WHITE PAPER

ON THE POSSIBLE PRESENCE OF CRY9C PROTEIN

IN PROCESSED HUMAN FOODS MADE FROM FOOD FRACTIONS PRODUCED
THROUGH THE WET MILLING OF CORN

I. Introduction

StarLink is a variety of Bt corn that has been genetically engineered to produce a protein, Cry9C,
intended to be toxic to certain insect pests of corn. Following a thorough scientific review of the safety
of this product, EPA concluded that, other than an unresolved issue regarding the potential for Cry9C
to pose an allergenic risk to humans, StarLink would pose no risks to public health or the environment.
Therefore, EPA issued a registration for the Cry9C protein and the genetic material necessary for its
production (called a plant-pesticide) in 1998 to AgrEvo (now Aventis CropScience). EPA limited the
registration by requiring that all StarLink corn only be used in domestic animal feed and for industrial
purposes. EPA did not approve the use of StarLink corn in foods destined for human consumption
because of unanswered questions about the potential allergenicity of the Cry 9C protein.

Because of Aventis’ continuing interest in obtaining approval for use of StarLink in the production of
human food and the novel scientific issues raised concerning the assessment of potential allergenicity,
EPA called a meeting of the FIFRA Scientific Advisory Panel (SAP), on February 29, 2000 regarding
Cry9C protein. (The SAP is an advisory committee, chartered under the Federal Advisory Committee
Act, composed of independent, external experts in the science of assessing the risks of pesticides.) The
February 29, 2000 SAP report stated that with the data available, it could not be determined whether
or not Cry9C is an allergenic protein.

In September 2000, cry9c DNA was detected in a finished food product - taco shells. Subsequently,
the DNA and protein have been found in corn grain and other corn products in the food supply. These
detections indicated that, despite the EPA restrictions, some quantities of StarLink corn had directly
entered the human food chain.

On October 12, 2000, Aventis requested that the registration for their StarLink corn product be
voluntarily cancelled. As a result, StarLink corn is not authorized for planting in future years. On
October 25, 2000, Aventis amended its petition for a food tolerance exemption under the Federal
Food, Drug, and Cosmetic Act (FFDCA) to ask for a temporary tolerance of four years to cover any
Cry9C protein and cry9c DNA that may be present in human food made from StarLink corn planted in
1998, 1999, and 2000. Aventis submitted additional information with its petition to support its
contention that the Cry9C protein posed no allergenic risk to public health. EPA convened another
SAP meeting on November 28, 2000 to consider the question of the potential of the Cry9C protein to
be an allergen, whether there is an adequate amount of the protein in corn to cause sensitization, and
what amount of Cry9C might be in the human food supply if this time limited tolerance exemption were

1
to be approved. More information including the Aventis submission, EPA’s papers for SAP review,
background information, and the SAP final reports can be found on the following web sites:

http://www.epa.gov/pesticides/biopesticides/cry9c/index.htm
http://www.epa.gov/scipoly/sap/index.htm

The final report from the November 28, 2000 SAP meeting was issued on December 5, 2000 which
expressed the consensus of the Panel that while Cry9C has a “medium likelihood” to be an allergen, the
combination of the expression level of the protein and the amount of corn found to be commingled
poses a “low probability” to sensitize individuals to Cry9C. The Panel report noted that the likelihood
of the protein being detected in different corn products varied considerably, especially depending on the
method of processing and whether the product was from white or yellow corn. The cry9c DNA was
only engineered into certain yellow corn varieties. The SAP report called on EPA to only include in our
dietary assessment those ingredients from corn that contain protein after processing. The SAP report
states that items such as corn syrup, corn oil, and starch contain virtually no protein.

EPA did not include corn syrup and corn oil in its dietary assessment presented at the November 28,
2000 SAP meeting because protein is absent or virtually undetectable in these food products.
However, the Agency has decided to further review wet milling methods to further address potential
dietary exposure to the Cry9C protein. EPA’s review is based on published literature, comments from
the SAP members during the November SAP meeting, software on dietary exposure, and information
from corn industry representatives. EPA examined some of the raw data from an industry performed
starch study, but did not examine any of the other raw data forming the basis of the information and
conclusions from the cited literature and industry representatives.

