[Federal Register: October 26, 1999 (Volume 64, Number 206)]
[Rules and Regulations]
[Page 57699-57733]
From the Federal Register Online via GPO Access [wais.access.gpo.gov]
[DOCID:fr26oc99-21]
[[Page 57699]]
_______________________________________________________________________
Part II
Department of Health and Human Services
_______________________________________________________________________
Food and Drug Administration
_______________________________________________________________________
21 CFR Part 101
[[Page 57700]]
DEPARTMENT OF HEALTH AND HUMAN SERVICES
Food and Drug Administration
21 CFR Part 101
[Docket No. 98P-0683]
Food Labeling: Health Claims; Soy Protein and Coronary Heart
Disease
AGENCY: Food and Drug Administration, HHS.
ACTION: Final rule.
-----------------------------------------------------------------------
SUMMARY: The Food and Drug Administration (FDA) is authorizing the use,
on food labels and in food labeling, of health claims on the
association between soy protein and reduced risk of coronary heart
disease (CHD). Based on its review of evidence submitted with comments
to the proposed rule, as well as evidence described in the proposed
rule, the agency has concluded that soy protein included in a diet low
in saturated fat and cholesterol may reduce the risk of CHD by lowering
blood cholesterol levels.
DATES: This regulation is effective October 26, 1999, except for
Sec. 101.82(c)(2)(ii)(B), which contains information collection
requirements that have not been approved by the Office of Management
and Budget (OMB). Upon approval, the FDA will publish a document in the
Federal Register announcing the effective date of those requirements.
FOR FURTHER INFORMATION CONTACT: Susan M. Pilch, Center for Food Safety
and Applied Nutrition (HFS-465), Food and Drug Administration, 200 C
St. SW., Washington, DC 20204, 202-205-4500.
SUPPLEMENTARY INFORMATION:
I. Background Information
On November 8, 1990, the President signed into law the Nutrition
Labeling and Education Act of 1990 (the 1990 amendments) (Public Law
101-535). This new law amended the Federal Food, Drug, and Cosmetic Act
(the act) in a number of important ways. One notable aspect of the 1990
amendments was that they provided procedures whereby FDA is to regulate
health claims on food labels and in food labeling.
In the Federal Register of January 6, 1993 (58 FR 2478), FDA issued
a final rule that implemented the health claim provisions of the act
(hereinafter referred to as the 1993 health claims final rule). In that
final rule, FDA adopted Sec. 101.14 (21 CFR 101.14), which sets out
rules for the authorization and use of health claims by regulation.
Additionally, Sec. 101.70 (21 CFR 101.70) establishes a process for
petitioning the agency to authorize by regulation the use of health
claims about a substance-disease relationship (Sec. 101.70(a)) and sets
out the types of information that any such petition must include
(Sec. 101.70(f)).
In response to the 1990 amendments, FDA also conducted an extensive
review of the evidence on 10 substance-disease relationships. As a
result of its review, FDA has authorized claims for 8 of these 10
relationships, one of which focused on the relationship between dietary
saturated fat and cholesterol and reduced risk of CHD. CHD is the most
common, most frequently reported, and most serious form of
cardiovascular disease (CVD) (58 FR 2739, January 6, 1993). Further,
although the agency denied the use on food labeling of health claims
relating dietary fiber to reduced risk of CVD (58 FR 2552), it
authorized a health claim relating diets low in saturated fat and
cholesterol and high in fruits, vegetables, and grain products that
contain dietary fiber (particularly soluble fiber) to a reduced risk of
CHD.
In the proposed rule entitled ``Health Claims and Label Statements;
Lipids and Cardiovascular Disease'' (56 FR 60727, November 27, 1991)
(hereinafter referred to as the saturated fat/cholesterol proposed
rule), FDA set out criteria for evaluating evidence on diet and CVD
relationships. The agency focused on those aspects of the dietary lipid
and CVD relationship for which the strongest scientific evidence
andagreement existed. FDA noted that, because of the public health
importance of CHD, identification of ``modifiable'' risk factors for
CHD had been the subject of considerable research and public policy
attention. The agency also noted that there is general agreement that
elevated blood cholesterol levels are one of the major ``modifiable''
risk factors in the development of CHD. FDA cited Federal Government
and other reviews that concluded that there is substantial
epidemiologic and clinical evidence that high blood levels of total and
low density lipoprotein (LDL)-cholesterol are a cause of
atherosclerosis and represent major contributors to CHD. Further,
factors that decrease total blood cholesterol and LDL-cholesterol will
also decrease the risk of CHD. FDA concluded that it is generally
accepted that blood total and LDL-cholesterol levels are major risk
factors for CHD, and that dietary factors affecting blood cholesterol
levels affect the risk of CHD. High intakes of dietary saturated fat
and, to a lesser degree, of dietary cholesterol are consistently
associated with elevated blood cholesterol levels. FDA tentatively
concluded that the publicly available data supported an association
between diets low in saturated fat and cholesterol and reduced risk of
CHD (56 FR 60727 at 60737), and it confirmed that conclusion in the
saturated fat/cholesterol final rule (58 FR 2739 at 2751).
Based on its review using the stated criteria, and on its
consideration of comments received in response to the proposed rule
entitled ``Health Claims; Dietary Fiber and Cardiovascular Disease''
(56 FR 60582), FDA concluded that the publicly available scientific
information supported an association between diets low in saturated fat
and cholesterol and high in fruits, vegetables, and grain products
(i.e., foods that are low in saturated fat and cholesterol and that are
good sources of dietary fiber) and reduced risk of heart disease (58 FR
2552 at 2572). In the 1993 dietary fiber and CVD final rule, in
response to a comment regarding the apparent hypocholesterolemic
properties of specific food fibers, FDA again articulated its criteria
for evaluating diet and CHD relationships (58 FR 2552 at 2567). FDA
agreed that the effectiveness of naturally occurring fibers in foods in
reducing the risk of CHD may be documented for specific food products.
Further, the agency indicated that if manufacturers could document,
through appropriate studies, that dietary consumption of the soluble
fiber in a particular food has a beneficial effect on blood lipids
predictive of CHD risk, they should petition for a health claim for
that particular product. In response to two petitions that documented
such evidence, FDA has authorized health claims for soluble fiber from
certain foods and reduced risk of CHD in Sec. 101.81 (21 CFR 101.81)
(62 FR 3600, January 23, 1997, and amended at 62 FR 15344, March 31,
1997, and 62 FR 8119, February 18, 1998).
In the Federal Register of November 10, 1998 (63 FR 62977), and in
response to a petition from Protein Technologies International, Inc.
(Ref. 1 and Ref. 2), the agency proposed Sec. 101.82 to provide for
health claims on the relationship of soy protein and reduced risk of
CHD (hereinafter referred to as the soy protein proposed rule). In the
soy protein proposed rule, FDA considered the relevant scientific
studies and data presented in the petition as part of its review of the
scientific literature on soy protein and CHD. The agency summarized
this evidence in the soy protein proposed rule and presented the
rationale for a health claim on this food-disease relationship as
provided for under the significant scientific
[[Page 57701]]
agreement standard in section 403(r)(3)(B)(i) of the act and
Sec. 101.14(c) of FDA's regulations.
Proposed Sec. 101.82(c)(2)(ii)(A) identified the substance that is
the subject of the proposed claim as soy protein from the legume seed
Glycine max. The soy protein proposed rule included qualifying criteria
for the purpose of identifying soy protein-containing foods eligible to
bear the proposed health claim. The proposal also specified mandatory
content for health claim statements; identified additional, optional
information for such statements; and provided model health claims.
In its evaluation of the scientific evidence for a relationship
between consumption of soy protein and blood total and LDL-cholesterol
levels, the agency found the data suggestive but not sufficient to
establish a dose-response for this relationship. However, the agency
did find consistent, clinically significant reductions of total and
LDL-cholesterol levels in controlled trials that used at least 25 grams
(g) of soy protein per day. Thus, the agency proposed to base the
qualifying level of soy protein on a total daily intake of 25 g, as
suggested by the petitioner. Therefore, in Sec. 101.82(c)(2)(iii)(A),
FDA proposed the qualifying criterion for a food to bear the claim as
6.25 g of soy protein per reference amount customarily consumed (RACC)
(i.e., 25 g divided by 4 eating occasions per day).
In the soy protein proposed rule, FDA had tentatively indicated its
intention to use a specific analytical method to measure soy protein
for assessing compliance with the qualifying criterion. Comments
persuaded the agency that the method would be inadequate for many
products. Therefore, in the Federal Register of August 23, 1999 (64 FR
45932), FDA issued a proposed rule to provide for an alternative
procedure for assessing compliance (hereinafter referred to as the soy
protein reproposal). In the soy protein reproposal, in
Sec. 101.82(c)(2)(ii)(B) FDA proposed that it would rely on measurement
of total protein and require manufacturers, when soy is not the sole
source of protein in foods, to maintain records that document the
amount of soy protein in products and to make these records available
to appropriate regulatory officials for inspection and copying upon
request.
II. Summary of Comments and the Agency's Responses
In response to the soy protein proposed rule, the agency received
approximately 130 submissions, each containing one or more comments,
from consumers, consumer organizations, professional organizations,
government agencies, industry, trade associations, health care
professionals, and research scientists.
About half of these submissions supported the proposed rule without
providing grounds for this support other than those provided by FDA in
the preamble to the soy protein proposed rule. The majority of the
remaining comments were generally supportive, but requested
modification of one or more provisions of the proposed rule. Some
comments provided additional data on the relationship between soy
protein and CHD, including one submission, originally submitted as a
health claim petition and converted to a comment on the soy protein
proposed rule (Ref. 3), that included a comprehensive review of
available scientific evidence about the relationship. Some of the
comments that disagreed with the soy protein proposed rule provided
specific support for their positions. Some of the comments were
received after the date for submitting comments had passed. Although
the agency is not obligated to respond to late comments, in the
interest of assessing the totality of the available data, it has
considered each of these comments to the extent that it provided
complete information for review or references accessible to the agency
and addressed issues not raised in earlier comments. The agency has
summarized and addressed the relevant issues raised in the comments in
the sections of this document that follow.
In response to the soy protein reproposal, the Agency received
approximately 10 submissions, each containing one or more comments. The
agency has summarized and addressed these comments in section II.C.2 of
this document.
A. Eligibility of Soy Protein as the Subject of a Health Claim
In the soy protein proposed rule, the agency assessed whether soy
protein satisfied the preliminary requirement that a substance that is
the subject of a health claim is associated with a disease for which
the U.S. population is at risk (63 FR 62977 at 62978). Based on
analyses presented in earlier rulemakings and its review of data on the
mortality, morbidity, and costs of CHD and prevalence of ``high risk''
and ``borderline high'' total and LDL-cholesterol levels in the United
States (Refs. 4 through 8), the agency tentatively concluded that, as
required in Sec. 101.14(b)(1), CHD is a disease for which the U.S.
population is at risk. One comment reviewed additional sources of
information and reached the same conclusion.
In the soy protein proposed rule, FDA also tentatively concluded
that soy protein from Glycine max satisfied the preliminary requirement
of Sec. 101.14(b)(3)(i) that the substance be a food that contributes
taste, aroma, or nutritive value (63 FR 62977 at 62978). Sources of soy
protein identified in the soy protein proposed rule included foods
composed of or derived from whole soybeans and foods that contain
processed soy protein ingredients: Isolated soy protein (ISP), soy
protein concentrate (SPC), soy flour (SF), texturized soy protein, or
texturized vegetable protein (TVP). In addition to protein, these foods
and ingredients contain other naturally occurring soy constituents,
such as isoflavones, fiber, and saponins. The specific processing steps
employed determine the extent of retention of such naturally occurring
constituents in the final product.
In assessing whether the petitioner had demonstrated that soy
protein is safe and lawful at the level necessary to justify the claim,
FDA noted that the petitioner stated that soy protein ingredients were
in common use in food before January 1, 1958, and that they are
generally recognized as safe (GRAS) by self-determination (63 FR 62977
at 62978). Because the fractionation procedures used to convert
vegetable flours to vegetable protein isolates and concentrates were
commonplace prior to 1958, the petitioner also asserted that ISP and
SPC can be defined as soy flour ``subject only to conventional
processing as practiced prior to January 1, 1958.'' In addition, FDA
reviewed information submitted by the petitioner about potential risks
of consuming soy products: allergenicity (Refs. 9 and 10), exposure to
trypsin inhibitors (Refs. 11 through 16), reduced bioavailability of
minerals (Refs. 13, 17, 18, 19, and 20), and hormonal disturbances due
to soy isoflavones (Refs. 21 through 26). Based on the totality of the
evidence and, in particular, its common use in food, the agency did not
take issue with the petitioner's view that the use of soy protein is
safe and lawful as required in Sec. 101.14(b)(3)(ii). Thus, FDA
tentatively concluded that the petitioner provided evidence that
satisfied the requirement in Sec. 101.14(b)(3)(ii) that use of soy
protein at the levels necessary to justify a claim is safe and lawful
under the applicable food safety provisions of the act (63 FR 62977 at
62979).
Several comments agreed with the agency's conclusion and some
provided the rationale for their support. A number of comments disputed
the
[[Page 57702]]
petitioner's assertion of GRAS status for soy protein and raised
questions about the safety of soy protein-containing foods. The
specific aspects of disagreement are summarized and discussed in the
following sections of this document.
