The media has inflamed public fears about the risks of genetically modified crops for human health and biodiversity. But many responsible scientists agree on the need for more research to identify potential long-term problems.

The declared position of the world's major regulatory and scientific agencies is that, in principle, genetically modified (GM) crops pose no greater threat to human health than those produced by traditional breeding.

"I don't see any problems at all for genetically modified plants in terms of human health. Researchers are being asked to prove negatives," says Robert McKinney, director of the division of safety at the US National Institutes of Health in Bethesda, Maryland.

But this view is being seen increasingly as an oversimplification by critics of gene technology -- and by some of its supporters. They argue that it could stifle serious consideration of the 'remote but real risks' potentially associated with the acceleration and broadening of plant breeding brought about by the development of genetically modified plants.

Most scientists believe that such risks are largely hypothetical, and that current safeguards are adequate. Even among ardent supporters of GM foods, however, calls are being increasingly heard for more research on health risks, and for the introduction of monitoring systems that would allow the early detection of any long-term problems.

For example, although the question of whether to label GM foods is usually considered to be an issue of consumer choice rather than public health, several scientists say there is also a strong argument for labelling to facilitate epidemiological studies to detect any increases in allergies or diseases that might be linked to GM foods.

The need for careful monitoring is urgent, given that the introduction of thousands of GM foods on a global scale appears imminent, says Suzanne Wuerthele, a risk assessor at the US Environmental Protection Agency, speaking in a personal capacity.

This view is supported by Ben Miflin, former director of the Institute of Arable Crops at Rothamsted, near London, who is a proponent of the potential benefits of genetic modification of crops. He argues that, under current monitoring conditions, any unanticipated health impact of such foods would need to be a "monumental disaster" to be detectable.

Miflin points out that a general increase in gastrointestinal disorders, for example, would be difficult to attribute to a particular food, given the diverse possible origins of such symptoms. "So, yes, there is an advantage for going slowly."

Some researchers have proposed specific monitoring strategies. Hans-Jörg Buhk, director of the Robert Koch Institute in Berlin, has called for the creation of a 'gene register' to track which genes and constructs have been introduced into the crop gene pool. This precautionary measure, he argues, would improve the ability of researchers to predict interactions between genetic modifications.

Whereas the genetic traits currently being introduced tend to be relatively simple, monogenic ones, the next generation of GM crops may require a rethink, says Buhk. These could include 'functional foods', such as plants with increased vitamin levels, or pharmaceutical-producing plants. He predicts that, before the full-scale introduction of such plants, additional safety trials may be needed, perhaps analogous to the clinical trials used to assess the safety of drugs.

Scientists "do not know everything in advance," says Buhk, adding that unexpected events cannot be ruled out. He points to a tomato developed by the British company Zeneca that was designed to have an increased shelf life, using antisense technology directed against the polygalacturonase gene that causes ripening. According to Buhk, the best performing line was in fact caused by an unpredicted 'sense' event (gene activation). "This was a rare event, either a contamination or a chance turnaround [in the genome]," he says.

In a similar way, virus-resistant plants have been genetically engineered using an approach where expression of a viral coat protein confers a virus-resistant phenotype on the plant. This strategy has been demonstrated against many viruses. But it has been occasionally observed that expression of the viral coat protein is unnecessary, and that translation of parts of the gene alone can confer the desired protection. "This is gene silencing," says Buhk. "There is interaction going on at the RNA level that we do not understand."

The likelihood and impact of such unexpected occurrences seem to be the major area of uncertainty among scientists working on the potential health impacts of GM foods. At the same time, many are highly critical of recent health scares, claiming that the arguments for and against such foods have been obscured and exaggerated, if not distorted.

The general view of such researchers, for example, is that, although the safety of antibiotic marker genes has received prominent media attention, it is ultimately a secondary issue as far as potential impact on human health is concerned. Whatever the perceived risks of such markers, there is a broad consensus that their use is unnecessary, and that innocuous alternatives are available.

Another public concern in genetic engineering in general is the safety of the vector used. But many scientists believe that the genetic constructs used to engineer plants -- such as the Agrobacterium tumefaciens vector and the cauliflower mosaic virus (CMV) promoter -- are inherently safer than, for example, the attenuated viral vectors used in gene therapy.

"You get massive doses of CMV each time you eat broccoli," says Marc Van Montagu, professor of genetics at the University of Ghent in Belgium.

What is clear, however, is that crop plants contain many compounds that have the potential to damage human health, and that even conventional plant breeding has on more than one occasion come close to producing plant varieties that could pose serious health risks.

"There are several well documented examples where breeding has inadvertently introduced higher toxin levels," says Miflin, pointing to cases of high levels of psoralen in celery, and of solanine in potato. "There are hidden risks in any form of plant breeding."

Such toxic varieties never came to market, however, as they were detected by the quality control procedures of the seed companies themselves. Miflin and others believe that, if anything, the generally stricter approval procedures required for crops produced by genetic modification reduce the likelihood of such mishaps.

