Even if bio-pharmed crops were to contaminate food crops, how likely is it that anyone would find
harmful amounts of prescription drugs in his corn flakes, pasta, or tofu? A combination of factors-
including natural selection, farmers pursuing their own commercial self-interest, liability concerns, and the vast size of the U.S. food supply - all militate against such a possibility.

Gene flow is a biological fact of life. It is ubiquitous. All crop plants have wild relatives somewhere, and some gene flow commonly occurs if the two populations are grown sufficiently close together. Thus, although genes could be transferred from a crop that has been modified to synthesize a pharmaceutical, the recipient plant is likely to proliferate only if a certain gene that has moved confers a selective advantage. Such occurrences should be uncommon with biopharming because, most often, the added drug-producing gene should not confer on the recipient any selective advantage and could even place it at a selective disadvantage. Thus, if such a gene were to be transferred into a food
crop, it might persist at a low level in the affected crop population for many generations, but we would expect its ability to proliferate and to cause significant contamination of the food crop to be limited.

Another relevant question is the persistence of post-biopharming volunteers. Michael Crawley and
his co-workers found in a study published in Nature (February 8, 2001), which compared the
performance of four different gene-spliced versus conventional crops (rapeseed, potato, corn, and sugar beet) in natural habitats, that in no case were the gene-spliced plants (which were engineered for traits other than synthesis of pharmaceuticals) found to be more invasive or more persistent than their conventional counterparts. They also found "that arable crops are unlikely to survive for long outside cultivation." By the end of four years, of all the varieties cultivated in the study, only one
variety of conventional potato persisted.

Gene transfer is an age-old consideration for farmers. Farmers in North America and elsewhere,
who grow many hundreds of crops virtually all of which (save only wild berries) have been genetically improved in some way, have meticulously developed strategies for preventing pollen cross-contamination in the field - when and if it is necessary for commercial reasons. Traditionally, plant breeders' guidelines have called for keeping distinct varieties of corn, a wind-pollinated crop, at least 660 feet apart. At this distance, the two corn varieties will not hybridize to any great extent, even if small
amounts of pollen might still drift between the fields. Even without government oversight, biopharmers themselves strive to keep their specialty corn sufficiently far from ordinary cornfields, lest their highly valuable drug-producing crops suffer contamination from the food crops.

Canola and rapeseed provide a good example of two crops that are rigorously segregated with minimal government interference. The original rapeseed oil, used as a lubricant, caused heart disease when ingested because of high levels of a chemical called erucic acid. Conventional plant breeding led to the development of rapeseed varieties with low concentrations of erucic acid, which came to be known as canola. In 1985, fda approved canola oil for food use, provided that it contained no more than 2 percent erucic acid. But since rapeseed oil is still used as a lubricant and plasticizer, farmers and processors must carefully segregate these distinct high- and low-erucic acid crops in the field and thereafter, a task they accomplish routinely and without difficulty. What makes these successes particularly compelling is that, unlike certain other crops, such as wheat and barley, which tend to self-fertilize and are less likely to pick up foreign genes, rapeseed/canola is "one of the more problematic in terms of gene flow . . . a worst-case scenario," according to Danish plant geneticist Rikke Jørgensen of the Riso National Laboratory.

Under this system, small quantities of rapeseed occasionally may get mixed into the canola, but
this is of no consequence as long as the finished product meets the federal safety standard. Although it might be politically unrealistic to expect that people will unquestioningly eat tiny quantities of biopharmed crops the way they regularly consume erucic acid, there is no scientific or medical objection to their doing so.

Federal regulators could establish non-zero tolerance levels for biopharmed contaminants in the food supply. In some cases, such as for drugs that are neither orally active nor likely to be allergenic, one might simply conclude that contamination at any level poses negligible risk (not unlike the level of concern about small amounts of pollen from a variety of yellow sweet corn pollinating white sweet corn in a nearby field).

For situations in which risk is uncertain or known to be non-negligible, one would base tolerances on animal toxicology studies, as regulators do for pesticide residues. Before approving a new pesticide, the Environmental Protection Agency requires the manufacturer to examine how much of the chemical mice, rats, rabbits, and chickens can absorb without suffering any observable long-term effects
following both acute and chronic exposure. Using highly conservative assumptions about both safety
margins and the relevance of extrapolating high dosing of animals to very low exposures in humans, the epa then builds in a safety margin of several orders of magnitude to allow for differences between animals and humans and for possible enhanced susceptibility of children. With these kinds of assumptions, regulators create a huge safety margin - excessively huge, according to many experts - when they determine the maximum safe dose for humans. An analogous approach, which would substitute performance standards - that is, non-zero tolerances for carryover into food - for usda's current design standards, also could work for pharmaceutical contaminants, at least from a medical standpoint.

Although potentially workable, the outcome of this conservative approach to establishing tolerances - like the epa's determination of acceptable pesticide residues, from which it is derived - will likely be overly risk-averse. Even in a worst-case scenario, by the time a food contaminated with a biopharmed substance passes a consumer's lips, it is unlikely to exert a significant effect. Recall that in the ProdiGene case, some 500,000 bushels of ordinary soybeans allegedly came into contact with a very small amount of biopharmed corn stalks and leaves. Not all the data necessary for a detailed analysis of that situation are publicly available, but we do know that for personal injury to occur, several highly improbable events would have to happen.

First, the active drug substance would have to be present in the final food product - say, tofu or salad dressing made with soybean oil - at sufficient levels to exert an adverse effect, the result of either direct toxicity or allergy. But there would have been a huge dilution effect as the tiny amounts of biopharmed corn stalks and leaves were pooled into the massive soybean harvest. With very few exceptions (e.g., peanuts), even an allergic reaction requires more than a minuscule exposure. Second, the active agent would need to survive milling and other processing, and then cooking. Third, it would need to be orally active; to take the example of ProdiGene's corn, the synthesized "drug" is not
pharmacologically active, except in the sense that it elicits antibodies that are intended to confer immunity to E. coli. The probability that all of these events would occur is extremely low.

Moreover, it is essential to consider the broader context of the kinds of chemicals that are commonly in our diet. We routinely consume hundreds of thousands of chemicals of all sorts - proteins, fats, carbohydrates, and minerals, among others. Bruce N. Ames and Lois S. Gold at the University of California at Berkeley have estimated that each day, "on average, Americans ingest roughly 5,000 to 10,000 different natural pesticides and their breakdown products," as well as about 2,000 milligrams of "burnt material, which is produced in usual cooking practices" and contains many rodent carcinogens and mutagens.

These observations emphasize the primary principle of toxicology - that the dose makes the poison. Unless we have the misfortune to eat something to which we are highly allergic, a poisonous mushroom, or a poorly dissected puffer fish, the chemicals present in food do not cause acute harm. The possible risks of adding one more chemical moiety to the diet, especially in a minuscule amount, must be considered in that context. Except for extraordinary circumstances (for example, biopharming of an extremely potent toxin), there is no scientific justification for the kind of rigorous oversight that usda imposes on biopharmers today.

Henry I. Miller, M.D.
The Hoover Institution
Stanford University
Stanford, CA 94305-6010
U.S.A.

Phone: 650.725.0185
Fax: 650.723.0576
E-mail: miller@hoover.stanford.edu