1. Concerns About the Safety of Soy Protein-Based Infant Formulas
(Comment 1). Many of the comments that raised concerns about the
safety of consuming soy protein-containing foods addressed the safety
of soy protein-based infant formulas. The observed or hypothesized
detrimental effects of such formulas discussed in these comments
included: hormonal disturbances due to estrogenic effects of soy
isoflavones; thyroid abnormalities; altered mineral balance, especially
for zinc; and diabetogenic effects in infants.
FDA is aware of concerns raised about the safety of soy infant
formulas, but notes that these are speculative at this time, pending
the results of definitive research. FDA also notes that the American
Academy of Pediatrics (Ref. 73) and the New Zealand Ministry of Health
(Ref. 74) have recently issued guidelines for the safe and suitable use
of soy-based infant formulas. Some issues regarding effects of infant
formula are unique because infants may be entirely dependent on formula
as a sole source of nutrition and the relevance of such issues for soy
protein consumed as part of a mixed diet by the general U.S. population
is not clear.
In any case, concerns about effects of soy protein specific to
infant formulas are beyond the scope of the current rule, which
authorizes a health claim about the relationship of soy protein and CHD
for foods intended for use by the general population. Health claims are
not permitted on foods represented or purported for use by infants and
toddlers less than 2 years of age unless specifically provided for in
the authorizing regulation (21 CFR 101.14(e)(5)). Diets restricted in
fat, saturated fat, and cholesterol are not recommended for infants and
young children, and the current rule (Sec. 101.82) contains no
provisions for use of the health claim about the relationship between
soy protein and CHD on foods for infants and toddlers.
2. Comments on Petitioner's Self-Determination of GRAS Status for Soy
Protein
(Comment 2). One comment specifically agreed with the petitioner's
assertion that soy protein-containing food ingredients are generally
recognized as safe (GRAS) by self-determination and based on common use
in food before January 1, 1958, in conformance with Sec. 201(s) of the
act. The comment also noted that, although soy protein is not listed as
GRAS or prior sanctioned in Title 21 of the CFR, FDA has noted that
these lists ``do not include all substances generally recognized as
safe for their intended use'' and, as stated at 21 CFR 182.1, ``[i]t is
impracticable [for FDA] to list all substances that are GRAS for their
intended use.'' This comment also agreed with the petitioner's
conclusion that fractionation procedures used to convert vegetable
flours to vegetable protein concentrates and isolates were commonplace
in various sectors of the grain industry, such as corn processing, well
before 1958. Therefore, SPC and ISP can be defined as soy flour
``subject only to conventional processing as practiced prior to January
1, 1958.'' The comment concluded that SF (including steam-treated SF),
SPC, and ISP all fall within the category of ingredients that are GRAS
through experience based on their common use. Several comments objected
to the petitioner's self-determination of GRAS status, citing a variety
of reasons. As stated previously, FDA does not take issue with the
petitioner's self-determination of GRAS status, and the comments,
discussed below, have not convinced the agency to change that
conclusion.
(Comment 3). Some comments raised objections on the basis that FDA
has not approved the GRAS status of soy protein.
Although FDA has not ruled formally on the GRAS status of soy
protein ingredients, it has not challenged determinations that soy's
use as dietary protein is GRAS. Food ingredients whose use is generally
recognized as safe by qualified experts are not required by law to
receive FDA approval. Under the health claim petition process, FDA
evaluates whether the substance is ``safe and lawful'' under the
applicable food safety provisions of the act (Sec. 101.14(b)(3)(ii)).
As discussed in greater detail below, FDA did not receive sufficient
evidence from comments to challenge the petitioner's assertion that soy
protein ingredients are GRAS by self-determination. The petitioner met
the showing required by Sec. 101.14(b)(3)(ii) that the substance be
``safe and lawful.''
(Comment 4). One comment claimed that the Center for Food Safety
and Applied Nutrition recently returned a petition requesting GRAS
recognition for soy protein.
The document referred to by the comment was a notification by
Archer Daniels Midland Company (GRN 000001), rather than a petition for
FDA action, and the subject of the notification was soy isoflavone
extract, rather than soy protein. At the company's request, FDA ceased
evaluation of the GRAS Notification pending the company's updating of
the file (Ref. 75). Thus, this comment was incorrect.
(Comment 5). A comment asserted that petitioner's basis for GRAS
self-determination of the use of soy protein as a dietary protein
ingredient (i.e., common use in food before January 1, 1958) was
incorrect. Because the 1979 Select Committee on GRAS Substances (SCOGS)
report (Ref. 76) determined that, at the time of the report, likely
average dietary exposure to soy protein isolate was only about 150
milligrams (mg) from food items, the comment asserted that soy protein
isolates could not have been in common use before 1958.
FDA finds that this comment is groundless and inaccurately
characterizes the findings of the SCOGS. The 1979 SCOGS report includes
the background statement ``Edible soy protein isolates for food uses
appeared about 1957 as a major article of commerce.'' The 1979 SCOGS
Report also cited a 1972 National Research Council survey of GRAS
ingredients that listed 14 food categories in which soy protein
isolates were used and calculated an average daily intake of several
grams. Soy protein isolates represent only one of several possible
sources of soy protein in foods. In addition, for purposes of
determining if a substance is GRAS, common use is not restricted to
common use in the United States.
(Comment 6). A comment supporting the petitioner's self-
determination of GRAS status noted that use of soy as a food dates to
about the 11th century BC in the eastern half of north China. From
about the first century AD to the 15th-16th century, soybeans were
introduced in Korea, Japan, Indonesia, the Philippines, Vietnam,
Thailand, Malaysia, Burma, Nepal, and northern India. Soybeans first
grew in the United States in 1765 and were used then to manufacture soy
sauce and vermicelli (soybean paste) (Ref. 77). A comment that disputed
the petitioner's self-determination of GRAS status speculated that the
species of soybean grown early in its history in Asia may have differed
significantly in its content of nutrients and other active components
from the modern species that is cultivated in this country.
FDA does not find this comment compelling. Although the composition
of soybeans has likely changed over time, modern soybean species and
[[Page 57703]]
cultivars are, in any case, encompassed within the period of common use
of soy and soy protein in food.
(Comment 7). One comment questioned whether the Asian experience
could provide assurance that soy is safe. Drawing parallels with herbal
medicine in terms of attitudes, monitoring deficiencies, and the
general difficulty in detecting toxicities with long latency, this
comment concluded that the long history of apparent safe use of soy
products cannot assure they are without risk (Ref. 78).
The comment did not provide evidence to document that soy products,
consumed at levels necessary to justify the claim, are not generally
recognized as safe. Moreover, considerable research is underway at this
time because of the hypothesized benefits of the historical use of soy
products by certain population groups. FDA supports the ongoing
research to clarify the effects, both potentially beneficial and
potentially adverse, of soy and agrees that any effects due to changes
in the conditions of use should be monitored. However, the information
currently available does not lead FDA to object to the petitioner's
self-determination of GRAS status of soy protein.
(Comment 8). Several other comments asserted that the proposal did
not adequately establish the GRAS status of soy protein food
ingredients in that the proposal did not include a thorough evaluation
of the safety of potentially harmful components, e.g., lysinoalanine,
nitrites and nitrosamines, trypsin inhibitors, phytate, and
isoflavones.
FDA notes that the 1979 SCOGS report (Ref. 76) discussed several of
these components extensively and recommended that it would be prudent
to develop food grade specifications for soy protein isolates that
would set acceptable limits on the levels of lysinoalanine, nitrites,
and nitrosamines. But, the possible presence of these components in soy
protein isolates did not lead the SCOGS panel to recommend against GRAS
status of soy protein isolates.
As noted above, the agency finds the petitioner met the showing
required by Sec. 101.14(b)(3)(ii) that soy protein is ``safe and
lawful.'' The agency lacks documented evidence of adverse effects in
humans and has received no information about actual levels of
potentially harmful components or about threshold levels for adverse
effects in humans. Accordingly, the agency has no basis to conclude
that soy protein is not safe and lawful. The specific comments about
potentially harmful components of soy are discussed below.
3. Lysinoalanine: Potential Toxic Effects
(Comment 9). A few comments noted concerns about the presence of
lysinoalanine in soy protein isolates and cited the SCOGS report (Ref.
76), which indicated that lysinoalanine was implicated as a renal toxic
factor in rats.
FDA finds that the comments inaccurately reflected the findings of
the SCOGS report. The SCOGS report noted that the relatively severe
alkali treatment used to modify viscosity and adhesive properties of
soy protein isolates used as sizing and coating adhesives in the
production of paper and paperboard products can cause formation of
lysinoalanine. The report evaluated the risk of lysinoalanine exposure
from soy protein adhesives and binders used in paper and paperboard
food packaging. The 1979 SCOGS report noted that, ``For edible isolated
protein production, extraction is usually carried out at a pH below 9
to avoid hydrolytic or rheological changes'' and concluded that, while
relatively low levels of lysinoalanine had been reported in some
samples of food grade soy protein isolate, available information
indicated that the levels of lysinoalanine in food grade soy protein
isolates pose no hazard to the consumer (Ref. 76).
FDA notes that the comments that expressed concern about
lysinoalanine in soy protein ingredients did not provide any
information about lysinoalanine levels in food grade soy protein
ingredients nor about use of alkali-processed soy protein as a food
ingredient. FDA finds that the potential presence of lysinoalanine in
soy protein isolates used for sizing and coating adhesives in paper and
paperboard products is not relevant to the safe and lawful use of soy
protein in food. FDA also notes that the production of small amounts of
lysinoalanine during alkali processing has also been documented with
casein and lactalbumin, so it is not unique to soy. Good manufacturing
practices are and should be employed to minimize the production of
lysinoalanine because of its deleterious effects on protein quality.
4. Nitrites and Nitrosamines: Potential Carcinogenic Effects
(Comment 10). Some comments expressed concerns about the potential
presence of nitrites in soy protein and the potential their presence
poses for the in vivo formation of nitrosamines, which have been shown
to be carcinogenic in experimental animals.
FDA notes that many natural and processed foods contribute to the
total human intake of nitrite. In an appendix titled ``Health Aspects
of Nitrites in Soy Protein Isolates,'' the SCOGS report (Ref. 76)
presented an estimate of the consumer exposure to nitrite contributed
by soy protein in perspective to nitrite from other dietary sources and
that formed in the gastrointestinal tract by reduction of salivary and
dietary nitrate. The SCOGS report estimated the maximum daily nitrite
consumption for a vegetarian eating meat alternatives prepared from soy
protein to be 0.04 mg/kilogram (kg) body weight (or 2.8 mg for a 70-kg
person). The report estimated daily per capita intake of nitrite from
other foods of plant origin and cured meats to be about 2.4 mg and
daily exposure to nitrite from saliva to be 15 mg. The report estimated
that nitrite formed in the intestine from reduction of ammonia or
organic nitrogen compounds contributed about 90 mg/day. Given the
relatively minor potential contribution of soy protein to total nitrite
exposure, and the fact that no data were submitted to document the
current levels of nitrites or nitrosamines in soy protein isolates, FDA
is not persuaded of the necessity for establishing specifications for
acceptable levels of these compounds.
5. Trypsin Inhibitors: Potential Effects on Pancreatic Function
(Comment 11). A number of comments presented evidence that modern
heat treatment and other processing do not entirely eliminate the
activity of trypsin inhibitors in soy protein-containing products.
Additional references provided in comments (Refs. 79, 80, 81, and 82)
suggested that the mechanism of feedback regulation of pancreatic
enzyme secretion may be responsible for deleterious effects on the
pancreas--hyperplasia and formation of nodules--seen in animal studies.
Further, Leiner (Ref. 80) demonstrated that infusion of high levels of
isolated trypsin inhibitor in humans can evoke this mechanism but noted
that further research was needed to assess whether frequent exposures
to low levels of trypsin inhibitors consumed in the diet could have the
same effect. Other comments cited evidence for potential
anticarcinogenic effects of these and other protease inhibitors (Ref.
83). Leiner (Ref. 82) hypothesized that any anticarcinogenic effect of
protease inhibitors would likely be manifested at levels too low to
evoke their adverse effects on the pancreas.
FDA notes that the observed adverse effects have been limited to
animal
[[Page 57704]]
studies. To date, deleterious effects of consumption of low levels of
soybean trypsin inhibitors have not been documented in humans. For
example, Mills et al. (Ref. 84) conducted a prospective study of fatal
pancreas cancer among 34,000 California Seventh-day Adventists, a group
with high soy consumption. Compared to all U.S. whites, Adventists
experienced decreased risk from pancreas cancer death, which was not
statistically significant. Although there was a suggestive relationship
between increasing meat, egg, and coffee consumption and increased
pancreatic cancer risk, these variables were not significantly related
to risk after controlling for cigarette smoking. However, increasing
consumption of vegetarian protein products, beans, lentils, and peas as
well as dried fruit was associated with highly significant protective
relationships to pancreas cancer risk.
Therefore, FDA finds that the information presented in these
comments has not documented deleterious effects of dietary intake of
trypsin inhibitors from soy in humans and, thus, does not lead the
agency to take issue with the petitioner's conclusion that the use of
soy protein is safe and lawful as required by Sec. 101.14(b)(3)(iii).
