John B. Fagan, Ph.D.


The use of genetic engineering in agriculture and food production has impacts, not only on the environment and biodiversity, but also on human health. Therefore, thorough biosafety assessment requires, not only evaluation of environmental impacts of genetically engineered organisms, but also assessment of the risks that genetically engineered foods pose for the health of consumers.


The hazards that may be introduced into foods through genetic engineering are three, (1) allergens, (2) toxins, and (3) reduced nutritional quality. This paper begins with a discussion of how genetic engineering may introduce these risks into foods, and then outlines the procedures for assessing whether or not a given genetically engineered food contains such hazards. In this discussion, foods, food ingredients, and food additives produced through recombinant DNA technologies will be called "genetically engineered," "recombinant," or "transgenic" foods, and the term food will be used to refer collectively to foods, food ingredients, food additives, and nutritional supplements.

Three sources of risk

The genetic engineering of foods involves the introduction of new genetic information into a food-producing organism. Some of the health risks associated with genetically engineered foods can be anticipated on the basis of what we already know about the characteristics of the food-producing organism in its unmodified state (called the unmodified organism UMO), from which the genetically engineered organism (GEO) is to be generated. Other aspects of the risks associated with genetically engineered foods can be deduced from the characteristics of the organism that is the source of the genetic information introduced into the food-producing organism (called the gene source or GS). For instance, if a gene derived from peanuts is introduced into a tomato, food produced from the resulting genetically engineered tomato might cause allergic reactions in people that are allergic either to tomatoes (the unmodified organism) or to peanuts (the gene source).

In addition to risks that can be foreseen by considering the characteristics of the UMO and gene source, there is another source of potential risks. This is the procedure of genetic engineering itself. Current recombinant DNA methods and those likely to be developed in the future are all capable of accidentally introducing unintended changes in the function and structure of the food producing organism. As a result, the genetically engineered food may have characteristics that were not intended by the genetic engineer, and that cannot be foreseen on the basis of the known characteristics of the unmodified organism or gene source. The mechanisms by which gene modifications can lead to such changes are discussed briefly below and described in detail in Section C.

Scope of safety testing

To protect the health and safety of the consumer, it is necessary to ascertain that all genetically engineered foods are free of allergens and toxins, and are unaltered in nutritional value before they are placed on the market. To assure the safety of genetically engineered foods, it is essential to test for health hazards derived from all three sources of risk presented above, (1) risks foreseeable based on the characteristics of the unmodified organism, (2) risks foreseeable based on the characteristics of the gene source, and (3) risks due to unintended changes in functioning of the food-producing organism caused by genetic manipulations, themselves.

At present, regulations in most countries governing the safety testing of genetically engineered foods focus almost entirely on health hazards that can be anticipated from the characteristics of the unmodified organism, and the gene source. However, accidental introduction of allergens and toxins through genetic manipulations and unintended alterations in nutritional value resulting from genetic manipulations constitute a very real source of health risk from genetically engineered foods. Therefore safety testing should be structured in such a way as to detect and eliminate products that contain these hazards. The discussion below presents science-based arguments that establish the actuality of these genetic engineering-induced hazards, and presents testing strategies capable of protecting consumers from these as well as other hazards.


1. Genetic engineering introduces into foods new proteins that can either directly or indirectly threaten health.

Genetic engineering introduces new genes, new genetic information, into the cells of a food producing organism. Since a gene is the blueprint for a protein, that new genetic information causes the organism to produce one or more new proteins. In turn, the food produced by that genetically engineered organism will contain those new proteins. Thus, genetic engineering introduces new ingredients, new constituents into foods.

The new proteins that genetic engineering introduces into foods can come from virtually any organism on earth, and most of these new proteins will never have previously been present in significant amounts in human foods. Because people have never before eaten these proteins, the effects that they might have on health will not be known. Thus, the only way to be sure that these foods are safe is to test them thoroughly.

What might be their possible harmful effects? These new proteins could, themselves, cause allergies or be toxic. Alternatively, they could alter the cellular metabolism of the food-producing organism in unintended and unanticipated ways, and in turn, these alterations in metabolism could cause allergens or toxins to be produced in the food.

Another possibility is that, as a result of these alterations in metabolism, the food-producing organism might fail to make some important vitamin or nutrient. Consequently, the genetically engineered food would lack important nutrients that are normally present in the corresponding, natural, non-genetically engineered food.

In summary, the new proteins produced in genetically engineered foods could:

a. Themselves, act as allergens or toxins

b. Alter the metabolism of the food producing organism, causing it to produce new allergens or toxins, or causing it to be reduced in nutritional value.

All of these possibilities and examples are discussed in greater detail in Section C.

2. Genetic engineering can create dangerous foods by generating mutations in the DNA of the food-producing organism.

Inserting a recombinant gene into the DNA of a food-producing organism disrupts the natural sequence of genetic information within that DNA. Thus, the process of genetic engineering causes mutations to the food-producing organism. These mutations are a second source of potential damaging effects of genetic engineering.

The location at which these mutations occur will be random, because genetic engineers cannot, by and large, control the site at which a recombinant gene is inserted into the DNA of the organism. They can cut and splice genes in the test tube with considerable precision, but the process of inserting those recombinant genes into the host is very imprecise.

These mutagenic events can cause damage in the following ways. First, the recombinant gene may be accidentally inserted into the middle of one of the more-than-one-thousand genes found in most food-producing organisms. This will disrupt that gene, and the organism will no longer be able to produce the protein for which that gene is the blueprint.

That protein might be an enzyme that is important in cellular metabolism. Disrupting the gene for that enzyme could alter cellular metabolism, possibly causing the organism to produce a toxic compound that accumulates in the food produced by the organism. Disrupting metabolism could also prevent the organism from producing certain vitamins or nutrients, and therefore reduce the nutritional value of the food.

Another possibility is that mutations caused by genetic manipulations could alter the expression of the genes of the food-producing organism. This could happen in at least two ways. First, mutationally induced alterations in metabolism, described above, could change the expression of other genes. Second, a genetically engineered gene might be inserted into the DNA very close to an important gene of the food-producing organism, thereby altering the expression of that gene.

