CU Comments: Draft “Guidance for Industry: Drugs, Biologics, and Medical Devices Derived from Bioengineered Plants

of Consumers Union on Docket No. 02D-0324:
Draft "Guidance for Industry: Drugs, Biologics, and Medical Devices Derived
from Bioengineered Plants
for Use in Humans and Animals


Consumers Union* welcomes the opportunity
to comment on the Food and Drug Administration’s (FDA’s) and U.S. Department
of Agriculture (USDA) proposed draft guidance on the growing of plants that
have been genetically engineered to produce drugs. We support FDA’s goal of
insuring that such plants do not get into human food or animal feed. However
we do not believe that the industry guidance will achieve that important goal.
In fact, for the reasons we outline, we think that even if industry follows
this guidance, but takes no other measures, significant contamination of the
food supply is inevitable.

We believe FDA should establish a
zero tolerance for drugs, biologics and medical devices in food and feed. To
achieve this, FDA/USDA guidance should indicate that drugs, biologics and medical
devices should not be genetically engineered into food crops. Rather, to achieve
the potential benefits of using this technology FDA should guide companies to
investigate engineering of such substances into non-food crops, if they can
be grown indoors under controlled conditions.

FDA Should Set a Zero Tolerance for Drugs in Food

We strongly agree with FDA’s statement
in section II. A. that, in regard to using "a food crop species engineered
to produce non-food materials" that "the presence of any such material
[drug, biologic or medical device] in food or feed could render such products
adulterated under the FD&C (12 U.S.C. 342)." We believe that any drug,
biologic or medical device produced in genetically engineered food crop species
should be considered an adulterant if present in food or feed at any level.
Drugs, biologics and medical devices are highly regulated products that FDA
does not allow on the market without a prior safety review. They are designed
to be biologically active substances. In part because of risks associated with
these substances, consumers are not allowed to purchase prescription drugs without
a physician sanctioning such use and indicating the dosage and time period of
administration, and manufacturers are tightly regulated as to production practices.

Consumers do not expect to have drugs,
biologics or medical devices in their food, and such presence would raise complex
and important safety issues. To take just one example, few drug safety evaluations
conduct testing to evaluate the possible effects of consumption of low levels
of the drug continuously over a lifetime, beginning in early childhood or even
infancy. Nor do such evaluations examine the synergistic or cross-reactive effects
of consumption of the hundreds of drugs which eventually could be produced in
plants. In fact, it is not possible to do a scientific credible risk assessment
of these potential public health hazards, given currently available data or
any reasonable scenario for near-term research. In the face of such great scientific
uncertainty, the FDA should set a zero tolerance for such "non-food (or
non-feed) material" in food.

FDA Should Not Allow Food Crops Species as Source Plants

The FDA states in II.A. that a concern
that must be addressed is "the measures to ensure that non-food (or non-feed)
material will not get into food or feed." We agree with that goal. However,
we believe that the present guidance will not achieve that goal. Indeed, for
the reasons discussed below, we believe that the measures described in this
guidance are so minimal and ineffectual, that even if they were followed by
industry, and no other measures were taken, it is inevitable that drugs and
biologics would get into food or feed. The measures described in the FDA/USDA
proposal specifically do not address prevention of additional incidents like
those that occurred in 2002 involving Prodigene field trials.

In our view, the only way "to
ensure that non-food (or non-feed) material will not get into food or feed"
is to indicate in the guidance that companies should not use of food crop species
as a source of "the desired regulated product." If the agency allows
companies to use a food crop species as a source-whether grown outdoors or indoors
in a greenhouse-contamination of food and feed will be inevitable, through biological
means and/or human error. Biological means of contamination include gene flow-via
pollen and seeds-which can occur through vectors ranging from bees and wind
to tornadoes. Human error includes many things, such as the accidental mixing
of seeds, or insufficient diligence in removing volunteers in the year(s) subsequent
to the planting of pharmaceutical crops. Effects of possible deliberate sabotage-whether
terrorist or vandalism-can also not be discounted. Gene flow is an especially
serious concern, given that some two-thirds of the field trials to date of pharmaceutical
crops have involved the use of corn, a wind-pollinated crop, as the source crop.
Corn is planted on tens of millions of acres.