6. Phytate: Effects on Mineral Balance
Comments raised concerns about the potential deleterious effect of
soy protein and its phytate content on mineral status. Phytate, the
salt of phytic acid or inositol hexaphosphate, is a natural plant
constituent containing six negatively charged phosphate groups that can
form strong complexes with divalent cations such as calcium, magnesium,
iron, zinc, and copper. Concerns relative to soy have concentrated
mainly on iron and zinc, based primarily on studies of the absorption
and bioavailability of these minerals.
(Comment 12). One comment cited a study in which a soy protein-
based purified diet induced iron deficiency in monkeys (Ref. 85). The
same comment also noted two studies in humans--one that found
inhibition of the absorption of nonheme iron from both semisynthetic
meals and meals comprising conventional foods by various soy protein-
containing ingredients (Ref. 86), and one that found increasing
inhibition of nonheme iron absorption with increasing amounts of
phytate in liquid formula meals that contained soy protein isolates
(Ref. 87). In a study cited in another comment, the substitution of
some meat in a mixed meal by soy protein caused a decrease in the
absorption of nonheme iron and an increase in the absorption of heme
iron (Ref. 88), so that overall iron absorption was not compromised.
Another comment reported that human feeding studies with soy protein
that have examined measures of iron status have not shown detrimental
effects (Ref. 89).
A comment raised concerns about the effect of soy protein on zinc
status based on studies of absorption of zinc from soy infant formula
(Ref. 90) and a study that showed decreased serum thymulin in subjects
fed a low-zinc, soy protein-based experimental diet designed to produce
mild zinc deficiency (Ref. 91). As noted earlier, issues specific to
infant formula are outside the scope of this rulemaking and the
experimental diet in the latter study (Ref. 91) is of limited relevance
to the likely conditions of consumption of soy protein in the
population that is the target of the health claim. Another comment
cited two studies (Refs. 92 and 93) showing no adverse effects of soy
protein on absorption of zinc from meals in subjects with adequate zinc
status.
One comment provided additional information on the mechanism of
phytate interference with zinc homeostasis (Ref. 94) and characterized
the problem as more than a matter of decreased bioavailability of the
zinc consumed in a meal. The comment noted that phytate can remove from
the duodenum zinc that is mainly derived from pancreatic secretions,
that is, zinc that may have been consumed 1-2 weeks earlier. Although
these data are derived from animal studies, the comment indicated that
the physiology of zinc homeostasis is not qualitatively different
across species.
This comment expressed concern that high consumption of soy protein
might exacerbate marginal zinc deficiency, which is difficult to
diagnose, and suggested that labeling should include the content of
both zinc and phytate so consumers can be educated that a molar ratio
of phytate:zinc of less than 10 is needed to avoid detrimental effects
on zinc status, as suggested by research in animals (including Ref.
95). The comment acknowledged that education would be needed for the
public to utilize such labeling. The agency recognizes that adequacy of
iron and zinc status in largely plant-based diets is a legitimate
concern.
FDA finds that the evidence of potential adverse effects of soy
protein on iron and zinc status is equivocal. Interpretation of the
evidence is difficult because findings in human studies are often
inconsistent with results of animal studies. Moreover, many factors
affect the absorption of these minerals, including the amount consumed
in a meal, the enhancing and inhibiting effects of other components of
the meal, and the nutritional status of the subject. Animal studies
suggest that zinc status is a strong determinant of effects of phytate/
soy on zinc absorption: zinc absorption is more impaired with zinc
deficiency, in contrast to the effect of low iron status, which
enhances iron absorption. However, given the lack of documented
evidence for impaired iron and zinc status in humans consuming soy
protein as part of a mixed diet, FDA is not persuaded of the necessity
for the suggested labeling with respect to the phytate: zinc molar
ratio. Nor is it persuaded that many consumers would find the suggested
information, which is highly technical, useful at this time.
7. Soy Isoflavones: Estrogenic Effects
Many comments addressed concerns about the possible deleterious
consequences of phytoestrogen effects of the soy isoflavones, genistein
and daidzein. Most of these addressed proliferative (and potentially
carcinogenic) effects on estrogen-sensitive tissues, effects on
circulating hormone levels and potential deleterious effects on
fertility, and potentially adverse effects on sexual development.
a. Proliferative effects. (Comment 13). Several comments cited a
number of studies of in vitro effects of individual isoflavones on
proliferation of estrogen-sensitive cells. For example, Dees et al.
(Ref. 96) found that genistein increased a number of indices for
proliferative activity in MCF-7 human breast cancer cells. As the
authors noted, these findings are consistent with the conclusion that
dietary estrogens at low concentrations do not act as antiestrogens,
but act like estradiol to stimulate human breast cancer cells to enter
the cell cycle. However, many other studies (reviewed in Refs. 97 and
98) have found that the phytoestrogens present in soybeans inhibit
breast cancer cell proliferation in vitro (at lower concentrations,
closer to physiological levels) and inhibit mammary cancer development
in various animal models. FDA concludes that studies in transformed
cells cannot predict with certainty whether effects will be beneficial
or detrimental in humans consuming soy protein.
(Comment 14). Comments argued that two reports showed effects of
dietary intake of soy isoflavones on breast tissue in women. Petrakis
et al. (Ref. 99) studied 24 normal pre- and postmenopausal white women,
ages 30
[[Page 57705]]
to 58 years, who underwent monthly nipple aspiration of breast fluid
and gave blood and 24-hour urine samples for biochemical studies. The
women consumed no soy in months 1-3 and 10-12. During months 4-9 the
women ingested daily 38 grams (g) of soy protein isolate containing 38
mg of genistein (daidzein content was not reported). This study's
findings indicated that prolonged consumption of soy protein isolate
had a stimulatory effect on the breast of premenopausal women,
characterized by increased secretion of breast fluid and elevated
levels of plasma estradiol. The study also detected evidence of
epithelial proliferation (hyperplasia) in 7 of the 24 subjects during
consumption of soy. McMichael-Phillips et al. (Ref. 100) examined the
effects of dietary soy supplementation on the proliferation rate of
premenopausal, histologically normal breast epithelium and the
expression of progesterone receptor. Women (n = 48) with benign or
malignant breast disease were randomly assigned to receive their normal
diet either alone or with a 60-g soy supplement (containing 45 mg
isoflavones) taken daily for 14 days. Serum concentrations of the
isoflavones genistein and daidzein increased in the soy group at 14
days. The proliferation rate of breast lobular epithelium significantly
increased after soy supplementation when both the day of menstrual
cycle and the age of patient were accounted for. Progesterone receptor
expression increased significantly in the soy group. The authors
concluded that further studies are required to determine whether the
short-term stimulation of breast proliferation is due to estrogen
agonist activity and to examine the long-term effects of soy on both
the pituitary gland and breast.
FDA finds that the detection of proliferative effects in these two
studies suggests the need for additional research. The findings do not,
however, establish that the observed effects are detrimental and are
not supported by the findings of epidemiologic studies of soy intake
and risk of premenopausal breast cancer (Ref. 101).
b. Fertility and Hormone Levels. (Comment 15). Some comments
referenced a number of studies that reported reduced fertility in
animals exposed to phytoestrogens (including Refs. 102, 103, and 104).
Some of these studies involved phytoestrogens other than those found in
soy or consumption of soy under extreme or unusual conditions. FDA is
not convinced of the relevance of these studies to human consumption of
soy protein.
(Comment 16). Comments cited the study of Cassidy et al. 1994 (Ref.
105) as suggesting the potential for deleterious effects on human
fertility. These investigators examined the influence of a diet
containing soy protein on the hormonal status and regulation of the
menstrual cycle in six premenopausal women. Soy protein (60 g
containing 45 mg isoflavones) given daily for 1 month significantly
(p<0.01) increased follicular phase length and/or delayed menstruation.
Midcycle surges of luteinizing hormone (LH) and follicle-stimulating
hormone (FSH) were significantly suppressed during dietary intervention
with soy protein. Plasma estradiol concentrations increased in the
follicular phase and cholesterol concentrations decreased 9.6 percent.
The authors concluded that responses to soy protein are potentially
beneficial with respect to risk factors for breast cancer and may in
part explain the low incidence of breast cancer and its correlation
with a high soy intake in Japanese and Chinese women. One of the
comments that cited this study acknowledged that it is unclear whether
these soy effects are beneficial or adverse. FDA notes that the study
found that soy did not interfere with ovulation and the study did not
assess effects on fertility.
In a similar study with a longer duration, Duncan et al. (Ref. 106)
studied effects of isoflavone consumption in 14 premenopausal women.
The women consumed isoflavones in soy protein powders (control diet,
10; low isoflavone diet, 64; high isoflavone diet, 128 mg/day) for
three menstrual cycles plus 9 days in a randomized cross-over design.
The low isoflavone diet decreased LH and FSH levels during the
periovulatory phase. The high isoflavone diet decreased free T3 and
dehydroepiandrosterone sulfate levels during the early follicular phase
and estrone levels during the midfollicular phase. No other significant
changes were observed in hormone concentrations or in the length of the
menstrual cycle, follicular phase, or luteal phase. Endometrial
biopsies performed in the luteal phase of cycle 3 of each diet period
revealed no effect of isoflavone consumption on histological dating.
FDA notes that although this study's findings varied somewhat from
those of Cassidy et al. (Ref. 105), it also did not directly address
the effect of soy on human fertility. FDA finds that these two studies
do not provide sufficient evidence to address the effect of soy protein
on human fertility.
c. Developmental Effects. (Comment 17). One comment cited the study
of Faber and Hughes, 1993 (Ref. 107) as showing alterations in LH
regulation following developmental treatment with genistein, suggesting
that during pregnancy in humans, isoflavones could be a risk factor for
abnormal brain and reproductive tract development. This study involved
injection of 0, 1, 10, 100, 200, 400, 500, or 1,000 micrograms of
genistein into neonatal rats on days 1-10. Because of the differences
in developmental stages between rodents and humans, this type of
experiment is used as a model for prenatal (third trimester) effects of
diethylstilbestrol (DES). Increased exposure to genistein led to
decreased LH secretion; the volume of the sexually dimorphic nucleus of
the preoptic area increased compared to controls only in animals that
received the two highest doses of genistein. An earlier paper by Faber
and Hughes 1991 (Ref. 108) showed that effects elicited by neonatal
injections of 1000 micrograms of genistein were similar to those of 0.1
micrograms of DES. The comment also cited studies using a similar
experimental model by Medlock et al. (Refs. 109 and 110) as
demonstrating that equol (a metabolite of daidzein in some individuals)
acts as an endocrine disruptor during development. FDA finds that the
relevance of these studies to an assessment of potential prenatal
effects of dietary soy protein during pregnancy is uncertain.
(Comment 18). One comment cited the study of Harrison et al. (Ref.
111) that showed pregnant Rhesus monkeys fed genistein had serum
estradiol levels 50 to 100 percent higher than the controls in three
different areas of the maternal circulation. The comment also noted the
finding that the fetuses of genistein fed monkeys had a 70 percent
higher serum estradiol level than did the controls. In this study, five
monkeys were fed genistein (amount not specified) during pregnancy and
compared to five controls. No differences were reported in maternal
weight gain, fetal weights at delivery, or placental weights.
Significant differences in estradiol levels (but not progesterone) were
noted at delivery in maternal peripheral blood, uterine veins, ovarian
veins, and the fetus, and in maternal blood during pregnancy, but
values were not reported. FDA received only an abstract describing this
study. Without more complete documentation, the merits or weaknesses of
this study cannot be evaluated. Therefore, FDA has not used this study
to evaluate the concerns raised in this comment.
FDA notes that, in another study that examined dietary effects,
Fritz et al. (Ref. 112) fed female rats genistein from
[[Page 57706]]
conception to day 21 postpartum in the diet at concentrations of 0, 25
and 250 mg genistein/kg diet. They found that genistein in the diet at
``physiological levels'' (equivalent to those in Asians consuming a
traditional high soy diet) enhances cell differentiation, resulting in
programming of mammary gland cells for reduced susceptibility to
chemically induced mammary cancer, with no observed toxicity to the
fertility of dams or the reproductive tract of female offspring. FDA
finds that these dietary studies in animals do not provide evidence for
detrimental developmental effects in humans.
(Comment 19). Another comment raised the possibility that soy
phytoestrogens could be responsible for inducing premature puberty and
cited the case-control study of estrogenic exposures by Freni-Titulaer
et al. (Ref. 113) of patients with premature thelarche seen in Puerto
Rico between 1978 and 1981. In subjects 2 years of age or older at the
onset of thelarche, the study found no statistically significant
associations. In subjects with onset before 2 years of age,
statistically significant positive associations were found with a
maternal history of ovarian cysts, consumption of soy-based formula,
and consumption of various meat products. A statistically significant
negative association was found with consumption of corn products. The
authors concluded that these statistical associations were not
sufficient to explain the reported increase in premature thelarche
because in over 50 percent of the case subjects there was no exposure
to any of the risk factors for which statistical associations were
found.
