Nature
October 7, 1999
Showing that a genetically modified food is chemically
similar to its natural counterpart is not adequate evidence that it is safe for
human consumption
Whenever official approval for the introduction of genetically modified (GM) foods has been given in Europe or the United States, regulatory committees have invoked the concept of 'substantial equivalence'. This means that if a GM food can be characterized as substantially equivalent to its 'natural' antecedent, it can be assumed to pose no new health risks and hence to be acceptable for commercial use. At first sight, the approach might seem plausible and attractively simple, but we believe that it is misguided, and should be abandoned in favour of one that includes biological, toxicological and immunological tests rather than merely chemical ones.
The concept of substantial equivalence has never been
properly defined; the degree of difference between a natural food and its GM
alternative before its 'substance' ceases to be acceptably 'equivalent' is not
defined anywhere, nor has an exact definition been agreed by legislators. It is
exactly this vagueness which makes the concept useful to industry but
unacceptable to the consumer. Moreover, the reliance by policymakers on the
concept of substantial equivalence acts as a barrier to further research into
the possible risks of eating GM foods.
Acceptable daily intake
The concept of substantial equivalence emerged in response
to the challenge confronting regulatory authorities in the early 1990s.
Biotechnology companies had developed several GM foods and, to reassure their
customers, wanted official approval for their introduction. But government statutes
did not cover GM foods, nor provide the authority to regulate these
innovations. Legislation could be amended, but that would not address the core
problem of how to assess the risks. One obvious solution at that time would
have been for legislators to have treated GM foods in the same way as novel
chemical compounds, such as pharmaceuticals, pesticides and food additives, and
to have required companies to conduct a range of toxicological tests, the
evidence from which could be used to set 'acceptable daily intakes' (ADIs).
Regulations could then have been introduced to ensure that ADIs are never, or
rarely, exceeded.
From the point of view of the biotechnology industry, this
approach would have had two main drawbacks. First, companies did not want to
have to conduct toxicological experiments, which would delay access to the
marketplace by at least five years, and would add approximately $25 million per
product to R&D costs. Second, by definition, using ADIs would have
restricted the use of GM foods to a marginal role in the diet. An ADI is
usually defined as one-hundredth of the highest dose shown to be harmless for
laboratory animals. Thus, even if laboratory animals show no adverse effects on
a diet consisting exclusively of a test material, human intake would still be
restricted to 1% of the human diet. The biotechnology companies want to market
GM staples, such as grains, beans and potatoes, which individually might
account for as much as 10% of the human diet, and collectively might provide
more than half of a person's food intake.
The adoption of the concept of substantial equivalence by
the governments of the industrialized countries signalled to the GM food
industry that as long as companies did not try to market GM foods that had a
grossly different chemical composition from those of foods already on the
market, their new GM products would be permitted without any safety or
toxicology tests. The substantial-equivalence concept was also intended to
reassure consumers, but it is not clear that it has served, or can serve, that
purpose. Although toxicological and biochemical tests, and their
interpretation, are notoriously problematic and contested, and are slow and
expensive, they can provide information vital to consumer protection.1
Trying to have it both ways
The challenge of how to deal with the issue of risk from
consuming GM foods was first confronted in 1990 at an international meeting,
consisting of officials and industrialists but no consumer representatives, of
the UN Food and Agriculture Organisation (FAO) and the World Health
Organisation (WHO).2 The FAO/WHO panel report makes intriguing reading, because
what it fails to mention is as important as what is discussed. It does not use
the term 'substantial equivalence' or mention ADIs. It implies that GM foods
are in some important respects novel, but it then argues that they are not
really novel at all ? just marginal extensions of traditional techniques. These
inconsistencies are inevitable, given that the industry wanted to argue both
that GM foods were sufficiently novel to require new legislation and a major
overhaul of the rules governing intellectual property rights to allow them to
be patented, yet not so novel that they could introduce new risks to public or
environmental health.
The biotechnology companies wanted government regulators to
help persuade consumers that their products were safe, yet they also wanted the
regulatory hurdles to be set as low as possible. Governments wanted an approach
to the regulation of GM foods that could be agreed internationally, and that
would not inhibit the development of their domestic biotechnology companies.
The FAO/WHO committee recommended, therefore, that GM foods should be treated
by analogy with their non-GM antecedents, and evaluated primarily by comparing
their compositional data with those from their natural antecedents, so that
they could be presumed to be similarly acceptable. Only if there were glaring
and important compositional differences might it be appropriate to require further
tests, to be decided on a case-by-case basis.
Unfortunately, scientists are not yet able reliably to
predict the biochemical or toxicological effects of a GM food from a knowledge
of its chemical composition. For example, recent work on the genetics of
commercial grape varieties shows that, despite detailed knowledge going back
for centuries of the chemistry and flavour of grapes and wines, the
relationship between the genetics of grapes and their flavour is not yet
understood3. Similarly, the relationship between genetics, chemical composition
and toxicological risk remains unknown. Relying on the concept of 'substantial
equivalence' is therefore just wishful thinking: it is tantamount to pretending
to have adequate grounds to judge whether or not products are safe.
The results of Arpad Pusztai's experiments with GM potatoes
and their interpretation remain a matter of controversy (see Nature 398,
98;1999), but his starting hypothesis was that GM potatoes would be
substantially equivalent to non-GM potatoes. Pusztai interpreted his still
unpublished results as indicating that the GM potatoes exerted adverse
biochemical and immunological effects, which could not have been predicted from
what was known of their chemical composition. The kinds of experiments which he
conducted are not legally required and are therefore not routinely conducted
before GM foods are introduced into the food chain.