The examples of the ProdiGene contamination
events last fall suggest that containment is virtually impossible. Last November,
the USDA and FDA made public the discovery of problems of contamination of food
crops-soybeans and corn-from two ProdiGene field trials, one in Iowa and one
in Nebraska. According to ProdiGene, both trials involved corn engineered to
produce TGEV, a pig vaccine (Gillis, 2002a). The FDA, however, put out a press
statement saying that in the Nebraska case, an unapproved human drug had been
engineered into the corn: "The pharmaceutical material being produced in
the corn plants was being studied under an Investigational New Drug (IND) application"
(FDA, 2002). Regardless of which is accurate, the contamination is still a recognized
fact. With more than 300 experiments already, and untold more planned, plus
likely commercial scale planting of some, if this technology goes ahead on current
track our food will be massively and most likely irreversibly contaminated.

Detailed Comments

III. Environmental Considerations

C. Confinement measures

1. General Considerations

The agency suggests, that when developing
a bioengineered pharmaceutical plant, that companies "should implement
procedures to ensure that such a plant line is used only for its intended purpose
as a source material for a regulated product" and then suggests that the
company keep detailed records "documenting the handling and transfer of
such materials." But detailed records will not ensure that no contamination
of food or feed will occur. Human error can occur. Records can be incomplete,
or even falsified. Further, paper records can be lost and computer records can
be accidentally erased. In addition, fires can destroy such records, particularly
if multiple back-up copies are not kept in several locations.

The agency also suggests that companies
"should consider the use of strategies-[such as using genetic markers that
alter the physical appearance of the plant, e.g. novel color or leaf pattern]-that
allow the bioengineered pharmaceutical plant line to be readily distinguished
from its food or feed counterpart" or "strategies to reduce the likelihood
of unintended exposure to the regulated product by" restricting where (via
the use of tissue-specific promoters) or under what conditions (via use of inducible
promoters) the product will be expressed. We note that while such suggested
mitigation concepts are good ones, none of them are foolproof. Such measures
help clarify that contamination has occurred. However, it may not prevent it
in the first place. Also, since such measures are voluntary, there is no assurance
that they will be used. Indeed, companies have indicated a reluctance to alter
color or leaf pattern because they believe that if the product is readily identifiable,
it might become the target of vandalism. In fact, we have seen no evidence to
date that any of these measures have been utilized in the hundreds of field
tests done so far.

FDA further suggests planting outcrossing
plants "only in regions of the country where little or none of its food/feed
counterparts are grown." We note that so far this is not industry practice,
as evidenced by the two examples of contamination of soybean crops with volunteer
pharmaceutical producing corn that took place last fall. In these two ProdiGene
examples, the pharmaceutical producing corn had been grown in Nebraska and Iowa,
both of which are part of the corn belt. Further, this advice has already apparently
been rejected by industry. A few weeks before the ProdiGene contamination stories
appeared in the press, BIO (Biotechnology Industry Organization) announced a
voluntary moratorium on planting pharmaceutical corn (the majority of field
tests have used corn as the source plant) in the corn belt (Gillis, 2002b; Brasher,
2002). This would have been a positive, although limited step– it should be
pointed out that there are some 20 million acres of corn grown outside the corn
belt. However this policy was short-lived. In early December, BIO sent a letter
to Senator Charles Grassley (R-Iowa) in which they rescinded this voluntary
moratorium and stated, instead, that they would defer to whatever future guidance
or regulations were proposed by FDA and USDA (DTN, 2002).

In addition, there is the potential
for pollen flow from such pharmaceutical plants (e.g. corn) to non-pharmaceutical
plants during the seed production stage; this has been shown to occur with StarLink
corn (USDA, 2001). Even if one could find regions of the U.S. where little or
no food/feed counterparts are grown (a doubtful proposition, particularly for
corn plants, which account for the bulk of the field trials), the possibility
exists that bags of such pharmaceutical seeds might be mis-labeled or inadvertantly
mixed with non-pharmaceutical seeds, and then unknowingly planted.

