Comments on the Assessment of Scientific Information
Concerning StarLink Corn (Cry9C Bt Corn Plant-Pesticide)
EPA Docket Number OPP-00688
Michael Hansen, Ph.D.
Consumer Policy Institute/Consumer Union
November 28, 2000
Consumers Union* thanks the EPA for the opportunity to comment on
Aventis' petition for a temporary exemption from the requirement of a
tolerance for the genetically engineered plant-pesticide materials (Cry9C
protein and the genetic material necessary for the production of this
protein) in StarLink. Although Aventis has narrowed the scope of their
petition from asking for an exemption covering both the Cry9C DNA and Cry
9C protein in all food commodities to one asking for an exemption only for
food products made from StarLink corn, and only for a limited time of four
years, we firmly believe that the EPA should not grant Aventis petition,
for both scientific and legal reasons.
As explained below, we believe Aventis has shown neither that the
Cry9C protein is not an allergen nor that StarLink corn could meet the
"reasonable certainty of no harm" safety standard under the Food Quality
Protection Act (FQPA). Furthermore, we have a number of criticisms of the
methodology Aventis uses to estimate human exposure to the Cry9C protein.
Indeed, we believe it is an open question whether the Cry9C protein is an
allergen, especially an inhalant allergen and feel that more research is
needed in this area.
On a policy level, we think it would be very dangerous for EPA to
grant Aventis request and feel that the agency's credibility is on the
line. We note that the EPA Science Advisory Panel was asked earlier this
year (February 2000) whether Cry9C was an allergen and was unable to give a
positive or negative answer. Instead, they called for more research.
What has changed between February 2000 and the present? As EPA
notes in its Federal Register Notice, Aventis has asked that the exemption
be granted for a limited time of four years, which time, Aventis contends,
is necessary to allow all processed foods potentially made from StarLink
corn grown in 1999 or 2000 to pass through channels of trade (65 FR 65248).
When EPA first granted approval for StarLink corn in 1998, Aventis agreed
to a restricted registration for animal feed (and industrial uses) and
vowed that they could ensure through a number of mechanisms (such as
restrictive farmer contracts, use of 660 foot buffer zones, etc.) that none
of the StarLink corn would make it into the human food chain or into export
channels. From 1998 until September 2000, EPA did virtually nothing to
ensure that Aventis was following the rules they had agreed to, and it
appears that Aventis did not abide by the rules. Instead, an environmental
organization--Friends of the Earth-- tested products and was able to
demonstrate that StarLink was in the human food chain. Furthermore, since
the story broke in late September, further testing by citizen groups, as
well as by industry and the FDA, has shown that the contamination was not
restricted to the one batch of taco shells, but was more widespread than
initially believed. Testing by a consumer group in Japan revealed that
corn products contaminated with StarLink had been exported from the U.S. in
clear violation of StarLink's limited registration (Strom, 2000). Other
stories in the press have revealed that some farmers appear not to have
been told that StarLink corn could not go into the human food chain, or
that corn in the buffer zone (within 660 feet of the edge of StarLink
fields) could not be sold for human food use, nor required to sign
contracts stipulating all the restrictions (Eichenwald, 2000; Feder, 2000).
Even though Aventis (1998, Agro Evo) had agreed to these restrictions as a
condition for partial registration of StarLink it is now clear that they
have not abided by these restrictions. Rather than face the consequences
of their failure to adhere to these restrictions, Aventis, instead, is
functionally asking the EPA to bail them out by granting their temporary
exemption from a tolerance for Cry9C DNA and protein. EPA should not
reward Aventis for their failure to follow the law.
It should be pointed out that the world is watching what action the
EPA will take. Indeed, EPA's decision could have effects overseas. While
citizen organizations were testing corn products in September 2000 and
revealing the presence of StarLink corn, Aventis, in the same month, was
applying for approval to grow StarLink corn in South Africa. According to
a story a couple of week ago in a South African newspaper, Department of
Agriculture spokesperson Magriet Engelbrecht says the Aventis application
to bring StarLink to South Africa is being reviewed by an advisory
committee. "'What happens in other countries is going to be taken into
account by the committee,' Engelbrecht adds, 'The experience in the U.S.
will play a role.'" (Maclead, 2000: 7). The article also points out that
Aventis "says it has new data to demonstrate that fears about allergies are
unfounded and has asked for temporary approval of the maize from the U.S.
authorities" (Maclead, 2000: 7).