Thus, FDA concludes that this study provides no convincing evidence
that soy was responsible for premature thelarche. Moreover, FDA notes
that the study documents no deleterious effects of consuming soy
protein at the levels necessary to justify the health claim in
population groups that are the target of the claim.
d. Other. (Comment 20). One comment cited a study associating
intake of tofu in mid-life by Japanese-American men in Hawaii with
vascular dementia and brain atrophy in old age (Ref. 114). This comment
hypothesized that isoflavone inhibition of aromatase, which catalyzes
the conversion of testosterone to estradiol, may provide a mechanistic
explanation for this finding. The report cited (Ref. 116) is an
abstract that indicates the researchers found an association of high
tofu intake with low cognitive test scores and with Alzheimer's
disease, rather than vascular dementia.
FDA finds that this abstract does not provide a sufficient basis to
evaluate the merits and weaknesses of this study. As such, it is not
useful in evaluating the safety concerns at issue. Moreover, the report
does not provide information on total soy intake or what variables were
controlled in the analysis. If tofu or soy were implicated in
Alzheimer's disease, its prevalence would be expected to be higher in
Japan than in Hawaii, but White et al. (Ref. 115) found the prevalence
of Alzheimer's disease was higher in Hawaii than in Japan. Therefore,
FDA is not persuaded by the comment raising concerns about potential
adverse effects of soy protein in dementia and brain atrophy in older
persons.
(Comment 21). One comment addressed the general issue of threshold
effects for estrogenic compounds, citing a study (Ref. 116) that showed
no threshold dose for estradiol-induced sex reversal of turtle embryos.
It also cited a study (Ref. 117), available in abstract form, that
reviewed 31 dose-response curves for hormone-mimicking chemicals that
also failed to show a threshold. The report of this study did not
include mention of soy isoflavones and did not specify the estrogenic
effects examined. FDA does not find this evidence particularly useful.
The relevance of the turtle model to humans is uncertain and the other
cited evidence was available only in abstract form.
e. Conclusion. Soy isoflavones and other dietary phytoestrogens are
known to exert hormonal effects--both estrogenic and antiestrogenic--
depending on the amount and type consumed and endogenous hormonal
status of the organism studied; they are much less potent than
endogenous estrogen or synthetic estrogens such as DES. There is
considerable variability from person to person in the absorption,
metabolism, and disposition of the soy isoflavones, genistein and
daidzein (Ref. 118), and researchers have found that their metabolism
and excretion depend on the duration of ingestion and the subject's sex
(Ref. 119).
Overall, the evidence for proliferative effects, effects on
fertility and hormone levels, and developmental and other effects in
humans due to the estrogenic effects of soy isoflavones is very
limited. Both possible beneficial effects and possible detrimental
effects are still hypothetical. FDA finds that the information
presented in the comments has not adequately documented deleterious
effects of dietary intake of soy isoflavones in humans.
8. Soy Isoflavones: Goitrogenic Effects
(Comment 22). Comments noted that isoflavones are inhibitors of the
enzyme thyroid peroxidase (TPO), which produces the thyroid hormones T3
and T4, and indicated that its inhibition can be expected to generate
thyroid abnormalities. Other comments, however, noted the lack of
evidence for consequential effects of TPO inhibition (i.e., high
prevalence of goiter) in populations with high soy consumption.
One comment noted that there exists a body of animal data that
demonstrates goitrogenic and even carcinogenic effects of soy products
and cited the study by Kimura et al. (Ref. 120). These researchers
developed malignant goiter in rats by feeding diets containing 40
percent defatted soybean and no iodine. No deleterious effects were
seen in controls fed the same diet with iodine added.
Comments noted the existence of a number of case reports in the
older literature of soy inducing goiter in infants (Refs. 121 through
125). Van Wyk et al. (Ref. 121) studied one infant who developed goiter
on a soybean formula and tested the same product in 12 adults. In
adults, the product did not interfere with iodine absorption, impair
iodine uptake, interfere with oxidation of iodine in the thyroid, or
(in most subjects) interfere with the release of protein-bound iodine
into the blood. Hydovitz (Ref. 12) provided a single case report;
Shepard et al. (Ref. 123) described three cases and presented evidence
that soybean goiter was caused by iodine deficiency. Pinchera et al.
(Ref. 124) reported on a case of a congenitally hypothyroid infant and
found high fecal losses of thyroxine. Addition of adequate iodine to
soy-based infant formulas in the 1960's generally resolved or prevented
goiter. However, Chorazy et al. (Ref. 125) more recently reported on a
hypothyroid infant who was semi-refractory to thyroid hormone therapy
while consuming soy formula.
Several comments cited the study of Ishizuki et al. (Ref. 126) as
evidence for goitrogenic effects of soy in adults. This study is
published in Japanese and the available English abstract is poorly
translated. As described in that abstract, the design and findings are
unclear: goiters were said to occur in half the subjects eating 30 g
soybeans daily for 3 months, though ``various parameters of serum
thyroid hormones remained unchanged by taking soybeans.'' The soybean
preparation used (reported in some comments to be roasted, pickled
soybeans), iodine intake, and other dietary changes were not reported.
[[Page 57707]]
In one comment, researchers indicated that they had identified
genistein and daidzein as the goitrogenic isoflavonoid components of
soy and defined the mechanisms for inhibition of TPO-catalyzed thyroid
hormone synthesis using in vitro studies of the pure isoflavones (Refs.
127 and 128). The comment noted that the observed irreversible
inactivation of TPO by isoflavones, through covalent binding to TPO,
raises the possibility of neoantigen formation. The comment also noted
that anti-TPO is the principal autoantibody present in autoimmune
thyroid disease and proposed that this hypothetical mechanism is
consistent with the reports of Fort et al. (Refs. 129 and 130) of a
doubling of risk for autoimmune thyroiditis in children who had
received soy formulas as infants compared to infants receiving other
forms of milk. However, the studies of Fort et al. were retrospective
case-control analyses of early feeding practices in children with
diabetes (Ref. 129) or autoimmune thyroid disease (Ref. 130). The
studies did not establish a cause-and-effect relationship or assess
medical indications for use of soy formula in these children.
FDA notes that no data or other information presented in the
comments documents deleterious effects on thyroid function of consuming
soy protein at the levels necessary to justify the health claim in
population groups that are the target of the claim.
9. Allergenicity of Soy Protein
(Comment 23). One comment disputed the statement in the soy protein
proposed rule that soy allergies are often outgrown. FDA finds that the
comment cited data that did not directly address this issue but
documented the following with respect to soy: a case report of an
anaphylactic reaction to soy in an adult (131); severe reactions to soy
in several Swedish children and adolescents, who had known severe
reactions to peanuts and asthma but had not reacted previously to soy
(Refs. 132 and 133); cross reactivity of some soy and peanut allergens
(Ref. 134); and an outbreak of gastrointestinal illness associated with
consumption of an improperly processed soy protein tuna salad extender
in which only a few individuals exhibited signs of true
hypersensitivity reactions (Ref. 135).
(Comment 24). One comment noted that use of soy protein health
claims will highlight the presence of soy protein in foods. Another
comment noted that any food protein can stimulate a food allergy and
that such allergies are commonly due to milk, egg, and nut proteins.
This comment noted that infants who develop cow's milk allergies or
intolerance are frequently prescribed soy substitutes and a small
subset of these high-risk children also develop soy protein allergy.
FDA finds that the comments that noted concerns about the
allergenicity of soy protein cited these concerns as evidence that
consumption of soy is unsafe, but did not propose that any particular
action be taken by the agency as a consequence to protect consumers
with soy allergies. FDA does not believe that, because some persons may
have allergic reactions to a food, it is unsafe. FDA has previously
stated that the declaration of an allergenic substance in the
ingredient statement on the food label provides adequate information
for consumers regarding the presence of the allergenic ingredient in
the product (63 FR 8103 at 8113), and sees no reason to change this
view with respect to soy. FDA notes, in agreement with one of the
comments received, that authorization of a health claim for soy protein
and CHD will highlight the presence of soy protein in those food
products that bear the claim. The agency, therefore, anticipates that
persons with known soy allergies will be able more easily to avoid soy
protein based products.
B. Updated Review of Scientific Evidence and Issues Related to the
Evidence
In the soy protein proposed rule, FDA conducted a comprehensive
review of the human studies submitted in the petition (Refs. 27 through
66) (63 FR 62977 at 62980). Of these, the agency gave particular weight
to 14 clinical trials (Refs. 27, 28, 30 (1 trial), 31, 36, 37 (1
trial), 40 (2 trials), 44, 49, 51, 54, 58, and 59). These 14 trials met
the criteria for selection set out by the agency (63 FR 62977 at
62980): they included subjects representative of the general U.S.
population; were well controlled; reported information on intakes of
saturated fat and cholesterol; and avoided problems associated with
small sample size, lack of a placebo, and other design problems. The
agency summarized these studies in Table 1 of the soy protein proposed
rule (63 FR 62977 at 62998). The agency also summarized seven clinical
trials in adults (Refs. 33, 35, 46, 55, 56, 60, and 64) and three
trials in children (Refs. 34, 42/45, and 63) with type II or familial
hypercholesterolemia in Table 2 of the soy protein proposed rule (63 FR
62977 at 63011). In addition, FDA reviewed the results of one
epidemiological study (Ref. 65 and 63 FR 62977 at 62986) and a meta-
analysis (Ref. 66 and 63 FR 62977 at 62987) that included a number of
the soy protein studies submitted in the petition.
Based on these studies, FDA concluded there was scientific evidence
for a consistent, clinically significant effect of soy protein on blood
total and LDL-cholesterol levels (63 FR 62977 at 62989). The
hypocholesterolemic effect of soy protein was seen in addition to the
effects of a low saturated fat and low cholesterol diet. The degree of
lowering of blood total and LDL-cholesterol was consistently and highly
dependent on initial levels, within and across studies of subjects with
normal, moderately elevated, and severely elevated blood lipid levels,
with persons having higher blood lipid levels showing greater effects.
Soy protein consistently caused only statistically nonsignificant
effects or slight elevations in high density lipoprotein (HDL)-
cholesterol levels. The intervention studies indicated that a minimum
level of approximately 25 g of soy protein was needed to have a
clinically significant effect on total and LDL-cholesterol levels.
1. Additional Data Submitted With Comments and New Studies
(Comment 25). Several comments included submissions of additional
studies of the effects of soy protein on total and LDL-cholesterol or
directed FDA to studies published since it issued the soy protein
proposed rule. FDA reviewed these studies and found that two (Refs. 136
and 137) meet its criteria for consideration.
One comment included an unpublished paper by Teixeira et al., 1999
(Ref. 136) that examined the effects of feeding four graded levels of
soy protein in moderately hypercholesterolemic men. After a three-week
lead-in on a National Cholesterol Education Program (NCEP) Step 1 diet,
subjects were randomly assigned to one of five experimental groups.
Each group received 50 g protein daily, provided in a variety of baked
goods and ready-to-mix beverages, from ISP or casein in different
proportions for 6 weeks. The proportions of protein were 50, 40, 30,
20, and 0 g (for control) as ISP and 0, 10, 20, 30, and 50 g as casein,
respectively. At 3 weeks, statistically significant (p<0.05) reductions
in total and non-HDL-cholesterol were seen only in the groups consuming
40 and 50 g of soy protein. At 6 weeks, statistically significant
reductions (p<0.05) from baseline were found for non-HDL cholesterol
levels in all soy protein-consuming groups and, in all except the 40 g
soy protein group, for total cholesterol level. Although a reduction in
total cholesterol was noted in this
[[Page 57708]]
latter group, it was non-significant (p=0.07). The authors noted that
neither non-compliance with the diet nor alterations in blood
isoflavone content could account for this result. The study also showed
that levels of HDL-cholesterol were not affected by dietary treatment
at any soy consumption level investigated.
FDA also noted the recently published study by Wong et al., 1998
(Ref. 137), who conducted a well designed and controlled trial using
NCEP Step 1 diets with most protein provided by soy (50 g/day of soy
protein) or animal protein. Subjects were 13 normocholesterolemic and
13 hypercholesterolemic men aged 20-50 years and the trial was a
randomized, 2-part, crossover study. Subjects were fed either an NCEP
Step I soy protein-containing diet or an NCEP Step I animal protein
diet for 5 weeks. After a washout period of 10-15 weeks, the subjects
were fed the alternate diet for 5 weeks. The study found the
hypocholesterolemic effect of soy protein to be independent of age,
body weight, pretreatment plasma lipid concentrations, and sequence of
dietary treatment. Regardless of plasma lipid status, the soy protein
diet was associated with a statistically significant decrease in the
plasma concentrations of LDL cholesterol (p=0.029). FDA finds these two
studies supportive of the relationship of soy protein to reduced risk
of CHD.
(Comment 26). One comment cited two metabolic ward studies by
Fumagalli et al. 1982 (Ref. 138), designed to examine fecal steroid
excretion in adults with familial type II hypercholesterolemia, that
had not been reviewed by FDA in the soy protein proposed rule, as
supportive of the ability of soy protein to lower total cholesterol
levels. However, FDA finds these studies had a very small number of
subjects, short duration of treatment, and reported insufficient
information to determine the amounts of soy protein in the diets
consumed. These studies failed to meet FDA's selection criteria for
review and, so, FDA has not considered them further.