These changes in gene expression could cause the food producing organism to produce ten times more or ten times less of an important protein or enzyme. This could cause a variety of problems. First, a protein that is not toxic or allergenic, when present at normal levels, might become toxic or allergenic if present at ten times higher levels. Second, if an important enzyme is produced at a level ten times higher or ten times lower than normal, this could drastically alter cellular metabolism, leading to the production of a toxin or an allergen, or to the inability to produce an important nutrient. Third, if the gene encodes an important regulatory molecule, such as a peptide hormone, producing it at higher or lower levels could disrupt important physiological processes, again leading to changes in food quality or safety.

It must be pointed out that the DNA of most food-producing organisms is very complex. Some of this DNA is in the form of genes, but many parts of the DNA of these organisms do not contain genes. At present the functions of these other portions of the DNA are not understood by science. Genetic engineers usually assume that genetic manipulations will be harmless if are restricted to the DNA of unknown function and avoid genes. However, this argument is really no better than saying that what you do not know cannot hurt you.

The fact that we do not currently understand the function of these DNA sequences does not mean that they do not have important functions. Therefore disrupting these sequences could have unanticipated, long-term or subtle effects that might not be immediately obvious, but that could be damaging to the species or to the quality of the food that it produces. Nature is parsimonious. Therefore, it is likely that these sequences have important functions, even though we do not presently know what they might be. It is unwise to assume that insertions into these sequences will be harmless.

In summary:

a. Mutations can damage genes naturally present in the DNA of an organism, leading to altered metabolism and to the production of toxins, and to reduced nutritional value of the food.

b. Mutations can alter the expression of normal genes, leading to the production of allergens and toxins, and to reduced nutritional value of the food.

c. Mutations can interfere with other essential, but yet unknown, functions of an organism's DNA.

The molecular mechanisms by which genetic engineering generates mutations and by which these mutations can generate allergens and toxins in food are delineated in more detail in Section C. Concrete examples of such allergens and toxins are presented, as well.

3. The damaging effects of genetic engineering cannot be predicted or controlled

The ability of genetic engineering to introduce unanticipated health hazards into foods derives from the fact that, although genetic engineers can cut and splice DNA molecules with base-pair precision in the test tube, when an altered DNA molecule is introduced into the genome of a living organism, the full range of its effects on the functioning of that organism cannot be controlled or predicted.

What this means is that, in addition to the changes in biological function intended by the genetic engineer, the introduced DNA may bring about other, unintended changes, some of which may alter the properties of the food produced by the organism in a manner that makes it damaging to health.

Although the potential health hazards of genetically engineered foods are not different from those associated with other foods, namely allergens, toxins, and reduced nutritional value, we show in Section C that the process of genetic engineering, itself is responsible for generating these dangers. That is, the use of the genetic engineering process introduces hazards into the resultant food. Thus, the use of genetic engineering in the development of a new food-producing organism, in itself, constitutes a valid regulatory trigger. Stated in another way, because there is a distinct class of risk that is directly and uniformly associated with the process by which genetically engineered foods are produced, that process-genetic engineering-can be used as a reliable flag for identifying foods that should undergo safety testing.

Proponents of biotechnology argue that the risk associated with genetically engineered foods is very small. However, there is no scientific evidence that this is the case. If one holds to the standards of the science of risk assessment, the existing body of data allows one only to state that, for a given genetically engineered food, the risk is finite, but of unpredictable magnitude. A real risk, especially one of unpredictable probability and severity, is something that requires testing.

To support the contention that risks are small, proponents attempt to infer the safety of future transgenic foods from the properties of genetically engineered foods now on the market. However, this is also not consistent with established principles of the science of risk assessment. Furthermore, even if such comparisons were valid, the handful of examples now available do not provide a sufficient database for such estimates; the diversity of possible genetic manipulations that could be carried out in the future, and the diversity of food-producing and gene-source organisms that could be employed in the genetic engineering of future foods is extremely large. Current examples simply are not representative of the range of possibilities that will emerge in the future. Thus, to assure safety, genetically engineered foods should be considered on a case-by-case basis and each should be tested thoroughly before it is placed on the market.


This section presents technical arguments regarding the following three points:

1. The inability to predict and control the outcome of gene manipulations.

2. The mechanisms by which genetic manipulations can generate allergens in foods.

3. The mechanisms by which genetic manipulations can generate toxins in foods.

This section may be of primary interest to those wishing to understand the technical details regarding the impact of genetic engineering on food safety. The arguments presented in the final section of this paper, which discusses procedures for testing the safety of genetically engineered foods, are not dependent on the details presented in the present section. Therefore it is not necessary to absorb this section before moving on the section on safety testing.

1. The inability to predict and control the outcome of gene manipulations

In the previous sections we point out that it is not possible to fully control and predict the outcome of genetic modifications of food organisms. The basis of this inability can be traced to the following three things, which are discussed in detail below.

a. The complexity of even the simplest food-producing organisms makes it impossible to predict the full range of effects resulting from changing even a single gene. Thus recombinant genes have unpredictable effects on the characteristics of the genetically engineered organism and the food it produces.

b. Recombinant DNA manipulations induce mutations at random locations within the genome of the recipient organism. These mutations have unpredictable effects on the characteristics of the genetically engineered organism and the food it produces.

c. Although structural genetic information is universal in its "meaning," regulatory genetic information differs in "meaning" depending on the cell type and type of organism into which it is introduced. Thus, the same regulatory genetic information will have different effects on the functioning of different organisms. Since these diverse effects have not been sorted our scientifically, the results from introducing a new gene into an organism cannot be fully predicted.

a. Biological complexity leads to the inability to control or predict the effects of recombinant DNA manipulations on food quality and safety

An important contributor to the unpredictability of genetic engineering is the complexity of food-producing organisms. Whether we examine the simplest single-celled microorganism, or a human being, or the global ecosystem, we find a huge number of complex components. These take part in extremely intricate, coordinated interactions, all as part of one, vast, integrated living phenomenon.

Within any one of the trillions of cells that make up the physiology of a food-producing organism there is another vast world of complex subcellular structures, organelles, molecular networks, and metabolic pathways, each composed of a variety of biomolecules. All work together in an integrated, interdependent manner.

Because all of these components are interconnected in their functioning, changes do not happen in isolation. The effects of adding just one gene to the system will ramify throughout the whole system, potentially influencing every aspect of cellular functioning and thereby every aspect of the organism as a whole. Because of the complexity of the system, it is impossible for the genetic engineer to even begin to predict how that initially small change will influence the functioning of the cell, not to mention the physiology and behavior of the organism as a whole.