Field corn that is made into processed foods for human consumption undergoes milling. There are two
primary types of milling: “dry milling” which produces flour and meal primarily and “wet milling” which
produces high fructose corn syrup, oil, starch, some animal feed products, and ethanol primarily. Corn
products produced by wet milling contain varying levels of protein. Those products intended for human
food consumption contain no or extremely low levels of intact protein. The exception is the use of one
protein, corn zein, specifically extracted from corn. Corn zein will be discussed further in Section III.
In contrast, products intended for animal feed may contain high levels of protein. The wet milling
process effectively separates these products and industry standards control protein content in food
products to a very low or undetectable level. The much more simple dry milling processes do not
remove protein from products intended for human consumption. Although additional processing such
as alkaline treatments and cooking may also affect the level of protein in the finished food, this paper
focuses on the wet milling process, although dry milling is discussed to provide a more complete
understanding of the milling processes.

2
II. Corn Milling

Nearly 2 billion bushels of corn are typically produced annually in the United States for food and other
industrial purposes. This represents approximately 20% of the corn grown in the U.S. annually with the
remaining 80% typically used for animal feed. The majority of corn earmarked for food or industrial
use is subjected to wet milling (Figure 1). The remaining corn is subjected to either dry milling or
alkaline cooking plus dry milling (Masa processing). In general, only yellow corn is used for wet milling,
while both white and yellow corn are used in dry milled products. The predominate type of yellow corn
used in wet milling is a “dent-type” corn. Approximately one-third of the yellow corn used for starch
production in the wet milling process is a “waxy-type” maize. StarLink corn is a “dent-type” of maize,
not the waxy-type.

Figure 1. Percentage of Corn Used in Each Milling Process

Source: http://www.ianr.unl.edu/pubs/fieldcrops/g1115.htm

A. Corn Wet Milling

The wet milling process involves a series of steps by which corn is separated into various
components, which are then further processed and/or used for animal feed. The basic steps for
wet milling include steeping, germ separation, fine grinding, starch separation, fermentation, and
syrup conversion. Corn processed by wet milling is typically separated into 5 basic
components: starch, germ, gluten, fiber and steep liquor (Blanchard, 1992). By-products of
wet milling along with corn germ and most corn gluten are used for animal feed products. A
very small amount of corn gluten is subjected to acid hydrolysis resulting in amino acids and/or
short peptide units called hydrolyzed vegetable protein (HVP).

The wet milling process involves a series of steps which produce the various fractions
described below (Corn Refiners Association, 2000; AAC, 1998; Blanchard, 1992; Jackson,
1996; May, 1987). Figure 2 provides a basic overview of the wet milling process. Corn
received at a refinery is inspected and cleaned twice to remove cob, dust, chaff and foreign
materials prior to steeping.

3
 Figure 2. Corn Wet Milling Process

Source: Minnesota Corn Processors, LLC: http://www.mcp.net/info/wetmill.html

Wet Milling Process (as described: Corn Refiners Association, 2000; Minnesota Corn
Processors, 1999):

1. Steeping:
Steeping takes place in stainless steel steep tanks which hold approximately 3,000
bushels of corn each. Corn is stored in these tanks for approximately 30 to 40 hours in
50° C soaking water which contains approximately 0.1% sulfur dioxide (to prevent
excessive bacterial growth). During the incubation period, kernel moisture levels
increase to between 15 and 45%, which also results in an increase in kernel size (up to
2X). As the corn is stored in the mildly acidic steep water, the gluten bonds within the
corn loosen, which allows for starch release. After steeping, the corn is coarsely
ground to break the germ loose from the other components. The steep water is
condensed to capture nutrients and this water is used for animal feed and future
fermentation processes. The resulting ground corn is contained in a water slurry which
flows to cyclone germ separators.

2. Germ Separation:
Corn germ contains approximately 85% of the oil found in corn. The cyclone
separators, which are similar to centrifuges, spin the low density corn germ out of the

4
 slurry and pump the germ onto screens where the germ is washed repeatedly to remove
any residual starch. The germ is finally subjected to a combination of mechanical and
solvent processes which extract the oil from the germ. The resulting germ residue is
saved to be used as a component of animal feed.