Thus, if EPA does grant Aventis request, the end result may be the
approval to grow StarLink in South Africa, possibly without any
restrictions, and perhaps approvals elsewhere. We could then potentially
see Starlink in imported corn products, perhaps indefinitely.
Problems with Aventis Submission
Data on Commercial Bt Sprays Do Not Support Safety
Aventis argues that Cry9C is not an allergen based in part on our
experience with commercial Bt sprays. Although there have been no
commercially available Bt microbial sprays which contain the Cry9C protein,
Aventis notes that there are microbial sprays (Agree, Xen Tari) with the
Cry9B protein, which is highly homologous with the Cry9C protein (Aventis
Submission, pg. 14). Aventis then argues that after more than 30 years of
commercial use of Bt sprays containing a variety of Cry proteins, including
proteins from the Cry9 class, no allergy has been attributed to Cry
proteins. While true in a narrow sense, this could be the result of not
looking for such data. In fact, data exist which suggests that some Cry
proteins may be allergenic--particularly when inhaled.
According to a paper published last year, "potential allergic
reactions associated with the use of Bt have not been considered"
(Bernstein et al., 1999: 575). That article noted, "In 1992 the use of Bt
in an Asian gypsy moth control program was associated with classical
allergic rhinitis symptoms, exacerbations of asthma, and skin reactions
among exposed individuals reporting possible health effects after the
spraying operation. Unfortunately, there was no follow-up to determine
whether these events were Bt-induced hypersensitivity or toxic reactions or
merely due to common aeroallergens coincidental to the season during which
the spraying occurred. Similar findings occurred during another Bt
spraying in the spring of 1994" (Bernstein et al., 1999: 575). In other
words, even though there was suggestive evidence that Bt sprays may be
aeroallergens, no one bothered to follow-up on such evidence at the time.
To address this question, researchers carried out a health survey
of farm-workers before and after exposure to Bt pesticide sprays. The
study found that a number of farm workers reacted positively to skin prick
tests involving Bt spore extracts, with the number of positive skin tests
being significantly higher in farm workers with high exposure to Bt sprays
compared to those with low or medium exposure (Bernstein et al. 1999).
Furthermore, the presence of IgE antibodies against the Bt sprays was more
prevalent in the high Bt exposure workers compared to low-or moderate Bt
In an attempt to find out which proteins of the Bt sprays were
causing the IgE and hypersensitivity (e.g. positive skin prick test), four
different types of BtK spore extracts were tested, along with two
vegetative extracts. One of the four BtK spore extracts, labeled J-PROTOX,
consisted of the pro-delta-endotoxin, derived from a BtK product called
Javelin, which was the main product used during the growing season during
which the study took place. Two workers had a positive skin prick test to
J-PROTOX. Molecular identification, using polymerase chain reaction (PCR)
technology, demonstrated the presence of Cry1Ab and Cry1Ac genes in the
Javelin product. Thus, the purified pro-delta-endotoxin in J-PROTOX
probably consists of Cry1Ab and Cry1Ac.
The authors concluded that "the fact that skin and serologic tests
of immediate hypersensitivity developed in some workers indicates that
adverse IgE mediated health effects could develop if repetitive exposure
continues in some of these workers"(Bernstein et al., 1999: 581). Although
clinical disease was not seen in the farm workers studied, the authors note
that such results "should be interpreted with caution because of the
healthy worker effect, which might be more prevalent among migrant farm
workers who, upon associating clinical symptoms with a particular crop or
farm job, would likely seek employment in a different agricultural area"
(Bernstein et al., 1999: 580).