In such a situation, surprises are inevitable, and many of those surprises will not be advantageous. For food-producing organisms, this translates into the possibility of unpredictable changes in food quality and safety. The mechanisms by which genetic manipulations can lead to increased allergenicity and toxicity, described below, provide examples of such surprises.

b. Mutations through Recombinant DNA Manipulations

The second source of uncertainty regarding the effects of recombinant DNA manipulations stems from the extremely crude nature of current gene transfer techniques. The genetic information introduced into the organism may be precisely defined in sequence, but it is inserted at random into the genome of the recipient organism. Each insertional event is in fact a random mutagenic event. Stated another way, gene transfer as it is commonly done is a mutagenic process that can disrupt any of the processes in which DNA and RNA participate. The sites at which such mutations occur will be random. Therefore, there is no way to predict which gene or regulatory processes will be disrupted as a result of gene transfer-induced mutagenesis.

By inactivating or altering the expression of genes encoding enzymes that catalyze important biosynthetic processes, mutagenic events could alter the allergenicity of a food or make it toxic, as described in detail below. These mutagenic events could also alter the nutritional qualities of a food. Furthermore, by altering regulatory sequences present normally in the recipient organism's genome, the same variety of regulatory sequence-related problems described below could be generated.

It should be pointed out that with most gene transfer methods used in eukaryotes, this mutational process will occur, not just sometimes, but every time a recombinant gene is inserted into the genome of an organism. Each such insertional event disrupts some native DNA sequence. Many such disruptions will, fortunately, be silent or inconsequential. However, there is a finite chance that one of these will alter the structure or function of the organism in a manner that significantly influences the properties of the foodstuff derived from it. That is, genetic alterations have a finite probability of altering the properties of the organism such that the properties of the food derived from it will be hazardous to health. In most cases, the procedures used in modification of food-producing organisms insert, not one, but several copies of a gene into the genome of the recipient organism. Thus, multiple random mutagenic events may occur, greatly increasing the probability of a damaging some gene important to food quality.

The risks related to manipulating the genomes of food-producing organisms are inherent in the mechanisms by which recombinant DNA techniques bring about genetic change. These risks cannot be discounted by pointing to the FlavrSavr tomato (the first genetically engineered crop to be commercialized) and saying that there have been no problems with it and therefore other transgenics will probably be safe, too. Each transgenic food-producing organism will undergo different mutagenic events, and respond to the genetic information introduced into it differently, leading to the range of unexpected alterations described above. Therefore there is no scientifically valid justification for such extrapolations.

c. Ambiguities of Genetic Information

Genes contain two distinct kinds of information, structural and regulatory. Structural information specifies the amino acid sequence of proteins, and consists of the genetic code, which was elucidated in the 1960's. This code is, with a few exceptions, identical for all terrestrial organisms. Thus, the structural information contained in a given piece of DNA is predictable.

However, the story is quite different for regulatory information. Transcription, translation, replication, recombination, and other processes involving DNA and RNA are controlled by regulatory information encoded in DNA or RNA sequences. The regulatory "code" is much more complex and diverse than the structural code. Furthermore, it is different in different organisms, and is even different in different cell types of the same organism.

For instance, there are many examples in the molecular biological literature in which recombinant genes, characterized in one cell type, are expressed at 100- or even 1000-fold higher levels in another cell type from the same organism. Such differences cannot be predicted simply by knowing the nucleic acid sequence of a recombinant gene. The only way to know is to gather empirical information-to actually introduce the gene into the second cell type and examine the result. If this is the case for different cell types within a single organism, the level of unpredictability will certainly be as great or greater for cross-species transfers of the kind commonly carried out in agricultural genetic engineering.

The underlying mechanism involved in the "reading" of regulatory information is well understood. Regulatory proteins exist in the cell, each of which is capable of scanning DNA (or RNA) molecules. Each can recognize and bind to a single, specific nucleic acid motif. That binding reaction triggers biochemical events leading to modulation of a process such as transcription, translation, replication, recombination, etc. In any particular cell, a given sequence can influence one of these processes only if the protein that recognizes that sequence is also present. Since different regulatory proteins are expressed in different cell types and in different species, a given DNA sequence will function as a regulatory signal only in some cell types and some species, and not in others. Our knowledge of the "regulatory code" is extremely incomplete. Therefore we cannot examine the sequence of a nucleic acid molecule and predict its regulatory function in a given organism.

Inserting DNA sequences that possess unanticipated regulatory activities into the genome of a food-producing organism could disrupt any of the cellular processes in which DNA or RNA participate, including replication, transcription, translation, recombination, transposition. Disruption of transcription or translation could alter the level of expression or the timing of the expression of any protein that is normally expressed in a food-producing organism. This could alter the allergenicity or toxicity of the food derived from that organism, as described below, and could also alter its nutritional or other characteristics. Disruption or alteration of replication, recombination, or transposition mechanisms could, among other things, alter the plasticity or stability of the recipient organism's genome, leading to increased rates of mutagenesis and consequently to a range of problems, as described below.

2. Allergens generated in recombinant foods

There exist several mechanisms by which allergens could be expressed in foods through genetic engineering. A number of molecular mechanisms have also been identified through which the genetic manipulation of food producing organisms could generate new allergens or increase the allergenicity of proteins normally present in food producing organisms. Because allergen-carrying transgenic foods will in most cases maintain the appearance of their natural, non allergenic counterparts, they pose a serious hazard to the consumer. Consumers will not be able to avoid these allergenic foods, because they will not be able to distinguish them from the corresponding natural foods. The labeling of all genetically engineered foods would, of course, solve this problem and would also make it possible for health authorities to trace allergen problems that arise.

At present, empirical evidence regarding the generation of allergenic foods through genetic engineering is sparse, since few of the genetically engineered foods now under development have been thoroughly tested for allergenicity. However, one example has already come to light: Pioneer Hybrid has developed soybeans with nutritionally balanced amino acid composition. They accomplished this by engineering into these beans the gene for a brazil nut storage protein. However, this protein turns out to be allergenic to a significant proportion of the population. Pioneer Hybrid has wisely decided to terminate plans to commercialize this product.