(Comment 27). Comments included information on two studies by
Jenkins et al. 1999 (Refs. 139 and 140) that assessed the effects of
inclusion of soy protein and soluble dietary fiber in an NCEP Step II
diet in hypercholesterolemic subjects in a randomized crossover design.
Dietary saturated fat (less than 7 percent of energy) and cholesterol
(less that 80 mg/day) did not differ in the test and control metabolic
diets (Ref. 139). Compared with the control diet, the test diet (which
provided 33 grams of soy protein from a variety of commercially
available foods) resulted in a 6 percent decrease in total cholesterol
and a 7 percent decrease in LDL-cholesterol levels. The second study
(Ref. 140) used a similar design but was only available as an abstract
that contained too little detail for the agency to evaluate it.
FDA finds that neither of these studies can provide support for a
hypocholesterolemic effect of soy protein per se because both soy
protein and soluble fiber were varied concurrently. However, these
studies do suggest that inclusion of these specific components can
further enhance the lipid-lowering effect of a low saturated fat, low
cholesterol diet.
(Comment 28). A comment also submitted the recent study by Washburn
et al., 1999 (Ref. 141) for consideration. In this randomized, double-
blind crossover trial, 51 normocholesterolemic, perimenopausal women
consumed supplements for 6-week periods of 20 g of complex
carbohydrate, 20 g of soy protein containing 34 mg of phytoestrogens
given in a single dose, and 20 g of soy protein containing 34 mg of
phytoestrogens split into two doses. Significant declines in total
cholesterol level (6 percent lower) and LDL- cholesterol level (7
percent lower) were observed with both soy treatments compared to the
carbohydrate placebo control. However, no dietary assessments were
performed; thus, FDA cannot determine whether the women may have
modified their usual dietary intake in response to the supplements and
whether and how intake of dietary constituents may have differed among
the treatment groups.
FDA identified two additional recently published studies for
consideration. Nilausen and Meinertz, 1998 (Ref. 142) employed liquid
formula diets containing a very high level of protein (150 g/day) with
soy or casein as the sole protein source to examine individual
variability in lipemic response in a small metabolic study of
normocholesterolemic men. In most subjects effects of soy protein on
both LDL- and HDL-cholesterol levels were favorable, but considerable
variability in response was observed. Duane, 1999 (Ref. 143) also
conducted a small metabolic ward study in normocholesterolemic men that
compared effects of (1) a control diet with ``standard'' amounts of
dietary cholesterol, (2) a diet with essentially no dietary cholesterol
and all animal sources of protein substituted by TVP, and (3) a diet
similar to the second one with eggs isocalorically substituted for
protein and fat to bring dietary cholesterol levels to the moderate
range. Diets containing soy protein decreased LDL-cholesterol but the
effect was of borderline statistical significance. FDA notes that the
small number of subjects and the unusual dietary conditions employed in
these two studies limit their usefulness in adding to the body of
evidence about the effects of soy protein on circulating lipid levels.
In summary, although most of the new studies considered had flawed
or unusual designs that compromised their evaluation, the two better
designed and controlled studies (Ref. 136 and Ref. 137) provide
additional support for the cholesterol lowering effects of inclusion of
reasonable amounts of soy protein in diets low in saturated fat and
cholesterol.
2. Interpretation of the Clinical Trial Data for Soy Protein
(Comment 29). One comment raised concerns about the apparent
inconsistency in FDA's application of its review selection criteria,
especially with respect to giving the greatest weight in evaluation of
the health claim to those studies that reported information about the
dietary intake of constituents known to have the greatest influence on
total and LDL-cholesterol levels. The comment noted that values for
dietary saturated fat and cholesterol were not reported for some
studies and that an outmoded description of polyunsaturated fatty acid
to saturated fatty acid ratio was reported for some studies.
FDA agrees that values for these dietary constituents were not
reported explicitly in all of the studies selected for review. In such
cases, FDA relied upon other documentation contained in the study
publications regarding the contents of the test and control diets, such
as sample menus and reported manipulations of sources of saturated fat
and cholesterol, for assurance that dietary saturated fat and
cholesterol did not differ significantly in the test conditions.
(Comment 30). One comment questioned the appropriateness of
including studies in which only total cholesterol levels were measured.
As noted above, in earlier rulemakings on diet and CHD
relationships, FDA concluded that it is generally accepted that blood
total and LDL-cholesterol levels are major risk factors for CHD, and
that dietary factors affecting blood cholesterol levels affect the risk
of CHD. FDA notes that a few of the older studies that it considered
and reviewed in the soy protein proposed rule, and in previous
[[Page 57709]]
rulemakings, measured only total cholesterol levels. FDA concluded that
inclusion of these studies for review was desirable in order to assess
the totality of the publicly available scientific evidence on the
relationship of soy protein and risk of CHD, even though LDL-
cholesterol levels are now considered to be a more powerful risk factor
than total cholesterol levels.
(Comment 31). A few comments disagreed with FDA's tentative
decision to authorize a health claim for the relationship between soy
protein and CHD because not all of the studies reviewed in the soy
protein proposed rule showed significant reductions of total and plasma
cholesterol levels.
A recent review and meta-analysis of the effectiveness of NCEP Step
1 and Step 2 dietary interventions in free-living subjects by Yu-Poth
et al. (Ref. 144) noted an appreciable range of response to the dietary
interventions with the maximal effect being more than twice the average
response reported in controlled feeding studies with Step 1 diets. The
interventions reviewed were designed to achieve reduction of dietary
saturated fat and cholesterol and weight reduction, factors known to
have a major impact on circulating cholesterol levels. (The
hypocholesterolemic effects of soy protein, like those of soluble fiber
from whole oats and psyllium seed, are of a lesser magnitude than those
of reduced dietary saturated fat and cholesterol.) Denke (Ref. 145), in
an editorial comment on the study by Yu-Poth et al., notes that
cholesterol-lowering dietary therapy is subject to profound individual
variation in response. In metabolic ward studies of subjects with
unselected cholesterol levels, 5 percent of individuals had no
cholesterol-lowering response to dietary modification and the
percentage of nonresponders increased to 10-25 percent in outpatient
studies (Denke, 1995, Ref. 146). Such nonresponse can result in a
significant underestimation of the effectiveness of dietary
intervention when only the mean response is considered. The small
metabolic ward study of Nilausen and Meinertz (Ref. 142), described
above, documented evidence for considerable inter-individual
variability in the response of cholesterol levels to diets containing
soy protein.
Based on the studies reviewed in the soy protein proposed rule and
the new studies reviewed in this document, FDA concludes that the
totality of the available scientific evidence supports a consistent, if
not universal, hypocholesterolemic effect of soy protein included in a
low saturated fat and low cholesterol diet. The degree of consistency
is notable in light of the different experimental designs and diets
studied, the different forms and amounts of soy protein tested, and the
variability in initial cholesterol levels of the subjects. The modest
lowering of total and LDL-cholesterol levels generally observed in
these studies can effect a significant reduction in CHD risk.
(Comment 32). Other comments reviewed various possible mechanisms
for the cholesterol-lowering effects of soy protein and some argued
that until the mechanism of action of soy protein is clearly
established, no health claim should be authorized. FDA notes, however,
that such knowledge is not necessarily required for authorization of a
health claim.
3. Role of Soy Isoflavones in and Effect of Processing on the
Hypocholesterolemic Effect of Soy Protein
In the soy protein proposed rule, FDA examined the limited evidence
that addressed whether the hypocholesterolemic effects of soy protein
intake were dependent, as suggested by the petitioner, on concomitant
intake of a specified level of naturally occurring soy isoflavones,
i.e., 2 mg isoflavones per g of soy protein (Refs. 22, 28, 31, 70, and
71). FDA also took note of a letter to the editor from Sirtori et al.
(Ref. 72), who conducted a number of trials in which soy protein
exhibited hypocholesterolemic effects and asserted that the products
used in those trials were essentially devoid of isoflavones. Given the
limited number of studies and the contradictory outcomes, FDA was not
persuaded that the isoflavone component of soy protein was a relevant
factor to the diet-disease relationship. Rather, FDA tentatively
concluded that the evidence from a wide range of studies using
differently processed soy protein was supportive of a relationship
between soy protein per se and reduced risk of CHD.
(Comment 33). Several comments reviewed and discussed the animal
and human studies that examined effects of isoflavones directly or that
compared the effects of ISP processed with and without alcohol
extraction that can remove essentially all isoflavones. Some of these
studies examined effects on parameters in addition to cholesterol
levels, such as measures of lipid-related gene expression,
atherosclerosis, and vascular reactivity. Because the health claim for
soy protein and CHD is based on the hypocholesterolemic effect of soy
protein, only that aspect of the studies is summarized below.
In one study, Balmir et al. (Ref. 147) fed male rats diets
containing protein from ethanol-acetone extracted ISP, nonextracted
ISP, casein, or casein to which the ethanol-acetone extract was added.
Rats fed either ISP diet had lower serum total cholesterol
concentrations compared with those fed either casein diet. Lower serum
LDL-cholesterol concentrations were found in rats fed either ISP diet
and in rats fed casein plus extract compared with those fed casein.
Sugano and Koba (Ref. 148) found that a methanol-extracted soy fraction
was not as effective as the unextracted fraction in maintaining low
plasma cholesterol levels in rats. Kirk et al. (Ref. 149) showed that a
soy protein-based isoflavone-containing diet resulted in a reduction in
cholesterol levels in C57BL/6 mice compared to a diet containing
alcohol-washed soy protein, although it had no effect on cholesterol
levels in transgenic mice that lacked the LDL receptor. In another
study, Balmir et al. (Ref. 147) fed male Golden Syrian hamsters diets
containing protein from ISP, ISP with added ethanol-acetone extract,
casein, or casein with added extract. Lower serum total cholesterol and
LDL cholesterol concentrations were observed in hamsters fed ISP, ISP
with extract, or casein with extract compared with those fed casein.
Addition of the extract to casein at higher levels did not lower serum
lipids relative to casein. Tovar-Palacio et al. (Ref. 150) fed gerbils
one of five experimental diets containing either casein or alcohol-
washed ISP provided alone, or ISP supplemented with one of three
different levels of an alcohol extract of isolated soy protein
contributing either 2.1, 3.6 or 6.2 mg isoflavones/g protein. Gerbils
fed all of the soy-based diets had significantly lower total and LDL +
very low density lipoprotein (VLDL)-cholesterol levels than those fed
casein. The additions of the alcohol extract to ISP did not reduce
serum cholesterol levels any further. This study suggests that, in
gerbils, consumption of an isoflavone-containing extract does not
contribute to the hypocholesterolemic effect of alcohol-extracted soy
protein. These reports did not characterize the nature of the extracts
used in the studies. Overall, FDA finds that studies in these animal
models do not clarify the role of isoflavones in the
hypocholesterolemic effect of soy protein.
Comments noted a series of studies conducted in monkeys that
examined the effect of removal of isoflavones and other alcohol-
extractable compounds from soy protein on its cholesterol-lowering
activity. Anthony et al. (Ref.
[[Page 57710]]
22) fed peripubertal male and female rhesus monkeys moderately
atherogenic diets in which the source of dietary protein was a soy
protein isolate, either containing isoflavones or with the isoflavones
removed by alcohol extraction, in a crossover design with each period
lasting for 6 months. The intact soy protein (compared with the
extracted soy protein) significantly reduced LDL+VLDL-cholesterol
levels in both males and females and significantly increased HDL-
cholesterol levels for females. Honore et al. (Ref. 23) fed young adult
rhesus monkeys with pre-existing diet-induced atherosclerosis one of
two soy-based diets, which were identical in composition except that
the isoflavones were extracted from one and intact in the other, for 6
months. Total and LDL-cholesterol levels were significantly lower in
females fed the intact soy protein than in those fed the extracted soy
protein. The same trend was seen in males, but the difference was not
statistically significant for total cholesterol. Anthony et al. (Ref.
70) studied young male cynomolgus macaques fed one of three moderately
atherogenic diets for 14 months. The groups differed only in the source
of dietary protein, which was either casein/lactalbumin, soy protein
with the isoflavones intact, or soy protein with the isoflavones mostly
extracted. Animals fed intact soy protein had significantly lower total
and LDL+VLDL-cholesterol levels compared with the other two groups. The
animals fed intact soy protein had the highest HDL-cholesterol level,
the casein group had the lowest level, and the group fed the extracted
soy protein was intermediate. Anthony et al. (Ref. 151) randomized male
and female macaques to groups fed a casein-containing diet or diets
with soy protein with the isoflavones intact or extracted. Fat and
cholesterol were identical in all diets. The LDL+VLDL-cholesterol
levels were highest in the casein group, slightly lower in the group
fed extracted soy protein, and significantly lower in the group fed
intact soy protein. The HDL-cholesterol levels were significantly
higher in both soy protein groups than in the casein group. FDA notes
that the alcohol extraction procedure used by these researchers, which
was not characterized in the study reports, appeared to diminish the
hypocholesterolemic effect of ISP.