3. Fine Grinding:
The corn and water slurry are moved from the germ separator into an impact or
attrition-impact mill to release the starch and gluten from the fiber in the kernel. The
suspension of starch, gluten and fiber flows over fixed concave screens which catch
fiber but allow the starch and gluten to pass through. The fiber is then collected,
slurried and screened again to reclaim any residual starch or protein, then piped to the
feed house to be used as a major component of animal feed. The starch-gluten
suspension (mill starch) is piped to the starch separators.

4. Starch Separation:
Mill starch is passed through a centrifuge which allows for the gluten to be spun out,
mostly for use in animal feed (gluten has a lower density than starch). At this point, the
starch has only approximately one to two percent of protein remaining. The starch is
diluted 8 to 14 times, rediluted and washed again in hydroclones to remove the last
trace of protein and produce high quality starch (usually > 99.5% pure). Most of the
corn starch is converted into corn syrups, dextrose, high fructose corn syrups and
crystalline fructose. Some of this starch is dried and marketed as unmodified corn
starch and some is modified into speciality starches.

5. Syrup Conversion:
Starch is suspended in water and liquified in the presence of acid and/or enzymes. The
resultant product is a low-dextrose solution. The solution is enzymatically-treated
further to continue the conversion process of starch into syrup. Throughout the
process, refiners can halt acid or enzyme actions at necessary points to produce the
proper mixture of sugars (e.g., dextrose and maltose) for syrups. The syrup is refined
in filters, centrifuges and ion-exchange columns and excess water is evaporated.
Syrups are sold directly, crystallized into pure dextrose, or processed further to create
high fructose corn syrup.

6. Fermentation:
Corn starch is also used to produce feedstock suitable for traditional yeast or bacterial
fermentation methods. Enzymes are used to modify corn starch to produce the
feedstock and the resulting fermentation product is ethanol. Alcohol production by wet
milling accounts for approximately 306 million bushels of corn annually. This ethanol
product is distilled to remove excess water and sold for use in industry and beverages.

5
 A by-product of the fermentation, carbon dioxide, is also sold to beverage
manufacturers to be used in carbonated beverages.

B. Corn Dry Milling

Dry milling for food use (including Masa processing) represents approximately 165 million
bushels annually in the U.S. Both white and yellow corn are processed by dry milling to
produce food products, with approximately 50 million bushels (about 30%) of this corn being
white corn (David Shipman, Personal Communication, 2000). Unlike the elaborate wet milling
process, corn dry milling is basically a simple grinding procedure. As in the wet milling process,
the corn is initially cleaned. Once clean, the moisture content of corn is raised to about 20%.
The corn germs are then removed for oil extraction and the remaining corn is ground and sieved
into many fractions which vary in particle size and composition. The primary products of dry
milling are flour, cornmeal and grits. Additional products include corn bran and feed mixtures.
These products are used in brewing, foods, building products (binders), fermentations
(pharmaceuticals and fuel), and animal feeds. Dry milling for alcohol production accounts for
approximately 161 million bushels annually in the U.S.

Alkaline-cooked corn is used in tortillas, tortilla chips, corn chips and other similar items.
Whole kernel corn is cooked in near-boiling water containing 1% lime for approximately 20
minutes. The corn is allowed to soak for 8-12 hours (steeping). The corn is then drained from
the steep water and washed with clean water to remove excess lime and the pericarp which has
been loosened. The washed corn is now at about 45-50% moisture and is subjected to stone
grinding to form a dough. If the dough is formed into strips and fried, corn chips are produced.
If the dough is formed into thin pancake-like sheets and baked, corn tortillas are produced. If
the baked tortillas are subsequently fried, tortilla chips are produced.

III. Protein Content of Fractions Resulting from Wet Milling

The wet milling process effectively separates protein-containing and non-protein-containing products.
Table 1 below provides the approximate percentages of protein found in each of the various wet-milled
fractions/products.