In addition to this limited empirical evidence regarding potential allergenicity of recombinant foods, a consideration of the mechanics of recombinant DNA manipulations and the fundamental principles of immunology, biochemistry, cell biology, molecular biology, and physiology uncovers a number of straightforward mechanisms by which genetic manipulations can alter the function of food producing organisms such that they express allergens. Based on this analysis, it is clear that genetic manipulations are capable of generating allergenic foods. The probability of such events is small but it is finite. Thus the risk of encountering dangerous allergens in novel transgenic foods is a very real one. The following are the most obvious mechanisms by which the use of recombinant DNA techniques could generate new food allergens:

(a) It is known that many foods contain components that are low level allergens or immuno-irritants. At these low levels they produce negligible or minor problems. As discussed earlier, when recombinant DNA techniques are used to introduce new genes into a food producing organism, those manipulations can inadvertently alter the levels of proteins that are normally present in that organism. If the expression of an allergen or immuno-irritant increases substantially, it may reach concentrations within the food that could induce serious allergenic responses.

In some cases, genes will be used in the development of transgenic food producing organisms that are derived from sources known to commonly cause allergic reactions. In such cases, foods derived from such GEOs should be assumed to be allergenic unless empirical evidence demonstrates that this is not the case. If positive evidence excluding allergenicity cannot be obtained, it is obviously necessary to label the transgenic food as potentially allergenic to alert sensitive consumers. In cases where the allergic reaction might be life threatening, such foods should not be introduced into the market place.

(b) Many of the recombinant proteins that will be expressed in food producing organisms are proteins that are not normally present in the food supply. Thus their potential allergenicity is unknown. There is a reasonable probability that some of these proteins will be allergenic. It should be pointed out that, since even trace amounts of some allergens are sufficient to induce powerful allergenic reactions, the fact that a genetically engineered substance may be present in only trace amounts does not necessarily eliminate the possibility that it could be allergenic. Because individuals will probably not have been previously sensitized to these new allergens, they will probably not elicit a powerful allergenic response on first exposure. However, if such an allergen becomes a common component of the food supply, allergenicity will develop as exposure continues.

(c) Recombinant modifications could alter the primary or secondary structure of some proteins in such a way as to increase their allergenicity or cause them to become allergenic. Thus, even though a protein may be commonly present in food and not allergenic, when recombinant DNA techniques are used to alter the gene encoding that protein, the resultant recombinant protein could be allergenic.

This is especially the case with fusion proteins, expressed from genes generated by linking pieces of coding sequences from two or more sources. These are proteins consisting of peptide segments derived from two or more proteins. This is accomplished by using recombinant DNA techniques to fuse together pieces of the genes for those proteins. The potential allergenicity of fusion proteins cannot be deduced from the properties of the parental proteins from which they are derived, and thus cannot be predicted or modeled. This is because the domains where two proteins are joined can assume conformations very different from those of either of the original proteins. Furthermore, the likelihood of generating allergenicity in fusion proteins is increased by the fact that the junctions at which two proteins are fused often assume secondary and tertiary structures that are not common in natural proteins, and are, therefore, more likely to be allergenic.

(d) Even though recombinant proteins will often be expressed in food-producing organisms at low levels compared to the total protein content of a food, it is likely that even these levels will be substantially higher-orders of magnitude higher in some cases-than the naturally-occurring levels of those proteins. Thus, even if previous research fails to uncover evidence for allergenicity, a protein may behave as an allergen when expressed at higher levels through genetic engineering.

(e) Different organisms possess different biochemical mechanisms for the processing of newly synthesized proteins. Therefore, a recombinant protein may be processed differently in the genetically engineered recipient organism than it was in the organism from which the gene for that protein was isolated, and in which that protein is naturally expressed. These differences in processing could result in the transgenic form of the protein having different allergenic properties than the naturally occurring form.

The risk that new allergens can be generated during the process of genetically engineering food-producing organisms leaves the developer of such organisms in a difficult position with regard to safety testing, because there is no adequate approach for generalized testing for potential allergenicity short of human testing.

In 1994, the US Food and Drug Administration, Environmental Protection Agency, and the Department of Agriculture hosted a "Conference on Scientific Issues Related to Potential Allergenicity in Transgenic Food Crops." The scientists selected by these agencies to attend this meeting and to advise them on the issue of allergenicity of genetically engineered foods indicated that it is likely that the use of recombinant DNA techniques in developing new crop varieties carries with it a significant possibility of generating unanticipated allergens.

They further pointed out that this is quite problematic for safety testing. While methods are available to assess whether a genetically engineered food contains hazardous amounts of known allergens, there are no direct or comprehensive methods for assessing the potential allergenicity of proteins derived from sources that are not known to produce food allergies. Scientists are reduced to comparing the general characteristics of such proteins to proteins that are known to cause allergic reactions, rating their similarities in terms of characteristics such as amino acid sequence, resistance to enzymatic or acidic degradation, heat stability, and molecular weight. These are of course extremely general characteristics and are unlikely to yield definitive information regarding the ability of a protein to elicit a biological reaction as complex as the immune response.

In this situation, the only viable approach to eliminating significant risk of allergenicity in a new transgenic food is to require that human testing be carried out before commercialization, and that commercialization be accomplished in two phases, a limited test market phase which includes careful monitoring of potential allergenicity, followed by full scale commercialization. This strategy assures that all except extremely rare allergens will be detected before the public is widely exposed to a new transgenic food.

3. Toxins and irritants generated in recombinant foods

Most substances that will occur in foods as a result of genetic engineering will be proteins that will be present in only trace concentrations. Never-the-less, those added components, in even trace amounts, may substantially alter either the nutritional or other biological characteristics of the food. Examples of this are presented above with regard to allergenicity. In addition to allergenicity, recombinant proteins could manifest a variety of other biological activities, and, in the case of recombinant enzymes, could catalyze the production of other compounds with biological activities not normally present in a particular food. For instance such substances could act as toxins, irritants, hormone mimetics, etc., and could act at the biochemical, cellular, tissue, or organ levels to disrupt a range of physiological functions.