This study clearly suggests that the Cry1Ab and/or Cry1Ac proteins
are inhalant allergens. Although this study did not address the question
of whether the Cry9C protein is an inhalant allergen, the authors pointed
out that the workers had been exposed in the previous year to the Bt
product Agree, which contain Bacillus thuringiensis awazi, or Bta. As
Aventis pointed out in their submission Agree does contain the Cry9B
protein which is highly homologous with the Cry9C protein. Furthermore,
the authors of the farm worker study note that Bta (found in Agree and
Xentari) is among the most commonly used strains of Bt.
Thus, we suggest that a farm worker study, similar to the one
conducted by Bernstein et al. (1999), should be carried out on farm workers
exposed to Bta to determine if these workers have positive skin prick tests
and IgE antibodies to the Cry9B protein. If the answer is yes, one could
probably assume that such workers would also exhibit similar reactions to
the Cry9C protein. Such a study would be far more accurate way to
determine if the Cry9C protein is an allergen, compared to relying on a
comparison of the physiochemical properties of the Cry9C protein to those
of known food allergens.
We are particularly concerned about mill workers and other workers
such as those who work in grain elevators who are exposed to high levels of
corn dust in their job. This would particularly be the case for such
workers in the areas, such as parts of Iowa, with, the largest acreage of
As an alternative to the farm worker test suggested above, one
could develop an antibody test for the Cry9C protein either by getting
purified samples of the Cry9C protein from Aventis, or isolating it from
Bacillus thuringiensis tolworthi.. One could then use such a test on mill
workers or grain elevator workers in those areas where the plantings of
StarLink have been the highest. Such workers may have been exposed to
StarLink corn dust since 1998, when it was first approved. Furthermore,
Aventis points out in their submission that the 1999 corn crop will not be
cleared from the bins/elevators until October, 2001 (pg. 4 of Aventis
submission). Thus, the 2000 corn crop will not be cleared from the bin
until October 2002. Thus, mill workers and grain elevator workers in some
regions of Iowa, may be exposed to StarLink dust for five years
(1998-2002). Mill workers, especially, may be exposed to corn dust year
round, while farm workers are usually exposed to Bt sprays only during the
growing season. Such mill workers may be at an increased risk, due to
continuous exposure over a long time period (up to five years) and so may
develop IgE mediated disease.
Cross Reactivity Study
Among the "new" studies submitted to EPA by Aventis, was one which
involved using sera from individuals allergic to the common food allergens
wheat, rice, buckwheat, soy, peanut, milk, eggs and shrimp and seeing if it
cross reacts with the Cry9C protein. Using the technique known as RAST
(radioallergosorbent test), they demonstrated a lack of cross-reactivity of
these sera to the Cry9C, concluding that such a study "provides additional
evidence of the low probability that Cry9C is a food allergen"(pg.25,
Aventis Submission). We disagree; such a study provides only weak evidence
Protein Abundance and Potential Allergenicity
Aventis argues that proteins must be present at particularly high
levels in a food before they can trigger an allergic response. They note
that induction (i.e., becoming sensitized) of an IgE response is
dose-related, with a larger dose needed to induce a response in a naive
person compared to the dose that elicits a response in an already
Aventis refers to a study involving "highly peanut allergic
patients" that were exposed to peanut protein levels between 10 micrograms
to 50 milligrams to determine the threshold of response. The lowest level
of peanut protein that elicited a response albeit one with mild short-lived
reactions was 100 micrograms (Hourihane et al., 1997). Aventis then
calculates that the highest possible level of exposure to Cry9C in the
human diet is 8.6 micrograms for 2000 and then maintains that this is much
lower than the dose that results in mild, subjective symptoms in highly
peanut allergic patients.
There are a number of flaws with this analysis. First, Aventis
ignores the fact that much lower doses of an allergen, either food or
inhalant, can elicit an allergic response in already sensitized people. As
pointed out in the report of EPA's Scientific Advisory Panel, (SAP) that
met last February, and which discussed scientific issues related to the
potential allergenicity of the Cry9C protein/endotoxin, "it is known that
nanogram levels of house dust mites are capable of causing dust mite
allergy symptoms. With peanut allergies, ingestion of nanogram quantities
can cause anaphylaxis" (pg.6,
according to EPAs own SAP, very low levels of peanut allergens can cause a
fairly serious reaction--anaphylaxis. Yet Aventis argues, based on the
Hourihane et al. (1997) study that 100,000 nanograms of peanut allergen
only elicits mild, subjective symptoms in highly peanut allergic patients.