6
 Table 1. Protein Contents and Uses of Products of Wet Milling

Fraction Approximate Percent Protein Uses

Content 1
Steep Liquor 45-48% Protein Animal Feed
(condensed)

Corn Germ 20% Animal Feed

Bran/Gluten Feed 18-22% Protein Animal Feed2

Gluten Meal $ 60% Protein Animal Feed2

Starch 0.3-0.35% Protein unmodified corn

(high amylose corn - up to 1%)3 starch, speciality
corn starch, corn
syrups, and
dextrose

Syrup Not Detectable 4 pure dextrose,

(made from corn starch) corn syrup, and
high fructose corn
syrup
Alcohol Not Detectable 4 ethanol
(made from corn starch)

Corn Oil Not Detectable cooking or salad

oil
1
Source: Blanchard,1992; Kyd Brenner, 2000
2
Most corn gluten is used for animal feed. Some is treated to extract corn zein and some is
subjected to acid hydrolysis to produce hydrolyzed vegetable protein as described in Section
III. B.
3
High amylose corn does not contain cry9c DNA or protein
4
Further processing of corn starch removes remaining protein

A preliminary study was performed using both the EnviroLogix (EnviroLogix, Inc) and SDI (Strategic
Diagnostics, Inc) Cry9C ELISA Well Tests to determine if any Cry9C protein was present in 74 starch
samples. Some of the raw data were provided for EPA review. In this study, the control samples are

7
from whole corn grain, not starch so the test system has not been validated for detection of Cry9C in
starch. The results of this study show no Cry9C protein detected in any of the 74 starch samples
(Charles Conner, Personal Communication, 2001).

In addition, a report on DNA analysis in maize (corn) starch and starch hydrolysates (various corn
sugars coming from the wet milling process) shows that corn DNA and DNA from Bacillus
thuringiensis (Bt) engineered into corn can be detected in starch, but not in the starch hydrolysates
(AAC, 1998). The Bt DNA could be detected in wet milling products such as the germ and fibers
where corn proteins also are commonly found. However, five independent laboratories were unable to
detect any corn or Bt DNA in maltodextrin, glucose syrup produced by three separate processes,
crystalline dextrose, and crystalline fructose. In addition, no DNA was detected in refined corn oil.
Such tests usually detect fragments of DNA and the heating and other milling process are likely to
degrade and denature any proteins. The authors were unable to obtain analytical methods for the
detection of the Bt proteins and therefore could not conduct these analyses. In general, a DNA test is
more sensitive than a test for protein. Although not always conclusive, the inability to detect DNA adds
to the weight of evidence that protein is not present in corn syrup and corn oil. Corn proteins are found
in food grade starch at very low levels. Whole kernel corn contains about 8.5 to 12 % protein, but
food grade starch has protein levels two orders of magnitude lower at 0.3 to 0.35% (Kyd Brenner,
Personal Communication, 2000; Blanchard, 1992).

A. Animal Feed Products

Four major animal feed products are produced from different combinations of steep water,
corn germ residues, fiber and corn gluten (Corn Refiners Association, 2000). These products
include gluten meal, gluten feed, corn germ meal and condensed fermented corn extracts (steep
water). Each of the products contains a relatively high percentage of protein. Corn gluten meal
supplies vitamins, minerals and energy in such products as poultry feed. Steepwater is a liquid
protein supplement for cattle and is also used as a binder in feed pellets. Corn gluten feed
provides protein and fiber for beef cattle. All of these products are strictly limited to animal
feed use and would not be present in the human food supply.

B. Gluten-Derived Products

Corn zein is an insoluble protein which is contained in the corn gluten fraction (see
http://www.arserrc.gov/es/zeinextratech.htm). Corn zein is used as a glazing and coating agent
in the food and pharmaceuticals industries. In contrast, Cry9C is a water soluble protein. The
high volumes of water used in the extraction/purification process to obtain zein should therefore
eliminate the presence of any Cry9C and other water soluble proteins from corn zein.