An example of a class of genetically engineered foods that are of particular concern are those that have been modified to produce biological control agents, such as the family of insecticidal Bt enterotoxins. Each of the Bt toxins is specific for a certain class of insects. The Btk toxin, which has been used topically in organic farming for many years, has not been reported to cause toxic reactions in consumers when used in this way. However, it would not be surprising if a compound, such as Btk toxin, that has powerful biological activity in one class of organisms might also have some biological activity even in a distant phylum such as the vertebrates. Such activity might become apparent if the toxin is consumed in larger amounts, as will occur in transgenic foods derived from organisms engineered to express this toxin constitutively at high levels.

Normally when used topically Bt toxin is degraded to undetectable levels by solar UV light and other mechanisms in just a few days. However, Bt-engineered plants produce this toxin continually, resulting in much higher steady-state levels. Furthermore, the toxin will be present, not only on the surface of the plant but internally, where it may be protected from degradation by UV light and accumulated. The result is that consumers of these foods may consume much larger amounts of Bt toxin than is the case with foods derived from plants treated topically. Consequently, the excellent safety record of topically applied Bt toxin does not constitute reliable evidence indicating that foods derived from plants genetically engineered to produce Bt toxin will be safe.

Some of the potential biological characteristics of genetically engineered proteins or their metabolites can be assessed by simple in vitro tests. However others cannot, and by the very diversity of possible effects and the complexity of the physiology, it is impossible to carry out laboratory experiments that will exhaustively, thoroughly, and conclusively establish that a genetically engineered food is free of such toxins, and therefore safe. In all cases, a finite probability will remain that some toxin or other biologically active molecule has been generated in the recombinant food for which no adequate test is available. Therefore, it is crucial that each transgenic food be tested for toxicity in human volunteers. Without these experiments, there will always be an appreciable probability that the toxic properties of some genetically engineered foods will not become apparent until that food is placed on the market and the health of consumers is damaged.

An example of this has already come to light, although there is controversy regarding details. The company Showa Denko genetically engineered a microorganism to produce L-tryptophan at high levels. The enzymes expressed in this bacterium through genetic manipulations were not present in massive amounts, but they altered the cellular metabolism substantially, leading to greatly increased production of tryptophan. This organism was immediately used in commercial production of L-tryptophan, and the product placed on the market in the USA. Within two months, 37 people died and 1500 were permanently disabled from using this product. This was, evidently, due to the presence of traces of a toxic contaminant. This contaminant was extremely powerful, since the preparation was at least 98.5% tryptophan.

This contaminant was later identified as a dimerization product of L-tryptophan. Based on fundamental chemical and biochemical principles, scientists have deduced that this compound was generated within the bacteria when internal tryptophan concentrations reached such high levels that tryptophan or its precursors began to undergo side reactions that led to dimerization. Thus, it appears that genetic manipulations led to increased tryptophan biosynthesis, which led to increased internal tryptophan levels. At these high levels, side-reactions occurred, which generated a deadly toxin. Being highly similar to tryptophan, itself, this toxin co-purified with tryptophan, contaminating the final commercial product at levels that were highly toxic to consumers.

Some ambiguity remains regarding this story, since the company destroyed all samples of the recombinant organism as soon as the problem was recognized, and since the company acknowledges cutting corners on the purification procedure used in preparing the batches of tryptophan that turned out to be toxic. Since the recombinant bacterium is no longer available, the definitive experiments cannot be done to resolve which variable-recombinant manipulations or purification procedure-was primarily responsible for the presence of the toxin in commercial batches of L-tryptophan. However, in addition to the biochemical rational presented above, two additional arguments point to recombinant manipulations at the culprit: (1) No evidence exists indicating that the parental bacterium produces this toxin. Thus the ability to produce it must have been conferred by genetic modifications. (2) The tryptophan produced by other manufacturers, who used natural bacteria, was not toxic, even though it is likely that they were also cutting corners in their purification procedures from time to time. We conclude that it is likely that genetic engineering was the determinate factor in generating this toxin.


In the preceding section we have reviewed in brief, how genetic engineering is capable of introducing dangerous allergens and toxins into foods and reducing nutritional quality (see Section C for further details). It is, of course, important to test for these hazards before a genetically engineered food is placed on the market.

Recognizing that a range of safety testing needs will be encountered, Flowchart VII presents three schemes for testing genetically engineered foods for allergens, toxins, and alterations in nutritional quality. These are intended to illustrate the range of possible approaches for assessing the safety and nutritional quality of recombinant foods. By selecting various elements from this range of possibilities, the regulator can adjust factors, such as stringency of safety assurance, timeliness of assessment, and cost of assessment, to design a protocol that meets local needs and expectations.

Known food allergens, known toxins, and nutritional quality can all be evaluated in a straightforward manner, employing well established in vitro analytical methods. All three testing strategies use the same procedures for this purpose. These are summarized in Flowchart VII and presented in detail in Flowcharts VIIIA (common food allergens), VIIIB (common toxins), and IX (assessment of nutritional quality).

Assessing unanticipated allergens and toxins is more challenging. It is in this area that the three testing strategies described below differ. These strategies employ various combinations of the following three approaches to detecting and characterizing allergens and toxins:

1. In vivo testing using small animals and human subjects for the purpose of screening broadly for allergens and toxins,

2. Molecular characterization of the genetic alterations induced through recombinant DNA modifications,

3. Controlled and monitored commercial release of recombinant foods.

Strategies IA and II both emphasize in vivo testing, and monitored marketing, but differ in the extent of in vivo testing, and thus in the degree of certainty with which one can claim that a given genetically engineered food is safe for human consumption. Strategy IA employs a graduated sequence of in vivo tests that will yield a high degree of certainty that a given genetically engineered food will be free of both long term and short term damaging effects for a high proportion of the population. Strategy II employs much smaller subject populations and less extensive in vivo testing, but still provides substantial protection for the consumer when genetically engineered food is cleared for full-scale marketing.

Strategy IB uses the same level of in vivo testing required in Strategy II, but includes extensive molecular characterization of the GEO and the genetically engineered food. This characterization is designed to assess the extent to which genetic engineering has altered the expression of other genes native to the organism, and/or changed the function or regulation of the metabolic and biosynthetic pathways of the GEO. It is felt that this combination of molecular and in vivo studies provides a safety margin similar to that attained with Strategy IA.