The agency also suggests that "[M]easures
should be in place to ensure that there is no inadvertent mixing of the bioengineered
pharmaceutical plant with plant material intended for food or feed (including
inadvertent mixing with seeds for food or feed crops)" and basically suggests
companies use a HACCP approach to "determine where in the process inadvertent
mixing could occur and establish appropriate control measures." Such a
HACCP system is currently being used to control microbial contamination in meat
production facilities and a recent GAO report has pointed out numerous short-comings
in such a system (GAO, 2002a). Since HACCP has only made modest progress in
controlling microbial contamination of meat products in slaughterhouses and
other processing facilities, it is difficult to believe that a HACCP system
would be foolproof in preventing contamination of food and/or feed with materials
from bioengineered pharmaceutical plants.

The FDA has also "strongly suggested"
development of tests "that can detect the presence of the target gene and
the protein product in the raw agricultural commodity." We believe that
such tests should be required as a precondition to the use of any plant species
as the source of the pharmaceutical product. However, having such tests available
won’t prevent contamination-it will simply allow the existence of contamination
to be detected. Since development of such tests is optional-neither FDA nor
USDA have the regulatory authority to require them-there is no assurance that
such a system would be able to reliably detect a contamination event.

The details of the two ProdiGene
examples, from Nebraska and Iowa, show that FDA and USDA cannot rely on industry
self-policing in this area. In the Iowa case, which was discovered in September,
2002, ProdiGene did not come forward and tell the USDA that volunteer pharmaceutical
corn plants existed in the field of soybeans planted in 2002. (In 2001, this
field had been planted to an engineered pharmaceutical corn.) A USDA inspector
visiting the Iowa location in September found 20 volunteer corn plants standing
in the soybean field in addition to a pile of rogue corn plants at the side
of the field (NGO mtg. with APHIS/BRS staff, Jan. 14, 2003). The plants had
flowered, so that pollen could have contaminated nearby corn fields. USDA acted
in a conservative manner and ordered some 155 acres of corn to be destroyed
(Gillis, 2002c). USDA did this, in part, because they did not have the ability
to test the neighboring corn for the presence of the pharmaceutical transgene.

In part because ProdiGene had not
notified USDA of the volunteer pharmaceutical corn plants in Iowa, USDA decided
to check all the other ProdiGene field sites (NGO meeting with APHIS/BRS staff,
1/14/’03). In early October, 2002, a USDA inspector found 211 pharmaceutical
corn volunteers-some of which had flowered-in a Nebraska soybean field that
had been planted in 2001 to pharmaceutical corn (NGO meeting with APHIS/BRS
staff, Jan. 14, 2003). Again, ProdiGene had not previously notified the USDA
about the presence of such volunteers; this constituted a violation of their
permit. The USDA inspector told the company to destroy the volunteers and the
company supposedly relayed this message to their consultant and to the contract
farmer, but the action wasn’t fully completed (Ibid). Indeed, some of the soy
had already been harvested and mixed with a silo of soybeans, thereby potentially
contaminating the whole silo. USDA responded by quarantining some 500,000 bushels
of soybean (Brasher, 2002). Blame was variously placed at the feet of ProdiGene,
the contract farmer, and consultants. The lesson for the FDA guidance must be
that human error, failures of judgment, and lack of accountability must be expected
and guarded against.

2. Control of Seed Stocks

The agency recommends that the companies
should "maintain careful control over the inventory and disposition of
viable seeds to preclude the possibility that such seeds will be used to produce
material that could be used for food or feed production" and further suggests
prominently labeling such seed stocks. However, we have already seen that the
use of labels is unlikely to ensure that seed for drug production does not contaminate
seed designed to be grown for food or feed.

