiotic resistance is that of cross-resistance. For example, resistance to kanamycin may be accompanied by resistance to new generation aminoglycoside antibiotics such as tobramycin and amikacin (Conner, 1995; Smirnov et al, 1994).

The Report states that, "The Consultation was not aware of any reports of genes from animal, plant or microbial origin into epithelial cells, except for infectious agents, such as viral DNA." But there is already evidence that viral DNA can enter the bloodstream and into many kinds of cells in mice. Again, vector DNA is modified viral DNA in many cases, and in the absence of results from actual experiments carried out to investigate this possibility, it is not legitimate to conclude that DNA cannot enter epithelial cells, or the blood stream and from there, gain access to other cells. One major immediate danger in this regard is the genetically engineered baculoviruses developed as insecticides, which are also simultaneously developed as vectors for human somatic gene therapy (see Section 4) which the Report has not even mentioned.

The Report has avoided any discussion of horizontal gene transfer to microbes and other organisms in the general environment, for which substantial evidence has emerged within the past three to four years. But there is still no explicit requirement to monitor for horizontal gene transfers during field releases. It is a blattant omission in view of already existing evidence that transgenic plants can transfer transgenes and marker genes horizontally to microbes in the soil (Schluter et al, 1995; Hoffman et al, 1994) and reviews of recent findings by many authors (dealt with in detail in Section 7.5) have shown that there is essentially no barrier to gene transfer between microorganisms. The microbes in the environment, in turn, serve as a gene transfer highway and reservoir for multiplication and recombination, from which the genes can spread to practically all other species. Particularly significant are new findings indicating that genetically "crippled" microorganisms can survive, or go dormant an d reappear, after having acquired genes horizontally from some species in the environment to enable them to grow and multiply; that naked DNA can survive for long periods in all environments and retain their ability to transform; that transformation frequencies are high in all environments. These findings have large implications for the safety of the releases from contained use, which is itself urgently in need of a full reassessment (Ho, 1997b). It is significant that the Norwegian Government has banned the imports of 2 rabies vaccines and 4 transgenic plants containing antibiotic resistance marker genes in September 1997, in recognition of the hazards arising from horizontal gene transfer and recombination.

On account of the possibilities of horizontal gene transfer, it is paramount that no organism containing antibiotic resistance marker genes, and in particular, unknown, uncharacterized foreign gene sequences, should be considered for release.

"Genetically Engineered Food - Safety Problems"
Published by PSRAST