Clearly, the Hourihane study is not as definitive as Aventis would lead one
Furthermore at the October 20 meeting of EPA's SAP, the Aventis
representative made the basic argument that fairly high levels of an
allergenic protein were needed to sensitize individuals and that such
levels were far higher than the level of Cry9C protein in StarLink corn.
One scientist on the SAPóDr. Rick Helm, a scientist who's specialty is
peanut allergies-- pointed out that in some severely peanut allergic
children , even the odor of peanuts can elicit asthma symptoms. Dr. Helm
further noted that in such individuals, we have no idea how small a dose of
peanut allergen is needed to sensitize a person. In essence, Dr. Helm said
that we don't know enough even about peanut allergies to be able to talk
with any confidence about what a safe level might be.
Second, there are problems with Aventis calculation of 8.5
micrograms as the maximum level of Cry9C protein in the diet of certain
sub-populations in the U.S. But even if we accept Aventisí calculation,
8.5 micrograms equals 8,500 nanograms. Compare that level to the SAP
finding that ìingestion of nanogram quantities [of peanut allergen] can
cause anaphylaxis.î Thus, according to Aventis own calculation, the level
of Cry9C protein in the diet of a segment of the Hispanic population may be
orders of magnitude higher than the level of peanut protein needed to cause
anaphylaxis, according to EPAís own SAP.
Finally, the February 29, 2000 meeting of EPA's SAP mentions the
need to look at allergenicity via both inhalation and ingestion and
concludes that more research is needed: "The exposure and/or monitoring
of Cry9C as an inhalant or potential food allergen source in the population
has not taken into consideration the exposed population demographics, the
exposure concentrations or the exposure time in hours, days or subsequent
seasonal exposures. In essence, well designed scientific studies are not
available to critically assess Cry9C as a potential food allergen" (pg. 9
wholeheartedly agree with this conclusion and do not think that the "new"
data submitted by Aventis constitutes "well designed scientific studies."
Indeed, we think the Aventis argument provides very weak evidence.
In sum, the data presented in Aventis' submission are not strong
enough for them to conclude that there is reasonable certainty that Cry9C
protein is not and will not become an allergen (pg.23, Aventis Submission).
Until such time as Aventis can supply the type of well designed scientific
studies that EPA's SAP says are needed to critically assess this question,
EPA should not grant Aventis present request.
*Consumers Union is a nonprofit membership organization chartered in 1936
under the laws of the State of New York to provide consumers with
information, education and counsel about goods, services, health, and
personal finances; and to initiate and cooperate with individual and group
efforts to maintain and enhance the quality of life for consumers.
Consumers Unionís income is derived solely from the sale of Consumer
Reports, its other publications and from noncommercial contributions,
grants and fees. In addition to reports on Consumers Unionís own product
testing, Consumer Reports, with approximately 4.5 million paid circulation,
regularly carries articles on health, product safety, marketplace economics
and legislative, judicial and regulatory actions which affect consumer
welfare. Consumers Unionís publications carry no advertising and receive no
Bernstein, IL, Bernstein, JA, Miller, M, Tierzieva,S., Bernstein, DI,
Lummus, Z, Selgrade, MK, Doerfler, DL and VL Seligy. 1999. Immune
responses in farm workers after exposure to Bacillus thuringiensis
pesticides, Environmental Health Perspectives, 107(7): 575-582.
Eichenwald, K. 2000. New concerns rise on keeping track of modified corn.
New York Times, October 14.
Feder, B.J. 2000. Farmers cite scarce data in corn mixing. New York
Times, October 17.
Madeod, F. 2000. Bid to grow harmful GM crop in SA. Mail and Guardian,
November 10-16, pg. 7