A very small amount of corn gluten is subjected to acid hydrolysis resulting in a hydrolyzed
vegetable protein (HVP). Acid hydrolysis using extremely concentrated (up to 6 normal)
hydrochloric acid for 24 hours degrades a protein into its constituent amino acids and/or short
peptides. These HVPs are more commonly produced from soybeans and wheat. Even though

8
 corn, wheat and soybeans are known as food allergens, there have been no well documented
cases of adverse reactions to any hydrolyzed vegetable protein (Talyor, Personal
Communication 2001).

C. Starch and Starch-Derived Products

Starch and starch-derived products account for approximately 74% of the products obtained
by wet milling as a percentage of the raw corn (May, 1987). Figure 3 shows the approximate
distribution of food use of starch-related products resulting from wet milling.

Figure 3. Food Use of Starch Resulting from Wet Milled Corn

(17.32%) (21.02%)
Starch
Glucose/
Dextrose

(23.52%)
Alcohol
(38.14%)
High Fructose
Syrups

Source:http://www.ianr.unl.edu/pubs/fieldcrops/g1115.htm

The amount of corn used for starch and starch-derived chemicals approaches nearly 1 billion bushels
annually in the U.S. with approximately 250 million of these bushels used for corn starch production
(Table 2). About 15% (37.5 million bushels) of this corn starch is for food and pharmaceutical use and
the remaining 85% (212.5 million bushels) is designated for industrial use (David Shipman, Personal
Communication, 2000). Approximately 33% of the food starch market is comprised of corn starch
made from waxy maize (Cry9C has not been bred into this speciality corn [Torres, Personal
Communication, 2000]). For corn starch produced from most corn, residual protein levels must be
below 0.5% according to industry standards (NAS food chemicals CODEX [Brenner, Personal
Communication, 2000]). Generally

9
, this level is reduced even further to between 0.3-0.35% (Blanchard, 1992; Corn Refiners Association,
2000b). The only exception to these levels of protein in starch occurs when the starting material is high
amylose corn. High amylose corn is a minor specialty corn (Cry9C has not been bred into this
speciality corn), where protein levels must be below 1.0% in the finished starch. In addition, only about
0.01% of the protein in finished starch is water soluble protein due to the large number of water washes
in the wet milling process (Corn Refiners Association, 2000a). Again, Cry9C is a water soluble protein
and any Cry9C that remained in starch would be as a fraction of this 0.01% soluble protein. Because
of the low levels of soluble protein found in corn starch, the levels of protein in corn starch-derived
products would be less than the 0.01% and are undetectable in such products as corn syrup and
alcohol because of further processing. In addition to food uses, starch is routinely used as an adhesive,
for manufacture of papers, wallboard, adhesives, anticaking agents, dusting powder, thickening agents
and as a filler for pharmaceuticals (Jackson, 1996).

Table 2. Summary of Corn Disposition Going Into Food Starch

Corn Use Approximate Amounts

Total Corn Wet Milled Annually 1.35 Billion Bushels

Percent of Wet Milled Corn Bushels 1 Billion Bushels (74%)

Used to Make Starch and Starch-
Derived Products

Percent of Starch Output Used 250 Million Bushels (25%)

Specifically to Make Corn Starch

Percent of Corn Starch 37.5 Million Bushels (15%)

Used for Food and Pharmaceuticals

D. Corn Oil

As described earlier, corn germ contains approximately 85% of the oil found in corn. Once
corn germ is separated as part of the wet milling process, corn oil is further refined (Rich
Torres, Personal Communication, 2000). Crude oil is degummed in the presence of
phosphoric acid which removes proteins, phospholipids, gums, etc. Following degumming, the
oil is alkaline treated which removes fatty acids, neutralized and bleached. Finally, the oil is
deodorized which removes residual proteins/amino acids and color bodies, resulting in a refined
oil product. Approximately 50% of U.S.-produced corn oil is used for cooking or salad oil and
another 25% is used for corn oil margarine (Corn Refiners Association, 2000). Through the
corn oil refining process, proteins are removed and are not detectable in food grade corn oil
(Brenner, Personal Communication, 2000; Torres, Personal Communication, 2000). In
addition, Bt DNA could not be detected in the corn oil (AAC, 1998).