Experiences with StarLink corn and
with the FDA bovine spongiform encephalopathy feed rule demonstrate the shortcomings
of labels as a means of preventing contamination. The Environmental Protection
Agency (EPA), as part of the approval process for StarLink corn, allowed the
corn on the market but stated that it could be used only for animal feed purposes.
Seed bags were supposed to be clearly labeled that StarLink could not be used
for human food. In addition, the farmers were supposed to be required to sign
a contract that stipulated a number of items-such as not growing StarLink within
660 meters of corn destined for use as human food, use of buffer rows of corn,
etc. These measures were supposed to ensure that none of the Starlink corn got
into the human food chain. However, many farmers now claim that they never saw
the contracts that they were supposed to have signed. In addition, the labeling
on the many bags of seed did not explicitly say that it could not be used as
human food (Ryberg, 2000). Despite EPA’s labeling requirements and other precautions,
the human food chain was contaminated with StarLink corn. In the end, companies
had to spend over $1 billion recalling all the products that had been contaminated.
Contaminated exports were identified as late as December, 2002 (Fabi, 2002).
Even a significant percentage of corn seed was contaminated with genetic material
from StarLink, demonstrating that gene flow had occurred (USDA, 2001).

To take another labeling example, in 1997, the FDA promulgated a feed rule designed
to minimize the chance that mad cow disease would become a problem in the U.S.
Part of the feed rule required that any animal feed that contained ruminant
or other banned proteins be labeled "do not feed to cattle and other ruminants."
A Government Accounting Office (GAO) study published in September 2000 found
that 28 percent of all the facilities that handled ruminant meat and bone meal
failed to put such a label on their product (GAO, 2000); a follow-up study found
that labeling problems still persisted (GAO, 2002b). We have no confidence that
the biotechnology industry, facing voluntary guidance, will succeed when other
industries, under a regulatory mandate, have massively failed.

3. Field-grown plants

On the question of field-grown plants,
USDA and FDA "recommend that you [the company] consider the use of perimeter
fencing to help exclude wildlife and escaped livestock." Such a step, while
perhaps marginally useful, will certainly not stop birds and insects from entering
the fields and consuming or transporting seeds and/or pollen out of the field
site, potentially contaminating food crops. Such animals, in addition, could
be affected by the pharmaceutical products themselves, including beneficial
species. Furthermore, a fence is useless against major acts of nature such as
flooding and tornados. The floods that ravaged the mid-west in the late 1990s
destroyed some test plots (NGO meeting with APHIS/BRS staff, 1/14/03). There
was also at least one incident where a tornado destroyed a plot of tobacco containing
transgenic proteins in Kentucky in the mid-1990s. This Kentucky case, which
involved the transgenic tobacco mosaic virus (TMV), could have resulted in the
spread of the virus. Although a tornado would destroy the experimental tobacco
plants, TMV can be transmitted by direct contact to other solanaceous plants
such as tomatoes and potatoes. Thus, a piece of tobacco infected with transgenic
TMV that floated back down to earth after the tornado could infect a tomato
plant if the tobacco fragment touched a tomato plant. Such acts of nature could
easily serve to spread the pharmaceutical plants far beyond the bounds of the
test plots.

4. Control of harvest material

The agency notes that APHIS "requires
[that] [D]uring transport, containers of harvested material should carry a label
that clearly indicates that the material, including but not limited to seeds,
leaves, roots, and stems is not to be used for food or feed." Such labeling
cannot be relied upon, as the case of StarLink, as discussed above, clearly

5. Control at Processing Facilities

The agency proposes potentially allowing
pharmaceutical plants to be processed at facilities that produce food or feed.
The guidance states that "[S]ource plant materials should not be processed
at facilities that are also used for the production of food or feed, such as
grain mills, without prior consultation with USDA/APHIS/BRS or FDA." We
strongly disagree and believe that under no circumstances should engineered
pharmaceutical plants be processed at facilities that also produce food or feed.
Such processing would inevitably result in cross-contamination of food or feed
products. To achieve zero tolerance, pharmaceutical crops should have a completely
separate production chain.

6. Control of Waste Material

This section would allow, under certain
circumstances, that process waste or residual source plant material could potentially
be used in human or animal food: "In process wastes . . . , rejected in-process
material, and residual source plant material from the purification process .
. . should be disposed in a manner to ensure that the material will not enter
the human or animal food chain unless you have specifically consulted with FDA
for the use of this material in food or feed products" italics added. We
strongly disagree with the idea that process wastes could be fed to food animals.
We urge that for safety reasons, this be prohibited, since just as in terms
of human safety, the risk assessment process would be impossible to execute
in a scientifically sound manner.