Strategy III is modeled after current regulations in the US. In this strategy, the developer is only required to carry out routine nutritional analyses and to test for common allergens and toxins that are suspected to be present on the basis of the identity or the UMO or on the basis of the source of the recombinant gene(s) used in generating the GEO. This testing scheme fails to assess the possibility of unanticipated but potentially dangerous allergens or toxins, because it does not include procedures, such as in vivo feeding tests with animal or human subjects, that are capable of detecting a wide range of toxins or allergens. Only if concrete evidence is available implying that unanticipated allergens or toxins may be present does this strategy require the developer to carry out further characterization.

In addition, Strategy III does not employ monitored marketing. Thus, genetically engineered foods carried through this testing scheme enter the market without any testing for novel allergens and toxins. Since, in this scheme genetically engineered foods need not be labeled, the consumer is even deprived of the choice of avoiding these inadequately tested genetically engineered foods.

Each of these testing strategies is discussed below. The stages of testing common to all three testing strategies are presented first, followed by discussion of the unique features of each strategy.

Flowchart VIIIA-In vitro screening for common food allergens

This flow chart presents a plan for testing for known food allergens. Three basic questions will be asked:

1. If food derived from the unmodified organism (the UMO), from which the transgenic organism was generated, is known to contain allergens, are the levels of these allergens in the GEO within the norms expected for the UMO?

2. If the transgene is derived from an organism that expresses allergens, were those allergenic determinants transferred to the GEO via the transgene?

3. Are other common food allergens present in the food derived from the GEO?

These questions will be answered using the standard laboratory test for allergens, which involves assessment of the reactivity of the test substance with immunoglobin E or sera active against the allergen of interest. These procedures are limited by the fact that they cannot provide information on novel allergens.

Positive responses lead to more detailed characterization of the level/activity of the antigenic material detected in the food. The results of these studies, along with clinical data regarding this common allergen, obtained from the literature, are then used to decide whether the food is acceptable for human use, and to formulate labeling and use instructions.

Flowchart VIIIB-In vitro screening for known toxins

This flow chart presents a plan for testing for known food toxins. Two basic questions will be asked:

1. If food derived from the unmodified organism (the UMO), from which the transgenic organism was generated, is known to contain toxins, are the levels of these toxins in the GEO within the norms expected for the UMO?

2. If the transgene is derived from an organism that expresses a toxin, was that toxin transferred to the GEO via the transgene, or were genes for enzymes critical to the synthesis of that toxin transferred?

These questions will be answered using specific analytical tests for that toxin. If the toxin is detected, further analytical work will be done to quantitate the level/activity of the toxin in the genetically engineered food. In conjunction with clinical data, these analyses will serve as the basis for deciding if the genetically engineered food is appropriate for commercialization, and to formulate labeling and use instructions.

Flowchart IX-Evaluating Nutritional Quality of Transgenic Foods

If the transgenic food lacks known allergens and toxins, its potential acceptability for commercialization is high. Thus, it would be justified to invest the time and resources necessary to carry out thorough nutritional analysis.

Alterations in metabolism due to genetic modifications may lead to changes in the nutritional composition of the transgenic food, compared to norms for the corresponding natural food. Some changes may be direct and intended consequences of a given genetic modification. In such cases, specific measurements should be carried out to quantitate the extent to which the developer has succeeded in accomplishing those intended changes. In other cases, secondary, untended alterations in nutritional content, composition, or bioavailability will occur. Before the transgenic food is placed on the market, it is incumbent upon the developer to detect and quantify such changes in common nutrients and vitamins, at least if they are substantial. For this purpose, the following studies are recommended:

1. Standard quantitative methods will be used to assess common nutrients contained in the transgenic food. This analysis will include: quantity, composition, and bioavailability of protein, fats, carbohydrates, major vitamins, and trace elements.

2. The nutrient content of the transgenic food will be compared to that of the corresponding natural food and to norms for that foodstuff. For instance we would want to know if a genetically engineered tomato contained vitamin C levels equivalent to the unmodified variety of tomatoes from which it was derived and contained levels within the range that is typical of other tomatoes. Significant differences in nutrient content between the transgenic and natural food should be stated clearly on the label, and if radical changes are found, the transgenic food should be given a common name that distinguishes it from the corresponding natural food.

3. Some foods are recognized as primary sources for certain nutrients. Transgenic forms of those foods will be tested in detail to assess possible changes in quantity or quality of such nutrients. Significant differences should be highlighted on the label.

4. Nutritional questions of particular relevance to a given GEO will also be explored in more detail. For instance, the FlavrSavr tomato was promoted on the basis of extended shelf life. It is therefore necessary to quantitatively evaluate the persistence of important nutrients in this tomato over the full range of the claimed storage life of this product. That is, because this tomato appears attractive to the consumer for a long period of time, it is necessary to objectively ascertain whether or not nutritional value is preserved, as well, throughout that period of time. If nutritional value is not preserved, this should be stated on the label.

Flowcharts XA and XB-Strategy IA: High stringency in vivo testing to detect possible unanticipated allergens and toxins

After the in vitro tests described in Flowcharts VIII and IX have been used to assess the presence of common allergens and toxins, the in vivo tests, described in Flowchart XA and XB, can be carried out to establish that a given genetically engineered food is free of novel, unexpected allergens and toxins.

Most governments specify rigorous, standardized protocols for testing the toxicity and allergenicity of novel substances categorized as drugs and food additives. Because genetic engineering introduces new genetic material, and therefore new constituents, into foods, it is reasonable to test all transgenic foods with the same rigor required for these other novel substances. This point is discussed in greater detail in earlier sections.

The testing strategy presented in Flowchart XA and XB is designed to accomplish this aim. It is adapted from typical governmental standards for testing novel drugs and food additives. These standards have been modified to meet specific needs unique to the evaluation of the safety of transgenic foods. This strategy relies on in vivo animal and human studies as a method of testing allergenicity and toxicity more broadly than is possible with in vitro, biochemical or immunological tests. In order to minimize the risk to human test subjects, safety tests progress from animal studies, to small scale human studies, to larger scale trials, and finally to test-marketing of the transgenic food in selected locations with careful monitoring. Successful completion of these studies lead to full-scale commercialization.

Information from each step of this evaluation will be used in two ways. First, it will contribute to the decision whether or not to commercialize the transgenic food under study. Second, it will provide information relevant to the labeling and use of the final product, if commercialization is permitted.