10
IV. Dietary Exposure to Corn Starch

As noted earlier, the FIFRA Scientific Advisory Panel met on November 28, 2000 to offer advice on
EPA’s “Assessment of Scientific Information Concerning StarLink Corn.” The SAP’s final report of
that meeting contains several references to their expectation that human food fractions produced from
wet milling of corn will not contribute significantly, if at all, to potential human exposure to Cry9C
protein. For example, the SAP wrote: “As is entirely appropriate, both the Agency and Aventis count
only those ingredients that contain protein after processing in assessing dietary exposure. Thus foods
containing corn bran and corn endosperm are counted, while corn syrup, corn oil, starch and other
food forms made from corn grain are not counted since they contain virtually no protein.” (SAP Report
p.21).

Because corn starch used in human food is likely to have a very low level of corn protein (typically
about 0.3% total protein) and some of that total protein could be Cry9C, EPA has performed a
quantitative assessment of the potential for human exposure to the Cry9C protein from corn starch.
EPA has developed an estimate of exposure from consumption of food containing corn starch based on
EPA’s earlier estimates of exposure presented to the (SAP) for its November 28, 2000 meeting and
new information gathered since November. The original paper (referred to as the November Exposure
Assessment) may be found at:
http://www.epa.gov/scipoly/sap/2000/november/prelim_eval_sub102500.pdf.

EPA’s November Exposure Assessment depended on the three variables: (1) the amount of corn
product consumed, (2) the percentage of the corn used in making a food item that was StarLink, and
(3) the level of Cry9C in the StarLink corn portion of the food. The SAP agreed with EPA’s basic
approach and, therefore, EPA has used a similar approach in calculating an estimated exposure to
Cry9C protein solely from corn starch. EPA concluded that it would not change its “mixing”
assumptions, i.e., the assumptions regarding the percentage of commingled corn that is StarLink. The
other two variables were modified based upon the available information for corn starch produced by
wet milling and starch consumption data from TAS-DIET software leased by FDA.

Consumption of cornstarch was estimated using the TAS-DIET software (see Table3), which is based
on the USDA 1989-91 Continuing Survey of Food Intakes by Individuals (CSFII). Although
consumption of cornstarch, per se, is not reported in the CSFII, TAS-DIET allows for estimating
consumption of raw agricultural commodities (RACs) from the survey food codes. Cornstarch
consumption was estimated by using the RAC code for corn grain endosperm only (which represents
the ingredient cornstarch as well as cornmeal and corn flour) but restricting the survey food codes to
only those that would likely contain cornstarch as opposed to cornmeal or corn flour.

EPA has estimated the potential exposure to Cry9C protein for the general US population, as well as
for various age groups of infants and children. (See Table 3.) Because data indicated that Hispanics
generally consumed higher levels of many corn flour-based and corn meal-based foods than the general

11
population, the November Exposure Assessment also estimated exposure to Hispanics and different
age groups of Hispanic children. Data, however, do not show that the Hispanic population are more
likely to consume corn starch produced through the wet milling than are any other portion of the US
population. So while EPA’s November assessment specifically considered these subpopulations, this
assessment of exposure to corn starch produced from the wet milling process does not include specific
assessments for Hispanic subpopulations.

Table 3: Per Capita Estimates of Daily Consumption of Corn Starch in g/day.

Population Subgroup 95th Percentile 99th Percentile 99.5 Percentile

US Population 20 57 81

All infants (<1 year) 7 11 12

All children, 1-6 years 6 15 20

All children, 7-12 years 9 25 33

Source: TAS-DIET software based on the Continuing Survey of Food Intake by Individuals from 1989
to 1991.

As described above, starch is highly refined through the wet milling process to remove protein and EPA
has used a value of 0.01% protein in corn starch (0.0001 grams protein/gram starch). This value
reflects the maximum amount of water-soluble protein potentially remaining in corn starch (May, 1987).
A limited number of samples indicated that Cry9C protein could not be detected in corn starch
(Conner, 2001). In addition to reviewing the industry data on Cry9C protein levels in corn starch, EPA
has calculated the amount of Cry9C potentially present in corn starch. This calculation relies on several
key conditions. First, the USDA and Aventis program for StarLink corn insures that all of the StarLink
and buffer corn will only be moved to appropriate operations where the grain will be fed to animals or
used for industrial purposes. StarLink corn will not be channeled into the wet milling process even
though the food products from wet milling have little or no protein of any kind. Next, that corn grain at
the elevators will be tested for the presence of Cry9C protein, and if detected, that corn will be
channeled to domestic animal feed or industrial uses only.