But this does not reduce the need for vigilance. Nor of the importance of research to better understand the impact of introduced genetic constructs on plant metabolism, and to improve existing techniques for evaluating toxicity and allergenicity.

Underpinning health risk assessment throughout the world is the concept of 'substantial equivalence', a term coined by the Organization for Economic Cooperation and Development in 1993 under which GM foods are compared with analogous conventional foods in terms of toxicity, nutritional qualities and other characteristics.

In line with this approach, applications to the UK Advisory Committee on Novel Foods and Processes require companies to show that their products are 'substantially equivalent' with respect to known toxins and allergens. A company submitting a modified potato would need to show that the genetic modification had not inadvertently increased alkaloid levels, for instance.

The first issue addressed by approval procedures is the safety of the introduced genes -- for example their purity -- and the proteins they express. "If you introduce a gene for a poison into strawberries, you shouldn't be surprised if the plant is toxic," quips Axel Kahn, former head of the French committee that approves GMOs.

Where proteins might be degraded by plant enzymes, the toxicity of any breakdown products is also considered in this procedure. Introduced genes coding for enzymes can in principle also modify the metabolism of plants, leading, for example, to an increase in toxins. Mutagenesis by genes randomly inserted into the genome also raises the risk that 'silent' or lowly expressed genes might be switched on or upregulated with damaging results.

The need to assess the potential risks of such unpredictable occurrences cannot be ignored. Plans to market Canola and Vicia (bean) plants, genetically modified to boost their low cysteine and methionine content using the methionine-rich 2S storage albumin of Brazil nuts, were abandoned when it was discovered that the protein was highly allergenic (see The New England Journal of Medicine 334, 688-692; 1996).

Regulatory authorities are confident that the allergenicity of untested proteins can usually be reliably predicted by structural analysis and testing. Most allergenic proteins tend to have a molecular weight between 10 and 70 kilodaltons, often share characteristic sequence stretches, and resist degradation by heat, as well as acid and peptinase conditions mimicking those found in the stomach. Some scientists, however, believe that regulatory agencies overestimate their ability to predict allergenicity.

More difficult is the assessment of the potential presence of unknown toxins. Animal feeding experiments are plagued with problems, such as high levels of variation in results between experimental groups. It is also difficult to feed relevant doses of GM foods to laboratory animals for sufficient periods to allow reliable extrapolation of the results to humans.

Another problem, as several scientists point out, is that such research is unattractive to researchers, as it tends to yield negative results that are difficult to publish and to account for to funding agencies.

A potential alternative to feeding experiments is to subject plants to a battery of chromatographic and spectroscopic analyses, to create a molecular 'fingerprint'. Any difference between the fingerprint of a modified crop and an existing one would indicate a need for more extensive safety testing.

However efficient the techniques used to pinpoint potential problems, critics such as Wuerthele, and many consumer groups, remain sceptical of the reliability of safety assessment procedures, and the judgement of expert committees -- especially following the failure of experts to prevent the bovine spongiform encephalopathy crisis.

Foreign proteins that have never been in the human food chain will soon be consumed in large amounts, she points out. "It took us 60 years to realize that DDT might have oestrogenic activities and affect humans, but we are now being asked to believe that everything is OK with GM foods because we haven't seen any dead bodies yet."

Many researchers point out that such unknowns are common in classical plant breeding, however. "I honestly don't see what you could do to a plant that would be so different to the traditional approach," says Mike Gale, director of the John Innes Centre in Norwich, England. He argues that, although transgenes are incorporated in a random fashion into the plant genome, the introduced gene is well characterized.

In contrast, in conventional breeding, Gale points out, in addition to introducing a desired trait into a crop from a wild relative, breeders have no idea what other changes they may have introduced through the integration of large chunks of the donor genome. "Surely putting in one gene is better than shuffling around tens of thousands at random," he says.

Similarly, genetic engineering may actually help to improve food safety by removing toxic compounds naturally present in many plants. Cassava, a major staple food, contains cyanogenic glycosides, for example, that must be eliminated by fermentation before the plant can be eaten. Kidney beans require soaking to remove toxic lectins -- poor food preparation still results in deaths.

A major goal of plant breeding throughout the ages, points out Miflin, has been to reduce or eliminate the many toxic compounds present in wild plants that have evolved, for example, as defence mechanisms against predators.

But, whatever the validity of such arguments, there is growing awareness -- even among major agrofood companies such as Monsanto and Novartis -- that consumer concerns about health cannot simply be wished away if the fruits of biotechnology are to be realized.

"It is clear [at least] in Europe that the consumer backlash means that more investment in research [and monitoring] is needed to clarify the scientific uncertainties," says Gerry Moy, head of food safety at the World Health Organization. "Society needs to begin a broad discussion on the potential risks and benefits of GMOs."

** NOTICE: In accordance with Title 17 U.S.C. Section 107, this material is distributed for research and educational purposes only. **