Comments submitted three human studies of isolated isoflavones that
examined their role in cholesterol lowering. In a study published only
as an abstract, Colquhoun et al. (Ref. 152) administered daidzein and
genistein to 23 male and female subjects with an average cholesterol
level of 243 mg/deciliter (dL) in a blinded crossover design. Nestel et
al. (Ref. 52) studied 21 women in a randomized cross-over design with
two active treatment periods (80 mg of isolated soy isoflavones) and
one 5-week placebo period, while they consumed a soy-free diet. Hodgson
et al. (Ref. 153) conducted a randomized, blinded, placebo-controlled
trial of 8 weeks duration and a two-way parallel design that tested the
administration of 55 mg of soy isoflavones to 46 men and 13
postmenopausal women. Plasma lipid levels were not affected by soy
isoflavones in any of these studies. FDA notes that these studies do
not support a role for isolated isoflavones in cholesterol lowering.
Three studies submitted in comments examined the effects of
variation of isoflavone content in soy protein-containing diets in
human subjects. Cassidy et al. (Ref. 154) conducted metabolic ward
studies of the effects of various soy products with and without
isoflavones in small numbers of healthy, nonvegetarian, premenopausal
women. During one (control) menstrual cycle, the women ate a constant
diet containing no soy products. Then, over a second complete cycle six
subjects consumed a similar diet into which 60 g TVP/day, containing 45
mg conjugated isoflavones, was incorporated. Three participants had 50
g miso, containing 25 mg unconjugated isoflavones, added daily to their
diet over a menstrual cycle, and six others consumed 28 g TVP/day,
containing 23 mg conjugated isoflavones. Five participants completed a
third diet period in which they were randomly assigned to consume
either the control diet over a cycle, or a similar diet incorporating
60 g of a ISP from which the isoflavones had been chemically extracted.
A significant reduction in total cholesterol was found with 45 mg
conjugated isoflavones, but not with 23 mg conjugated isoflavones or
isoflavone-free ISP.
As previously reviewed in the soy protein proposed rule (63 FR
62977 at 62988), the study of Baum et al. (Ref. 28) investigated the
impact of soy protein as ISP containing different levels of isoflavones
in hypercholesterolemic, postmenopausal women. Adjusted mean
differences in the change from baseline for total serum cholesterol
level did not differ in the two soy groups and the control group.
However, there was a statistically significant reduction of 8-9 percent
in non-HDL (LDL+VLDL)-cholesterol in both of the ISP treatment groups
(p<0.05) compared to the control group. HDL-cholesterol was also
significantly increased (p<0.05) in both soy groups compared to the
control. The level of isoflavones did not affect any of the blood lipid
levels measured.
FDA also previously reviewed the unpublished study by Crouse et
al., which was subsequently accepted and published (Ref. 31), in the
soy protein proposed rule (63 FR 62977 at 62987). This study examined
the effect of soy protein containing different levels of isoflavones in
hypercholesterolemic men and women. Subjects with qualifying serum
lipid levels (LDL-cholesterol greater than 140 mg/dL) after one month
and who were compliant with the study regimen were randomized into one
of five treatment groups. The treatment groups received 25 g protein
from ISP prepared from soy with different levels of isoflavones (either
1.0, 1.6, or 2.5 mg total aglycone isoflavones/g protein), or 25 g
protein from alcohol-washed ISP that contained essentially no
isoflavones (0.2 mg total aglycone isoflavones/g protein) or 25 g
protein from casein (no isoflavones) in beverages for 9 weeks. Results
indicated that compared to casein the ISP containing the highest level
of isoflavones significantly lowered total (p<0.05) and LDL-cholesterol
(p<0.05), by 4 percent and 6 percent, respectively, while HDL-
cholesterol was not altered. In subjects with LDL-cholesterol in the
top half of the study population, serum total and LDL-cholesterol were
reduced by 9 percent (p<0.03) and 12 percent (p<0.03), respectively, by
the ISP with the highest isoflavone content, and by 8 percent (p<0.03)
and 9 percent (p<0.03), respectively by the ISP with the second highest
isoflavone content, while HDL-cholesterol concentrations were
maintained. The authors reported a dose-response effect of increasing
amounts of isoflavones on total and LDL cholesterol level. One comment
included a reanalysis of the dose-response data that did not include
data for the casein diet, in order to control for an independent effect
from soy protein itself, and found no significant effect based on
isoflavone content. A comment from the petitioner disagreed with this
analysis. It also indicated that the study did not eliminate the
possibility that isolated soy protein per se has cholesterol-lowering
properties, but rather suggested that soy protein with higher levels of
isoflavones might have even greater effects. FDA finds that the
disparity in these comments does not clarify the equivocal nature of
the available evidence. FDA finds that these studies do not provide
sufficiently consistent results to cause the agency to
[[Page 57711]]
change the conclusion reached in the soy protein proposed rule.
(Comment 34). One comment objected to FDA's consideration of the
letter to the editor by Sirtori et al. (Ref. 72) because the reference
substantiating the technique for processing the soy protein product was
missing from the letter, the products were not tested for isoflavone
content at the time of the studies, different soy products (isolate and
flour) were used to manufacture the textured soy protein used in the
studies, and the references for studies cited in the letter did not
match the ones cited by FDA in the soy protein proposed rule. FDA
agrees that the reference for the patented procedure for the production
of the TVP, described as making use of rapid heating under high
pressure, was omitted in the letter by Sirtori et al. (Ref. 72) and
that the isoflavone content of the products reported (Cholsoy and
Croksoy) was not measured at the time the studies in which they were
used were conducted.
The letter by Sirtori et al. (Ref. 72) cites two older studies--
Sirtori et al., 1979 (Ref. 55) and Sirtori et al., 1979 (Ref. 155)--as
well more recent studies--Sirtori et al., 1995 (Ref. 156)--conducted by
their group. The five studies of Sirtori's group that FDA reviewed and
cited in the soy protein proposed rule as using products that contained
essentially no isoflavones (Refs. 33, 34, 35, 46, and 56) are included
in the reference list of Sirtori et al., 1995 (Ref. 156), which is a
review article. The agency did not review Sirtori et al., 1979 (Ref.
155) in the soy protein proposed rule, and it did not cite Sirtori et
al., 1977 (Ref. 55) because it specifically indicated use of a soy
protein product different from those tested for isoflavone content. FDA
gives some credence to the knowledge of the investigator about the
products used in his studies, but agrees that the letter to the editor
does not provide sufficient documentation to permit an unequivocal
conclusion that the products found to be devoid of isoflavones were
identical to those used in the clinical studies.
(Comment 35). One comment asserted that most of the studies
reported by Sirtori's group were performed using a textured soy protein
based on steam-treated soy flour; this treatment would be expected to
remove isoflavones. The comment also included a letter from Sirtori
(Ref. 157) stating that essentially all of his group's studies
beginning in 1980 were with products without isoflavones. However, the
patent referenced in this letter was not included with this submission.
Thus, FDA cannot verify that the process used to produce the products
used in Sirtori's studies over time was the same used to produce the
products analyzed recently for isoflavone content.
(Comment 36). The interpretation of the data available on the role
of soy isoflavones in and the effects of processing on the
hypocholesterolemic effect of soy protein varied widely in the
comments. Several comments agreed with FDA's conclusion that the
evidence did not support a significant role for soy isoflavones in
cholesterol-lowering effects of soy protein. One comment supported the
petitioner's original conclusion that a level of 2 mg aglycone
isoflavones per g soy protein was necessary for cholesterol lowering.
In a comment, the petitioner agreed with FDA ``that a relationship
exists between soy protein per se and reduced risk of CHD.''
The additional evidence about the role of isoflavones is
contradictory and inconclusive and has not persuaded FDA to alter its
original conclusion about the inability to identify a specific
contribution of soy isoflavones to the cholesterol-lowering effects of
soy protein. At the same time, the evidence shows a clear relationship
between soy protein and reduced risk of CHD despite lack of a clearly
defined mechanism for its effect.
(Comment 37). Several comments interpreted the evidence as showing
that alcohol extraction used in the processing of certain soy protein
ingredients (to the extent that they are rendered essentially devoid of
isoflavones) impairs or eliminates the hypocholesterolemic effects of
soy protein and recommended that the health claim not be allowed for
alcohol-washed products. Comments also raised some questions about the
extent to which extensively alcohol-washed products, such as those used
in the animal studies, are available commercially. One comment asserted
that some of ISP products used in the primate studies were subjected to
additional alcohol extraction by the investigators, but the agency
could not independently verify this assertion. This comment also stated
that all commercial sources of soy protein contain some isoflavones.
FDA examined the recently compiled USDA-Iowa State University
Isoflavone Database (Ref. 158), which documents the following ranges of
total isoflavone content for various soy protein-containing
ingredients, and found that most, but not all, contained levels of
isoflavones higher than those that would result from harsh alcohol
extraction procedures:
Table 1
------------------------------------------------------------------------
Aglycone isoflavones
Product (mg/100 g edivle
portion
------------------------------------------------------------------------
Soy flour, textured............................... 4.40-295.55
Soy flour, defatted............................... 73.72-168.09
Soy flour, full-fat, raw.......................... 59.80-264.84
Soy flour, full-fat, roasted...................... 131.70-260.50
Soy protein concentrate, aqueous washed........... 61.23-167.00
Soy protein concentrate produced by alcohol 2.08-31.82
extraction.......................................
Soy protein isolate............................... 46.50-199.25
Instant beverage, soy powder...................... 100.10-125.00
------------------------------------------------------------------------
FDA agrees that the data from the animal studies reviewed suggest
that alcohol washing of soy protein can reduce its hypocholesterolemic
effects. With respect to human studies, FDA finds the available
evidence is insufficient to permit any conclusions about the impact of
processing by alcohol extraction on the hypocholesterolemic effect of
soy protein. Thus, FDA concludes it would be premature to exclude
alcohol-washed products from eligibility to bear the health claim.
(Comment 38). One comment noted that several clinical trials
designed to resolve questions about the impact of processing and
isoflavone content are
[[Page 57712]]
currently in progress. Many of the comments on these issues urged that
FDA proceed with the health claim regulation as proposed, but monitor
research developments and make changes in the regulation as warranted
by the results.
As noted above, FDA finds that, in light of the evidence that soy
protein processed in various ways, containing unknown amounts of
isoflavones, has hypocholesterolemic effects, FDA is not applying any
criteria for inclusion of naturally occurring isoflavones or excluding
alcohol-washed products from eligibility to bear the health claim on
soy protein and CHD.
(Comment 39). A few comments suggested that, regardless of the
conclusions about the significance of soy isoflavones to the reduction
of CHD risk, food products that bear the soy protein health claim be
allowed or required to state the isoflavone content of the product on
the label. The comments did not provide any evidence that persuaded the
agency that consumers would find this information helpful in making
healthful dietary choices. Accordingly, the agency is not adopting this
suggestion.
4. Amount of Soy Protein Required for Significant Effect on Cholesterol
Levels
Based on the limited data reviewed that supported a dose-response
and the data that showed clinically significant reductions in total and
LDL-cholesterol with soy protein ingestion in the range of 17-31 g/day,
and recognizing that the hypocholesterolemic effects of soy protein
were dependent on initial blood lipid levels, the agency tentatively
concluded that 25 g/day represented a reasonable, effective amount of
soy protein (63 FR 62977 at 62992). In addition, the agency noted that
an amount of 25 g/day of soy protein represents half of the Reference
Daily Intake (RDI) of 50 g for protein and is a reasonable level of
consumption in the context of the total daily diet. Thus, FDA
tentatively concluded that the amount of soy protein associated with
reduction in total and LDL-cholesterol levels and, thus, with reduced
risk of CHD was 25 g or more of soy protein per day (63 FR 62977 at
62992).
(Comment 40). Many comments agreed with the agency's conclusion
that 25 g or more of soy protein per day was associated with reduction
in total and LDL-cholesterol levels. Several comments raised concerns
about the adequacy of the available data to support an assessment of
dose-response. One comment expressed concern that higher levels of soy
protein are needed to modify cholesterol levels in normocholesterolemic
individuals and that this should be indicated as part of the claim.
FDA agrees that the available data on the hypocholesterolemic
effects of soy protein do not permit a dose-response assessment.
However, FDA notes that dose-response data are not required to
establish the qualifying criteria for a substance that is the subject
of a health claim. Under Sec. 101.70, which describes the requirements
for health claim petitions, the petition must address whether there is
an optimum level of the particular substance to be consumed beyond
which no benefit would be expected (Sec. 101.70(f)(B)(1)). This
information may or may not be based on dose-response data. For example,
in its evaluation of the scientific evidence for a relationship between
consumption of soluble fiber from psyllium seed husk and blood total
and LDL-cholesterol levels, the agency found no reliable data to
establish a dose-response for this relationship (62 FR 28234 at 28240).
However, the agency did find that, in placebo-controlled studies that
tested an intake of 10.2 g of psyllium seed husk per day as a part of a
diet low in saturated fat and cholesterol, there were consistently
significant effects of psyllium husk on blood total and LDL-cholesterol
levels. Therefore, the agency based the qualifying level of soluble
fiber from psyllium seed husk on a total daily intake of 10.2 g husk or
about 7 g of soluble fiber.
The qualifying level of 25 g/day has been demonstrated to have a
consistent, clinically significant effect on total and LDL-cholesterol
levels. This 25 g/day level of intake for cholesterol lowering is
confirmed by the new study of Teixeira et al. (Ref. 136), which showed
significant hypocholesterolemic effects of 20 g/day of soy protein.