Finally, it is important to note the SAP concluded that the high end estimates in EPA’s November
Exposure Assessment were likely to overstate potential exposure. Specifically, the SAP wrote that:
“The Agency’s analysis results in an upper bound estimate that is considerably high and could be
justifiably reduced if several of the issues cited were incorporated. However, this conservative
approach results in an estimate with a significant safety factor” (SAP Report page 19). The SAP cited
a number of factors that could result in estimated values that are higher than are likely to occur. These
factors included: (1) a greater degree of mixing of StarLink and non-StarLink corn than assumed by

12
EPA’s high end estimates; (2) the industry practice of preferring corn varieties other than StarLink for
producing processed human foods, thus reducing the likelihood that StarLink would be directed to
human food channels; and (3) the effects of processing on levels of Cry9C protein in processed food.
EPA agrees with the SAP that our November upper bound estimates overstated potential exposure to
a considerable degree. For example, the upper bound estimate for mixing in the year 2000 results in
about 1.5% StarLink corn. For 1999, the upper bound estimate is about 1.2% StarLink corn, only
slightly less than 2000 as discussed in EPA’s November Exposure Assessment for the November SAP
meeting. These values are three to four times higher than the percentage of the overall U.S. acreage
planted to StarLink corn in those years. Because EPA has not modified the methodology used to
generate our earlier estimates to take these factors into account, the estimates below retain the same
significant safety factor.

First, EPA calculated the amount of Cry9C protein in cornstarch as follows: total protein in cornstarch
(0.0001 grams total protein per grams cornstarch) times 0.0000129 (grams of Cry9C protein per
grams corn kernel) divided by 0.08 (grams total protein per grams corn kernel). Thus, the amount of
Cry9C assumed to be in corn starch is 0.00000001 grams Cry9C in each gram of corn starch, or 1.61
x 10-2 µg/g. Using this estimate and the consumption estimates from Table 3, EPA calculated the
amount of Cry9C protein potentially in the diets of adults, infants, and children in the U.S. See Tables
4 and 5 below. Note that these numbers are given in micrograms in order to emphasize the extremely
low amounts of Cry9C protein that might be present.

Table 4. Estimated Upper Bound Exposure for Various Population Groups for 2000 Assuming Food
Containing Corn Starch Was Made from Grain Containing 1.5% StarLink Corn

Group Potential Daily Exposure of Cry9C Protein from Corn Starch

Upper Bound Exposure for 2000 (1.5%)

Percentile: 95 99 99.5

US Population 0.004 ug 0.013 ug 0.019 ug

Infants 0.001 ug 0.002 ug 0.002 ug

Children 1 to 6 yrs 0.001 ug 0.003 ug 0.004 ug

Children 7 to 12 yrs 0.002 ug 0.006 ug 0.007 ug

*Data obtained from FDA TAS-DIET Analysis
ug=micrograms

13
Table 5. Estimated Upper Bound Exposure for Various Population Groups for 1999 Assuming Food
Containing Corn Starch Was Made from Grain Containing 1.2% StarLink Corn

Group Potential Daily Exposure of Cry9C Protein from Corn Starch

Upper Bound Exposure for 1999 (1.2%)

Percentile: 95 99 99.5

US Population 0.003 ug 0.011 ug 0.015 ug

Infants 0.001 ug 0.002 ug 0.002 ug

Children 1 to 6 yrs 0.001 ug 0.002 ug 0.003 ug

Children 7 to 12 yrs 0.001 ug 0.004 ug 0.006 ug

*Data obtained from FDA TAS-DIET Analysis
ug=micrograms

In summary, EPA believes that the upper bound estimates of potential exposure to Cry9C protein as a
result of consumption of corn starch are extremely low (approximately one-hundredth of a microgram a
day). Based on the review by the SAP and the limited data available indicating Cry9C cannot be
detected in commercial corn starch, this extremely low amount is likely to overestimate exposure
significantly. After public and scientific review, EPA will evaluate the impact that this new information
has on assessing the potential exposure to StarLink corn from eating food manufactured through the
wet milling process.