Stage I-Animal studies

Short term animal tests are first carried out to eliminate genetically engineered foods containing powerful toxins or allergens before humans subjects are exposed to them. This series of studies will test the novel food in mice for (a) acute effects observed from feeding at maximum feasible doses for 48 hours and, if possible, for up to 2 weeks and (b) sub-acute effects resulting from feeding at levels proportional to maximum dietary levels in humans for up to 90 days. It should be possible to complete these tests within a period of 120 days. In these studies, standardized protocols will be followed which will evaluate the following parameters:

Maximum tolerated dose, based on autonomic signs

Central nervous system effects

Cardiovascular effects

Metabolic effects

Allergy/Inflammatory effects

Gastrointestinal effects

Because there are physical, chemical, and physiological limitations to the amount of a food that can be administered through feeding, it will not be possible in short term experiments such as these to detect toxins and allergens with the same sensitivity as can be done in toxicological experiments, in which extremely large doses can be administered. Thus, it is of paramount importance to carry out longer term experiments to be assured that a given food is free of significant toxins and allergens. Three dosage levels will be used in in vivo studies: (1) normal dietary level (NDL)-the amount of the food typically consumed by humans in a single meal; (2) maximum dietary level (MDL)-the maximum amount that can be consumed daily on a long term basis (limited, among other things, by the need to consume other foods to meet nutritional requirements, which can be calculated on the basis of nutritional data for the genetically engineered food); (3) maximum feasible intake (MFI)-the maximum amount that can be consumed short term (usually limited by the physiological capacity of the subject). For animal studies, NDL and MDL will be calculated from human doses scaled down proportionally to body weight.

Stage II-Short term, high dose human studies

The objective of this work is initial assessment of the safety of the transgenic food in humans. This work will be done using 20-50 normal volunteers. In an in-patient unit, these volunteers will be administered escalating single and multiple doses of the food until toxic effects are observed or the maximum feasible level, described above, is attained. That dose is continued for 48 hours, monitoring vital signs, physiological parameters, blood chemistry, etc. If toxic effects are not observed during that period, dosage will be dropped to the maximum dietary level, defined above, and the experiment continued for up to 90 days. If at any time serious negative effects are observed, the experiment will be terminated.

These studies will not only define toxic or allergenic levels of the food, but will also provide information concerning the nature of any toxic or allergenic effects that occur in humans. This work can be completed within 4 months.

Stage III-Medium term, moderate dose human studies

These studies assess toxic or allergenic responses that manifest within 6 months, and that occur in the general population at a frequency of greater than 1%. In these studies, 100 to 500 subjects will be fed the genetically engineered food at maximum dietary level for up to 6 months. Vital signs, physiological parameters, blood chemistry, etc. will be monitored, and, if necessary, dosage will be manipulated in response to changes in tolerance of the subjects. This phase of safety testing should be completed within a period of 8 months.

Stage IV-Long term human studies

These studies will assess longer-term effects, and because a larger subject population will be used, will be capable of identifying small subpopulations (less than 0.1% of the population) that may have special problems with the transgenic food under study. Depending on the nature of the food and on the outcome of earlier stages of testing, 1000 to 3000 human volunteers will be fed the genetically engineered food at normal dietary levels daily for 1.5 to 2 years. In addition to vital signs, physiological parameters, and blood chemistry, this study may also provide information on the effects of the food on reproduction and cancer incidence.

Stage V-Test-Marketing with Health-Impact Monitoring

If human trials indicate the safety and desirability of a transgenic food, it will next be test-marketed in selected areas, with careful monitoring to detect impacts on the health of consumers. The monitoring system will include two important elements. (1) Hospitals and other medical facilities in the area will be alerted to the trial, and asked to report any health problems that might be related to consumption of the experimental, genetically engineered food. (2) The transgenic food will be labeled. The label will : (a) clearly designate the food as genetically engineered; (b) specify the source species from which genetic material was obtained to assemble the recombinant DNA molecule(s) used in constructing the transgenic organism; (c) describe the unique nutritional or other characteristics of the transgenic food; (d) tell the consumer that the product is experimental and ask the consumer to report any possible health impacts, minor or major, to the developer; (e) provide a mechanism that will allow the consumer to report health impacts to the developer conveniently and without incurring cost (such as a toll-free telephone number, or local contact address). This phase of safety testing will continue for 2 to 3 years.

Full-scale marketing

Even after test-marketing is complete, labeling is required, not only for continued monitoring of safety, but also to provide the consumer with sufficient information to make informed purchasing decisions.

Labeling should: (1) specify that the product is genetically engineered; (2) indicate any unique characteristics of the transgenic food relative to the natural counterpart; (3) provide a mechanism for consumer feed-back to the developer; (4) provide information on special handling or preparation requirements.

Flowchart XC-Strategy IB: Safety assessment emphasizing molecular characterization of the genetically engineered organism.

Strategy IB, presented in Flowchart XC uses a combination of in vivo tests and molecular characterization of the GEO to eliminate several classes of unanticipated allergens and toxins that might be present in the genetically engineered food. This molecular characterization should reveal if genetic engineering has disrupted the normal gene structure or normal gene expression of the GEO in obvious ways. If such disruptions have not occurred, then unanticipated changes in the quality or safety of the genetically engineered food are less likely. In the light of such evidence, it should be possible to reduce the extent of in vivo testing without compromising the safety of human subjects and consumers.

The first step in this analysis is to consider the nature of the transgene. If it encodes a protein(s) that is known to be non-toxic and non-allergenic, and if it is unlikely to catalyze reactions that modify cellular metabolism in such a way as to generate toxins, then one class of risks is eliminated.

The possibility remains, however, that the transgene or its protein product, might modify cellular gene expression, thereby causing the GEO to produce new toxins or allergens or higher levels of toxins or allergens than the UMO (via mechanisms discussed in detail earlier and in even greater detail in Section C). This possibility is explored by investigating the following three questions:

1. Does the insertion site of the transgene interrupt one or more open reading frames within the DNA of the GEO? If an open reading frame is interrupted, then the expression of at least one gene is blocked. The question then arises as to the identity of that gene and the function of the protein that it encodes, and the actual effect of the loss of that protein on the metabolism and regulation of the GEO.