Therefore, the agency disagrees with the comments suggesting that dose-
response data are needed before the agency can authorize a health
claim. The totality of scientific data, which establish a clinically
significant reduction in blood cholesterol based on an intake of at
least 25 g/day of soy protein, provides an adequate basis for
establishing a qualifying level for soy protein-containing products.
The agency agrees that the available data indicate that the
hypocholesterolemic effect of soy protein may be dependent on initial
cholesterol levels, but notes that moderately hypercholesterolemic
individuals are generally more responsive to dietary interventions than
normocholesterolemic individuals. As the leading cause of death in this
country, CHD is a disease for which the general U.S. population is at
risk. The risk of dying from CHD is related to serum cholesterol levels
in a continuous and positive manner, increasing slowly for levels
between 150 mg/dL and 200 mg/dL and more rapidly when the cholesterol
level exceeds 200 mg/dL (Ref. 37). The public health policy articulated
by the NCEP, National Heart, Lung, and Blood Institute, is to extend
the benefits of cholesterol lowering to the population as a whole by
promoting adoption of eating patterns that can help lower the blood
cholesterol levels of most Americans (Ref. 67). A dietary intervention
that lowers blood cholesterol levels only in persons with high levels
would, like an intervention that lowers cholesterol levels across the
entire population range, cause a shift in the population distribution
of blood cholesterol levels resulting in a decrease in the mean value
for the blood cholesterol level in the general population (Ref. 67).
The anticipated effect of such a shift would be to reduce the morbidity
from CHD and to produce a continued or accelerated decline in the CHD
mortality rate in the United States. The agency is persuaded by the
evidence it has reviewed in this rulemaking that the consumption of soy
protein, as part of a low saturated fat and cholesterol diet, can be a
useful public health measure to assist in the national policy of
promoting eating patterns that will help in achieving or maintaining
desirable blood cholesterol levels in the general population.
Therefore, it concludes that the health claim need not indicate that
hypercholesterolemic individuals may be more responsive to consumption
of soy protein than normocholesterolemic individuals. In addition,
consistent with the agency's conclusions in rulemaking on the dietary
saturated fat and cholesterol/CHD claim (58 FR 2739 at 2745, January 6,
1993), the wording of the health claim as `` `may' or `might' reduce
the risk of heart disease'' adequately represents the fact that not all
persons will realize the same magnitude of benefit from adopting the
dietary change.
5. Summary of the Scientific Evidence
FDA reviewed human studies submitted by the petitioner and in
comments that evaluated the effects on serum cholesterol and LDL-
cholesterol levels of dietary interventions with soy protein in
subjects with normal to elevated serum cholesterol levels and that met
the agency's criteria for selection.
Most intervention trials in subjects with total cholesterol levels
less than 300 mg/dL found that soy protein
[[Page 57713]]
reduced total and/or LDL-cholesterol levels to a clinically significant
degree (Refs. 31, 28, 27, 51, 44, 37, 49, 30, 58, 29, 43, 136, and
137.). Moreover, HDL-cholesterol levels were unchanged (Refs. 31, 27,
51, 40, 37, 49, 36, 53, 136, and 137) or slightly increased (Refs. 28,
44, 58, and 59). In some cases (Refs. 27, 44, and 49), decreases in
total and LDL-cholesterol were statistically significant only in
subsets of subjects with the higher initial blood lipid levels. Results
in normocholesterolemic subjects (Refs. 30, 36, 58, 59, and 53) were
more variable than those in hypercholesterolemic subjects (Refs. 31,
28, 27, 51, 44, 40, 37, 49, 54, 29, 43, and 136) except in the study of
Wong et al. (Ref. 137), in which normocholesterolemic and moderately
hypercholesterolemic subjects were equally responsive. The outcome of
an epidemiologic study (Ref. 65) also supported a relationship between
higher levels of soy protein intake and lower blood lipid levels.
Most of the studies in subjects with total cholesterol levels less
than 300 mg/dL used low saturated fat and low cholesterol diets (Refs.
31, 28, 27, 51, 44, 30, 36, 53, 29, 43, 136, and 137), but some used
``usual'' diets (Refs. 37, 49, 54, 36, 58, and 59). Although soy
protein was found to lower blood lipid levels in some of the studies
using ``usual'' diets, hypocholesterolemic effects of soy protein were
more consistently observed with diets low in saturated fat and
cholesterol. Given the variability of amounts and forms in which soy
protein was provided in the diets, the response of blood lipid levels
appears robust and notably consistent, particularly in subjects with
moderate hypercholesterolemia.
Data from studies of adults with type II and familial forms of
hypercholesterolemia (and total cholesterol levels in excess of 300 mg/
dL) (Refs. 55, 33, 64, 56, 64, 46, and 35) were also consistent in
showing large and statistically significant decreases in total and LDL-
cholesterol, accompanied by no changes or slight increases in HDL-
cholesterol levels. Nearly all of the subjects in these trials consumed
low saturated fat and low cholesterol diets during the studies and had
consumed such diets prior to studies with soy protein. Soy protein was
tested in a variety of foods but produced fairly consistent results
regardless of the food form fed and apparent differences in processing
techniques.
The FDA concludes, based on the evidence submitted and reviewed,
that soy protein, included in a diet low in saturated fat and
cholesterol, can lower blood total and LDL-cholesterol levels, without
adversely affecting HDL-cholesterol levels. The agency also concludes
that the effect is due to soy protein per se and is not consistently
related to the presence or absence of isoflavones. The evidence
currently available, as reviewed in section II.B.3 of this document,
does not permit a conclusion regarding how significantly alcohol
processing may affect the hypocholesterolemic effects of soy protein.
The intervention studies reviewed indicate that a minimum level of
approximately 25 g of soy protein per day results in a clinically
significant effect on total and LDL-cholesterol levels.
With respect to the scientific data and information about the
relationship of soy protein and CHD, the relevant data are provided by
well controlled and well designed studies. Soy protein, the food
substance that is the subject of the claim, is measured in those
studies. The relationship of the biomarkers evaluated--total and LDL-
cholesterol--to the risk of CHD is validated and the studies measured
the biomarkers appropriately. Finally, a consistent body of evidence
from a variety of studies is available. Accordingly, the agency is able
to conclude, based on the totality of the publicly available scientific
evidence, that there is significant scientific agreement that soy
protein, included at a level of 25 g/day in a diet low in saturated fat
and cholesterol, can help reduce total and LDL-cholesterol levels, and
that such reductions may reduce the risk of CHD.
C. Nature of the Food Eligible to Bear the Claim
1. The Qualifying Amount of Soy Protein
Using 25 g of soy protein as the qualifying amount for a CHD claim,
the petitioner suggested that a single serving of a soy protein-
containing product (i.e., one RACC) should provide \1/4\ of this amount
(based on four servings a day). Thus, a soy protein-containing product
would have to contain at least 6.25 g soy protein (\1/4\ x 25 g) per
RACC. The petitioner stated that this approach was reasonable because
it would permit a wide variety of low fat, soy protein-containing
products to bear the health claim. The petitioner provided a list of
products on the market that currently meet the proposed requirements
and a list of products that could be modified to meet them (Ref. 1,
Appendix V). The agency has generally made the assumption that a daily
food consumption pattern includes three meals and a snack (see 58 FR
2302 at 2379, January 6, 1993). The agency tentatively concluded in the
soy protein proposed rule that the assumption of four servings per day
of soy protein-containing foods was reasonable. Therefore, the agency
found that use of the qualifying criterion set forth in the petition
would be appropriate (63 FR 62977 at 62992).
Most comments agreed that the qualifying level of 6.25 g soy
protein per RACC was appropriate. Many of these comments also indicated
that a sufficient number and variety of soy protein-containing foods
are available to enable consumers to select suitable products to
consume a total of 25 g soy protein per day.
(Comment 41). Several comments suggested rounding the qualifying
level to 6 or 7 g of soy protein per RACC, in keeping with the
requirements for the labeling of protein in the Nutrition Facts panel.
FDA, however, notes that the Nutrition Facts panel contains the
amount of total protein per serving of the product, regardless of the
source ingredient. For many products that may bear the claim, soy
protein may not be the sole contributor to total protein. Therefore,
FDA finds that the amount of soy protein in a serving of a food that
may bear the health claim will neither be required nor permitted to
appear in the Nutrition Facts panel. The qualifying level need not
conform to requirements specific to the Nutrition Facts Panel.
(Comment 42). One comment received in response to the soy protein
reproposal indicated that food processors will be required to declare
the corrected amount of protein and the percent Daily Value of protein
on the Nutrition Facts panel, in accordance with 21 CFR 101.9(c)(7)(i).
This comment noted that, in nearly all cases, the amount of protein
declared will be lower than the quantity of protein present in the
product and may, in some instances, be lower than the qualifying amount
of soy protein.
FDA notes that compliance with the requirements of this regulation
will be based on the actual amount of soy protein present in the food
and not on the amount of protein declared on the nutrition label.
(Comment 43). One comment suggested that the qualifying level
should be increased to 12.5 g soy protein per RACC because of concerns
that consumers would not choose soy protein-containing foods frequently
enough during a day to reach a total of 25 g and might believe that the
health benefit may be attained by eating a single serving of a food
that provided no more than 6.25 g soy protein. Several
[[Page 57714]]
other comments also raised concerns that consuming soy protein-
containing foods up to four times daily would represent a significant
change from the typical American diet that might not be selected by
many consumers.
FDA cannot assess how many consumers would be interested in making
such a change, but it is persuaded that it will be feasible for
motivated consumers to do so. Doubling the qualifying level of soy
protein per RACC would greatly and unnecessarily restrict the number of
foods potentially eligible to bear the health claim. Because
Sec. 101.82(c)(2)(i)(G) requires that the claim specify both the daily
dietary intake of soy protein that is necessary to reduce the risk of
coronary heart disease and the contribution that one serving of the
product makes to the specified daily dietary intake, consumers will not
be misled about the amount of soy protein needed for the health effect.
(Comment 44). A number of comments suggested that greater
flexibility in meeting the recommended total daily intake of 25 g soy
protein per day could be achieved by permitting a lower qualifying
level on the basis of increasing the number of servings or eating
occasions per day from four to five or six or more. Several of these
comments proposed that the qualifying level of soy protein should be
reduced to 4 g per RACC; one suggested lowering the qualifying level to
2.5 g per RACC. Most of these comments indicated that 4 g soy protein
per RACC is the maximum amount of soy protein from soy flour that can
be incorporated in baked products that consumers find palatable and
acceptable. These comments suggested that lowering the qualifying level
would stimulate manufacturers to develop a wider range of products and
indicated that use of ISP in baked products would be prohibitively
expensive. One comment challenged FDA's assertion that consumers would
be able to consume an effective amount of soy protein from a variety of
products, including baked goods. FDA based the assertion on its
observation that baked products had been used to provide soy protein in
some studies the agency relied upon to justify authorization of the
health claim (Refs. 27, 28, and 51); in one study (Ref. 27), the
authors indicated that 25 g soy protein daily was provided in four
muffins. ISP was the source of soy protein in the baked products used
in these studies. Some comments stated that FDA need not base the
qualifying level on four eating occasions per day as the agency had
done for other health claims for substances (beta-glucan soluble fiber
from whole oats and soluble fiber from psyllium seed husks).
FDA finds that these comments did not provide a compelling
rationale for selecting an appropriate number of eating occasions on
any other basis. The agency has not limited its previous determinations
of an appropriate qualifying level of a substance that does not have a
Daily Value in a food to be eligible to bear a health claim to
consideration of the number of individual foods or classes of food
products then available that might bear the claim. Rather, in
determining what constitutes a level of the substance sufficiently high
to justify the claim, FDA considers factors such as the number of
servings likely to be consumed and the feasibility of developing a
variety of foods that contain a significant proportion of the total
daily intake needed for the claimed benefit. For example, when the
psyllium claim was authorized, FDA was aware of only one conventional
food product that would have been eligible to bear the claim and
concluded that, if various psyllium-containing foods were available,
consumption of four servings daily could be achieved. Based on
experience with that claim and other health claims, FDA believes that
manufacturers will be encouraged by the availability of a health claim
for soy protein and CHD to develop new products that will be eligible
to bear the claim. The agency is not persuaded by the comments received
that it should abandon its assumption that a daily food consumption
pattern includes three meals and a snack (see 58 FR 2302 at 2379,
January 6, 1993) and that one serving of a soy protein-containing
product could reasonably be consumed at each eating occasion. As noted
in the discussion above of the comments that expressed concerns about
the willingness of consumers to select soy protein-containing foods as
many as four times a day, such an eating pattern represents a
considerable change from a typical American diet. Although one of the
comments included detailed menus that illustrated the possibility of
consuming more than one soy protein-containing product per eating
occasion, FDA has concluded that it should not lower the amount of soy
protein required for a food to be eligible to bear the health claim.
(Comment 45). One comment suggested that the amount of soy protein
required for eligibility to bear the health claim be permitted to be
determined on the basis of serving size as well as RACC.
This comment is outside the scope of this rulemaking. Current
regulations (21 CFR 101.12(g)) require that, ``The reference amount
[i.e., the reference amount customarily consumed] * * * shall be used
in determining whether a product meets the criteria * * * for health
claims.'' In a previous rulemaking, FDA had considered permitting this
option, but comments persuaded the agency that the most reasonable
approach was to base claim evaluations on the reference amount (58 FR
2229 at 2287). FDA agreed with the comments that claims should reflect
the true characteristics of a product, and that those characteristics
do not change if the product is packaged in a different size container.