V. Recommendation

Based on the above evaluation, EPA believes it is reasonable to conclude that there is virtually no
Cry9C protein in wet milled products and that there is no likely health concern for the public associated
with the consumption of any food fraction produced by wet milling of corn as long as reasonable steps
are taken to ensure that StarLink corn is not diverted to wet milling. Data show that corn protein will
not be present in high fructose corn syrup, corn oil, or alcohol (ethanol). Data also indicate that corn
starch will contain, at most, such extremely low levels of corn protein that there is virtually no potential
human exposure to Cry9C protein from consumption of corn starch.
Continued testing of corn grain for Cry9C protein prior to entry into the food processing chain and
diverting any shipments testing positive to domestic animal feed or industrial purposes will insure that
food fractions from wet milling contains virtually no Cry9C protein. Such testing will minimize the
possible occurrence of shipments of corn containing StarLink from entering the wet milling process.

A possible scenario is for testing whenever a wet milling operation produces food starch. The method
would require testing for the presence of Cry9C protein prior to processing using a representative

14
sample of 400 kernels of corn from selected conveyances (e.g., rail car, barge, truck, etc.) using
methods validated by the USDA’s Grain Inspection, Packers, and Stockyards Administration (GIPSA)
and accepted by FDA. If StarLink corn is determined to be present in the grain, it would have to be
diverted to appropriate channels. More details on the methods recommended for dry milling and
related assistance can be found at the following web sites:

http://vm.cfsan.fda.gov/~dms/starguid/html
http://www.usda.gov/gipsa/reference-library/handbooks/grain-insp/grbook1/gihbk1.htm
http://www.usda.gov/agency/gipsa/pubs/primer.pdf.

15
 References

AAC Study on DNA Analyses (1998). Starch and Starch Derivatives Produced from BT-Maize.
Association Des Amidonneries De Cereales De L’U.E. (21 pp).

Blanchard, Paul H. (1992). Technology of Corn Wet Milling and Associated Processes. Industrial
Chemistry Library, Volume 4 (pp. 69-125). Elsevier (Amsterdam-London-New York-Tokyo).

Brenner, Kyd (2000). Corn Refiners Association - Personal Communication (Brenner to Phil Hutton).
November 14, 2000.

Conner, Charles F. (2001). Corn Refiners Association - Personal Communication to Janet L.

Andersen regarding Quantitation of Cry9C in Corn Wet Milling Unmodified Starch Products Using
Commercial ELISA Cry9C Well Tests. January 8, 2001 and January 16, 2001.

Corn Refiners Association (2000a). Comments from the Corn Refiners Association to OPP Docket
PF 867B. Attachment 3.

Corn Refiners Association (2000b). The Corn Refining Process:

http://www.corn.org/web/process.htm

Jackson, David S. (1992). Corn Quality for Industrial Uses. Cooperative Extension, Institute of
Agriculture and Natural Resources, University of Nebraska:
http://www.ianrwww.unl.edu/pubs/fieldcrops/g1115.htm

May, James B. (1987) Wet Milling: Process and Products. Corn: Chemistry and Technology (Watson
and Ramstad editors) 12:377-397. American Association of Cereal Chemists, Inc., St. Paul, MN.

Minnesota Corn Processors, LLC (1999). Corn Wet Milling Process:

http://www.mcp.net/info/wetmill.html

Shipman, David , USDA. Personal Communication-email (Shipman to Janet Andersen), December

18, 2000.

Taylor, Steve, University of Nebraska, Personal Communication (Taylor to Michael Watson),

February 14, 2000.

Torres, Rich, Assistant Vice President - Merchandising Manager, Cargill. Personal Communication-
FAX (Torres to Janet Andersen), December 18, 2000.

16