In sum, we think that the confinement
measures laid out in the Guidance, and listed above, will not guarantee that
the goal of zero tolerance-a proper goal in our view-will be reached. A zero
tolerance is needed because acceptable risk levels simply cannot be established
based on current scientific knowledge. We believe that the only way to achieve
the goal of zero tolerance is to ban the use of food crop species as a source
plant material for pharmaceutical production. This means that food crop plants
should not be permitted to be engineered for pharmaceutical production, regardless
of whether they are grown outdoors or indoors. Even if the crops were grown
indoors, there is the strong possibility of a mistake which could mix up the
seeds of food crop engineered with pharmaceuticals with seeds of the same food
crop destined for human consumption, or a natural disaster, which would disseminate

As for non-food (or non-feed) plants
engineered to produce pharmaceuticals and grown outdoors, while there would
be no potential contamination of human or animal foods or feeds, we believe
it is vitally important to consider the effects on non-target organisms and
the environment, such as soil flora and fauna. Since pharmaceutical products,
by their very nature, are designed to be biologically active, the risks need
thorough and cautious assessment. Currently, the data are not available to conduct
a valid assessment.

Consider the following hypothetical
examples. One could engineer non-food or non-feed plants to produce the botulinum
toxin that is used in BoTox for cosmetic purposes. The bacteria that produces
botulinum only produces it under anaerobic conditions. If the gene were placed
in a non-food plant and grown widely outdoors, the environment could be contaminated
and numerous non-target organisms such as mammals could be adversely affected.
There is also the example of various animal protein hormones such as bovine
or ovine growth hormone, or epidermal growth factor. Such compounds, if grown
in non-food plants, could have effects on wild bovids or various wild ruminants
such as deer, elk, bison, etc. if they accidentally consumed plants that contain
these substances (The plant matrix may ensure that these protein hormones partially
or fully survive digestion and could have effects). There is also the example
of powerful antibiotics, such as chloramphenicol or others. While there are
bacteria in the environment that may produce such antibiotics, engineering the
antibiotic in a non-food plant and growing in on a large scale could potentially
lead to far far higher levels in the environment than occur naturally. Indeed,
all the above biologically active compounds could have negative effects on the
environment and on non-target animals.

Thus, we believe production of pharmaceuticals
in non-food or non-feed plants should not be permitted outdoors. However, we
also recognize that there may be advantages to producing pharmaceuticals in
plants. We therefore believe that companies should explore the use of non-food
and non-feed plants in controlled conditions indoors-such as in greenhouses
or phytotrons.


Brasher, P. 2002. Biotech firm under
fire has link to Iowa: The company accused in a Nebraska case paid to burn Iowa
corn that may have been contaminated. Des Moines Register, November 14.

DTN. 2002. BIO changes mind about
pharmacrops. December 12. Available at

Fabi, R. 2002. Japan finds StarLink
in US corn cargo -U.S. exporters. Reuters news story, Dec. 27.

GAO. 2000. FOOD SAFETY: Controls
Can Be Strengthened to Reduce the Risk of disease linked to unsafe animal feed.
GAO/RCED-00-255. Government Accounting Office, Washington, D.C.

———2002a. Better USDA Oversight
and Enforcement of Safety Rules Needed to Reduce Risk of Foodborne Illnesses.
GAO-02-902. Government Accounting Office, Washington, D.C.

Improvements in the Animal Feed Ban and Other Regulatory Areas Would Strengthen
U.S. Prevention Efforts. GAO-01-188. Government Accounting Office, Washington,

Gillis, J. 2002a. Corn for Growing
Far Afield? Washington Post, November 16, 2002.

—————–b. Soybeans mixed
with altered corn, Washington Post, November 13, pg. E01.

—————–c. Biotech firm
mishandled corn in Iowa. Washington Post, November 14, pg. E01.

Ryberg, W. 2000. Growers of biotech
corn say they weren’t warned: StarLink tags appear to indicate it’s suitable
for human food products. Des Moines Register, October 25.

USDA, 2001. USDA purchases Cry9C
affected corn seed from seed companies, June 15, 2001. See