2. If there are mRNAs actively expressed from the sequences within the 20 kb domains flanking the insertion site of the transgene, are the levels and patterns of expression of those mRNAs unchanged in the GEO, compared to the UMO? The genes whose expression is most likely to be disrupted by the inserted transgene are those nearby. This experiment evaluates the expression of these genes directly.

3. Is the transgene expressed in the parts of the GEO that are normally used for food?

If these questions all yield negative answers, then it is unlikely that the gene expression of the GEO is significantly different from that of the UMO, and there is sufficient confidence that the genetically engineered food is safe to advance to monitored marketing (Stage V).

If any one of these questions yields a positive answer, then it is necessary to assess gene expression more fully in the GEO. For this purpose we propose the use of the differential display technique or another method capable of exhaustively comparing the mRNA profile of the GEO to that of the UMO. If no significant changes in the mRNA profile are observed, it can be concluded that the genetic alterations carried out have not significantly disrupted gene expression in the GEO. These techniques require significant technical expertise, and are somewhat time consuming. However, in some cases the developer will prefer this approach to in vivo testing. If no significant alterations in mRNA expression profile are detected, the developer can proceed directly to monitored marketing. If significant alterations are observed, in vivo testing is required, both Stage I and Stage II, before proceeding to monitored marketing.

Flowchart XI-Strategy II: Safety assessment employing limited in vivo testing of the genetically engineered organism.

This strategy is similar to Strategy IA, except only short-term human testing is carried out with a small number of subjects. All of the stages included in this strategy are discussed in detail earlier. This strategy will eliminate those toxins and allergens that are most severe in their effects. Although this approach is less costly and time consuming, its use carries the disadvantage of placing a significantly larger number of people at risk than would be the case with Strategy I. Because fewer than 100 subjects are used, allergens or toxins to which fewer than 1% of the population are susceptible will not be detected. Furthermore, because Strategy II employs only short term studies, serious long term problems would also remain undetected.

Flowchart XII-Strategy III: Safety assessment employing primarily in vitro analysis.

As discussed earlier, Strategy III focuses only on common allergens and toxins. It investigates the possibility of other allergens and toxins only in cases where circumstantial evidence happens to arise indicating their presence. As implemented in the US, this approach does not require that genetically engineered foods be labeled. We strongly recommend as a minimal safety measure that all genetically engineered foods be labeled so that any problems that arise can be traced.



The Author

John Fagan has spent more than 24 years using cutting edge molecular genetic techniques in cancer research. He earned a B.S. (cum laude with distinction in chemistry) from the University of Washington and a Ph.D. in biochemistry and molecular biology from Cornell University. He then spent 7 years doing research in molecular biology at the National Institutes of Health, first as a postdoctoral fellow, and subsequently leading his own research group from 1980 to 1984. In 1984, Dr. Fagan moved his research laboratory from the National Institutes of Health to Maharishi University of Management (then Maharishi International University), where he is now Professor of Molecular Biology and Biochemistry, Chairman of the Department of Chemistry, Co-director of the Physiology and Molecular and Cell Biology Ph.D. Program, and Dean of the Graduate School.

During his years at Maharishi University of Management, Dr. Fagan has received more than $2.5 million in grants from the National Cancer Institute of the National Institutes of Health. These grants supported research whose long-term goal was to identify cancer susceptibility genes and to understand how carcinogens and environmental pollutants, such as dioxin, influence gene expression. He has authored more than 30 technical articles on these topics, which have been published in internationally recognized, peer-reviewed journals, including Molecular and Cellular Biology, The Journal of Biological Chemistry, and Biochemistry. From 1991 to 1995 Dr. Fagan was the recipient of a Research Career Development Award from the National Cancer Institute, which is given to enhance the research development of promising scientists.

In recent years Dr. Fagan has been increasingly concerned about the dangers of genetically engineered foods, the hazards of releasing genetically engineered organisms into the environment, and the risks of germ-line genetic engineering in humans. In November of 1994, he took an ethical stand against these applications, urging scientists to take safer, more productive research directions, and to focus more on prevention and less on high-tech therapeutics. He underscored these warnings by returning a $613,882 grant to the National Institutes of Health and withdrawing grant applications worth another $1.25 million. These would have further supported research that might have contributed indirectly to the development of germ-line genetic engineering in humans. He has now redirected his own research to study the natural health promotion and disease prevention strategies of Maharishi's Vedic Approach to Health. He recently published a book on this topic, Genetic Engineering: The Hazards, Vedic Engineering: The Solutions.

Dr. Fagan has served as a scientific consultant on health and environmental issues, and as an editorial advisor and reviewer for scientific journals. He has also served on committees for the peer-review of federal government-sponsored research grants.

He is a frequent speaker at international scientific conferences, and to organizations with interests in the environment, agriculture, and health, as well as to students and civic groups, and to international organizations. Topics include the hazards of genetically engineered foods, and the risks of genetic engineering in medicine and agriculture. He discusses, not only the health and environmental hazards, but also social, cultural, economic, and human rights impacts of genetic technologies, and the ethical issues associated with genetic engineering and biotechnology. His current research activities , which use rigorous biomedical approaches to evaluate preventive and natural approaches in health care and to assess the effects of genetic engineering on agriculture and food safety, are other topics.

Currently, Dr. Fagan is conducting a global campaign to alert the public to the hazards of genetically engineered foods. The goal of this campaign is to reshape national and international policy and regulations regarding the safety-testing, labeling, and importation of genetically engineered foods. In recent months this work has taken Dr. Fagan to the capital cities of Austria, Belgium, Croatia, Denmark, Finland, France, Germany, Ireland, Italy, Netherlands, Northern Ireland, Norway, Sweden, Switzerland, and the United Kingdom, as well as Canada and the United States, where he has conducted meetings with legislators, representatives of national governments, the food industry, the press, and the public. He has also made presentations to international regulatory bodies, such as committees of the Codex Alimentarius Commission and the Convention on Biological Diversity. These meetings and the numerous newspaper articles, and TV and radio shows resulting from them, have contributed to significantly increased public awareness of the dangers of genetically engineered foods and to moving policy in a safer direction.

Address: Maharishi University of Management, Fairfield, Iowa 52557-1078; Phone: 515-472-1111 or 472-8342; Fax: 515-472-5725; email