Failure to define 'substantial equivalence'
The concept of 'substantial equivalence' was first introduced
in 1993 by the OECD4, and was subsequently endorsed in 1996 by the FAO and WHO.
Given the weight the concept has been required to carry, it is remarkable how
ill-defined it remains, and how little attention has been devoted to it. The
OECD document states:
"For foods and food components from organisms developed
by the application of modern biotechnology, the most practical approach to the
determination is to consider whether they are substantially equivalent to
analogous food product(s) if such exist....The concept of substantial
equivalence embodies the idea that existing organisms used as foods, or as a
source of food, can be used as the basis for comparison when assessing the
safety of human consumption of a food or food component that has been modified
or is new."
That is the closest there has been to an official definition
of substantial equivalence, but the definition is too vague to serve as a
benchmark for public health policy.
GM glyphosate-tolerant soya beans (GTSBs) illustrate how the
concept has been used in practice. The chemical composition of GTSBs is, of
course, different from all antecedent varieties, otherwise they would not be
patentable, and would not withstand the application of glyphosate. It is quite
straightforward to distinguish, in a laboratory, the particular biochemical
characteristics which make them different. GTSBs have, nonetheless, been deemed
to be substantially equivalent to non-GM soya beans by assuming that the known
genetic and biochemical differences are toxicologically insignificant, and by
focusing instead on a restricted set of compositional variables, such as the
amounts of protein, carbohydrate, vitamins and minerals, amino acids, fatty
acids, fibre, ash, isoflavones and lecithins. GTSBs have been deemed to be
substantially equivalent because sufficient similarities appear for those
selected variables.
But this judgement is unreliable. Although we have known for
about 10 years that the application of glyphosate to soya beans significantly
changes their chemical composition (for example the level of phenolic compounds
such as isoflavones5), the GTSBs on which the compositional tests were
conducted were grown without the application of glyphosate6. This is despite
the fact that commercial GTSB crops would always be treated with glyphosate to
destroy surrounding weeds. The beans which were tested were, therefore, of a
type which would never be consumed, while those which are being consumed were
not evaluated. If the GTSBs had been treated with glyphosate before their
composition was analysed it would have been harder to sustain their claim to
substantial equivalence. There is a debate in the research community on whether
such changes to the chemical composition are desirable or undesirable, but it
is an issue which remains unresolved, and which has been neglected by those who
have deemed GTSBs and non-GM soyabeans to be substantially equivalent.
Acknowledged limitations
Only one official organization has recognized some of the limitations of the concept of substantial equivalence. A Dutch government team has acknowledged that "compositional analysis...as a screening method for unintended effects...of the genetic modification has its limitations...in particular regarding unknown anti-nutrients and natural toxins", and it has given a lead by trying to explore some alternatives7. The Dutch team accepts that comparisons of relative crude compositional data provide a very weak screen against the introduction of novel genetic, biochemical, immunological or toxicological hazards, and they have suggested a finer-grained screen to test for differences in some of the relevant biological variables, such as DNA analysis, messenger-RNA fingerprinting, protein fingerprinting, secondary metabolite profiling and in vitro toxicity testing. If the use of such a finer screen revealed that a GM food contained a relevant novelty then the case for further studies would be far stronger, and those studies might benefit from having some clues as to which end-points might be most worthy of investigation.
Substantial equivalence is a pseudo-scientific concept
because it is a commercial and political judgement masquerading as if it were
scientific. It is, moreover, inherently anti-scientific because it was created
primarily to provide an excuse for not requiring biochemical or toxicological
tests. It therefore serves to discourage and inhibit potentially informative
scientific research. The case of GTSBs shows, moreover, that the concept of
substantial equivalence is being misapplied, even on its own terms, within the
regulatory process. If policymakers are therefore to provide consumers with
adequate protection, and genuinely to reassure them, then the concept of
substantial equivalence will need to be abandoned, rather than merely supplemented.
It should be replaced with a practical approach which would actively
investigate the safety and toxicity of GM foods rather than merely taking them
for granted, and which could give due consideration to public health principles
as well as to industrial interests.
Erik Millstone, (SPRU) Science and Technology Policy
Research, Mantell Building, Sussex University, Brighton BN1 9RF (E-mail
e.p.millstone@sussex.ac.uk); Eric Brunner, Department of Epidemiology and
Public Health, University College, 1-19 Torrington Place, London WC1E 6BT; Sue
Mayer, GeneWatch UK, The Courtyard, Whitecross Road, Tideswell, Buxton,
Derbyshire SK17 8NY
1.'The ADI
Debate', Appendix I of Food additives and the consumer', European Commission,
1980, pp. 41-3;
2.Strategies for
assessing the safety of foods produced by biotechnology, WHO, Geneva, 1991
3.Bowers, J. et
al. Science 285, 1562-1565 (1999)
4.Safety
Evaluation of Foods Derived by modern biotechnology OECD, Paris, 1993.
5.Lydon, J.
& Duke, S.O. Pesticide Science 25, 361-374 (1989)
6.Padgette, S.
R. et al. Journal of Nutrition 126, 702-716 (1996).
7.Kuiper, H. A.
et al. Food Safety Evaluation of Genetically Modified Foods as a Basis for
Market Introduction (Ministry of Economic Affairs, The Hague, 1998).
NOTICE: In accordance with Title 17 U.S.C. Section 107, this
material is distributed for research and educational purposes only. **