The comment received in response to the soy protein proposed rule did
not provide a convincing rationale to justify a change in this
decision.
2. Method for Determining Qualifying Amount of Soy Protein in Foods
In the soy protein proposed rule (63 FR 62977 at 62992), FDA
proposed use of the Association of Official Analytical Chemists (AOAC)
official method of analysis No. 988.10 to measure soy protein in foods.
As described in the soy protein reproposal (64 FR 45932 at 45933), each
of the comments on this proposed analytical method disagreed with its
use and concluded that the method was unlikely to produce a reliable
measure of the soy protein content in every food. The comments noted a
variety of problems with the assay. These comments persuaded the agency
that AOAC official method of analysis No. 988.10 was not an appropriate
method for the quantitation of soy protein in many of the products that
may be eligible to bear the health claim.
In the soy protein reproposal, FDA discussed the alternative
approaches suggested in comments for assessing compliance with the
qualifying level of soy protein in products that bear the health claim.
Based on this information, the agency provided its tentative rationale
for a procedure employing measurement of total protein and, for
products containing sources of protein other than soy, calculation of
the soy protein content based on information contained in
manufacturers' records (64 FR 45932 at 45934). Thus, in the soy protein
reproposal, FDA modified previously proposed Sec. 101.82(c)(2)(ii)(B)
to provide for this alternative approach for compliance assessment that
relied, in some cases, on records that the agency could inspect.
The agency received approximately 10 comments in response to the
soy protein reproposal. One of the comments did not address the
proposed
[[Page 57715]]
procedure for compliance assessment but, rather, reiterated concerns
raised in comments on the soy protein proposed rule about the safety of
soy isoflavones. Among the materials it referenced were two documents
authored by FDA staff that the comment characterized as ``reports.''
FDA could not identify one of these documents from the citation given
and the other was a letter submitted as a comment to Docket 98P-0683 in
response to the soy protein proposed rule. Another comment raised
concerns about the GRAS status of soy protein. FDA has addressed the
issues raised in the earlier comments regarding GRAS status and safety
in Section II.A of this document. In addition to commenting about the
reproposal, one comment raised a technical issue about the nutrition
labeling declaration of protein that is addressed in Section II.C.1.
(Comment 46). Two comments objected to the 30-day comment period
allowed for the soy protein reproposal. FDA stated its rationale and
authority for selecting this period in the soy protein reproposal (64
FR 45932 at 45936 and 45937) and notes that these comments were
submitted and received in timely fashion. One of these comments
asserted that after the comment period for the soy protein proposed
rule had passed, no new submissions or evidence after that date other
than that of FDA origin (or from published scientific documents
accessed by FDA) was acceptable. As noted in the introduction of
Section II of this document, FDA disagrees with this assertion. FDA
considered comments received after the initial comment period,
regardless of source, to the extent that each provided complete
information for review or references accessible to the agency and
addressed issues not raised in earlier comments.
(Comment 47). A comment asserted that the issue of the method FDA
will use to verify that foods contain the qualifying amount of soy
protein is irrelevant because FDA was required to consider and evaluate
only the claims made for the substance identified in the petition, soy
protein with naturally occurring isoflavones.
This comment misunderstands FDA's responsibility to review and
evaluate the available scientific evidence and reach appropriately
supported conclusions about the substance-disease relationship based on
information provided in the petition, accessed in the public scientific
literature, and received in comments. FDA notes, for example, that in
response to a petition for oat bran and oatmeal, it proposed to
authorize a health claim on the relationship of those foods and CHD (61
FR 296). Comments received in response to that proposal persuaded FDA
to change the substance of its final rule to beta-glucan soluble fiber
from whole oats (62 FR 3584). The agency has addressed the earlier
comments on the role of isoflavones in the hypocholesterolemic effect
of soy protein in Section II.B.3 of this document.
(Comment 48). Two comments objected to any use of recordkeeping for
compliance assessment, questioning whether it could be an appropriate
substitute for analytical methods to assess the truthfulness of health
claims. One of these comments also reiterated objections to
authorization of the health claim, because of concerns about incomplete
scientific understanding of the biological activity of soy components,
in terms of both safety and contribution to the protective effect of
soy protein in CHD. The agency has addressed these concerns, which were
raised in comments on the soy protein proposed rule, in Sections II.A
and II.B.3, respectively, of this document.
The other comment asserted that an approved, scientifically
accurate methodology is needed for any health claim. However, it also
indicated that FDA should finalize its regulation as originally
proposed, but did not propose an alternative for compliance
verification other than suggesting that a manufacturer might
voluntarily share analytical data with the agency if questions about
compliance were raised.
FDA does not agree with the contention that an analytical method is
an absolute requirement for a health claim, even though it is the
preferred means for verifying compliance with the requirements of a
health claim regulation and substantiating the truthfulness of all
label statements.
(Comment 49). Many other comments supported continued work to
develop appropriate analytical methodology for measuring the content of
soy protein in foods, and urged FDA, in collaboration with other
government agencies, industry, and scientific organizations, to pursue
this effort. As noted in the soy protein reproposal, FDA intends to do
so, to the extent that resources permit. Also, as noted in the soy
protein reproposal, and as urged in a number of comments, FDA would
propose to amend its regulation to provide for compliance verification
based on one or more analytical methodologies when such methods have
been validated.
(Comment 50). Several of the comments specifically addressed the
method for assessing compliance set out in the soy protein reproposal.
None of these comments objected to use of an analytical method for
measuring total protein as a measure of soy protein in foods that
contain soy as the only source of protein. Absent an appropriate
analytical methodology, each of these comments supported the need for
manufacturers to have and keep records to substantiate the amount of
soy protein in a food that bears the health claim and contains sources
of protein other than soy, and to make such records available to
appropriate regulatory officials upon request. These comments noted
that in cases where records are needed to substantiate label claims,
food manufacturers have historically provided such records voluntarily
upon request to the FDA and could be expected to continue to do so in
the future. They argued that FDA need not assert broad records
inspection authority in order to obtain the information needed for
compliance assessment. They noted 21 CFR 101.13(j)(ii)(A), which
requires firms to have substantiation for the basis of nutrient
reference values in comparative nutrient content claims and to make
such substantiation available to appropriate regulatory officials upon
request, as a model for requests of records.
FDA agrees that a manufacturer must have substantiation that a
qualifying amount of soy protein is present in a product that bears the
health claim and that such records can serve as the basis for
substantiation of use of the health claim. FDA noted in the Federal
Register of February 2, 1996 (61 FR 3885 at 3886) several examples of
regulations that implemented the 1990 amendments in which the agency
could not independently, using analytical methodology, verify the basis
for statements on the food label, but instead would rely on access to a
manufacturers' information supporting its labeling claims. These
include access to:
(1) A detailed protocol and records of all data used to derive a
density-adjusted reference amount for aerated foods (58 FR 2229 at 2272
and Sec. 101.12(e));
(2) Information that provides the basis for deriving reference
nutrient values for comparative nutrient content claims such as
``light'' (58 FR 2302 at 2365 and Sec. 101.13(j)(1)(ii)(A));
(3) Specific information with respect to the caloric content of new
products with reduced digestibility (58 FR 2079 at 2087 and 2111 and
Sec. 101.9(c)(1)(i)(D)); and
(4) Information supporting nutrient content claims for restaurant
foods (58 FR 2302 at 2388 and Sec. 101.13(q)(5)(ii)).
[[Page 57716]]
In each of these cases, verification of the truthfulness of a label
claim can be assessed by FDA only with access to information known only
by the manufacturer. The same is true, in the absence of a validated
analytical method to measure the amount of soy protein in the presence
of other proteins, for verifying that the qualifying amount of soy
protein to bear the health claim is present in a food that contains
sources of protein in addition to soy. Thus, the agency concludes, in
agreement with these comments, that it is appropriate to require access
to manufacturers' records substantiating the ratio of soy protein to
total protein for foods that contain sources of protein in addition to
soy to assess their compliance with this regulation. Also, in agreement
with these comments, the agency concludes that it need not assert broad
records inspection authority to have access to appropriate records. The
agency disagrees, however, with comments that indicate that reliance on
the voluntary provision of records by manufacturers is sufficient to
meet the agency's need to verify compliance. Rather, the agency is
taking the approach of codifying a requirement for the manufacturer to
provide appropriate records, on request, as the agency has done
previously.
Although most of the comments supported the use of records, in
principle, for compliance assessment, they also raised concerns about
the types of records that FDA might request, the circumstances under
which FDA would request records, and the legal authority of the FDA to
require records and records inspection.
(Comment 51). Several comments indicated that FDA had used overly
broad and imprecise language in the soy protein reproprosal to describe
the types of records that FDA would request. They indicated that a
manufacturer is best able to determine the nature of the records that
would be needed to substantiate the amount of soy protein in its own
products and urged that manufacturers be allowed the flexibility to
determine how to document substantiation. One comment argued that a
recipe-based system would be too complex and burdensome for baked goods
in particular. Other comments expressed concern that FDA would, in all
cases, require inspection of a wide variety of records, including
nutrient data bases or analyses, recipes or formulations, purchase
orders for ingredients, and others.
FDA agrees that the manufacturer will be in the best position to
know which of its records provide documentation of the amount of soy
protein in its products, and specifically the ratio of soy protein to
total protein. By listing the types of records that could provide such
documentation in the soy protein reproposal, FDA did not intend to
indicate that it would request all of these records and subject them to
inspection, or even that it would specify any particular records when
it requests them. Instead, FDA intended to suggest the types of records
a manufacturer might use to substantiate the levels of soy protein in
its foods. Accordingly, FDA has modified Sec. 182(c)(2)(ii)(B) to
clarify that the manufacturer is to identify these materials.
(Comment 52). One comment questioned whether FDA might request
records for products in which soy is the only source of protein and
urged FDA to specify that it would not request records for such
products.
FDA agrees that, because measurement of total protein provides
adequate assessment of compliance for products in which soy is the sole
source of protein, that it would not, under the regulation, request
records for substantiation of the amount of soy protein in such
products. The agency believes that the proposed language adequately
communicates this point and has made no changes to the regulatory
language in response to this comment.
(Comment 53). One comment requested that FDA identify what
circumstances would precipitate a request for records. Although FDA
cannot specify all such circumstances, it notes, as did another of the
comments, that a substantial proportion of its enforcement actions are
undertaken in response to trade complaints.
(Comment 54). One comment asked that the agency specify that any
records requested could be provided on site without the need for
reproduction or duplication by the investigator. Another comment,
however, objected to FDA making requests for information on site,
arguing that most companies would have the necessary information at
headquarters rather than at production facilities. This comment urged
that FDA make any such requests in writing and allow the manufacturers
to provide appropriate substantiation within a reasonable period of
time. As FDA will not require inspection of records on site, the
concern about reproduction or duplication is moot. FDA agrees that
making a request for records in writing is appropriate and has modified
the regulation accordingly.
(Comment 55). Some comments objected to the alternative offered in
the soy protein reproposal that FDA would authorize the claim only for
products that contain soy as the sole source of protein, if it could
not proceed with a regulation to provide access to records for
compliance verification. These comments noted that such an action would
give unfair advantage to certain products, unfairly penalize products
that were equally beneficial, and dilute the potential benefit of the
health claim to consumers. Because the agency has authorized the claim
for any food that contains adequate amounts of soy protein, without
regard to other sources of protein, these comments are moot.
(Comment 56). One comment noted that, in addition to providing FDA,
upon request, information regarding substantiation of the claim, food
processors may, on a voluntary basis, present information on the food
label or in labeling that may support the eligibility of the product to
bear the claim and facilitate an FDA compliance review. Such
information might take the form of statements about the percentage
composition of soy protein in a serving of food. The agency agrees that
manufacturers may voluntarily provide such truthful and not misleading
information and that the provision of such information may aid consumer
understanding of the claim.
(Comment 57). Several of the comments strongly objected to the
proposal for records inspection on the basis that FDA lacks the
statutory authority to require access to records for foods. Another
comment argued that, once the agency determined that a substance-
disease relationship meets the standard of significant scientific
agreement, the act requires the agency to authorize a claim, and the
agency may not require that manufacturers maintain records or that FDA
be able to request or inspect them. This comment also asserted that,
were FDA to require recordkeeping, record production, or records
inspection, it would violate the First Amendment by conditioning the
exercise of speech rights on the recordkeeping, record production, or
records inspection requirement.
FDA disagrees with these comments. Other comments have convinced
the agency that, in this instance, it need not assert its rulemaking
authority to provide for inspection of records. This issue is therefore
moot. The agency maintains, however, that it has the legal authority,
using section 701(a) of the act, to promulgate record inspection
requirements for the efficient enforcement of the act. The requirements
that records be maintained and submitted to the agency upon request
pass the test in National Confectioners Association v. Califano, 569
F.2d 690, 693 & n.9 (D.C. Cir. 1979). First, these requirements are
limited to those records that the manufacturer
[[Page 57717]]
reasonably determines substantiate the level of soy protein in its
food, and only with respect to foods that contain a source of protein
in addition to soy. Second, the requirements assist in the efficient